Master Guide


StressCheck
MASTER GUIDE
Volume 1: User’s Guide
Volume 2: Modeling Guide
Volume 3: Analysis Guide
Volume 4: Advanced Guide
Release 7
August, 2005
Engineering Software Research & Development, Inc.

Master Guide - Table of Contents
1 Preface 1
What is StressCheck? 1
Why use StressCheck? 2
Who should use StressCheck? 3
StressCheck features 3
How to use this manual? 4
Frequently asked questions about the p-version 4
2 StressCheck Interface 9
Interface Layout 9
Standard File Extensions 18
File Menu 20
Edit Menu 25
Class Menu 31
View Menu 31
Display Menu 37
Tools Menu 48
General Interface Conventions 49
The COM interface 58
3 Model Input 61
Getting started 61
Length Units in StressCheck 62
Geometry Class 63
Mesh Class 64
Section Properties Class 65
Thickness Class 68
Material Class 72
Load/Flux Class 77
Constraint/Temperature Class 81
Solution ID Class 84
p-Discretization Class 85
h-Discretization Class 87
Tables Class 89
Sets Class 91
Parts Class 92
4 StressCheck Solutions 101

Solution dialog window 101
Linear Analysis 103
Solution Activation 105
Nonlinear Analysis 108
Modal Analysis 110
Buckling Analysis 111
Margin Check Analysis 113
Crack Path Analysis 114
Measurement Analysis 115
Adaptivity 119
5 Post-processing Operations 133

Results dialog window 133
Report window 135
Checking the overall quality of the solution 145
Plotting the data of interest 151
Computing minimal and maximal values 159
Computing data at points 163
Properties 170
Fracture Mechanics 172
Extraction using the Formula and Calculator
options 174
XY Graphing 182
Setting convergence criteria 189
6 The Handbook Framework 193

Handbook Library 193
Handbook Interface 195
Handbook Analysis 205
Handbook Results 213
Handbook Material 218
Handbook Constraint 219
Constructing a handbook model 220
1 Introduction 1
Geometric Modeling in StressCheck 1
Terminology 2
Important Limitations 4
2 Basic Geometry Construction 5

Length Units in StressCheck 5
Geometry Class 6
3 2D Geometry Creation 25
Geometric construction of 2D objects 25
4 3D Geometry Creation 43
Construction of 3D primitive objects 43
Performing 3D Boolean Operations 60
Performing 3D Clipping Operations 75
Performing 3D Edge Blending Operations 75
Performing 3D Face Blending Operations 81
Example: Construction of a Simple Bracket 82
5 Mesh Creation 91
Introduction to Meshing 91
Mesh Refinement 122
Automeshing 127
MeshSim automesher 128
PAM automesher 148
1 Planar Elasticity 1
Setting the analysis type and choice of units 1
Entering geometric data 2
Entering mesh data 5
Assigning thickness 9
Entering material properties 10
Entering load data 19
Entering constraint data 24
Defining the solution ID 28
Setting discretization parameters 29
Writing an input file 29
Executing a linear analysis 30
Quality assessment and extraction procedures 34
Model problem: Tension strip with a circular hole 46
Principal directions 53
Using beam elements 55
2 Axisymmetric Elasticity 71
Setting the analysis type 71
Entering geometric and mesh data 72
Model problem: Floyd’s vessel 89
3 Plate Bending 97
Assigning thickness 99
Model problem: Built-in square plate 128
4 Three-Dimensional Elasticity 137

Model problem: Plate in tension 168
Extrusion 176
Laminate Composite Analysis 186
Thin Solids 196
Global-Local Analysis 206
Post-processing 232
Using 3D-beam elements 244
5 Modal and Buckling Analyses 253

Input data preparation 253
Modal analysis 255
Modal analysis for beams 263
Buckling analysis 267
Using the eigensolver 273
6 Conduction Heat Transfer 277

Input data preparation 277
Model problem: Pipe with cooling fins 278
Verification example 292
Stresses caused by thermal loading 299
1 Nonlinear Analysis 1
Nonlinear materials: Elastic-plastic problems 2
Isotropic Exponential Material Properties 24
General nonlinear analysis 25
Coldwork analysis 44
Fastened structural connections 53
Margin Check analysis 68
Nonlinear heat transfer 78
2 Multi-body Contact 101
3 Fracture Mechanics 119

Computing stress intensity factors 119
Mesh design in the presence of cracks 120
Model problem: 2D-Cracked lug 127
Model problem:3D-Cracked panel 137
Computing the T-stress 142
Computing the J-integral 157
Model problem: Panel with multi-site damage 157
Crack Path Analysis 164
Eigenpairs and generalized intensity factors 178
4 Material Property Fitting 203

Fiber fitting procedure 203
Fitting in 2D 205
Fitting in 3D 209
5 Theoretical Background 213

Mathematical models and their approximate
solutions 213
Control of discretization errors 216
Control of modeling errors 222
Hierarchic models for structural plates 226
Meshing and boundary conditions 242
p-discretization: The trunk and product spaces 251
Generalized Hooke’s law in three-dimensions 254
Strains and stresses for planar elasticity 282
Fracture mechanics 283
Eigenpairs and generalized intensity factors 290
Summary of the main points 313
6 Solvers 315
Sequential linear system solver 315
Eigensolver 317
Network License Management 319

What is License Management? 319
License Server and Password Installation 321
License Management Trouble-shooting 321
Error and Warning Messages 323
References 337
USER’S GUIDE

StressCheck
Volume 1
USER’S GUIDE
Release 7
August, 2005
For Windows Operating Systems

Copyright 2005
COPYRIGHT NOTICE
Copyright  2005 by Engineering Software Research & Development, Inc. All rights
reserved, worldwide. No part of this manual may be reproduced, transmitted, transcribed,
stored in a retrieval system, or translated into any human or computer language, in any
form or by any means, electronic, mechanical, magnetic, optical, chemical, manual, or
otherwise, without the expressed written permission from Engineering Software Research
& Development, Inc., 10845 Olive Boulevard, Suite 170, St. Louis, MO 63141-7760,
U.S.A.
StressCheck includes portions of Raima Data Manager version 3.21. Copyright  1984-
1996 by Raima Corporation. 1605 N. W. Sammamish Road, Suite 200, Issaquah, WA
98027. All rights reserved.
StressCheck includes portions of FLEXlm license manager version 6.1. Copyright 

1997 by Globetrotter Software Inc., 1530 Meridian Avenue, San Jose, GA 95125. All
rights reserved.
Tech Soft America (www.hoops3d.com) supplied the following core technology:

HOOPS 3D Application Framework  1998
HOOPS 3D Graphic System  1998
Portions of StressCheck are owned by EDS, Inc. Copyright  1998. All rights reserved.
StressCheck incorporates MeshSim™ a product of Simmetrix Inc.
DISCLAIMER
Engineering Software Research & Development, Inc. makes no representations or warran-

ties with respect to the contents hereof and specifically disclaims any implied warranties
of merchantability or fitness for any particular purpose. Further, Engineering Software
Research & Development, Inc. reserves the right to revise this publication and to make
changes from time to time in the content hereof without obligation of Engineering Soft-
ware Research & Development, Inc. to notify any person or organization of such revision
or change.
1
1 Preface
What is StressCheck?
From the perspective of designers, StressCheck is a very advanced handbook that
provides reliable solutions quickly and conveniently.
From the perspective of analysts, StressCheck is a tool for advanced problem solv-
ing and a framework for communicating the results to designers.
From the perspective of managers, StressCheck is a tool for increased productivity

and better design in less time.
StressCheck is the first finite element analysis program to emphasize both

advanced technological features and ease of use for everyday design and analysis
problems. ESRD founders are pioneers in development of p-version FEA and have
built the most advanced features available into StressCheck: An advanced represen-
tation of surfaces utilizing recently developed technology; hierarchic models for
structural plates, including plates made of laminated composites; advanced imple-
mentation of superconvergent extraction procedures for the computation of stress
intensity factors in two and three-dimensions; efficient and reliable treatment of
material and geometric nonlinearities in the context of the p- and hp- versions;
multi-body contact including material nonlinearities; the option to employ either
User’s Guide Chapter 1: Preface 1

Why use StressCheck?
1
the trunk space or the product space in p-extensions, and capabilities related to
the analysis of fastened connections, including cold working analysis.
StressCheck improves the reliability of computed information while increasing

the productivity of analysts. Recognizing that the analyst’s time is usefully
spent only if the computed information is sufficiently accurate and reliable to
serve the purposes of engineering decision-making, StressCheck was designed
so that the reliability of the data of interest can be readily ascertained. For most
analysis tasks the largest cost component, typically more than 90 percent, is the
cost of time spent on data preparation and interpretation of the results. Stress-
Check was designed so as to minimize this cost. The user interface was
designed to permit quick generation of finite element meshes, entry of material
properties and boundary conditions.
There is an immediate visual feedback confirming that the data is properly

entered. Modification and editing tasks can be performed quickly and conve-
niently. With StressCheck, the desired information, such as displacements,
stresses and stress maxima, stress intensity factors, and stress resultants can be
conveniently extracted from finite element solutions.
Why use StressCheck?

StressCheck delivers the most advanced p-version stress analysis technology in
a convenient, easy to use, handbook style interface. With StressCheck, you
overcome both the limitations of engineering handbooks and the complexity of
conventional FEA. StressCheck provides information that enables its users to
verify solution quality in a fraction of the time that would be required for con-
ventional FEA.
By incorporating your proprietary technology into an everyday handbook style

tool for both analysts and design engineers, routine problems can be set up and
solved in minutes. The handbook framework makes it possible for users to
define frequently occurring problems parametrically which can be recalled
quickly and conveniently for analysis, even by non-specialists. Therefore
StressCheck provides solutions which are much more representative of the
parts that need to be analyzed than handbook solutions. The amount of time
required for analysis is about the same as for computerized handbooks but the
versatility and reliability are much greater.
2 Chapter 1: Preface User’s Guide

Who should use StressCheck?
1
StressCheck’s unique handbook capability is combined with an automated paramet-
ric analysis capability making it convenient to investigate the sensitivity of a solu-
tion to variations in critical design parameters.
StressCheck’s unique and advanced post-processing capability allows detailed

evaluation of engineering data anywhere in the model without expensive and time
consuming re-run of the problem.
StressCheck incorporates a Component Object Model (COM) programming inter-

face which makes it possible to utilize StressCheck as a “black-box” finite element
computational server from any COM enabled programming language or applica-
tion.
Who should use StressCheck?

StressCheck has been developed to facilitate analysis throughout the design pro-
cess, making it a valuable tool for both analysts and design engineers. The hand-
book framework provides designers with easy access to advanced finite element
technology within an easy-to-use intuitive interface. The handbook library can be
expanded by FEA analysts to incorporate commonly encountered parts and
designs. The problems can then be executed quickly and easily by design engineers.
With COM you can develop engineering design utilities, optimization tools, or
application integration tools outside the StressCheck environment which activate
StressCheck as a background FEA server to solve predefined parametric part mod-
els or to solve models that have been constructed on the fly via COM model con-
struction functions.
StressCheck features
StressCheck is based on the p-version of the finite element method: The errors of
approximation are controlled by increasing the polynomial degree of the elements.
The main features of the program for elasticity are summarized in the table on page
6.

How to use this manual?
1
How to use this manual?
An overview of the user interface is presented in Chapter 2. Chapter 3 intro-
duces the Input classes, including geometry, mesh, thickness, material, loads,
constraints, and more. Chapter 4 introduces the various solution methodologies
including linear, nonlinear, modal, and buckling analysis. Chapter 5 discusses
the post-processing operations provided by StressCheck, including error esti-
mation, plotting, point extraction, resultant computation and fracture mechan-
ics. Chapter 6 introduces the Handbook Framework and describes the
construction of a simple model. For details about model creation refer to the
Modeling Guide. For specific analysis types and procedures refer to the Analy-
sis Guide and the Advanced Topics Guide.
Frequently asked questions about the p-version

In this section some frequently asked questions about the p-version of the finite
element method, which is the technological basis of StressCheck, are
answered.
Why is the p-version important?

The finite element method provides an approximate solution. In engineering
practice it is important to know not only the information one wishes to com-
pute but also to have an indication about the size of the error of approximation.
The p-version makes it convenient to obtain error estimates in terms of the data
of interest very efficiently. Since the analyst is responsible for the computed
information, it is important to have tools available which make it possible to
exercise that responsibility.
When was the p-version developed?

Research on the p-version dates back to the late 1960's. Many important
advances occurred in the 1970's. The theoretical basis was established in 1981
and optimal meshing strategies appropriate for the p-version were developed in
the period 1984-1985. For details we refer to Szabo and Babuska, Finite Ele-
ment Analysis, John Wiley & Sons, Inc. (1991). Beginning in 1985, these

StressCheck features - Elasticity
Model Materials Boundary Solution Output

Conditions
Geometry: Linear: Loads: Reference: Standard:
Parasolid Isotropic Tractions Plane-Stress Error Esti-
kernel Point Load Plane-Strain mate
Orthotropic
System Body Force Axisymm. Equilibrium
Anisotropic
Point Check
Fitted Fiber Spring Displ. Plate Bending
Line Resultants
Temperature- Bearing Extrusion
Circle dependent Contour
Shear 3D-Solids
Fillet Plotting
Moment
Ellipse Deformed
Imposed Dis- Analyses:
Nonlinear: Shape
Spline placement
Linear Min/Max
Cylinder Elastic- Thermal
Nonlinear Extraction
Cone Plastic Formula Material Point/Line/
Plane Bilinear Point Load Edge Extrac-
Nonlinear
Torus Ramberg- Importation Geometry tion
Formula Osgood Modal Averages
Constraints:
Composite Hyperelastic Prestress Animation
more Rigid Body Modal
Nodal Constr. Eigenvalue
Elements: Boundary General Advanced:
Laminated: Buckling
Beam Face Constr. Cold Working Fracture
Cartesian Mechanics:
Fastener Spring Coeff. Margin
(Flat)
Link Built-In Check SIF
Cylindrical
Quadrilateral Soft-Simple Measurement T-Stress
General
Triangular Curvature Symmetry Crack Path
J-integral
Hexahedral Antisymmetry Contact
GSIF
Pentahedral Fastener to Fas- Global-Local
tener GFIF
Tetrahedral
Hinge TSIF
Meshing:
Formula
Manual Point Constraint
2D-automesh Importation
3D-automesh
Boundary Layer Meshing

1
developments were made available for use in professional practice. The p-ver-
sion is a more recent technology than the h-version, the development of which
began in the late 1950's.
Are error estimation procedures available in h-version programs as well?

Most h-version programs offer some form of adaptive capability. The theory of
adaptive mesh construction was developed in the 1970's by Babuska and Rhei-
nboldt. The objective of an h-adaptive process is to obtain a sequence of finite
element meshes in such a way that the error measured in energy norm is mini-
mal, or nearly minimal, for each mesh. Subsequently Zienkiewicz and Zhou
proposed an adaptive scheme, variants of which have been implemented in h-
version programs. In general, h-version programs do not provide convenient
and reliable means for making an assessment of the quality of computed infor-
mation, however.
Does the p-version have clear advantages over the h-version?

Yes. For typical design problems in mechanical and civil engineering practice
the errors of approximation are reduced at an exponential rate when the num-
ber of degrees of freedom are increased, provided that the finite element mesh
is properly constructed. The h-version can provide algebraic convergence rates
only. This makes error control much more effective in the p-version. Further-
more, a converging sequence of solutions is much more naturally and conve-
niently obtained with the p-version than with the h-version. This makes it
feasible to employ quality control procedures in the setting of practical engi-
neering decision-making processes.
Are there significant differences in p-version FEA programs?

Yes. There are several important differences. For example, proper implementa-
tion of the p-version requires that the mappings from the standard elements to
the “real” elements must be sufficiently accurate so that the error of approxi-
mation is controlled by the mesh and the polynomial degree of elements, not
by the mapping procedures. This is because, unlike in the h-version, the mesh
is not refined as the number of degrees of freedom is increased. Quadratic and
cubic polynomial mappings (also known as isoparametric mappings) should
not be used in connection with the p-version unless the maximum polynomial
order is restricted to 4 or 5. StressCheck has advanced mapping procedures
implemented. Many other important differences exist in such areas as enforce-

1
ment of constraints, specification of loading conditions, the availability of nonlin-
ear analyses, graphic user interfaces, post-processing operations, etc.
Are there areas of application which can be handled by the p-version but not by the h-
version or vice versa?
In principle, any problem which can be solved by the h-version can be solved by
the p-version and, conversely, any problem which can be solved by the p-version
can be solved by the h-version. There are large differences in convergence rates,
however. For example, it was demonstrated in one well-documented plane elastic
model problem that to achieve one percent relative error in energy norm (which is
similar to the root-mean-square measure of error in stress), approximately 1000
degrees of freedom were needed with the p-version and properly designed mesh,
whereas 10 million degrees of freedom would have been required with the h-ver-
sion, utilizing 8-noded quadrilaterals and uniform mesh refinement. For details we
refer to p.190 in Szabo and Babuska, Finite Element Analysis (1991). There are
other important areas where the p-version has clear and substantial advantages:
adhesively bonded joints (where very large aspect ratios are required), structural
plates and shells, fracture mechanics, etc.
What are the advantages of StressCheck over other FEA programs?

There are several important advantages. The most important advantage is that
StressCheck is the only FEA program in existence today which was designed for
controlling both the errors of discretization and idealization. The errors of discreti-
zation are the errors controllable by the finite element mesh and the polynomial
degree (h- or p-extensions). The errors of idealization are the errors associated with
the restrictions incorporated in mathematical model. For example, the basic
assumptions of the linear theory of elasticity are that the strains are much smaller
than unity; the stress is proportional to the strain independently of the magnitude of
strain; the deformed and undeformed configuration of the elastic body are virtually
identical, hence the equilibrium equations can be written for the undeformed con-
figuration. Inasmuch as these assumptions may not be applicable in particular
cases, errors of idealization are incurred. StressCheck was designed so that the lin-
ear solution is a potential starting solution for a geometric and/or material nonlinear
problem.
There are many other advantages as well: StressCheck incorporates advanced pro-
cedures for the computation of stress intensity factors in linear elastic fracture
mechanics; it can compute the natural straining modes and the corresponding gen-

1
eralized stress intensity factors in homogeneous and heterogeneous bodies.
StressCheck is the first FEA program to provide hierarchic models for homo-
geneous and laminated plates. StressCheck provides a number of unique post-
processing procedures as well.
What are the recommended quality control procedures in FEA?

The main idea in quality control procedures is that the exact solution is inde-
pendent of the mesh or the polynomial degree. Therefore the data of interest
cannot depend on the choice of mesh or polynomial degree. Furthermore, the
data of interest should not be sensitive to the restrictive assumptions incorpo-
rated in the mathematical model. The recommended quality control procedures
consist of the following steps:
a) Linear analysis: Control of the errors of discretization.

• Verify that the error in energy norm (which is related to the RMS mea-
sure of error in stress) is reasonably small (typically under 5 percent).
• Knowing that the data of interest are finite, show that the data of inter-
est are substantially independent of the polynomial degree of elements.
• Show that equilibrium is satisfied.
• Show that there are no significant jumps in the stress contours.
b) Nonlinear analysis: Control of the errors of idealization.

• Show that the data of interest are independent of the restrictive
assumptions incorporated in the linear model. This requires that geo-
metric and/or material nonlinear analysis be performed.

2 StressCheck Interface
2
This chapter contains a general overview of the user interface. It covers the following topics:
• Interface layout
• Standard file extensions
• File menu options
• Edit menu options
• View menu options
• Display menu options
• Tools menu options
• Interface conventions
Interface Layout
The StressCheck user interface is designed to simplify data entry and to standardize
program operation. This interface consists of a Main Menu Bar and Main Toolbar
at the top of the screen, a Views Toolbar at the bottom of the screen, a graphic
Model Window in the center, and three dockable tool bars that provide access to the
program’s Actions, Objects, and Methods. Tabbed dialog windows provide for data
entry. When a tabbed dialog window has more tabs than can fit on the screen, a con-
User’s Guide Chapter 2: StressCheck Interface 9

Interface Layout
venient pop-up menu can be activated by a right mouse button click. FIGURE
1 illustrates one form that the user interface will take.
Model Window
2 Main Toolbar
Reference, Theory, and Units Selectors
Main Menu Bar
Method Toolbar
Action Toolbar
Object Toolbar
Status Line
Views Toolbar Tabbed Input Dialog Window
FIGURE 1 StressCheck screen layout.
The primary components of the StressCheck interface are the Main Menu Bar,
the Main Toolbar, the Model Window, and various Dialog Windows. The Main
Menu Bar and Main Toolbar provide access to program options such as open-
10 Chapter 2: StressCheck Interface User’s Guide

Interface Layout
ing and closing files, changing display attributes, selecting an input class, etc. The
Views Toolbar provides a shortcut to the most frequently used menu options, such
as display manipulation. The Reference Selector may be used to choose whether to
model a problem as a membrane, as an axisymmetric solid, as a plate in bending, or
as a fully three-dimensional solid. The Theory Selector may be used to indicate
whether the problem to be solved is an elasticity or a heat transfer problem. The 2
Model Window is where the finite element model will appear for both pre- and
post-processing. A dialog window is where most text based user interaction will
occur.
There are five primary dialog windows: one for model information, one for model
input, one for solver options, one for model results, and one for interacting with the
handbook framework. Each of these have been placed in the Main Toolbar and are
briefly discussed below.
Model Info The Model Information window can be displayed on the screen by selecting
“Model Info” from the Main Menu “Edit” pulldown menu. It can also be activated
by selecting the icon from the Main Toolbar (FIGURE 2). The three tabs at the top
“Model Info” Icon
FIGURE 2 Model Info interface.

Interface Layout
of the Model Info window give the user access to model descriptions, design
variable definitions, and design variable rules.
Input The Input window is divided into three sections (FIGURE 3).
2
“Create Model” Icon
Class Tabs
Curve/Surface Selector
Surface/Solid Option
Input Fields
Command Buttons
FIGURE 3 Input interface.
At the top are tabs, which allow the user to select a category of input. These
tabs are ordered in a way consistent with a typical model creation sequence:
Create geometry, attach a mesh, assign attributes, apply loads and constraints,
etc. In the center are the input fields and combo-boxes which relate to the spe-
cific category of input chosen with the tab. At the bottom of the window is a set
of push-buttons which are used to invoke a command.

Interface Layout
Some dialog windows contain a summary of the data records corresponding to a

particular class of input. This listbox gives the user access to data previously
entered so that it may be altered and replaced. For geometry and mesh classes, this
listbox can be viewed by selecting the “Index” tab.
The Input window can be displayed on the screen by selecting “Input” from the 2
Main Menu “Edit” pulldown menu, or by selecting an input class such as Geome-
try, Mesh, Thickness, etc., from the Main Menu “Class” pulldown menu. Alterna-
tively, the input dialog may be activated by selecting the “Create Model” icon in the
Main Toolbar. Refer to the Modeling Guide for details about the use of the Input
window.
Note: In the Geometry tab of the Input interface (See FIGURE 3), there is a button
between the Action and Object combo boxes, which controls whether the object
combo presents a list of point/curve objects , or point/surface/solid objects
. Point objects are included under both settings for convenience. There is also a
button which identifies whether a box, sphere, cone, cylinder or torus
should be created as a surface or a solid.
Solution The Solution window can be displayed by selecting “Solution” from the Main
Menu “Edit” pulldown menu, or by selecting the “Compute Solution” icon in the
Main Toolbar. The Solver interface is shown in FIGURE 4.
“Compute Solution” Icon
Solver Option Interface Solver Execution Interface
FIGURE 4 Solver interface.

Interface Layout
The interface contains several tabs, one for each type of solution supported by
StressCheck, i.e. Linear, Nonlinear, Modal, Buckling, Crack Path, Measure-
ment, and Margin Check. Once a specific solution type is selected, and the per-
tinent options are chosen, the solution may be performed by choosing the
“SOLVE!” tab. This tab contains the various solution options that are common
2 to all solution types. Refer to the StressCheck Solutions chapter for details
about the use of the “Solution” window.
Results The Results window can be displayed by selecting “Results” from the Main
Menu “Edit” pulldown menu, or by selecting the “View Results” icon in the
Main Toolbar. The Results dialog window is shown in FIGURE 5. The results
“View Results” Icon
Class Tabs
Solution ID Selection
Computation Options
Command Buttons FIGURE 5 Results interface.
interface contains several tabs, one for each type of post-processing option

Interface Layout
supported by StressCheck, i.e. Error Estimation, Plot, Point Extraction, Min/Max

Extraction, etc.
Every time the Results interface is selected, a graph window (FIGURE 6) appears.
The window is designed to present the results of most post-processing computa-
tions. NOTE: The Graph window is not applicable for the Plot class. Refer to the 2
Post-processing Operations chapter for details about the use of the “Results” win-
dow.
FIGURE 6 Graph window: Chart and Table options.
Handbook The Handbook Library window can be displayed by selecting “Handbook” from
the Main Menu “Edit” pulldown menu, or by selecting the “Handbook Library”
icon in the Main Toolbar. The Handbook Library interface is shown in FIGURE 7.
The Handbook Library interface provides access to, and interaction with, pre-
defined models of frequently occurring mechanical design components. Its main
row of tabs provides access to different functions of the handbook framework. The
“Model Info” tab through its Browser Icon gives access to the Model Browser. The
Model Browser provides a list of the available handbook models from which to
choose. Click on the Browser Icon and the Model Browser will be displayed on the
screen (FIGURE 8). The three buttons at the right hand side of the Browser Icon
provide access to the Icon Window and a capability to capture, edit and save an
image of the model. The Icon Window provides an illustration for each handbook
problem which is useful for associating the design parameters with the model. The
Keywords help to identify the type of model during browsing. The Comments area

Interface Layout
“Handbook Library” Icon
Handbook Tabs
2
Browser Icon
FIGURE 7 Handbook Library interface. Icon Window
usually provides information about the model, the execution of the model, and/
or the interpretation of results.
The “Analysis” tab gives the user control of model dimensions and other
design properties, and provides a set of command buttons which automatically
perform a solution, plot results, and compute engineering data specific to each
handbook model. The Analysis tab also contains a Design Study feature which
makes it possible to evaluate design variations by selecting design variables
which will be systematically changed during a series of solution computations.
The “Results” tab provides a variety of post processing procedures that may be
performed very conveniently within the handbook framework. Computing an
estimate of the error in energy norm, plotting standard engineering quantities,
computing minimum and maximum engineering quantities, computing engi-
neering data at selected locations in the model, computing resultants, comput-
ing fracture mechanics quantities, or computing various engineering properties
such as deformed area/volume or distortion, are possible options when using
this interface.

Interface Layout
FIGURE 8 Model Browser.
The “Material” tab provides access to material property definitions for linear and
nonlinear isotropic, orthotropic, and laminate composites. The user may modify
existing material properties only.
The “Constraint” tab gives the user the possibility of changing the existing type of
constraint. It applies to Built-In, Symmetry, Antisymmetry, Soft Simple and Free
types.
Additional details about the Handbook interface are provided in Chapter 6.

Standard File Extensions

During a typical modeling session, StressCheck produces several files, each
having a unique file extension. The most important files are the StressCheck
input file (.sci) which contains an ASCII representation of all model input data,
2 the Parasolid transmit file (.x_t), and the model icon file (.bmp or jpg). All
other files are usually referred to as the StressCheck database (.scb, .sck, .sol,
etc.) and can be reconstructed from the input files. During a modeling session
it is recommended that you periodically save a copy of your model input with a
unique name different from the current database name, to preserve your model
changes. This is accomplished by selecting the "Save Input" icon. The “Save”
icon will save model changes in memory that have not yet been recorded in the
database file, and produce a backup version of the input data in files with the
extensions ".sci", “.x_t”, and “.jpg”. When reporting problems related to
StressCheck to technical support, it is usually much more helpful to provide a
copy of the .sci and .x_t files than to provide the complete set of database files.
NOTE: StressCheck makes a copy of your database files, called a “shadow”

database to serve as a backup copy of your work. These files have an under-
score added to the end of the filename, i.e. model_.scb. The shadow database is
updated whenever you select SAVE. When exiting from StressCheck if you
choose not to save your changes, the shadow database will become the active
database and all model changes made since the last SAVE operation will be
discarded.
The file extensions used by StressCheck, and a brief description of each is

given below. Some files are direct access, and therefore are not human read-
able. Those files which are human readable, are denoted with *.
Required files: *.sci - default extension for StressCheck input file (batch file).
.x_t - Parasolid transmit file.
.scb - database manager file for intermediate data.
.sck - database manager file for the data dictionary.
.qgd, .qsd, .qgk, .qsk - database manager files for query management.
.log - database manager log file.
.sol - solution data used for post processing.
*.pcf - process communication file.

Disposable files: *.bak - backup copy of the StressCheck input file produced automatically by Stress-
Check at regular intervals during program use.
.wrk - stiffness matrix and other intermediate solution data. Must be kept only if
additional downward p-level or nonlinear solutions are desired.
.tmp - scratch file used during the solution process. Usually deleted automatically. 2
*.und - data transaction log file used to perform the UnDo operations and data
recovery.
*.red - data transaction log file used to perform the ReDo operations.
*.lis - session log file (cpu time measurements).
*.err - error log file (error messages and disk space estimates).
*.dat - temporary graph data file
Special request files: .scg - graph data.

*.mtx - unconstrained stiffness matrix data.
*.vct - unconstrained load vector data.
*.svc - solution vector data.
*.con - constraint data.
*.nik - nickname data.
*.ps - postscript file.
*.eps - encapsulated postscript file.
*.igs - IGES neutral file.
*.rtf - report file.
*.chk - file produced by Check action
*.cmd - analysis command file.
*.jac - jacobian data.
*.mas - mass matrix.
. jpg - handbook bitmap icon file.
.bmp - handbook bitmap icon file.
.htm - handbook HTML file (INFO).
.sctm - autoreport template file.

File Menu
.tti - template tag input file
2
File Menu
The following sections provide a brief summary of the menu options found in
the Main Menu Bar, FILE pulldown menu.
New When you use StressCheck to begin the analysis of a new problem, you must
create a new database file. You will be expected to provide a primary filename,
and the program will attach special filename extensions to this name for each
file it creates. See discussion above for an explanation of each filename created
by StressCheck.
Open If you have created a StressCheck database in a previous session, then you may
re-open the database using the Open menu option.
Close Use this option to close the current database, but keep StressCheck active.
Save Use this option if you wish to save the current state of your StressCheck model
to the database file (.scb). This option may also be invoked by pressing the F2
function key. This operation also copies the current database to a shadow copy
that preserves the current state of your model in case you wish to return to that
state later. See User Preference settings to enable/disable backup database.
SaveAs Use this option if you wish to save the current state of your StressCheck model
to a new set of files, with a different name. All StressCheck files will automat-
ically be copied and assigned the new name. After copying the files, Stress-
Check will be connected to the new files. Files with the old name may be re-
opened at a later time using the File/Open option.
Save Input Use this option to save your model input data to a StressCheck input file (.sci)
with a new name. This file is an ASCII format file containing a basic descrip-
tion of your input data. It is recommended to use a different name than that of
the database name to save the input file.

File Menu
Read Input Use this option to read a StressCheck input file (.sci) into your current database.
The current contents of your database will be erased automatically before the new
data is read in.
2
Erase Database Use this option to delete all information from the currently open database. The
database will be in the same condition as if you had opened a new database.
Recover Database StressCheck automatically makes a record of each input data transaction performed
during each session. This file is used to facilitate the Undo/Redo operations, and
may be used to recover the model input data in case of a hardware or software fail-
ure that might interrupt the current StressCheck session. If you select Recover, the
program will automatically delete all database files having the primary name you
supply, then create a new database and load the data found in the “.und” file. This
procedure only recovers input data. All solution data will be lost. It may be advis-
able to first copy your “.und” file to a new name and attempt to open the original
StressCheck files. If these files are found to be corrupted, making it impossible to
perform any further analysis, then it will be necessary to rename the “.und” file
back to its original name and perform a recovery operation.
Delete Database Files You may remove database files by specifying the primary name, or by selecting a
file from the dialog box. The program will automatically delete all files having the
primary name supplied, except for files with the “.sci” filename extension.
CAD File Import StressCheck provides a facility for importing CAD geometry that may be used as
the basis for the construction of a finite element model. Currently supported CAD
import formats include Parasolid, CATIA, and IGES. Additional formats may be
supported in the future. Use the File > Import menu option, then set the file filter to
identify which type file you wish to import, as illustrated in FIGURE 9.
The type of file to be imported is determined by the file extension, so if the file
extension of your file does not match one of those provided in the filter selector,
you must change the file extension on your file to match one of the options pro-
vided.
As the file is imported, the native CAD file format is translated into a Parasolid rep-
resentation, and the result is displayed in the model window. Although the model
may have been parametric in the native CAD system, it is imported with static
dimensions, and with no feature information. It is not possible to change dimen-
sions or to alter any of the relationships between various components in the solid

File Menu
FIGURE 9 CAD File Import dialog window
model. On the other hand, it is possible to add new details to the model using
boolean, blend and clipping operations provided in StressCheck. And, of
course, you may construct a geometrically associative finite element model
using the solid model as the basis.
CAD File Export StressCheck provides a facility for exporting its part geometry to 3rd party
CAD systems using the File > Export menu option. StressCheck’s geometry
may be exported to Parasolid or IGES formats.
The geometry exported from StressCheck will represent a static snapshot of the
model, and will not provide parametric or feature level information to the
receiving CAD system.
Print Use this option to print the current contents of the main model window. The
standard windows Print dialog will appear so that you may select the desired
printer, orientation, and other print options.
Graphic file output There are two different ways to save a graphic image of the model window to a
graphic file. The first method is to use the camera icon from the Views Toolbar
to activate the graphic capture utility which uses the outline box to select the
area of the window to capture. This facility automatically deposits the image

File Menu
FIGURE 10 CAD File Export interface
on the Windows clipboard, and saves the image to a JPEG file with the same name
as the database and file extension .jpg ( ).
The second method is to use the "Graphic File Output..." menu option in the File
menu. This menu option will activate a file dialog window as shown in FIGURE
11.
From this window, browse to the desired directory, enter a filename, and select the
file type: Postscript, Encapsulated Postscript, HOOPS Metafile, Pict, or HPGL.
Click on the Save button to capture an image of the full model window and write
the image in the chosen format in the filename provided.
View Session Log StressCheck creates a session log file where a variety of runtime statistics are kept.
The logfile can be examined outside StressCheck, or loaded into an edit window
using this menu option.
View Error Log Use this option to load the latest error log into an edit window. The Error log con-
tains all error messages generated in the current session. You may clear old mes-
sages from the file by selecting the Edit/Clear menu option, then the File/Save
menu option.

File Menu
FIGURE 11 Graphic File Output dialog window.
Edit ASCII File You may load any ASCII text file or Rich Text File into an edit window inside
StressCheck.
Exit When you exit from StressCheck, if your current preference setting (see Tools
menu) is set to not save a backup copy of your database files, StressCheck will
simply ask you whether you are sure you want to exit.
If your current preference setting is to save a backup copy of your database

files, then a dialog window will appear giving you the opportunity to decide
whether or not to save the changes you have made to the current database.
StressCheck actually maintains two complete copies of your database files.

One is the active database where all operations that you perform during the
session are applied. The second database is a shadow database which is pre-
served in the original state that existed at the start of the current session. If you
choose to save your changes, the shadow database will be deleted, and the cur-

Edit Menu
rent database will be preserved as the permanent version of your model database. If
you choose not to save your changes, the current database will be deleted, and the
shadow database will be preserved as the permanent version of the model database.
NOTE: if you perform a SAVE operation during a modeling session, the currently
active database is copied to the shadow database, and the original state of the data-
base will be lost. 2
Edit Menu
The following sections provide a brief summary of the menu options found in the
Main Menu Bar, EDIT pulldown menu.
Undo Use this option to reverse the effect of the previous data transaction. The Undo
applies only to creation, deletion, and modification of geometric objects, and other
input records. It does not apply to selection, blanking, rotation, or other display
related operations. The Undo operation may be repeated indefinitely until the entire
sequence of input operations is reversed. Note: Solution data is not preserved after
an Undo operation.
Redo Use this option to re-apply a data transaction which has been reversed with the
Undo operation. Like the Undo, Redo applies only to creation, deletion, and modi-
fication of geometric objects and other similar input records. The Redo may be
repeated until all Undo operations have been reapplied. If you perform a new oper-
ation after an Undo operation, you will no longer have the operation to perform a
Redo.
Model Info The Model Info option provides access to the Model Browser and gives the user
control of the model dimensions and other design properties. When clicking on this

Edit Menu
option, the Model Information dialog window appears giving access to three
tabs: the Model Info tab, the Parameters tab, and the Rules tab.
If the preference setting for the Model Browser is not enabled (see Tools menu)
a standard window file browser will appear when you select the Model
2 Browser button.
If the preference setting for the Model Browser is enabled, a special Handbook
file browser will appear when you click the Model Browser button.
The “Model Info” tab through its Model Browser icon gives access to the
Model Browser. The Model Browser provides a list of the available models
from which to choose. The three buttons at the right hand side of the Model
Browser icon relate to the editing and viewing of the Icon window. The Icon
window provides an illustration for each problem which is useful for associat-
ing the design parameters with the model. The Keywords area is provided to
help in the search for a type of problem. The Comments area may be used to
describe various characteristics of the model for a future user of the model. It
can also provide specific instructions to assist in the execution of a model and
the interpretation of results.
The “Parameters” tab (shown in FIGURE 12) gives access to controls which
make it possible to change the values of all parameters. By activating the Auto
Step and clicking on the up or down arrows it is possible to increment or decre-
ment the parameter value by the Auto Step value and the model will be
updated. If Auto Step is not selected, only the parameter value will change, and
the model will not be updated until the Accept button is selected. The first field
in the grid is the parameter selection field. By holding down the Shift key, sev-
eral parameters can be selected by clicking the first field with the left mouse
button. Use the Delete button to remove one or more selected parameters. The
last column (Sort) allows sorting of the parameters by other than the default
(alphabetically) rule.
To define a new parameter, simply enter the following information:
Name: The name to be assigned to the parameter. Avoid using an underscore

(“_”) as the first character of the parameter name. An underscore indicates a
reserved parameter name. Additional restrictions are discussed in Handbook
chapter.
Description: This description of the parameter will appear in the Handbook
Analysis interface.

Edit Menu
FIGURE 12 Parameter interface.
Expression: A parameter may be defined in terms of other parameters or may be

assigned a value directly. If a parameter is to be defined in terms of other parame-
ters, the value will be computed automatically. Care should be taken not to create a
circular definition, i.e. making the parameter depend indirectly on itself. Rules for
parameter expressions are covered in detail in Handbook chapter.
Value: The value assigned to the parameter.
Limit: If the value of a parameter should be limited to a specific range in order to
properly maintain the integrity of the model, it is possible to supply limits for the
parameter value. Simply enter the limits using relational operators, e.g. >0, <=7,
etc. If two limits are required they may be separated by a comma.
Class: Three options are available: Boundary Condition, Property and General.
These options are filters to organize parameters. This information facilitates the
parameters update for the Handbook user.
When parameters are defined, StressCheck automatically saves the last valid set of
parameters in a configuration called “_Input_Save”, so that you can go back to
them anytime. After saving the new values or expressions, StressCheck evaluates
the parameters and expressions in the proper order. If any parameter is detected as
being invalid, or if there is a recursive definition, this invalid set of parameter defi-
nitions is saved as a configuration called “_Input_Error”, and the current parame-
ters are set back to the last valid definitions. You can retrieve the “_Input_Error”
settings and make your corrections if you wish.

Edit Menu
The “Rules” tab gives access to define relationships between parameters which
will be enforced each time new values are assigned to parameters, FIGURE 13
illustrates the Rules interface. A rule is composed of a relational expression
and a message that will be issued if the expression is false.
FIGURE 13 Rules interface.
For more information, refer to Chapter 6.

NOTE: When a parametric expression is used in an input field, the length of
the expression is limited. If your expression exceeds the length limitation of a
given input field, you may use a formula definition to represent the expression
and reference the formula from the parametric input field. Define a formula
with the desired expression, then enter the name of the formula in the input
field preceded by a pipe symbol (|).
For example:
formula: fx
(a+1)*(b+2)/3*sin(length/height)
point definition:
x: |fx
y: 2.0
z: 0.0
It should also be noted that formulas that are preceded by a the pipe symbol
may not reference spatial variables within the expression (i.e. x,y,z).
Input The Input option provides access to the various input classes, including Geom-
etry, Mesh, Section Properties, Thickness, Material, Load, Constraint, and

Edit Menu
Solution ID. The StressCheck Input dialog window provides a tab for each avail-
able input class. This interface is described in greater detail in Chapter 3.
Solution The Solution option provides access to the various StressCheck solver options,
including Linear, Nonlinear, Coldworking, Modal, Buckling, Crack Path, Margin 2
Check, and Measurement analysis. The StressCheck Solution dialog window pro-
vides a tab for each solver option. Once you have entered the required information
for the desired solver, select the SOLVE! tab to activate the corresponding solution
procedure. This interface is described in greater detail starting in Chapter 4.
Results The Results option provides access to the various output classes, including Error
Estimation, Plot, Min/Max, Points, Resultant, Properties, and Fracture Mechanics.
The StressCheck Results dialog window has a tab for each results option. This
interface is described in greater detail starting in Chapter 5.
Handbook The Handbook option provides access to the StressCheck Handbook Library inter-
face, including the Model Info, Analysis, Results, Material, and Constraint options.
This interface is described in greater detail beginning in Chapter 6.
Formulae The Formula(e) option provides access to the dialog window for entering and edit-
ing formula record definitions.
Formulae are expressions which may be used in the definition of geometry, thick-
ness, materials and boundary condition data records, and during various data
extraction operations. The formula interface is shown in FIGURE 14.
To define a formula, provide the following information:
Name: Supply the name to be assigned to the formula.

System Option: Choose Cartesian, Cylindrical or Spherical system type that will
be used when the formula references spatial variables such as X,Y,Z.
Angle Option: A formula may be defined as a function of an angle measured
counter-clockwise from the positive x-axis (0 to 2pi) or from the negative x-axis (-
pi to pi).
Constants: You may define constants and then reference the constants in the for-
mula using a C followed by the constant number, for example: C1. Up to 10 con-
stants may be defined.

Edit Menu
FIGURE 14 Formula interface.
Subexpressions: You may define sub-expressions, and then reference these

sub-expressions in the formula using curly brackets around the sub-expression
number, for example: {1}. Subexpressions are limited to 31 characters long.
Formula: define the formula using any combination of solution variables,
parameters, constants and sub-expressions. You may use the standard intrinsic
functions such as sin() and cos() and other special StressCheck intrinsic func-
tions such as deg() and rad(). StressCheck also recognizes intrinsic constants
such as PI. Intrinsic functions are listed near the end of this chapter (Table 1).
Graph The Graph window may be activated in order to read in and display previously
stored graph data. See Chapter 5 for a detailed description of the graph window
user interface.
Report The Report window may be activated in order to read in and display previously
stored report data. See Chapter 5 for a detailed description of the report win-
dow user interface.

Class Menu
Class Menu
The CLASS pulldown menu provides quick access to the various Input and Results
class interfaces. Simply select the Input or Results class of interest and a dialog
window will appear, containing a set of “property sheet” tabs, with the appropriate 2
tab automatically selected.
Classes provide the basic organizational structure for input and results interaction
in StressCheck. Input classes include Geometry, Mesh, Thickness, Section Proper-
ties, etc. Each Input class provides access to model objects and input data records
which define the finite element model.
Results classes include Error Estimation, Plot Solution, Points Extraction, etc. The
Results classes provide access to the various post-processing features of Stress-
Check. Post-processing requires selection of the desired results class, followed by
selection of the solution(s) of interest and various options related to the chosen
results class.
Grow/Shrink The Class Menu also contains an entry which provides a keyboard accelerator for
the “Grow/Shrink” button found in the Input and Results dialog windows. The
Grow/Shrink button can be used to reduce the size of the dialog window tempo-
rarily, then restore it to full size.
View Menu
The VIEW pulldown menu provides quick access the various StressCheck dock-
able tool bars. You may remove a tool bar from the display, or replace it again by
selecting the corresponding menu option from the View Menu.
Views Toolbar Choose View > Views Toolbar to obtain the icons for all the available view perspec-
tives and other display manipulation operations.
You may select a predefined view of your model from the Views Toolbar. You may
store a particular view of your model, and then restore the model to this precise ori-
entation at a later time using the Save and Restore options in the Views Toolbar.
You may fit the current orientation of the model into the screen by selecting Center
Model. To get a 3-dimensional view of your model click on the Isometric option
listed in the Pre-defined views pulldown menu. The Isometric view is a 3D view

View Menu
2
Pre-defined views
Center Model Capture Graphics
Rotate
Insert Window in Report
Center of Rotation Display Report
Translate Restore Saved View
Zoom Save View
Box Zoom View Controls
with a 45 degrees rotation about the x-axis and a -35 degrees rotation about the
y-axis.
Select the Rotate icon, then use the mouse to dynamically rotate your model
while holding the right mouse button down and dragging the mouse cursor
over the model window. Dragging the mouse near the edge of the window will
cause a rotation about the window Z-axis. Dragging up or down near the center
of the screen will cause a rotation about the window X-axis. Dragging right or
left near the center of the screen will cause a rotation about the window Y-axis.
You may define a new center of rotation by selecting the Center of Rotation
icon and then using the mouse to point to a node, point, or coordinate system.
Click the right mouse button to select the new center. The current dynamic
operation will automatically switch to rotation.
Translate the model by clicking the Translate icon, then hold the right mouse
button down while dragging the mouse cursor over the model window.

View Menu
Zoom the model by clicking the Zoom icon, then hold the right mouse button down
while dragging the mouse cursor up and down over the main window. Moving
upward zooms in, moving downward zooms out.
Use the Box Zoom to identify a region of the model that you want to zoom into.
Click the right mouse button and drag the mouse cursor over the model window to 2
trace the bounding rectangle defining the zoom area.
The View Controls icon will be described in a later section of this chapter.
View Icon It is frequently useful to be able to preserve a particular view of a model that may
be recalled for future use. StressCheck provides an icon in the Views Toolbar
shown in FIGURE 15. By clicking directly on the icon with the arrow pointing
downward into the red box, the current view is saved as the default view. This will
be the view that appears whenever you load the model into StressCheck. This is
also the view that is restored when you click the icon to its right which is shown as
a green box with a small arrow pointing upward.
Save Default View Recall Default View
FIGURE 15 Views Toolbar
It is also possible to save several different views, and to assign each view a unique
name. These named views may be retrieved on demand. FIGURE 16 illustrates the
procedure for saving a named view. The procedure for saving a named view is as
follows:
• Click the small arrow next to the Save View icon to activate the Edit option.
• Click the Edit button to activate the Saved Orientations dialog window.
• Supply a name.
• Click the Save Button.

View Menu
To retrieve a previously saved view, click the small arrow to the right of the
Restore Saved View icon to see a list of available named views. If you have not
already saved at least one named view, no list will appear. Click on the name of
the view you wish to retrieve, and the model window will immediately be
updated with the corresponding view.
2
Save a named view
FIGURE 16 Save and Retrieve a named view.
Retrieve a named view
Edit Toolbar The View > Edit Toolbar contains icons for quick access to object editing fea-
tures which may be used to cancel selected objects, blank selected objects,
unblank blanked objects, and to undo or redo previous operations.
Undo Display Reset

Redo Reveal Blanked
Cancel Highlighted Objects Unblank Objects
Cancel Specific Object Type Blank Objects
Invert Selection Select Any Object

View Menu
Use the Display Reset icon button to refresh the model window display. This oper-
ation will rescale boundary condition symbols if necessary.
Use the Select Any Object icon button to identify the object type you wish to select
from the adjacent pulldown menu. You may immediately select objects of that type
with the mouse cursor without having to navigate to an Action/Object/Method class 2
interface. The object type chosen becomes the new “current” object type for all
subsequent operations.
Attributes Toolbar The View > Attributes Toolbar contains icons allow you to toggle on and off model
attributes (loads, constraints, etc.) in the graphic display area. To control scaling of
the attribute symbols, you must interact directly with the corresponding property
sheet class tab of the Input dialog window.
Display Thickness Display Constraints
Display Section Properties Display Loads
Display Materials Display p-level
Display Objects The View > Display Objects Toolbar contains icons for controlling the objects
being displayed in the display window.
Display Objects
Display Points Display Text
Display Nodes Display Fasteners
Display Systems Display Elements
Display Curves Display Surfaces

View Menu
Input Grid The View > Input Grid > Toggle Toolbar contains three switches provided to
represent the YZ, XZ, and XY planes. You may change the grid point spacing
by supplying a new value of the spacing in the input field next to Grid.
Display Options The View > Display Options Toolbar contains icons for better visualization of
your model.
Surface Grid
Axis/Legend Surface Shade
Perspective Wetted Faces
Toggle Light Source Element Handles
Wireframe Element Edges
Hidden Lines Shrink Elements
Shade Cutting Plane
Cutting Plane The cutting plane dialog window provides the ability to introduce a cutting
plane into any model that contains a 3D solid geometric object.

Display Menu
The controls shown in the figure may be used to position and orient each of three
cutting planes. The “X” button activates a cutting plane in the YZ plane, ie. the
X=0 plane. The “Y” button controls the Y=0 plane, and the “Z” button controls the
Z=0 plane.
The “Show Grid” button activates a visual representation of the active cutting 2
planes. The “Spacing” input field gives control of the grid density used to represent
the cutting planes.
The “Rotate” controls will cause all active planes to be rotated about the corre-
sponding global coordinate axis.
The “Translate” controls will cause all active planes to be translated along the cor-
responding global axis.
Inactive cutting planes will not be affected by the rotations or translations.
To reset the cutting planes to their default locations, click the “Reset Planes” but-
ton.
For models that contain a finite element mesh, a surface will be constructed that
represents the intersection between the cutting plane and the solid. This surface is
subdivided into two dimensional facets and the finite element solution is interro-
gated at the vertices of each facet in order to display a fringe plot on the cutting
plane surface.
Display Menu
There are several ways to manipulate the contents and appearance of the graphic
display information. These options are contained in the DISPLAY pulldown menu
in the Main Menu Bar, in the Display Controls dialog box (Display > View Con-
trols).
Reset Reconstruct the main window display. Rescale attribute symbols if necessary.
Move You may change the orientation of the model on the screen, by translating, zoom-
ing, or rotating. These basic operations may be performed in several ways. The
function keys on your keyboard are defined as shown in FIGURE 17.

Display Menu
Function keys The function keys are active only when the display window has focus. The
rotation keys are for the angular rotation about the global coordinate axis. The
rotation step after each click can be adjusted in the Display Control box shown
in FIGURE 21. The default step size is 10 degrees. Translation and zoom are
self-explanatory.
2
F1 F2
HELP SAVE
F3 F4 F5 F6 F7 F8
Z+ Z- Y+ Y- X+ X- Rotation
F9 F10
Front Menu
F11 F12
IN OUT Zoom
Translation
FIGURE 17 Function key definitions.
Model orientation may also be manipulated dynamically using the mouse cur-
sor. First you must choose the type of orientation operation you wish to per-
form by selecting Display > Move in the main menu or clicking the appropriate
icon in the Views Toolbar. Translation, Rotation, and Zoom are self explana-
tory. Just press the right mouse button and drag the mouse while you hold
down the button.
The Display > Move > Box Zoom option is provided so that you may draw a
rectangle around the area of interest. This area will be expanded to fill the dis-
play window.

Display Menu
Accelerator Keys StressCheck makes several functions accessible by keyboard accelerators. Different
windows have their own accelerators. Those that are defined when the main
graphic display window has focus are as follows:
Ctrl + u – unblank all blanked objects
Alt + n – open a new database 2
Alt + o – open an existing database
Ctrl + p - reveal blanked
Ctrl + o - blank objects
Ctrl + i - invert selection
Ctrl + s – save the current contents memory to the database files
Ctrl + v – save the current contents of memory to a database with a new name
Ctrl + y – activate the redo procedure
Ctrl + z – activate the undo procedure
Alt + p – print the contents of the main window
Alt + r – initiate a database recovery
Alt +w – alternate between grow and shrink of current input dialog window
Alt + i – activate the Input interface
Alt + s – activate the Solve interface
Alt + x – activate the Extraction interface
Alt + m – activate the Model Info interface
Alt + l – activate the Handbook interface
d – restore the default saved view
v – save new default view
c – center the model in the window
r – switch to dynamic rotation
t – switch to dynamic translation
z – switch to dynamic zoom
b – switch to dynamic box zoom
a - show previous view
Those that are defined when an input dialog window has focus:
Ctrl + c – copy contents of current text input field

Display Menu
Ctrl + x – cut contents of current text input field

Ctrl + v – cut contents of current text input field
Shift + delete – cut contents of current text field
Shift + insert – paste contents to current text field
2 Alt + backspace – undo/redo within a text input field
Attributes The Display > Attributes menu contains options for displaying various model
attributes in the graphic display area, for example loads, constraints, materials,
thickness, etc. To control scaling of the attribute symbols, you must interact
directly with the corresponding property sheet class tab of the Input dialog
window. Attribute display may also be controlled using the View > Attributes
Toolbar discussed earlier.
Selection The Display > Selection menu provides for controlling the display of blanked
objects.
Assign Colors Display > Assign Colors provides options for defining 4 different color tables:
Default, Material, User Default, and User Fringe. These tables may be modi-
fied and made active in the Display > Assign Colors interface (FIGURE 18) as
described in the following paragraphs.
The Default color table is a set of colors for each object type displayed in
StressCheck, and is considered the “session” color table. This means that these
colors will be used for all models that you may load into StressCheck during
the current session. This color table is initialized from a color definition file
(.def) that is part of your user preferences. The Default color table is of fixed
size (20 colors), but you are free to change the entries in the default color table
if you wish. These new settings will be in effect for the remainder of your cur-
rent StressCheck session. You may preserve these settings for future use by
writing them to a color definition file (Write button). This file may be refer-
enced in your user preferences, (See Tools > User Preferences) or may simply
be read back into StressCheck during a future session (Read button). If refer-
enced in the preference file, this color table will automatically be used at the
start of each StressCheck session.
The Material color table is created for each StressCheck model, and is pre-
served with the input data as part of the StressCheck input file (.sci). You may
create new entries in the Material color table, and then reference these color
entries by name in the Input > Material > Assign interface.

Display Menu
FIGURE 18 Assign Colors.
The User Default color table is similar to the Default color table, except that it is
preserved in the StressCheck input file for the current model. Each model may have
its own User Default color table which may be activated from the Display > Assign
Colors interface as a substitute for the standard Default color table.
The User Fringe color table is provided so that you may override the internally gen-
erated fringe color table that StressCheck uses to display contour plots. If you
define your own User Fringe color table, and make it active, the number of entries
in the table will automatically control the number of fringe intervals used during
contour generation.
To change a color table, select the desired table from the combo box in the upper
left of the Assign Colors interface. The entries in the table will be displayed in the
scrolling list at the top of the interface. To change an entry in the table, select the
entry from the scrolling list. The color name and Red/Green/Blue values will feed
back into the interface. You may directly edit these values, or click the "Browse
Colors..." button to set them by choosing from a color palette. To save the new
color settings, click Replace. To add a new entry to the color table (Material and
User Fringe tables only), click Add. To save the current color table to a file, click
Write, and supply a file name (.def) where the color table will be saved. To read a

Display Menu
saved color table back into the Assign Colors interface, click Read and browse
to the saved file (.def). This saved color table will replace the current table
identified in the Table combo box.
One use for this Read/Write facility is to convert a Default color table to a User
2 Default table. First edit the Default table by selecting color entries and modify-
ing them until you are satisfied with the settings. Then write the settings to a
file. Next, change the combo to User Default, and read the saved file back.
Finally, click Replace to update the User Default table with the retrieved set-
tings.
The “Use” box at the right of the interface controls which color table is cur-
rently active. You may choose either Default or User Default as the current
table controlling object colors. You may choose whether the internally gener-
ated color table is used for fringes (User Fringe not checked), or you may acti-
vate the User Fringe table by checking the box.
The User Preferences interface is described in a later section of this chapter.
Model Icon Displays the Icon Window that belongs to the Handbook Library and provides
an illustration of the problem model selected.
FIGURE 19 Model Icon.
Objects The Display Objects dialog window shown in FIGURE 20 provides a mecha-
nism for controlling the display and labelling of each type of object. The label
check box will turn the object labels on or off. The display check box will

Display Menu
enable or disable the display of each object type. If you wish to view a specific
range of objects, or a specific set of objects, select the corresponding tab labelled
“Ranges” or “Sets”. Each object is assigned an object number, which may be used
to display a range of objects. Sets may be created using the Set tab under the Input
Class, and may be referenced in the Sets tab to display only the objects belonging to
the selected set. 2
Set
Range
FIGURE 20 Display Objects dialog window.
View Controls Model orientation may also be controlled by bringing the View Controls dialog box
shown in FIGURE 21 to the screen by activating Display > View Controls. This
box also contains parameters for controlling the size of each rotation step, transla-
tion step, and zoom step. In addition, you may control the shrink option for the ele-
ments. The resolution of geometric boundary objects and element edges may be
increased to improve display precision, or decreased to improve display speed. The
Display Format controls the precision of the data values displayed in the Geometry
input box and the input Check reports; this is a C language format specification.
It is possible to display the finite elements transparently by picking the Mesh but-
ton. It is possible to display geometric surfaces transparently by picking the Geom-

Display Menu
2
FIGURE 21 View Controls.
etry button. Choose the degree of transparency by entering a value from 0 to

100 in the input field, or by clicking the spinner arrows next to the input field.
A value of zero (0) represents complete opacity. A value of 100 represents
complete transparency.
Note on Face Displacement: When edges (or curves) are coincident, or very
nearly coincident with faces (or surfaces), the edges can be visually obscured
by the coincident faces. In order to reduce this effect, faces are displaced
slightly away from the direction of the point of view so that the edges will
appear slightly in front of a coincident surface. However, in the case of very
thin domains, where there are two surfaces which are themselves nearly coinci-
dent, the displaced location of the faces may be such that the edges which
should be hidden from view will appear to bleed through and become visible in
front of the front surface. When this occurs, it may be desirable to reduce the
amount of face displacement. By decreasing the value for Face Displacement
in the Display Control interface, this “bleeding” effect can be reduced. The

Display Menu
consequence is that edges which are in fact coincident with a surface may not be as
visible in the general case.
Model Summary You may obtain a summary of model information such as the number of elements,
number of nodes, and current coordinate range. The Model Summary window is 2
illustrated in FIGURE 22.
FIGURE 22 Model Summary window.
Material Summary You may obtain a summary of standard materials currently used in your model.
This summary carries useful information about the material including its type (iso-
tropic, orthotropic, anisotropic) its nonlinear behavior law (Ramberg-Osgood, Elas-
toplastic, Bilinear, etc.) and so on. Information presented in this summary comes
from a material database file provided with StressCheck. This material database can
be updated at a customer site to include materials of particular interest to the user.
An alternative material file may be supplied and identified by name in a Windows
registry key. Define the registry key named ESRD7_MATERIAL_FILE and assign
the value to be the complete path name of the material file.
The format of the material database file (/esrddev7/ui/material.dat) is as follows:

Each material has four records.
Record 1: Name, class, type, law, units

name - up to 31 characters
class - up to 16 characters
type - one of the following

Display Menu
(I)sotropic
(O)rthotropic
(A)nisotropic
(T)ransversely-Isotropic
2 (L)aplace
law - one of the following
(N)one
(E)lasto-Plastic
(B)ilinear
(5)Parameters
(R)amberg-Osgood
(H)yperelastic (Isotropic Exponential)
(O)rthotropic-Exponential
units - one of the following
(N)one
(S)I
(U)S
Record 2: Material properties. See Advanced Guide for the coefficients defini-
tion.
For Isotropic: Modulus of elasticity (E), Poisson’s ratio (v), coefficient of
thermal expansion (ath), specific weight, mass density.
For Orthotropic or Laminate Orthotropic: E11, E22, E33, v12, v23, v31,
G12, G23, G31, a11, a22, a33, density, thickness.
For Transversely-Isotropic: E11, E22, G12, v12, v23, a11, density.
For Anisotropic: Not supported.
For Laplace: Kxx, Kxy, Kyy, Kxz, Kyz, Kzz, Q.
Record 3: Six nonlinear coefficients: v, E1, E2, A1, A2, A3. Where v is the
Poisson’s ratio, E1 is the modulus of elasticity and the other coefficients
depend on the stress-strain law:
Elastic-Plastic: E2=0, A1=Yield strength, A2=A3=0
Bilinear: E2=Tangent modulus, A1=Yield strength, A2=A3=0

Display Menu
5-Param: E2=Tangent modulus, A1= Yield strength, A2 and A3 are the strain
and stress values at the beginning of the strain-hardening.
Ramberg-Ossgood: E2=0, A1=S70E, A2=n, A3=0
Record 4: Comment. 2
Example for an Isotropic material:
2014-T6 Extrusion, ALUMINUM, ISOTROPIC, Ramberg-Osgood, US <-Record 1
0.109e8, 0.397, 0.128e-4, 0.101. 0.262e-3 <- Record 2
0.397, 0.109e8, 0.53e5, 18.5, 0.0, 0.0 <- Record 3
Syield=58 ksi, S-UTS=66 ksi, E-UTS=13%, Kic=30 ksi in^.5 <- Record 4
The figure below shows the information presented to the user for this example.
Note that it only includes the first and last records.
The material properties provided in the material database are given in the units indi-
cated in the table below.
Quantity US SI
Length in mm
Force lb N
Time sec sec
Pressure lb/in2 MPa (N/mm2)
Density lb sec2/in4 N sec2/mm4

Tools Menu
Quantity US SI
Coef. Thermal Expansion 1/oF 1/oC
2 Coef. Heat Conduction BTU/ sec in oF W/mm oC
Convective Coef. BTU/ sec in2 oF W/mm2 oC
Tools Menu
The TOOLS pulldown menu provides access to a few additional features of
StressCheck that are used relatively infrequently. These are described in the
sections which follow.
Mode The Tools > Mode option may be set to either Novice or Expert. Novice mode
is the default and indicates to the program that it should automatically display
the Formula or Parameter dialog windows anytime you enter an input value
that contains an undefined parameter or formula. Expert mode enables shell
elements in 3D, suppresses the test for valid parameter limits of surface defini-
tions, suppresses the test for use of restricted parameter names in formulas,
suppresses the error that the number of parameters in the configuration file
does not match number of parameters currently defined when reading a param-
eter configuration file, enables product space in 3D and extrusion, and does not
disable the repeat toggle after Accept during creation.
Table Reset This is an advanced feature used only when table data is used to describe a fit-
ted material. The table data may be stored in standard element coordinates.
When loading a batch file (.sci) containing table data which has standard ele-
ment coordinates, StressCheck will assume that since the data has just been
read into the program, the coordinates have not yet been inverse mapped, and
will repeat the inverse mapping operation. If you select “Table Reset”, the
inverse mapping operation will be skipped. This is important when the table
data is stored in standard coordinates, and some parametric geometry is
changed, but the relative locations within each element should be preserved.

General Interface Conventions
Set Font The Set Font option is used to activate the Font dialog window shown in FIGURE
23. This window provides a mechanism for selecting the font used for graphic
object labels in the main window.
FIGURE 23 Font Selection window.
Command File The Command File option may be used to execute a StressCheck command file
which has been constructed manually with a file editor. This option is an unsup-
ported feature intended only for internal development use.
User Preferences The User Preferences option is used to activate the User Preferences dialog win-
dow (FIGURE 24). This window lets you control a variety of default program set-
tings such as the user mode (expert, novice) and interface option (input, solver,
results, etc.). You may choose whether or not various dialog boxes will automati-
cally appear on the screen, the path and file name of the file containing the default
color settings for the user interface, and the thickness of the lines used to display
your model in the main window. Preference files may be stored in any directory.

Since StressCheck is based on Windows graphic user interface development tools,
there are several standard conventions for interacting with the program. These and

FIGURE 24 User Preferences interface.
some important guidelines for navigating through the StressCheck interface are
described in the following sections.
On-line HELP On-line help is accessible from the HELP pulldown menu in the Main Menu
Bar or by pressing the F1 key while pointing to the interface control of interest.
This will open the Master Guide document in the relevant section. To access a
tool tip help, hold down the SHIFT key and then press F1. StressCheck docu-
mentation files (.pdf) are located in a subdirectory (/docs/1033) of the Stress-
Check installation directory. Any files (up to a maximum of 15 files) found in
this directory with the extension .pdf will appear in the StressCheck HELP
menu and will activate Adobe Acrobat when selected for viewing. This makes
it possible for each site to develop its own documentation for the use of Stress-
Check, and to make this documentation (such as company specific modeling
practices) readily available to users.
TAB navigation Another user interface convention is to use the Tab key to navigate between
text fields. Tab by itself navigates forward. Shift/ Tab moves through the fields
in reverse order. You should also note that a TAB operation will sometimes
trigger an interaction with the program. For example, if you enable the input
grid and change the grid spacing, the display will be updated automatically
when you press the Tab key.
Enter key Pressing the Enter key in the Input Class, Results Class, or Handbook Results
interface is the same as clicking on the Accept button.

Delete key The Delete key will perform the same object deletion operation as clicking the
Delete button.
Backspace key When an object being created requires more than two selections from the screen,
such as polyline, spline curve, spline surface, or element, the Backspace key makes 2
it possible to go back one step in order to select a different basis object.
Escape key The Escape key will interrupt any time-consuming operation such as model solu-
tion or animation.
Right mouse click Clicking the right mouse button over the model window will abort any construction
operation. It may also be used to refresh the display to reset highlighted objects.
Input evaluation StressCheck provides a mechanism for evaluating simple expressions and deposit-
ing the result in the input field. If you enter an equal sign (=), followed by an
expression (FORTRAN or Mathematica conventions), the result will appear after
pressing the Tab key. For example,
=sin(rad(45))^2
would convert 45 degrees to radians, evaluate the sine, and then square the result.
You also could have used the FORTRAN convention for the exponentiation as in
the following example:
=sin(pi/4)**2.
All computations are performed in double precision whether you supply a decimal
point or not. Input is not case sensitive. The expression may reference user defined
parameters, (e.g. pi, lo) and the intrinsic functions summarized in the following
table:
Table 1: StressCheck Intrinsic Functions
Function Syntax
square root sqrt(x)
sine sin(x)
cosine cos(x)
tangent tan(x)
hyperbolic sine sinh(x)
hyperbolic cosine cosh(x)

Function Syntax
hyperbolic tangent tanh(x)
arc sine asin(x)
arc cosine acos(x)
2 arc tangent atan(x)
log base 10 log10(x)
natural log log(x)
ex exp(x)
convert radians to degrees deg(x)
convert degrees to radians rad(x)
absolute value abs(x)
arc tangent 2 atan2(x)
modulus mod(x;y)
minimum of two values min(x;y)
maximum of two values max(x;y)
test for relational condition* if(condition;true;false)
less than relation x :lt: y
greater than relation x :gt: y
equality relation x :eq: y
inequality relation x :ne: y
less than or equal relation x :le: y
greater than or equal relation x :ge: y
and relation x :and: y
or relation x :or: y
not relation x :not: y
Note that the asterisked function “if” is intended for use only in special situa-
tions in the formula input dialog window. Also note that when an intrinsic
function requires more than one argument, the arguments must be separated by
a semi-colon (;) not a comma (,) as you might expect. This is because the
comma is used as a data delimiter in the StressCheck input file.
There is a special set of intrinsic functions for computing section properties

which may be called from formulas used in the application of loads. These
functions are evaluated on a set of elements or element faces. The functions are
referenced in a formula without any arguments and without parentheses (e.g.

c1*sparea). The following table contains a list of the special intrinsic functions for
section property computation.
Table 2: Section Property Intrinsic Functions

2
Function Description
sparea Area
spvol Volume
spctrx,spctry,spctrz X,Y,Z of centroidal axis
spmomx, spmomy X,Y Moment of inertia about centroidal axis
spcross Cross moment of inertia
spmom1, spmom2 Moment w.r.t. 1st and 2nd principal directions
spangle First principal angle
The intrinsic functions described below, interrogate the definitions of existing

points and nodes to obtain information about their location in space.
Table 3: Intrinsic Functions
$n_d(n1,n2) distance from node “n1” to node “n2”
$n_x(n1) x coordinate of node “n1”
$n_y(n1) y coordinate of node “n1”
$n_z(n1) z coordinate of node “n1”
$n_u(n1,b1) U offset of node “n1” projected to boundary “b1”
$n_v(n1,b1) V offset of node “n1” projected to boundary “b1”
$n_ud(n1,b1) (degrees) U offset of node “n1” projected to boundary “b1”
$n_vd(n1,b1) (degrees) V offset of node “n1” projected to boundary “b1”
$p_d(p1,p2) distance from point “p1” to point “p2”
$p_x(p1) x coordinate of point “p1”
$p_y(p1) y coordinate of point “p1”
$p_z(p1) z coordinate of point “p1”
$p_u(p1,b1) U offset of point “p1” projected to boundary “b1”
$p_v(p1,b1) V offset of point “p1” projected to boundary “b1”

$n_ud(p1,b1) (degrees) U offset of point “p1” projected to boundary “b1”
$n_vd(p1,b1) (degrees) V offset of point “p1” projected to boundary “b1”
2 Window sizing Windows may be resized by grabbing the edge or corner of the window and
dragging to obtain the desired dimensions. All graphic display windows will
automatically scale the size of the image to the new window size. The System
Menu (found in the upper left corner of the window title bar or obtained by
clicking the right mouse button on the title bar) contains a “Size” option which
may also be used to resize a window. After selecting the Size option, use the
arrow keys to identify the edge of the window to stretch, then use the arrow
keys to actually shrink or grow the window.
Display refresh The display may be refreshed at any time by clicking the right mouse button in
the model window while it has focus. This action will also interrupt any multi-
step operation such as creating a polyline.
Process abort You may abort any time-consuming computation prior to completion by press-
ing the <Esc> key while the model window has focus. The program will usu-
ally continue to perform the current task for a few seconds until it reaches a
safe place to interrupt the computation. The following message will appear on
the screen.
Reserved parameters StressCheck provides special parameters for user control of some advanced
program features. All internal parameters must begin with an underscore (“_”),
so that they will not be confused with your own parameters used, for example,
as design variables.

C/A/O/M The user interface frequently makes use of a Class/ Action/ Object/ Method con-
vention for command interpretation. This can be roughly compared with declara-
tive sentence construction. The Class tells the program what type of data you wish
to work with: Geometry, Mesh, Material, etc. The Action is the verb of the com-
mand telling the program what action you wish to perform: Create, Select, Edit, etc.
The Object will be acted upon when the command is performed, as when we create
2
a point, select a circle, move a node. The Method is how the Action will be per-
formed on the Object. For example, we might create a node as an intersection of
two boundaries. This approach is also used when specifying boundary conditions
and loads. You might Select (action) a Curve (object) to apply a load by Traction
(method).
Note that the program constructs a message at the bottom of the display area based
on the current action, object and method. This is how the program keeps you
informed of what input is expected from you in the graphic display area.
It is important to remember that the mouse cursor is always ready to perform the
current Action > Object > Method command when you press the left mouse button.
Types of feedback StressCheck provides several types of feedback to assist in the interpretation of
graphic display information. This is accomplished by varying the cursor icon, the
color of individual objects types, and by varying the type of lines used to display
objects.
Cursors: Each time you change the action in StressCheck, the cursor will change to
reflect the current action. Once you learn the different icons used by StressCheck,
you will be able to determine quickly what action the program is waiting for. For
example, the Select action uses the hand icon. As long as the hand icon appears on
the screen, the program is ready to mark the next object selected. No matter what
icon is displayed, you are free to make menu selections or to manipulate any but-
tons or text fields available to you in the user interface. The only information the
cursor icon conveys is the currently selected action, which will be invoked by a
graphic cursor pick (left mouse button click in the graphic display area).
Colors: Each type of object is displayed in a different color. For example, bound-
aries are displayed in one particular color, elements in another color, etc. When an
object is selected or blanked, it is displayed in yet another color. This makes it easy
to interpret the status of each object displayed.
Line Types: Line type is another way to distinguish objects. Objects are usually dis-
played with solid lines. However, when both elements and boundaries are dis-
played, the boundaries are displayed with dashed lines and the elements with solid

lines. This is so that when you select a boundary, it is still possible to see the
underlying element edge between the dashes of the selected boundary.
Input conventions One of the most important input conventions is related to the use of toggle but-
tons where there is an associated text input area. When you see a toggle button
2 turned on, the associated input field will be white, to indicate that manual key-
stroke input is enabled. When a toggle is turned off, the associated input field
will be shaded and will be insensitive to keystroke input. In most situations, a
disabled toggle button will indicate that the program is automatically assigning
a value to the input parameter represented by the corresponding input field.
This convention is followed throughout the program.
Selection Object selection is accomplished by clicking the left mouse button while the
mouse cursor is pointing to the desired object. If you pick another object with-
out holding any other keys, the current selections will automatically be can-
celled, and the new object will be selected. To cancel a single selected object
while retaining the selection status of other selected objects, depress the Ctrl
key while clicking the left mouse button. To select additional objects without
canceling current selections, hold the shift key while picking with the mouse
cursor and left mouse button.
If you simultaneously hold the Ctrl and Shift keys while selecting an object,
StressCheck will automatically search all sets to determine whether the
selected object is a member of one or more sets. If so, the members of the first
set encountered that contains the object will automatically be selected. If more
than one set contains the object, the user may index through each set by press-
ing the Page-Up or Page-Down keys while the main window has focus. The
objects belonging to each set will automatically be selected as the key is
pressed. The set name corresponding to the set currently highlighted will auto-
matically be displayed in the “Set” combo box in the current user interface.
Set browser The set browser tool may be used to select objects belonging to one or more
sets, to be subsequently incorporated into a new set for assigning thickness,
material, boundary conditions, or other modeling attributes. The set browser
may be activated by selecting the main menu Tools/Set Brower option. The
browser is also automatically activated when picking an object while holding
the Ctrl-Shift keys. See FIGURE 25.
The set browser interface consists of a list of candidate sets displayed at the
right, and a list of selected sets displayed at the left. If launched from the Tools

FIGURE 25 Set browser interface
menu, the candidate list will automatically be filled with the names of all sets that
match the current object type. If launched from a Ctrl-Shift-Pick, the candidate list
will be filled only with the names of those sets containing the picked object. The
Candidate set list may also be filled by the names of all sets of a particular type by
making a selection from the “Objects” combo box.
When a set name from either list is selected, the members of that set will automati-
cally be highlighted on the screen. A “candidate set” may be identified as a “selec-
tion set” by double clicking the set name (or selecting the set name and clicking the
Move button) from the list at the right, at which time it will be added to the selec-
tion list at the left. To remove a selected set name from the list at the left, double
click the set name or select the set name and click the Delete button. To remove all
sets listed in the selection list, click the Purge button. To highlight the members of
all selection sets, click the Select icon. The members of the selection sets will also
be highlighted whenever the Move, Delete or Select icons are picked.
Attribute assignment Assignment by region: Attributes such as thickness, material, p-discretization, and
some element oriented boundary conditions may be assigned according to whether
an element lies within the boundaries of a sheet body or within the volume of a geo-
metric shell. This feature is particularly useful for assigning attributes to automati-
cally generated meshes. Choose the AOM settings Select/Mesh Region/Selection
then select a geometric surface. All elements attached to the selected sheet body or
lying within the volume of the selected geometric shell will automatically be
selected.

The COM interface
Assignment by surface matching: Attributes such as material properties, ele-

ment boundary conditions, and p-discretization may be assigned to elements
that are found to be attached to a selected surface. Choose the AOM settings
Select/Any Element/Match and select a surface. The elements that are attached
to the surface will automatically be selected. The selection technique can also
2 be used to conveniently blank elements attached to selected surfaces.
Marquee Pick When you position the mouse at the location where you wish to pick an object,
then press and release the left mouse button, you perform a standard pick.
However, if you press the left mouse button, then drag the mouse, a marquee
selection rectangle appears. When you release the mouse button, all objects of
the currently selected type which lie within the marquee rectangle will be
selected. As with standard selection, you must press the Shift key in order to
add additional object to the set of objects already selected. Similarly, you must
press the Ctrl key while performing the selection in order to cancel previously
selected objects.
Dynamic operations All dynamic display operations can be performed by dragging the mouse
across the display area with the right mouse button depressed. This technique
is used for rotation, translation, dynamic zoom and box zoom.
The right mouse button is also used to select a point or node as the center of
rotation.
Clicking the right mouse button in the model window without dragging will
indicate that a multistep operation should be aborted.
Pressing the F1 key over a dialog control provides a tool tip information box.
Pressing Shift-F1 will activate online help for the selected control.
The COM interface

StressCheck provides a COM (Component Object Model) interface to its
model construction, solution, and post-processing functions so that Stress-
Check may be used as a “black-box” finite element solution engine for user-
developed applications that are developed using COM enabled tools such as
Visual Basic, Excel, MatLab, etc .

The COM interface
Documentation for StressCheck’s COM interface may be found by navigating to

the Windows Start menu, Programs, ESRD, Programming Help (). This on-line
documentation describes all the classes, objects, methods and properties of the
COM interface. Sample Visual Basic programs illustrating the use of the COM
interface may be found in the StressCheck COM installation directory (usually
C:\program files\ESRD\StressCheck\com). 2
FIGURE 26 COM inteface documentation.

The COM interface

3 Model Input
This chapter contains a general overview of the features of StressCheck used for creating a finite ele-
ment model. The features for model construction are conveniently grouped together in the dialog win-
dow, known as the Input dialog window. There are thirteen tabs in the window, one for each Input class:
Geometry, Mesh, Section Properties, Thickness, Material, Load, Constraint, Solution ID, p-Discretiza-
tion, h-Discretization, Tables, Sets and Parts. The following sections provide a general description of
each input class.
Getting started
The best way to become familiar with the techniques for model creation in Stress-
Check, is to follow the examples described in this manual. To get started, just select
the StressCheck icon from the Start > Programs option in the Windows taskbar.
When the program main window appears, select Open a new database and click on
the OK button. When the file selection dialog box appears (FIGURE 27), move the
cursor to the File name field and type the name of the StressCheck database
(sample_1, for example) then press the Return key or click on the Save button. A
set of files will be created in the current directory using the name sample_1 and
extensions assigned by the StressCheck program, as described in the StressCheck
Interface chapter. You can always exit from the program any time you wish by
User’s Guide Chapter 3: Model Input 61

Length Units in StressCheck
going to the File > Exit menu option. Don’t be afraid to browse through the
menus and dialog boxes. There are no hierarchic menus to get lost in.
FIGURE 27 New File interface.

With the adoption of Parasolid as the geometric modeling kernel, it is neces-
sary to consider the choice of units of measurement before creating any geom-
etry in StressCheck. Since Parasolid limits the definition of any geometric
object to fit within a cube that is 1000 by 1000 units, it is necessary to intro-
duce a system of units so that the objects represented in Parasolid will be prop-
erly scaled to fit within this cube. You may think of this cube as being
measured in meters, and that the largest object that can be created in Parasolid
as being one that is 1000 by 1000 meters.
Length units are now chosen as part of a standard system of units. The choices
are Other, inch/lbf/second, and millimeter/Newton/second. The "Other" option
indicates that the units are unknown. In all cases, it is the user's responsibility
to insure that units are given consistently throughout the program.
When StressCheck starts up, the default units are inch/lbf/second. The default
units may be chosen as part of the preference settings. When loading in an
existing StressCheck model file, the units will be assumed to be “Other” unless
the .sci file contains a specific record assigning specific units.
62 Chapter 3: Model Input User’s Guide

Geometry Class
To manually select your system of units, use the Edit > Units menu option or use
the Unit selector from the Reference/Theory/Units toolbar. You may change the
system of units only if no geometry has yet been defined. If you have an existing
model that is not Parasolid based (no CAD geometry was imported), you may insert
a record into the .sci file that will assign the units. The units selection will not affect
the values used to define the model. The record appears as shown below (see
"*13"). The options are: IN/LBF/SEC/F, MM/N/SEC/C and OTHER.
!STRESSCHECK Version: V7.0.1 - Professional Edition

!---------------------< Batch File: CIRCLES >----------------------
! Date: 04/11/29 Time: 14:12:40 3
!-------------------------------------------------------------------------------
!
*5
*13
INCH_LBF_SEC_F
*10000
No comment
*6
0
*7
Dummy
Geometry Class
From the Class menu, select Geometry (FIGURE 28). The options under Geome-
try provide for the definition of the solution domain using points, lines, circles,
ellipses, rectangles, etc. The program lets you separate the definition of boundaries
from the definition of the finite element mesh. Separation of geometric objects
from mesh objects provides for a great deal of flexibility in modeling. Geometric
objects may be defined parametrically so that the domain may be easily changed by
adjusting individual design variables. Since boundary conditions may be attached

Mesh Class
to geometric boundaries, the finite element mesh may be easily changed with-
out affecting boundary condition definitions.
Details about geometry construction and revision are available in the Modeling
Guide.
FIGURE 28 Geometry Input.
Mesh Class
From the Class menu, select Mesh. The options under Mesh provide for the
specification of the solution domain using nodes, elements, fasteners, etc.
Details about mesh construction and revision are available in the Modeling
Guide.

Section Properties Class

The Section Properties input class is used for assigning cross sectional properties to
beam elements. To access the Section Properties class, select the corresponding tab
in the Input dialog window. The section property interface contains three tabs: User,
Library and Database under the Planar elasticity reference, as shown in FIGURE
29.
FIGURE 29 Section Properties interface
User tab The option User means that the section properties will be provided by the user. The
required inputs are the cross sectional area (A), the moment of inertia about the
centroidal axis (I) and the third (M3) and fourth (M4) moments of the area which
are defined as follows:

3
dA
A =
∫ dA M3 =
∫ η dA
Area Area
η
O
2 4
I =
∫ η dA M4 =
∫ η dA
Area Area
M3 and M4 are needed for curved beams only and can be set to zero for
straight beams.
Library tab The option Library means that the section properties will be determined from a
set of predefined parametric sections available in StressCheck (FIGURE 30).
FIGURE 30 Section Properties: Library option.

Database tab The option Database means that the section properties will be obtained from the
AISC standard shape database. You can choose between US and SI units if you
FIGURE 31 Section Properties: Database option.
have selected the standard system of units: Other, as described above. If you
selected in/lbf/sec/F, then the US option will be the only one available. If you
selected mm/N/sec/C then the SI option will be the only one available. When US
units are used, all the section properties will be in inches to the appropriate power.
When SI units are used, the section properties will be given in mm to the appropri-
ate power.
After the Action > Object > Method is selected, you click on the Database tab, and
the AISC Section Database box will appear (FIGURE 31). As you click on any of
the entries in the scrolling window of the dialog box, the corresponding area and
moment of inertia will be displayed in the Section Properties region. Once you
select the section of interest click on the Accept button. Note that in the Section
Properties input box, the 3rd and 4th moments also appear. These two moments are
calculated by StressCheck automatically.

Thickness Class
Note that in the Section Database box you can select the moment of inertia so
that it is consistent with the orientation of the beam. The default is that the neu-
tral axis coincides with the local x-axis of the section.
Thickness Class
In the Input dialog window, select the Thickness tab (FIGURE 32). The thick-
ness can be given as: constant, parameter, formula, or table/fitting.
3
FIGURE 32 Thickness interface.
Constant Most frequently you want to assign a constant thickness to all elements. To do
so, select Select > All Elements > Selection and enter the value of the constant.
in the field next to Thickness. Click on the Accept button and StressCheck will

Thickness Class
confirm your entry in the scrolling list. If you wish to assign different thicknesses to
different elements, select Select > Any Element and click on the elements to which
the thickness is to be assigned. The elements will be highlighted. Alternatively,
make a marquee pick to select the desired group of elements. If you wish to remove
one or more elements from the selected group, hold the Ctrl key down while select-
ing the element to be cancelled. Enter the desired thickness and click on the Accept
button.
Parametric You can assign thickness using parameters (see the StressCheck Interface chapter
about how to define parameters). When you enter the parameter name in the Thick-
ness field, you are indicating that a constant thickness will be assigned to an ele-
ment or group of elements the value of which is computed from the parametric
3
expression entered in the corresponding input field. Any expression using already
defined parameters and constants can be used. For example, assume that two
parameters TH and A have been defined. The parametric thickness can be defined
as 2*A+TH which means that the selected element will have a constant thickness
equal to the evaluation of the current values of A and TH. If A and TH are modi-
fied, the thickness will be automatically updated.
Formula Once again, the assignment may be element by element or through a group of ele-
ments. Complete the Formula entry form as described in the StressCheck Interface
chapter. In the Thickness input field enter the Formula name and, if the formula is
defined in other than the global coordinate system, enter the System name. Click on
the Accept button and your entry will be confirmed in the scrolling list.
The Object: System is only applicable when Formula is used. It has been added to
the list of possible objects to facilitate the search of a system name. With System
selected, if you click on a system its name will appear in the Status Line. You can
then check that system name from the System field.
Table/fitting This feature allows you to enter measured thickness values at discrete points which
are then fitted by polynomials. To input tabular thickness data, select the Tables tab
from the Input dialog window as shown in FIGURE 33. Provide a name for the tab-
ular data then select the Type: Thickness. Three Object options are relevant for tab-
ular thickness data entry: Any Curve, Any Surface and Location. The Any Curve
(Any Surface) option is provided to let you quickly enter constant thickness values
along a boundary at any number of evenly spaced points.
Under the Any Curve (Any Surface) option, select a boundary or boundaries from
the model in the display area, assign a value for the thickness along those bound-
aries and the number of points to be used along each boundary (use the first field of

Thickness Class
FIGURE 33 Tables and Thickness Fitting interfaces.
Value for thickness data as shown in the figure) and click on the Accept button.
StressCheck will automatically compute the coordinates of the boundary
points, equally spaced along the boundary, and enter the information into the
database. To cancel a boundary selection, hold the Ctrl key down while select-
ing the boundary. Enter the desired thickness and click on the Accept button.
Use the DeLast button to delete the last set of tabular data records created.
Under the Location option, input the coordinates of the point and the value of
the thickness at that point and click on the Accept button. Repeat this operation
for all points with known thickness values. The values being entered are dis-
played in the scroll window of the Input area and on the finite element mesh.
To edit any entry created by boundary or location, simply click on the corre-
sponding entry of the scrolling list. The values of that entry will be displayed in
the Input area. To modify the data, reenter the new values and click on the
Replace button. To delete the data, click on the Delete button. If the Accept

Thickness Class
button is selected instead, the modified data will be added to the existing entries,
and the original entry is retained.
To assign thickness tabular data to the elements, switch back to Class: Thickness,
provide the name of the table created in the input field, and select the element or
elements to which the tabular data applies. Click on the Accept button and your
entry will be confirmed in the scroll window. The tabular thickness data cannot be
used directly for the solution of elasticity problems. A polynomial fitting is
required when tabular data thickness is assigned to the elements.
This fitting option allows you to compute a continuous polynomial function which
fits the input data over the entire mesh. The polynomial degree is user-defined and 3
the fitting finds the coefficients of the polynomial which minimize the difference
between the input data and the fitted values in the least square sense. A penalty
term can be applied to minimize the jumps in the first derivative of the fitted func-
tion across inter-element boundaries. The selection of the proper polynomial degree
(p-level) and penalty factor is problem-dependent. You should try different values
of p-level and penalty factor to find the best representation for your input data. In
general, when fitting with low p-levels (typically less than 4), the effect of sudden
transition in the input thickness spreads over larger areas than when the fitting is
performed using high p-levels. On the other hand, high p-levels induce oscillatory
behavior of the fitted function, which can be controlled by the use of a penalty fac-
tor.
The input thickness can be given as different constants over groups of elements,
tabular thickness data, formula thickness, parametric, or any combination of con-
stant, table, formula and parametric. The fitting option can be used only after the
input information has been completed for all elements in the mesh. To compute a
polynomial fitting, select the Fitting sub tab (FIGURE 33), provide an ID for the
fiiting, assign a p-level (Fit Order) and penalty factor and then click on theAccept
button and then on the Fitting button. The amount of time required to compute the
fitting depends on the size of the mesh and on the p-level. The status bar in the main
window area displays the progress of the fitting. Once the fitting is completed, turn
on the toggle switch in front of ID to have a visual feedback of the computed poly-
nomial fitting. Use the toggle Input/Fitting to display the tabular or fitted thickness.
Checking thickness Use Action: Check to check for consistency in the thickness input data for all ele-
ments in the mesh. It checks whether all elements have a thickness assignment and
whether there are multiple thickness values assigned for the same elements. The
result of the check is displayed in an “Edit Window”. The elements without assign-

Material Class
ment or with more than one assignment are identified by the element number
in the edit window.
Displaying thickness The pulldown menu under the Display Thickness button of the Attributes Tool-
bar can be used for visualizing the thickness in different ways. The Display
Values option applies to the input data only. It displays the formula/table/
parameter name or the value of the constant assigned thickness. The Outline
option displays an outline of the thickness in the Z-direction. By selecting the
Grid option, the thickness can be displayed at grid points inside the elements,
as well as at the element edges. To see the displayed thickness, select the rota-
Display Thickness tion icon from the Views Toolbar, click on the main window (any location) and
3 Button drag the mouse while holding down the right button, until the desired view is
obtained. The scale of the displayed thickness can be adjusted with the up and
down arrow buttons.
Note: If no thickness is assigned to one or more elements, a default value of

unity is assumed if the Display Thickness button is turned on. If a Formula was
assigned, and the evaluation of the formula gives zero thickness at any location
inside the mesh, a warning message is displayed.
After the thickness information is entered in the database, a summary of the

thickness assignment record is inserted in the scrolling list.
Add a new record To add a new thickness record if an existing record has already been selected
from the scrolling list, you first must choose Set: New Set or re-select the
action, object, or method. Then, you are able to create the new thickness
record as usual.
Material Class
In the Input dialog window select the Material tab (FIGURE 34). The material
properties module is handled in two parts: One is the definition of the material
properties and the other is the assignment of material properties to the ele-
ments. The Define tab is used for providing the material coefficients (FIGURE
34a) while the Assign tab is used for assigning the defined properties to the
elements in the mesh (FIGURE 34b).

Material Class
(a) Define mode. (b) Assign mode.
FIGURE 34 Material properties interface.
Define Properties First decide whether you wish to enter your own material property, or select mate-
rial properties from a standard list provided by StressCheck. To use material prop-
erties from the standard list, click on the Browser button located at the side of the
material ID and when a list of material designations appears in the window of the
Material Library form [FIGURE 35], click on the material name and the property
data will be entered automatically in the Material input window. Finally, click on
Accept.
In the material library interface or in the material input window, specify whether
the material is Linear Elastic, Elastic-Plastic or Nonlinear-Elastic material and
whether you wish to use standard U.S. units (lbf, in, sec, degrees Fahrenheit), S.I.
units (kg, m, sec, degrees Kelvin, force: MN, stress: MPa), or some other system of

Material Class
FIGURE 35 Material Library Form.
Note: The material properties on the Standard list of StressCheck are

average values taken from various reference books. Actual material
properties may vary substantially from these generally accepted aver-
age values. It is the users responsibility to ensure that the material
data are properly defined for the specific application.
units. If you choose your own system of units then you must enter your own
material properties. If you have selected either standard U.S. units or S.I. units,
and you wish to use a material from the material library, then click on the mate-
rial name and the property data will be entered automatically in the input form.
Specify the material type. For linear materials the choices are Isotropic, Trans-
versely Isotropic, Orthotropic, Anisotropic, Laminated Isotropic, or Laminated
Orthotropic. After you select the material type, the scroll window automati-
cally adjusts to provide the appropriate number of fields into which the mate-
rial coefficients are to be entered.
Specify whether the analysis is to be Plane Stress or Plane Strain (only applica-
ble for planar elasticity), and provide the material coefficients if not using a
material from the library. All values of the material coefficients can be constant
or parametric. Entering a parameter name or expression in the input field is all
it takes to make a coefficient parametric.

Material Class
Click on the Accept button to associate the material name (entered into the material
ID field) with the specified material properties.
The material data browser provides two tabs, one for materials defined in the “Stan-
dard” StressCheck material database file, the other for materials that may be
defined in a “Company” material database. By default, the Standard material data is
contained in a file called “material.dat” found in the /ui installation directory. By
defining a registry key named ESRD7_MATERIAL_FILE, the StressCheck
installation may be customized by substituting an alternative material file (using the
same format conventions as the StressCheck material.dat file) containing company
specific material definitions.
3
By default, the material definitions found under the “Company” tab are also
obtained from the standard StressCheck material file, but they could be obtained
from any company material database. This can be accomplished by re-program-
ming a function contained within the StressCheck OCX. This OCX is included with
the StressCheck installation and contains various functions, written in Visual Basic,
for customizing StressCheck. The material data is obtained by calling the OCX
function called ReadMaterial(). This function reads a material file and sends the
material definitions to StressCheck through a standard StressCheck COM function.
The OCX will read the material data from a file identified in a registry key called
ESRD7_OCX _MATERIAL_FILE.
One easy way to have access to both the Standard StressCheck material file and a
company material file (same format as material.dat), is to define the registry key
ESRD7_OCX_MATERIAL_FILE to point to the company material file. The
OCX is already programmed to read a material file that is based on the standard
StressCheck material file format.
Assign Properties To assign a material name to an element or group of elements which are to have the
specific properties, select the Assign tab, and the form will change to the assign-
ment form shown in FIGURE 34b. The form is self-explanatory: You may select all
elements, elements one-by-one, or a group of elements. If the assignment is ele-
ment by element then simply click on the elements to which the property is to be
assigned and choose Accept. If the assignment is to a group of elements then drag
the cursor to enclose the desired group of elements and choose Accept. The
selected elements will be highlighted. If you wish to remove one or more elements
from the selected group, hold the Ctrl key down and click on the elements to be
removed from the group.
The multi-layer toggle can be turned on if you want the program to select the ele-
ments that share the face of the opposite face selected in a given element. This cas-

Material Class
cade effect allows selection of “layers” of elements for the assignment of

laminate properties.
If the material properties are orthotropic or transversely isotropic and the ori-
entation of the material axes is not coincident with the global coordinate sys-
tem, you can enter the name of the local coordinate system in the System field.
To create a local coordinate system select Geometry > Create > System >
Locate. Enter the location and orientation of the system. The material 1-axis
will be aligned with the local x-axis of the system if the system is Cartesian.
For cylindrical systems the material 1-axis will be aligned with the circumfer-
ential direction and the material 2-axis with the radial direction. To display the
3
name of a coordinate system select Display > Objects from the Main Menu Bar
and turn the corresponding label on. To complete the assignment process, click
on the Accept button. To check whether a material property was assigned to
each element, select Action: Check, Object: All Elements, and then click on
the Accept button. A confirmation report will appear on the screen.
Material property fitting Property fitting refers to the possibility of describing the orientation of the
material 1-axis of an orthotropic or transversely isotropic material as smooth
polynomial functions. These polynomial functions are computed using a least
square fitting on data points provided by the user. This capability is described
in detail in the Advanced Guide, Material Property Fitting.
Input Checking When you define material properties, StressCheck automatically computes the
smallest and largest eigenvalues of the material stiffness matrix ( λ min and
λ max , respectively). When the condition number of the material stiffness
matrix, that is, the ratio λ max / λ min , is greater than 20 then a warning message
appears in the execution window. The reason for this is that the energy error
and the root-mean-square error in stresses are close only when λ max / λ min is
not large. For details see Section 5.3.1 in Ref. [3]. You may proceed with the
analysis but exercise care when computing stresses from the finite element
solution. For further information and an example see Section 11.4.2 of Ref. [3].
Add a new record To add a new material record if an existing record has already been selected
action, object, or method. Then, you are able to create the new material record
as usual.

Load/Flux Class
Load/Flux Class
In the Input dialog window, select the Load tab (depending on the Theory selector,
this tab will read Load for Elasticity, or Flux for Heat Transfer) and the Input area
will appear as shown in FIGURE 36. You have to specify a unique name which
identifies the loading case you are about to enter. In engineering practice often mul-
tiple load cases must be investigated. Each load case must be given an unique name
in the ID field. Several Methods of loading such as traction, spring, body forces or
FIGURE 36 Load interface.
point loads are available in StressCheck. The types of loading available for elastic-
ity problems are briefly discussed in the following. For additional details and for
flux boundary conditions refer to the Analysis Guide. Note also that the content of
the load/flux tab depends on the reference (planar, 3D, axisymmetric, plate).

Load/Flux Class
Select objects You may load an object (such as an edge, boundary, face, surface or element)
or a group of objects. An object is selected by pointing to it and clicking on the
left mouse button. A group can be selected by dragging the cursor until the
desired group is completely enclosed in a rectangular box. If the group
includes objects which were not to be selected then individual objects can be
removed from the group by holding the Ctrl key down while clicking on them.
Traction loading Traction loading means that a distributed load (in force per unit area) is being
applied. Traction is a vector quantity. Thus, in planar elasticity two vector com-
ponents must be given. These may be in the normal-tangent reference frame, in
the global system, or an arbitrary local system. Traction loads can be applied to
3 geometric boundaries of element edges in 2D or to surfaces or faces in 3D.
Spring loading Spring loading means that load is applied through a displacement imposed on a
distributed spring which has been specified on a boundary. Specifically, the
traction vector components in the normal-tangent reference frame in 2D, for
example, are dependent on the imposed spring displacements:
Tn a nn a nt δ n – u n
=
Tt a nt a tt δ t – u t
where Tn, Tt are traction vector components in the normal and tangent direc-
tions, respectively, ann, ant, att are spring coefficients (in units of force per
length cubed), δ n , δ t are imposed displacement components in the normal-
tangent directions. These are the loading terms. The symbols un, ut represent
the normal-tangent displacement vector components.
Spring loading can be applied in global or local xy-coordinates as well. The

only restriction is that the spring coefficients and the imposed spring displace-
ments must be in the same reference frame. For further information on spring
loading you can refer to pp. 75-76 of Ref. [3].
Body force loading Body force loading means that the body is loaded by some distributed load,
such as by gravity or inertia due to acceleration. The body force is in terms of
force per unit volume. If the acceleration vector a = {a , a , a } and the mass
x y z
density ρ are known, then enter the products:
X constant = ax ρ, Y constant = ay ρ, Z constant = az ρ
Body forces can be applied to elements only.

Load/Flux Class
Thermal loading Thermal loading means that a temperature distribution (in degrees) will be speci-
fied on elements. Temperature distribution obtained from a heat transfer analysis
can also be specified. Thermal loading can be applied to elements only.
Point loads Point load means that the elastic body is loaded by a concentrated force
F = {F x, F y, F }. Point loads can be applied to points, nodes or fasteners. The
z
points can be located within an element. In 2D you may specify Fx, Fy directly or
the absolute value of F and the direction of F in terms of the angle measured
counterclockwise from the positive x-axis.
Bearing load Normal sinusoidal traction applied to a circular hole by specifying the magnitude 3
and angular direction of the force.
Moment Nodal moment can be applied to beam elements only. Edge moments can be
applied to plate elements.
Shear Edge shear can be applied to plate elements only.
Interference loading Magnitude of the diametral interference that can be specified to a fastener element.
Positive interference means that the diameter of the fastener is larger than that of
the hole. If the interference is negative (gap), the diameter of the fastener is smaller
than that of the hole.
Direction of load You may select the Norm./Tan. reference frame or an XY reference frame, the
default of which is the global system. The positive normal is understood to be the
outward normal. In 2D, the direction of the tangent is such that moving in that
direction the positive normal is to the right. Depending on your choice for direction,
either the Normal and Tangent or the X and Y components will be highlighted. You
may select either or both. For example, if you wish to apply a load only in the tan-
gential direction then turn on the button in front of Tangent and turn off the button
in front of Normal. In 3D, if the reference frame is the normal-tangent then only
normal traction can be applied.
Data type The loading data may be a constant, a parameter or a formula. For the case of ther-
mal loading it is also possible to specify the solution vector corresponding to a Heat
Transfer analysis. Once you run the heat transfer problem, the corresponding tem-
perature distribution can be specified as thermal loading for the elasticity analysis.
The only thing you need to do is to enter the name of the solution and run number
from the heat transfer analysis.

Load/Flux Class
Entering constant data is straightforward. Simply click on the appropriate box

and key in the constant value.
Entering parametric data means that a constant load value will be computed
from the parametric expression entered in the input field. Any expression (up
to 15 characters long) using already defined parameters and constants can be
used.
Constant and parametric loading is constant within the domain of each ele-
ment.
If loading is specified by a formula then the formula name is entered into the
3 input area. If the formula is defined in a coordinate system other than the glo-
bal system, then the name of the coordinate system must be selected. A for-
mula load may be defined in terms of spatial variables X, Y, Z, or R,T,Z
depending on the coordinate system, and is evaluated at each element Gauss
point.
When a name is entered into one of the input fields, the program first checks
whether or not the name corresponds to any defined parameter. If it does, then
uses the value of the parameter for that input field. If the name is not a parame-
ter, then it checks if there is a formula with that name.
Check load StressCheck provides for checking the magnitude of the applied mechanical
loads. You may check the size of the mechanical load applied to any part of the
domain. Select Check from the Action combo box and Any Element or All
Elements from the Object combo box. Once the elements of interest are high-
lighted, click on Accept. The load check can be applied to any load case avail-
able in the scroll window of the Input box.
Add a new record To add a new load or flux record if an existing record has already been selected
action, object, or method. Then, you are able to create the new load or flux
record as usual.

Constraint/Temperature Class
In the Input dialog window, select the Constraint tab (depending on the Theory
selector, this tab will read Constraint for Elasticity, or Temperature for Heat Trans-
fer) and the Input Area will appear as shown in FIGURE 37.
FIGURE 37 Constraint interface.
You have to specify an unique name for the constraint data you are about to enter.
This is necessary because StressCheck allows more than one constraint case, and
each case must be identified by a unique name.
The types of constraints available for elasticity problems are briefly discussed in
the following. For additional details and for the available temperature boundary
conditions, refer to the Analysis Guide.

Select objects You may constrain an object (such as an edge, boundary, face or surface) or a
group of objects. An object is selected by pointing to it and clicking on the left
mouse button. A group can be selected by dragging the cursor until the desired
group is completely enclosed in a square. If the group includes objects which
were not to be selected then they can be removed from the group by holding
the Ctrl key down while clicking on the individual objects. The object selected
determines the type of constraint that can be applied. Methods of constraints,
such as General, Rigid Body, Spring Coefficient, etc. are available in Stress-
Check and described below. For additional explanation on object selection See
“Selection” on page 56.
3 General General means that you will prescribe one, two or three displacement compo-
nents to one or more boundaries. For example, if one of the boundaries of the
solution domain lies on a plane of symmetry, which in 2D elasticity appears as
a line of symmetry then you will prescribe the normal displacement to be zero.
The following commonly occurring cases have been implemented:
Symmetry means the normal displacement component is set to zero. Symme-

try constraints are applicable only to straight edges or faces.
Antisymmetry means the tangential displacement components are set to zero.

Antisymmetry constraints are applicable only to straight edges or faces.
Built-In means all displacement components are set to zero.
An advantage of using these cases is that they can be interpreted automatically

in 3D when the extrusion option is used.
Spring Coefficients When an elastic body is constrained by means of distributed springs, or loaded
by displacements imposed by distributed springs, then the spring coefficients
modify the stiffness matrix. For this reason the spring coefficients are entered
as constraints.
Rigid Body StressCheck provides a convenient way for specifying rigid body constraints.
In the case of planar elasticity the user selects two nodes. StressCheck will then
create a right-handed local coordinate system, the x-axis of which is directed
from the first node to the second. In this local coordinate system both displace-
ment components are constrained for the first node; only the displacement
component in the direction of the local y-axis is constrained for the second
node.

Single Node Single node constraint can be applied to individual nodes, one at the time. How-
ever, point constraints are admissible only for preventing rigid body displacement
and rotation. The body must be in equilibrium under the action of external forces.
Unlike the errors introduced by concentrated forces, errors introduced by point con-
straints are not localized, except when the point constraints are to prevent rigid
body displacement only, in which case the body must be in equilibrium under the
applied forces. In every other case the reactions at point constraints are dependent
on the mesh and the polynomial degree of elements. For this reason multipoint con-
straints should not be used.
To specify rigid body or nodal constraint, the object Node should be selected first.
3
Direction of constraint When appropriate, you may select the Norm./Tan. reference frame or an XYZ ref-
erence frame, the default of which is the global system. The normal is understood to
be the outward normal and the direction of the tangent is such that moving in that
direction the positive normal is to the right. Depending on your choice for direction,
either the Normal and Tangent or the X and Y components will be highlighted. You
may select either or both. For example, if you wish to apply a constraint only in the
tangential direction then turn on the button in front of Tangent and turn off the but-
ton in front of Normal.
Data type The data type may be fixed, constant, formula or parametric. Fixed means that the
selected displacement component is set to zero. If the displacement is constant
along a boundary then the constant value must be entered into the appropriate box.
The parametric option can be used also to specify a prescribed displacement. The
value of the displacement is computed from a parametric expression (up to 15 char-
acters long) entered in the corresponding input field. Any combination of previ-
ously defined parameters and constants can be used. Prescribed displacement
components may be specified by formulas. The formula name is entered into the
appropriate box. When a formula is defined in a coordinate system other than the
global system then the name of the coordinate system must be entered also.
Contact Zone To especify the spring constant for a contact problem. For additional details refer to
the Advanced Guide, Multi-body Contact.
Add a new record To add a new constraint or temperature record if an existing record has already been
selected from the scrolling list, you first must choose Set: New Set or re-select the
action, object, or method. Then, you are able to create the new constraint or tem-
perature record as usual.

Solution ID Class
Solution ID Class
In the Input dialog window, select the Solution ID tab and the Input area shown
in FIGURE 38 will appear. Because StressCheck allows more than one load
case and constraint case to be defined, it is necessary to associate a unique
solution name with each desired constraint and load name pair.
FIGURE 38 Solution ID interface.
Enter a solution name in the Solution ID field and select a Constraint ID and
Load ID from the existing records. Click on the Accept button and the solution,
constraint, and load names will appear in the scrolling list area at the top of the
dialog box.
You may disable a solution record by simply turning the check box off next to
the Solution ID. Select the desired record, enable or disable the check box, then
click Replace.

p-Discretization Class
Use of the Solution Configurations tab will be described later in the Parts Class sec-
tion.
In the Input dialog window, select the p-Discretization tab and the Input area will
appear as shown in FIGURE 39. If the object Element (or All Elements) is selected,
then only the p-Discretization is available for input. If the object face is selected,
then the q-discretization field is also available for input. 3
FIGURE 39 p-Discretization interface.
When a solution is initiated, the initial and final p-levels given in the SOLVE inter-
face are assigned to ALL elements as a default setting. Explicit p-discretization

settings may be assigned to individual elements that will override the default
settings. Individual elements may be assigned p-discretization attributes that
are Variable, Bounded, or Fixed. Any p-discretization attributes that are
assigned to ALL elements, will be automatically overridden by the Solve inter-
face.
In StressCheck, typically uniform p-extensions are used. Extension means a

systematic increase in the number of degrees of freedom. It is possible, how-
ever, to select any group of elements and limit or fix the polynomial degree for
those elements. If uniform p-extension, using the trunk space, is to be
employed then nothing needs to be done because this is the default option for
3 StressCheck. The different options are described in the following.
Space The options are: Trunk and product space. The trunk space is the default space
and is applicable to all element types. The product space is applicable for
quadrilateral and hexahedral elements only. The use of product space in 3D is
limited to models that have only hexahedral elements. In that case all the ele-
ments must be assigned the product space. Refer to the Theoretical Back-
ground chapter of the Advanced Topics Guide for an explanation on the
difference between the trunk and product space.
p-Discretization The options are: Variable, Bounded, Fixed, and Uniform. Variable means that
the p-level is allowed to change from the starting level in the corresponding
input field. Bounded means that the p-level for any element or group of ele-
ments is bounded between a lower and upper limit. Fixed means that the p-
level of the selected group of elements will remain fixed at the input value dur-
ing the p-extension. Uniform means that during an adaptive p-extension, the p-
level of all the selected elements will be set to the same value based on the
error indicator reading for the elements.
q-Discretization The only option is fixed. Allows to fix the polynomial order (q-level) in one
direction for hexahedral and pentahedral elements. Used in connection with
‘thin-solids’, the ‘thin’ direction is determined based on the selected element
faces. By selecting the face of an element, the thin direction is understood to be
in the direction normal to the selected face. Any fixed value of q between 1 and
8 can be assigned.
Add a new record To add a new p-discretization record if an existing record has already been
selected from the scrolling list, you first must choose Set: New Set or re-select
the action, object, or method. Then, you are able to create the new p-discreti-
zation record as usual.

h-Discretization Class
In the Input dialog window, select the h-Discretization tab and the Input area will
appear as shown in FIGURE 40.
FIGURE 40 h-Discretization interface.
StressCheck allows manual and automatic h-refinement. The refinement is per-

formed by subdividing each element using a user-specified parameter, the number
of “midsides”. For example, if the number of midsides selected is 1 then each quad-
rilateral element will be subdivided into 4 elements by locating a node in the mid-
point of each side and connecting the nodes on opposite sides. StressCheck
currently supports uniform h-refinement and simple graded refinement.

Uniform refinement To perform a uniform mesh refinement select Refine > Any Element > Uni-
form Div. > Midsides:1 and click on the element you want to refine. Alterna-
tively, chose Select > Any Element > Uniform Division > Midsides:1, and
click on all elements to which the refinement should be performed and then
click on the Accept button. If instead of Any Element you select All Elements,
the refine operation will be performed on all elements of the mesh.
FIGURE 41 illustrates the effect of a uniform refinement operation applied to a
straight sided hexahedral element.
FIGURE 41 Sample uniform element refinement.
Graded refinement To perform a simple graded refinement with one layer of elements select
Refine > Node > Simple Graded > Midsides: 1 > Grading: 0.15 and click on
the node to refine. You can select a group of nodes first using the Select/Accept
button combination as explained above.
Graded refinement may be applied to a node, edge, or face (3D only), and may
be performed using either Action: Refine or Action: Select as described above.
FIGURE 42, FIGURE 43, and FIGURE 44 illustrate the affect of applying a
grading factor of 0.15 to a node, an edge, or a face of a hexahedral element.

Tables Class
Note: When performing edge or face refinement, you may choose whether to apply
the refinement only to the parent element of the selected edge/face, or to all ele-
ments which share the selected edge/face.
FIGURE 42 Sample graded node refinement.
When the parent element has either curved edges or faces, the resulting elements
will inherit the curvature only on the edges or faces of the original parent element.
New edges and faces which are internal to the original parent element will always
be straight.
Mesh refinement is not supported for tetrahedral elements. Use the DeLast or Undo
button to delete the last set of refined elements created.
Tables Class
In the Input dialog window, select the Tables tab and the Input area will appear as
shown in FIGURE 33.

Tables Class
FIGURE 43 Sample graded edge refinement.
FIGURE 44 Sample graded face refinement.
An explanation of the use of tables is given in connection with the specifica-

tion of thickness by tabular data (see page 69). Other uses of tabular data are
considered advanced features. A description is available in the Advanced Top-
ics Guide.

Sets Class
Sets Class
In the Input dialog window, select the Sets tab and the Input area will appear as
shown in FIGURE 45(a).
(a) (b)
FIGURE 45 Sets interface.
Sets may be defined either by explicitly selecting individual objects (Method Selec-
tion), or by defining a box region that will be used to identify the objects for selec-
tion (Method Locate). Any object which lies within the box region, or which
intersects with the box surfaces, will be selected.
In most cases, explicit selection of objects is the best way to define sets, but there
are situations when regional sets are necessary. For example, applying thickness,
material properties and element-based boundary conditions to sub-regions of an
automatically meshed parametric model will require the use of regions sets. This is

Parts Class
because the model will be remeshed in response to a change in a parametric

dimension, and the attributes assigned to elements in the original model will no
longer be valid. By defining regions sets, attributes will automatically be re-
assigned to whatever elements (or objects) lie within the defined box regions.
To define a region set, choose the “Locate” method (FIGURE 45.b). In the cur-
rent release, only rectangular regions that are aligned with the global coordi-
nate system may be defined. Supply the X,Y,Z coordinates of the center of the
base of the box, and the width, height and depth of the box. The base of the box
is located in the XY plane at Z=0. As you tab through the input fields, you will
see the box appearing on the screen. When you have supplied the desired loca-
3 tion and dimensions, click Accept to complete the set definition.
If you enable the check box next to the set ID, all existing region sets will be
displayed in the model window. When you select a region set record from the
list box, the outline will be highlighted in the model window, and the input val-
ues will feed back to the user interface. You may modify these values and click
Replace to revise the definition of the region set.
All model input interfaces that create sets automatically now have a combo
box which permits the use of pre-defined sets. If you wish to use a pre-defined
set rather than to create a new set by explicit selection, choose the pre-defined
set name from the combo box before clicking Accept. To create sets by explicit
selection as usual, the combo box should be set to “New set”.
Parts Class
StressCheck provides a facility for organizing the components of a particular
model into parts and assemblies of parts. Parts and assemblies are just a gener-
alization of the more familiar set definition in StressCheck. While the members
of a StressCheck “Set” must be of the same type, for example a set of nodes or
a set of boundaries, a “Part” is a set of geometric or mesh objects of any type.
An “Assembly” is just a set of parts. This facility is particularly handy when
working with complex models where it is helpful to be able to activate and
deactivate subsets of the model during construction or during post-processing.
Parts and assemblies are also essential to the definition of solution configura-
tions, i.e. the definition of multiple model configurations for the purpose of
investigating model variations that are topologically different.

Parts Class
Parts and assemblies may be defined in the Parts input class. FIGURE 46 illustrates
the user interface for the definition of parts and assemblies.
FIGURE 46 Part input class interface.
The first step in creating a new part is to make sure the objects of interest are in
view in the model display window. Next, select the desired objects using the Select
action, combined with an appropriate Object selection.
When you are finished selecting objects, you may enter a name to identify this col-
lection of objects as a Part, and supply a description to further identify the part.
Click the Accept button when you are ready to save the new part definition. You
will see the name appear in the scrolling list in the Parts interface. If you wish to
review the definition of any part, simply select the part in the scrolling list. The
objects it contains will be highlighted on the screen, and will be listed in the scroll-
ing list below the list of parts. You are then free to cancel individual objects that
define the part, or to select additional objects that you wish to add to the part. Click
the Replace button to change the definition of the selected part, or click Accept to
create a new part (after supplying a new part name and description).

Parts Class
To create an assembly, switch to the Assemblies tab to obtain the interface

shown in FIGURE 47. The scrolling list at the top of the interface lists the cur-
rently defined assemblies. The scrolling list in the middle of the interface lists
FIGURE 47 Assembly construction interface.
the parts that have been included in the current assembly. The scrolling list at
the bottom of the interface lists all parts that are available to be included in an
assembly, but are not yet included in the current assembly. To define a new
assembly, enter a name and description in the input fields provided, and select
the desired parts from the bottom scrolling list. These selected parts will be
moved to the middle scrolling list. If you change your mind about a particular
part, simply click on the part in the current list, and it will be moved back to the
previous list. When you have finished selecting parts for the assembly, click on
the Accept button. To change the definition of an existing assembly, select the
assembly from the top scrolling list, edit the definition by transferring parts
between the middle and bottom lists. When finished, click Replace to change

Parts Class
the current assembly definition, or change the assembly name and click on the
Accept button to create a new assembly.
Once parts and assemblies have been defined, they may be chosen for selective
viewing using the combo boxes in the Part/Assembly Toolbar that may be activated
from the View menu in the main menu bar. The assembly combo will list all avail-
able assemblies, plus two additional options: All Assemblies and No Assemblies. 3
Depending on the setting of the assembly combo, the Part combo may list all avail-
able parts belonging to a particular assembly, or a list of all available parts in the
model, and the option All Objects. There are many possible combinations of set-
tings in these two combo boxes. By selecting specific combinations, you can exer-
cise a great amount of control over the contents of the display. Use the following
rules to control the contents of your model display window using the Assembly and
Part combo boxes.
Assembly Combo Part Combo Result

ALL OBJECTS Show all objects in the entire model.
ALL PARTS Show only objects that belong to parts
which are identified as a component of any
assembly.
ALL ASSEMBLIES
Specific Part Show only objects that belong to parts
which are identified as a component of any
assembly. The part chosen in the combo
will be highlighted (selected).
ALL OBJECTS Show only objects that do not belong to
any assembly. They may be in a part that is
not in any assembly.
ALL PARTS Show only objects that belong to parts that
NO ASSEMBLIES
are not part of an assembly.
Specific Part Show only objects that belong to the cho-
sen part.

Parts Class
Assembly Combo Part Combo Result

ALL OBJECTS Show only objects that belong to parts in
the assembly and objects that do not
belong to any other assembly. The objects
may be in a part which is not yet included
in any assembly.
Specific Assembly ALL PARTS Show only objects that belong to parts in
the assembly.
Specific Part Show only objects that belong to parts in
the assembly. Objects belonging to the
3 chosen part will be highlighted (selected).
As mentioned earlier, Assemblies and Parts are useful for controlling the con-
tents of the model window during model construction and post-processing. The
other use of parts and assemblies is to define two or more variations of a model
that each represent a complete and valid model that may be solved and post-
processed by StressCheck.
For example, suppose you have constructed a StressCheck model of a simple

attachment lug. You may wish to analyze more than one variation of this basic
lug design making different modeling assumptions. The first variation might
be a simple lug with a bearing load, the second might the same lug, but with a
bushing inserted into the hole. This second variation will require additional
elements, and different boundary conditions. Both variations can be incorpo-
rated into the same model by using parts and assemblies, and by creating solu-
tion configurations which define which boundary conditions should be applied
to which model variation.
FIGURE 48 and FIGURE 49 illustrate two model configurations for a simple

2D attachment lug, one without a bushing and one with a bushing. In the sim-
ple lug model without the bushing, the load is applied to the outer ring of ele-
ments. In the complete lug model with the bushing, the load (imposed
displacement on a spring) must be applied to the inner ring of bushing ele-
ments. These two variations must be managed using solution configurations so
that the two boundary condition descriptions are not in conflict.
FIGURE 50 illustrates the assembly interface after creating the assembly for
the simple lug. Note that the inner ring of elements has been assigned the part
name “bushing”, while the rest of the elements in the model have been
assigned the part name “lug”. The assembly “Simple Lug” is composed only of

Parts Class
FIGURE 48 Simple lug with bearing load (no bushing).
FIGURE 49 Complete lug with spring boundary condition (bushing).
the part “lug”. The assembly “Complete Lug” is composed of both parts “lug” and
“bushing”.
The boundary conditions for each model variation may be defined as usual, using
the Part/Assembly toolbar combos to control which model variation is visible, if
desired. The load and constraint records for each model variation are assigned
unique ID's if necessary. In this case, the bearing load is assigned ID=LOAD, and
the spring boundary condition is assigned ID=LOAD2.
In order to solve these two model variations separately, it is necessary to create a

solution configuration. The solution configuration interface for this model is illus-
trated in FIGURE 51. To create a solution configuration for each model variation,
simply assign a solution configuration name, select the appropriate Assembly and
Part options, and activate the solution records displayed in the table that are appro-
priate for this solution configuration. In this example, the “simple lug” configura-
tion activates the table entry where solution = SOL, load=LOAD, and
constraint=CONST, and deactivates the table entry for SOL2. The "bushlug" con-
figuration activates SOL2, and deactivates SOL.

Parts Class
FIGURE 50 Assembly definition of simple lug model.
After defining the solution configurations, you are ready to solve your model
variations. You may choose whether to analyze all available solution configu-
rations, or to select a particular solution configuration for analysis. Since each
configuration uses a unique solution ID, you can save solutions for all model
variations in the same database if you choose All, and post-process them all
without having to solve them separately and post-process them separately. This
makes comparison of model variations more convenient.

Parts Class
FIGURE 51 Solution configurations for lug model.

Parts Class

4 StressCheck Solutions
4
This chapter contains a general overview of the Solution interface. It covers the following topics:
• Solution dialog window

• Linear analysis
• Nonlinear analysis
• Modal analysis
• Buckling analysis
• Margin Check analysis
• Crack Path analysis
• Measurement analysis
• Adaptivity
Solution dialog window

Once you have a properly defined model, including geometry, mesh, materials,
loads, constraints, and solution ID definitions, you are ready to perform a Stress-
Check solution. The type of solution that you choose from the Solution interface
depends on the goal of your analysis. There is no need to change elements in order
User’s Guide Chapter 4: StressCheck Solutions 101

Solution dialog window
to perform one solution type or another. Elements only represent the geometry
and topology of the model. Analysis of specific attributes such as material
properties and boundary conditions are easily provided so that the same topo-
logical model may be used to perform several different types of analysis. In
fact, all attributes can be attached to a single model so that it is possible to
define the model for any type of analysis, then simply choose the type of anal-
ysis to perform at any time.
In any case, StressCheck will construct a system of simultaneous equations and

solve this system of equations to produce a set of solution vectors, one for each
load case or mode shape for each element. The solution vectors are then used
during post-processing to compute output quantities, such as stresses, strains,
and displacements for problems of elasticity, or temperatures and gradients for
problems of heat transfer.
Since StressCheck post-processes “on the fly” from basic solution vector data,
it is not necessary to plan ahead what data you will be interested in, or what
locations in the model you will want to examine. This flexibility is one impor-
4
tant advantage StressCheck has over other FEA programs, the ability to com-
pute any engineering quantity, at any time, at any location in the model.
Now that you know that you do not have to worry about the data you will even-
tually want to examine after the solution, you can imagine that there is not
much to do to prepare for a solution in StressCheck. All that is required is to
choose the type of solution, select a few options, and activate the solution pro-
cedure. These options are selected in the Solution dialog window shown in
FIGURE 52.
The solution dialog window provides access to the various model solution pro-
cedures offered by StressCheck, including: Linear (Elasticity or Heat Trans-
fer), Nonlinear (Elasticity or Heat Transfer), Modal, Buckling, Margin Check,
Crack Path, and Measurement Analysis. There are solution options specific to
each type of analysis, and a few options which are required for any type of
solution. The solution specific options may be viewed by selecting the tab cor-
responding to a particular solution type. The options which are appropriate for
any solution type may be viewed by selecting the “SOLVE!” tab, as illustrated
in FIGURE 53.
Each of these options will be discussed briefly in the sections which follow. In
addition, the options associated with each type of solution will be described in
greater detail in the Analysis Guide.
102 Chapter 4: StressCheck Solutions User’s Guide

Linear Analysis
Solution Icon Solution Activation
Solution Type
User options
FIGURE 52 Linear Analysis.
Linear Analysis
The simplest type of solution to perform in StressCheck is a Linear Analysis. One
of the powerful characteristics of StressCheck is that the linear analysis is just the
first step in a hierarchy of analyses. The same model used to perform a linear anal-
ysis may be used as the starting point for a nonlinear analysis. It is also possible to
use the result of a steady state heat transfer analysis to represent the temperature
distribution for a thermal elasticity analysis. These analysis extensions may be per-
formed without changing the basic finite element model. All that is required is that
additional attributes be defined, such as thermal coefficients or nonlinear material
properties.
Upward-p The only solution type specific information required for a linear analysis are the
“extension” and the range of p-levels used to compute a sequence of solutions. By
“extension”, we mean the procedure by which the program repeats the analysis at
different p-levels, or with different degrees of element refinement, in order to estab-

Linear Analysis
lish a convergence sequence from which an error estimate can be computed.

By default, StressCheck uses an “Upward-p Extension” and assigns a constant
p-level to all elements, increasing the p-level of all elements uniformly for
each solution in the sequence. For example, if you specify upward extension,
p=1 to 4, the program will automatically assign p=1 to all elements that do not
have a specific p-discretization assigned in the Input class, for the first solution
in the sequence. During the second solution, the p-level for all elements will be
set to p=2, and so on.
Downward-p The “Downward-p Extension” performs the highest p-level first and decreases
the p-level until the lowest p-level of the specified range is reached. The most
significant difference between upward and downward extension is that during
a downward extension the elemental stiffness matrices are saved on disk and
reused during subsequent solutions. The matrices can be reused because the
lower order terms are embedded in the higher order matrices and do not need
to be computed again. This can save a significant amount of computation time.
However, there is a price to be paid for computational efficiency in terms of
4 disk space to save the high order matrices. The stiffness matrix for a single 2D
quadrilateral element in elasticity (p=8) requires 4465 terms or 32 thousand
bytes. The stiffness matrix for a single 3D hexahedral element in elasticity
requires 166,176 terms (p=8), or almost 1 megabyte.
The range of p-levels assigned to elements from the Linear analysis tab will
affect only those elements which are considered to have a “variable” p-discret-
ization attribute. This attribute may be assigned in the p-Discretization input
class. All elements are “variable” by default. The p-discretization attribute for
each element may be assigned to be variable, uniform, fixed, or bounded.
Uniform-h “Uniform-h Extension” performs a sequence of solutions using a fixed distri-

bution of p-levels, but refines each element uniformly at each step. The first
step solves the model with the original mesh. The next step refines the model
uniformly, for example, dividing each quadrilateral element into four smaller
elements. In order to repeat a solution using h-extension, it is necessary to
reload the model from a StressCheck input file, or to go to the h-discretization
input class and use the DeLast button to return to the previous mesh definition.
Adaptive-p Choosing “Adaptive-p” means that the polynomial order will be determined
for each element and for each run based on an error indicator until convergence
is realized. The convergence options are described below.

Solution Activation
Upward model For plate bending analysis, you will also be given the opportunity to perform an
analysis during which the hierarchic model is allowed to change, while the p-level
is held constant. For other types of extension, the model is held constant while the
p-level or h-discretization is changed.
Solution Activation
When you are ready to activate a solution, whether it be a linear analysis, nonlinear
analysis, or any other type of analysis, simply select the “SOLVE!” tab in the solu-
tion dialog window as shown in FIGURE 53. This tab provides access to solution
Status Window
Option Combos
Action Buttons
FIGURE 53 Solution activation.
options which are the same no matter which type of analysis you are performing.
The precise interpretation of each option sometimes depends on which type of anal-
ysis you are performing. For this reason, you should refer to the chapters which dis-
cuss the specific analysis options in order to obtain precise instructions for using
each option in the context of a particular analysis type.

Solution Activation
To perform a particular type of analysis, first select the corresponding tab,

select the desired options, then select the “SOLVE!” tab. You may notice that
the currently selected analysis type appears in the title bar. The following para-
graphs provide a general explanation of the options found in the “SOLVE!”
tab.
Execution The “Execute” option may be either “Initialize” or “Restart”. Choose “Initial-
ize” when you wish to solve the problem for the first time, or you wish to
remove previous solutions and re-solve the problem as though this were the
first time. The meaning of “Initialize” depends somewhat on what type of anal-
ysis you are performing. If it is a linear analysis, Initialize if you wish to dis-
card all previous solutions and to start over. If it is a nonlinear analysis,
Initialize if you simply wish to compute a new nonlinear solution starting from
an existing linear solution.
Choose “Restart” if you have already solved the problem at least once and
wish to proceed with the solution sequence or solution iteration where you left
4 off. The meaning of “Restart” depends somewhat on what type of analysis you
are performing. If it is a linear analysis, Restart if you wish to solve the next p-
level or the next h-refinement in the sequence. If it is a nonlinear analysis,
Restart if you wish to perform the next iteration of the nonlinear solution.
Run Mode The “Run Mode” option may be either “Automatic” or “Stepwise”. An auto-
matic solution will proceed without an opportunity for user intervention. For
example, if you are running a linear solution, p=1 to 4, the automatic solution
will perform 4 solutions one right after the other. A stepwise solution will stop
after each p-level is complete. After each step completes you are free to post-
process, or even exit from the program. When you are ready to perform the
next solution in the sequence, simply click the “Solve” action button.
Method The “Method” option may be either “Iterative” or “Direct”. Method: Iterative
means that the Iterative Solver (default) is used. Method: Direct means that the
Direct Solver is used.
Converge Two options are available: “None” means that the number of runs is controlled
by the settings in the linear tab. “Error=2%” means that the number of runs will
be controlled by the global error of the solution. When the estimated relative
error in energy norm is less than or equal to 2% (5% for 3D problems), the exe-
cution will stop. The highest polynomial order cannot go beyond the limits
entered in the linear tab. Additional convergence criteria can be defined in the
Settings tab of the Results interface as described in the Results Chapter.

Solution Activation
Status Window During the solution, the current status of the solution process is reported in the
“Status Window”. This will indicate which run is currently in progress. The type of
information shown depends on the type of analysis that is being performed.
Results During a solution you may also monitor the progress of the solution by displaying
the “Sequence”. This will remove each element from the display as the elemental
stiffness matrix is computed. You may also display “Results” such as a contour
plot, the energy convergence, or any other post processing information. If you
choose to monitor results, the standard Results dialog window will appear so that
you may select the desired options. For more information, refer to the Post-process-
ing Operations Chapter.
Browse Use the “Browse” button to identify a directory path where you wish to have inter-
mediate work files created. These files are disposable files, which can be quite
large, generated during the solution procedure. These files may be directed to a
scratch device where there is sufficient space on a temporary basis. If you prefer to
simply type the path manually, you may enter it directly in the text field provided. 4
Solve The “Solve” action button will activate the currently specified solution. If you wish
to interrupt the solution after it has started, simply press the “<Esc>ape” key. The
solution computation will continue until the program reaches a convenient place to
stop. You may later restart the solution computation with the run that was inter-
rupted.
Check The “Check” action button will produce a report summarizing Parameters, Element
Distortion, Geometric Properties, Material Properties, Applied Load, and the Disk
Space required to perform the currently specified solution.
Settings Tab The SOLVE! Settings tab provides an interface that will enable the user to record
the current solution interface control settings for later use. When saving a new set-
tings record, the user should supply a unique name in the combo box provided to
identify each record of settings. To retrieve a previously stored settings record, sim-
ply select the desired name from the combo box. The user interface will automati-
cally be updated to reflect the settings stored in the chosen record.
If the handbook icon is enabled, the settings record will be used during a handbook
solution when clicking the Solve button in the Handbook interface. The records are
stored in the database in alphabetical order, and are therefore also processed in
alphabetical order when solving. It is important to name these records in such a way

Nonlinear Analysis
that the solutions will be performed in the desired order. If the handbook icon
is disabled, the record will be skipped during a handbook AutoRun solution.
Use the Save button to save the current settings in a record to be identified by
the name provided. If the name provided matches an existing record, the exist-
ing record will be replaced. Use the Delete button to remove the record identi-
fied by the name in the combo box. Use the Purge button to remove all settings
records from the database.
File Tab The SOLVE! File tab provides a similar capability to that provided by the Set-
tings tab, except that the interface settings are saved in an external ASCII file
instead of in the database.
Nonlinear Analysis
4
Each nonlinear analysis starts from a previously computed linear solution. It is
always important to verify the quality of the linear solution before attempting a
nonlinear analysis. If the linear solution is of poor quality, there is no point in
computing a nonlinear solution which will be of poor quality also. Starting
FIGURE 54 Nonlinear Analysis.

Nonlinear Analysis
from a poor quality solution may also lead to a nonlinear solution with poor conver-
gence characteristics. In other words, you may waste a lot of time trying to get a
nonlinear solution to converge when the real problem is that the initial linear solu-
tion was not accurate enough.
The following paragraphs describe the solution options briefly. For more detailed
information regarding nonlinear analysis, consult the corresponding chapter in the
Advanced Topics Guide.
Linear Solution List Each nonlinear solution must start from an existing linear solution. Select the
desired starting solution in the list box. It is not necessary to use the solution with
the most degrees of freedom if there is another solution with fewer degrees of free-
dom which is of sufficient quality. As a rule of thumb, you should use a solution
with less that 10% error in the global energy error norm. Choosing a solution with
fewer degrees of freedom will require less computation during each iteration, but
may require more iterations to converge. Selecting a solution from the scrolling list
will automatically fill in the Linear Solution ID and the Linear run number. You
may fill these text fields in manually as well (FIGURE 54). 4
Nonlinear ID Enter a nonlinear ID of your choice. This ID will appear in the post-processing dia-
log windows later.
Analysis Type Choose whether you wish to perform a Material Nonlinear Analysis, General Non-
linear Analysis (material and geometric nonlinear), or Coldworking Analysis.
Convergence Choose whether you want the program to use a “Stress” or “Energy” criterion for
evaluating solution convergence.
Technique Specify the integration procedure. For Material nonlinear analysis, Direct Integra-
tion is the only option. For General nonlinear analysis, Newton-Raphson is also
available.
Tolerance Specify the allowable convergence tolerance. You may change the tolerance during
the solution by performing a “Stepwise” solution (see discussion of SOLVE!
options), or by waiting for the specified number of iterations to complete.
Iteration Limit Specify the number of iterations to perform. You may increase the iteration limit if
the convergence information shown in the status window of the SOLVE! tab indi-
cates that the solution has not achieved the desired accuracy.

Modal Analysis
Load steps This option is used to apply the load in increments from the minimum to the
maximum specified load in equal increments. When the load steps toggle is
turned on, you will be able to select the load parameter and enter the maximum
value for the load (FIGURE 54). The minimum load value is displayed but
cannot be changed. You also will have the option of selecting the Save Load
Steps toggle. This option will save all converged intermediate load steps for
further post-processing. Each saved step will be identified with the name given
in the nonlinear ID field, an underscore ‘_’ followed by the run number of the
starting linear solution. For example, NSOL_8 means that the solution NSOL
was initiated from the linear solution #8. The run number associated with each
nonlinear solution will correspond to each load step.
Modal Analysis
4 Modal analysis is used for finding the natural frequencies and associated mode
shapes. StressCheck also provides a capability for taking into account the
effects of initial stresses on the computed natural frequencies and modes. The
following paragraphs describe the basic options which may be used to control
a modal analysis (FIGURE 55). Please refer to the Analysis Guide for addi-
tional information.
Extension As in a linear analysis, you may choose the type of extension to perform during
the modal analysis, Upward-p, Downward-p, or Uniform-h. These have the
same interpretation as in a linear analysis, as do the p-limits.
Type The type of modal analysis may be “Frequencies and Modes” or “Frequencies
Only”. If you choose Frequencies and Modes, the analysis will produce both
the frequencies and the mode shapes, otherwise only the frequencies will be
computed. Computation of mode shapes requires more CPU time and disk
storage.
Frequency You may choose to compute frequencies and modes in a specified “Interval” or
“Range”. Interval means that the requested output will be computed for all fre-
quency values greater than or equal to the lower limit (in Hz) specified in the
first Frequency value field and less than or equal to the upper limit specified in
the second Frequency value field. Range means that the requested output will
be computed from the Frequency number i to the frequency number j, where i

Buckling Analysis
FIGURE 55 Modal Analysis.

4
and j are selected in the Frequency number fields. Typically the first few natural
frequencies are of interest.
Pre-stress StressCheck has the capability to take into account the effect of pre-stress on the
natural frequency of vibration. To exercise this option you must be performing a
three-dimensional analysis and the body must be properly loaded and constrained.
StressCheck will compute the stresses due to the applied loads and make the appro-
priate corrections to the stiffness matrix.
Buckling Analysis
Buckling analysis is used for computing the load factor corresponding to a given
loading condition, and the corresponding buckling shapes. Buckling analysis is
available for three-dimensional problems only. To perform a buckling analysis,
StressCheck first solves the linear problem corresponding to the specified loads and
constraints, then, utilizing the stress field computed from the linear solution, com-
putes the geometric stiffness matrix, which is used for the eigenvalue computation.
Refer to the Analysis Guide for additional information.

Buckling Analysis
FIGURE 56 Buckling Analysis.

4
The input required for the buckling analysis is similar to that required for the
modal analysis (FIGURE 56).
Extension As in a linear analysis, you may choose the type of extension to perform during
the buckling analysis, Upward-p, Downward-p, or Uniform-h. These have the
same interpretation as in a linear analysis, as do the p-limits. NOTE: In the
case of “thin” bodies, locking will occur at low p-levels, so let p range between
3 to at least 6. Uniform-h extension should not be used with “thin” domains.
Type The type of buckling analysis may be “Buckling and shapes” or “Buckling
load only”. Buckling load only means that the natural buckling load factors
will be computed and stored, but not the mode shapes. Buckling and shapes
means that the buckling load factors and the corresponding buckling mode
shapes will be computed and stored. Computation of mode shapes requires
more CPU time and disk storage.
Load numbers The requested output will be computed from Load factor number 1 to Load
factor number j, j is entered in the second Load number field. Typically only
the first buckling mode is of interest. Letting the first and second fields to be 1,
will save computation.

Margin Check Analysis
Margin Check Analysis

The Margin Check is a nonlinear analysis capability (material or general nonlinear)
with failure criteria control (FIGURE 57). To solve a Margin Check problem a lin-
ear problem must be solved first. A nonlinear material stress-strain law must be
specified if a material nonlinear analysis is involved.
After the linear problem is solved, the nonlinear analysis is performed for the origi-
nal load until convergence is achieved. The load is then incremented according to
the specified step and the nonlinear solution is obtained for the new load. At the end
of each load step the program checks if any of the predefined failure criteria was
exceeded. A report is issued summarizing the load at which each criterion was
exceeded, and a solution is kept for further post-processing.
FIGURE 57 Margin Check Solver.
Refer to the Advanced Guide for additional information.

Crack Path Analysis
Crack Path Analysis

Crack Path Analysis is limited to use with 2D planar domains and requires that
a set of model construction rules be carefully followed. In contrast to other
analysis modules in StressCheck, Crack Path Analysis does not require that the
user construct a finite element mesh in advance. Rather, the analysis procedure
will automatically mesh the domain during the solution process.
After defining the model domain, locating the crack objects, and assigning all
attributes, you are ready to initiate the crack path analysis. Go to the Solve
interface and select the Crack Path tab (FIGURE 58). In this interface you will
see a list of all crack objects that are currently defined in the model, with the
default status set to not active. If you want the crack to actually be included in
the analysis, you must check the Active box. Next supply the crack increment
FIGURE 58 Crack Path Solver.
to be added at each end of each crack after each iteration. One or both values
may be zero. When a crack has an initial length of zero and if the crack object
lies on an external boundary, you should supply a non-zero crack increment for
both tips. This is because the direction of growth can only be determined plus
or minus 180 degrees. If only one tip is assigned a non-zero length, this tip may
immediately grow outside the problem domain. After the first iteration, the
crack will grow only inside the domain. If the crack is initially internal, it is not

Measurement Analysis
possible to predict which tip will be A and which will be B when starting from a
zero length crack, so assigning only one tip increment may or may not give the
desired result.
NOTE: It is not advisable to interrupt a crack path analysis using the ABORT fea-
ture (Esc key), unless you intend to INITIALIZE prior to the next analysis.
Although you can RESTART after an abort, the results obtained from the analysis
will be INCORRECT. This is because the abort request may have come at a time
after the mesh was regenerated, but before a new solution is computed and there-
fore the mesh may not match the solution information maintained in the program.
Under these circumstances, all extraction operations would be invalid. StressCheck
will display a warning message anytime you attempt to post-process a crack path
solution that was aborted.
Measurement Analysis 4
Measurement Analysis is an advanced feature used primarily in research. The mea-
surement analysis implemented in StressCheck consists of approximating either the
displacement components (elasticity problems) or the temperature distribution
(heat transfer problems) over a domain, given a set of measured displacements (or
temperatures) at NP points. The measured quantities must be provided in tabular
form using the Table Class Input described in the Model Input chapter. In the fol-
lowing, the explanation of the measurement analysis is outlined in the context of
three-dimensional elasticity.
Let ux(xi, yi, zi), uy(xi, yi, zi), uz(xi, yi, zi) be the measured displacement components
in a Cartesian coordinate system at point (xi, yi, zi), i=1, NP. The objective of the
measurement analysis is to seek a polynomial approximation for the displacement
field in the least squares sense. Consider the following approximation for each dis-
placement component (ux, uy, uz):

u x (x,y, z) =
∑a j φ j (x,y, z)
j=1
M
u y (x,y, z) = ∑a j+M φ j (x,y, z)

j=1
u z (x,y, z) = ∑a j + 2M φ j (x,y, z)
j=1 (EQ 1)
where M is the number of terms in the expansion, φj(x, y) are known polyno-
mial functions and aj are unknown coefficients which are computed from the
condition of minimizing the square of the error, that is the difference between
the measured and the approximated displacement at each point:
4
 NP 
∂  2
∑
∂ a j 
[ u x (x i,y i, z i) – u x (x i,y i, z i) ]  = 0

i = 1 
 NP 
∂  2
∑
∂ a j 
[ u y (x i,y i, z i) – u y (x i,y i, z i) ]  = 0

i = 1 
 NP 
∂  2
∑
∂ a j 
[ u z (x i,y i, z i) – u z (x i,y i, z i) ]  = 0

i = 1  (EQ 2)
The condition described by Eq. 2 results in three systems of simultaneous

equations of size M each in the unknown coefficients aj. Once the coefficients
aj are computed, the displacements at any point can be determined from Eq. 1,
and all post-processing operation supported for the standard analyses can be
performed for the measurement solution.
To perform a measurement analysis the only information required is the topol-

ogy of the object (finite element mesh) for which the point-wise displacements

are available. Material properties and boundary conditions (loads and constraints)
are not necessary. If material properties are provided for the model however, they
will be used during the post-processing operations if, for example, the extraction of
stresses is requested.
It is also possible to use the results of a measurement analysis to estimate the mate-
rial properties of the object if the boundary conditions are known. A procedure
called minimization of the traction residuals can be used in conjuction with the
measurement analysis as described in the following.
At the time of performing the measurement analysis, the candidate material proper-
ties and the known boundary conditions are specified for the topology of interest.
With this information, the corresponding global stiffness matrix [K] and load vector
{r} corresponding to the input data can be computed. After performing the mea-
surement analysis, the coefficients {a} of the expansion of the displacements in
Eq. 1 will be available, and can be used to determine a pseudo load vector {r0} as:
[ K ] { a } = { r0 } (EQ 3) 4
Note that the stiffness matrix is not used for computing the coefficients {a}, and
therefore the product of [K] and {a} is not necessarily equal to {r}. The difference
between {r} and {r0} can then be used to assess whether the provided material
coefficients are correct, assuming the loading, and therefore {r}, is known accu-
rately. The traction residual is defined as the L2 norm of the difference between {r}
and {r0}:
3M
∑ (r
2
k – r 0k )
k=1
Residual L2 = ---------------------------------- (EQ 4)
3M
∑r
2
k
k=1
Ideally, the traction residual should be identically zero. Its value is minimized by
proper selection of the material coefficients which modify the stiffness matrix in
Eq. 3.
To perform a measurement analysis, select the Measurement tab in the Solution

dialog window and the screen layout will be as shown in FIGURE 59.

FIGURE 59 Measurement Analysis.

4
Extension Upward-p and Downward-p are the only type of extensions available in Stress-
Check to perform measurement analysis.
p-limits These have the same interpretation as in a linear analysis. The maximum p-
level you can specify depends on the number of data points in each element
available for fitting and on the reference frame (2D or 3D). The minimum
number of data points per element should be: R(p+1)(p+1), where R depends
on the reference (R=2 for 2D and R=6 for 3D) and p is the p-level of the fitting.
Type The type of measurement analysis is intended for flagging the solution for
post-processing operations. You may choose to identify the measurement solu-
tion as Linear, Material Nonlinear or General Nonlinear, depending on the type
of measurement you performed.
Solution ID Provide a name to identify the measurement analysis. This name will be used
for designating the solution vectors for post-processing operations. The name
is arbitrary, but if the minimization of the traction residual option is utilized
then the solution ID should be the same as the load ID and constraint ID pro-
vided for the input data. All three ID’s should be the same. To activate the
computation of the traction residuals, create the parameter _check_measure.

Adaptivity
Create the parameter _true_residual to force the computation of the traction residu-
als only on those boundary segments where tractions were specified (it will ignore
free boundaries).
Table ID Provide the name of the table that contains the set of measured displacements (or
temperatures) at NP points.
Penalty This factor is used to minimize the jumps in the derivative across element bound-
aries. Penalty is only applicable for Heat Transfer problems.
When you are ready to begin the measurement analysis, simply switch to the
SOLVE! tab and start the analysis. The solution for each p-level will be saved for
subsequent post-processing.
Adaptivity 4
Intrinsic control of discretization errors by finite element programs is very impor-

tant to ascertain the quality of the numerical solution. In StressCheck the discretiza-
tion error is reduced by increasing the number of degrees of freedom N, which is
controlled by the polynomial degree of the approximating elemental shape func-
tions and by the number of elements in the mesh. Orderly sequences of discretiza-
tion by increase of the polynomial degree of elements is called p-extension (see
“Control of discretization errors” in the Advanced Topic Guide for additional
details).
For problems involving a few thousands degrees of freedom a good initial mesh
should be selected, based on the expected smoothness of the solution, and uniform
p-extension performed. For small problems adaptivity does not provide appreciable
benefits because selective assignment of the polynomial degree may carry an over-
head which would not provide a substantial reduction in the number of arithmetic
operations made possible by the adaptive process.
Adaptive p-extension means that the polynomial order for each element in the mesh
will be determined for each run based on an error indicator. Adaptivity makes good
sense for large problems when the data of interest are directly related to the poten-
tial energy. When the data of interest are not directly related to the energy then care
must be exercised. This is because adaptivity seeks to minimize the error in energy
norm for a given number of degrees of freedom. Smallness of error in energy norm

Adaptivity
does not guarantee that the errors in the data of interest are small. For this rea-
son, StressCheck provides the means for defining convergence criteria based
on any extraction operations (global error, point extraction, min/max extrac-
tion, etc.) to be used at the completion of each adaptive run in order to test
whether the solution is of sufficient quality.
For large problems, the following procedure is recommended:

• In the case of large problems it is usually possible to distinguish
between two regions, the region of primary interest and the region of
secondary interest. The region of primary interest is that part of the
solution domain where the data of interest, for example, the highest
stress or the largest stress intensity factor, is expected to be found. The
region of secondary interest is included in the solution domain only
because the boundary conditions on the region of primary interest are
not known. The region of secondary interest acts as a sophisticated
spring, conveying information about constraints and loads to the region
of primary interest. Since only the traction-displacement relations
4 along the boundary between the two regions are of interest, which are
directly related to the energy, adaptivity is sensible for the region of
secondary interest only.
• For large problems assign different p-discretization to the regions of
primary and secondary interest, such that adaptivity is performed in the
region of secondary interest and uniform p-extension in the regions of
primary interest. This point will be illustrated by an example in the fol-
lowing.
The adaptive p-extension can be defined for all the elements in the mesh or for
a group of elements only, based on the settings of the p-discretization interface.
For the selected group of elements in the mesh, an adaptive p-extension pro-
cess requires three basic components:
• An error estimator: What is the magnitude of the error?
• An error indicator: Where the error is coming from?
• An adaptive strategy: How to increase the number of degrees of free-
dom to minimize the error?
These are described in the following.

Adaptivity
Adaptive p-Extension
Adaptive p-extension is applicable for the linear solution of elasticity and heat
transfer problems. To perform an adaptive p-extension, select the Adaptive-p
option in the Linear tab of the solver interface (FIGURE 60). Adaptive-p means that
Linear tab SOLVE! tab
FIGURE 60 Input area for linear analysis.
the polynomial order of the elements with a variable or uniform p-discretization

assignment will be determined based on the values of the error indicator for each
run until convergence is realized. The Run limit input field refers to the maximum
number of adaptive runs the program will perform if convergence is not realized.
The p input field is to provide the initial p-level for the adaptive p-extension.
The convergence options are chosen under the SOLVE! tab. Two default conver-
gence options are available: “None” and Error=2%.
• None means that the number of runs is controlled by the settings in the lin-
ear tab. The adaptivity process will repeat until the number of runs reaches
the run limit provided or until all the elements in the adaptive region reach
p-level 8.
• Error=2% (5% for 3D analysis) means that the number of runs will be con-
trolled by the global error of the solution. When the estimated relative error
in energy norm is less than or equal to 2%, the execution will stop. The
number of runs cannot go beyond the limits entered in the linear tab, how-

Adaptivity
ever. Additional convergence criteria can be defined in the Settings tab

of the Results interface, as described in the Setting convergence crite-
ria section and in the example below.
The status window in the SOLVE! interface will display the error associated
with each run during the execution of an adaptive p-extension. If the conver-
gence option contains more than one criterion, the largest error will be
reported.
Selecting the adaptive region
In the p-Discretization interface, different elements in the mesh can be

assigned one or more of the following options: Variable, Bounded, Fixed, and
Uniform. When an adaptive p-extension is performed, these options have the
following interpretation (FIGURE 61):
• Variable means that the p-level of each element in this group will be
increased whenever the error indicator for the element is above a
4
threshold.
• Bounded means that the p-level for all the elements in this group is
increased by one per run until the upper limit is reached, regardless of
the error indicator reading.
• Fixed means that the p-level of the selected group of elements will
remain fixed at the input value during the p-extension.
• Uniform means that during an adaptive p-extension, the p-level of all
the elements in this group will be increased to the same value when-
ever the error indicator for any of the elements is above a threshold.
Error estimator
The relative error in energy norm is computed using the a posterior error esti-
mator described in Finite Element Analysis1, page 69. The error in other quan-
tities such as stress maxima, stress intensity factors, point deflection, etc., are
estimated using a similar algorithm for non-monotonic functions. Three solu-
tions must be available for the estimators to work. During an adaptive p-exten-
sion, the program will compute three solutions before it can estimate the
magnitude of the error(s) to decide whether it needs to increase the p-level
1. B. A. Szabo & I. Babuska, Finite Element Analysis, John Wiley & Sons, Inc., 1991.

Adaptivity
FIGURE 61 p-Discretization options.

4
according to the information provided by the error indicator (convergence is not

realized) or must stop (convergence was achieved).
Error indicator
The error indicator for the kth element (ηk) provides an indication of the relative
contribution of the element to the error of the solution. The error indicator is com-
puted based on the jumps in the stress resultants across any shared edges (2D) or
faces (3D). By definition, the jump (∆) is the sum of the absolute values of the dif-
ferences in stress resultants. For example, the jump for shared edge j of element k
for a planar (2D) problem is given by (FIGURE 62):
(k) (k) (k + 1) (k) (k + 1)

∆j = Fx – Fx j + Fy – Fy j

Adaptivity
An element (k) has as many jumps as there are shared edges/faces. The largest
(k+1)
Fy(k) Fy(k+1)
Fx(k) Fx(k+1)
(k)
edge j
4 FIGURE 62 Stress resultants along shared edges.
jump of the element (∆(k)) is then divided by the sum of all the largest jumps
(∆T) over all the elements (N) to obtain a relative jump for that element (Φk).
(k) (k)
∆ = max∆ j
(k)
∆
N Φ k = ---------
∆T
∑∆
(k)
∆T =
k=1
The error indicator for the kth element, is the ratio between Φk of the element
and the largest Φk for all the elements in the mesh (ηk = Φk/Φkmax). Note that
the error estimator of any element is bounded between zero and one (0 ≤ ηk ≤
1).
Extension strategy
During an adaptive p-extension, the polynomial order of an element with a

variable or uniform p-discretization assignment will be increased by one when-

Adaptivity
ever the error indicator for the element is larger than a threshold (ηk > ηth) and con-
vergence was not realized. The value of the threshold is set to ηth=0.75. If the
polynomial order of an element is less than 5 and the error indicator is larger than
0.875, then the p-level is increased by two.
Example
Consider the Parts Handbook problem: BracketTetra.sci, consisting of 330 tetrahe-

dral elements as shown in FIGURE 63.
FIGURE 63 Example problem for

adaptive p-extension.
The objective of the exercise is to compare three type of extensions:

• Case 1: Uniform p-extension without convergence control
• Case 2: Adaptive p-extension in the region of secondary interest without
convergence control plus uniform p-extension in the region of primary
interest.
• Case 3: Adaptive p-extension in the region of secondary interest with user-
defined convergence control plus uniform p-extension in the region of pri-
mary interest.

Adaptivity
Solution for Case 1 Consider the case of a uniform p-extension from p=1 to p=5 and convergence
set to none. In the solver interface complete as follows:
• Linear tab > Extension: Upward-p > p-limits: 1 to 5.
• SOLVE! tab > Execute: Initialize > Run Mode: Automatic > Method:
Iterative > Convergence: None > Click on the Solve button.
The results of the estimated error in energy norm, the contour plot of the von
Mises stress (Seq) and convergence of the maximum Seq are shown in FIG-
URE 64. Note that global error of the solution is estimated to be 2.78% for run
#5 (24492 DOF), and the maximum von Mises stress in the fillet region for run
#5 is Seq=28151 with an estimated error of 0.17%.
Solution for Case 2 Consider the case of an adaptive p-extension and no convergence control. To
define the discretization parameters, select the Create Model icon and then
select the p-Discretization tab. The region of primary interest in this problem is
located in the fillet area. Therefore, bounded p-discretization is assigned to the
elements in that region as shown in FIGURE 65. Note that in the p-Discretiza-
4 tion form there are two records:
• Trunk space > p-Discretization: bounded > p-level: 2 > p-limit: 6 >
SET3. This is the region of primary interest in which the polynomial
order will change from 2 to 6 during the adaptive run. A total of 107
elements are in this region.
• Trunk space > p-Discretization: variable > p-level 1> ALL. This record
covers all the elements except those in SET3, and the p-level of each
element in this group will be increased whenever the error indicator for
the element is above a threshold.
To perform the adaptive p-extension select the Compute Solution icon and
complete the solver interface as follows:
• Linear tab > Extension: Adaptive-p > Run Limit: 5 > p: 1
URE 66. Note that global error of the solution is estimated to be 0.82% for run
#5 (16131 DOF), and the maximum von Mises stress in the fillet region for run
#5 is Seq=28018 with an estimated error of 0.69%. The results are very close
to the ones obtained by uniform p-extension but using fewer degrees of free-
dom.

Adaptivity
FIGURE 64 Results for Case 1: Uniform p-

extension without convergence control

Adaptivity
4
FIGURE 65 Region of primary interest.
Solution for Case 3 Consider the case of an adaptive p-extension with user-defined convergence
control. The Settings tab in the Results dialog window is used for defining
convergence criteria records that will be used in the context of the SOLVE!
interface. These convergence criteria records identify extraction operations
which will be performed at the completion of each finite element solution in
order to test whether the solution is of sufficient quality. Each individual con-
vergence record is composed of the name of one or more extraction records,
and a corresponding numeric value which represents the level of error consid-
ered acceptable for the given extraction.
For the example problem, consider the creation of a convergence record com-
posed of two extractions: the estimated relative error in energy norm and the
maximum von Mises stress in the fillet region. The two required extraction
records are part of the StressCheck input file, so the only thing to do is to create
a convergence record by selecting Results > Settings tab as shown in FIGURE
67.

Adaptivity
FIGURE 66 Results for Case 2:

Adaptive p-extension without
convergence control.
The convergence record shown in the figure (GLerror) was defined by selecting the
two available extraction records (1-Error and 2-MaxSeq), entering the criterion
value for each (3%) and clicking on the Accept button.
To perform the adaptive p-extension select the Compute Solution icon and com-
plete the solver interface as follows:

Adaptivity
FIGURE 67 Setting convergence record for the example problem.
• Linear tab > Extension: Adaptive-p > Run Limit: 5 > p: 1

Iterative > Convergence: GLerror > Click on the Solve button.
URE 68. Note that only 4 runs were required to achieve the convergence goal
and that the global error of the solution is estimated to be 2.1% for run #4
(11208 DOF), and the maximum von Mises stress in the fillet region for run #4
is Seq=28078 with an estimated error of 0.77%. The results are again very
close to the ones obtained by uniform p-extension but using fewer degrees of
freedom.

Adaptivity
FIGURE 68 Results for Case 3: Adaptive

p-extension with convergence control.

Adaptivity
Summary The results for the 3 cases analyzed are summarized in Table 1. As can be seen
the results for the quantity of primary interest (the von Mises stress) is practi-
cally the same in all cases. Substantial savings in the number of degrees of
freedom are realized when adaptive p-extension is used for the region of sec-
ondary interest.
TABLE 1. Bracket problem: Comparison of results for different extensions.
Case Extension Convergence DOF Global error von Mises stress

1 Upward-p None 24492 2.78% 28151
2 Adaptive-p None 16131 0.82% 28018
3 Adaptive-p 3% global & 3% stress 11208 2.10% 28078

5 Post-processing Operations
This chapter describes the most commonly used post-processing operations available in StressCheck:
• Estimation of the relative error in energy norm

• Plotting engineering functions
• Computation of minimal and maximal values 5
• Computation of functions at points
• Computation of element equilibrium and resultants
• Computation of distortion and area/volume of elements
• Computation of fracture mechanics parameters
• Setting convergence criteria
You can access post-processing features of StressCheck by selecting Results from the Main Menu Edit
pulldown menu, or by selecting the View Results icon in the Main Toolbar (FIGURE 69).
Results dialog window

Each Results class interface provides three tabs: Input, Settings and File.
User’s Guide Chapter 5: Post-processing Operations 133

Results dialog window
5
FIGURE 69 Results interface.
The Input tab provides the input controls for performing post-processing oper-
ations. After providing the desired entries, click the Accept button to perform
the operation.
The Settings tab provides a facility to save/retrieve your input settings from a
record in your database that may be retrieved for later use. You may also indi-
cate whether or not the settings record is to be used from the handbook inter-
face. Each record is assigned a unique name and saved in alphanumeric order.
Records enabled for handbook use will be activated in the order determined by
their name.
The File tab makes it possible to save/retrieve input settings from a standard
ASCII file.
134 Chapter 5: Post-processing Operations User’s Guide

Report window
For the result classes that provide graphical output, a button is provided
which will bring the Graph window to the front if hidden behind the
model window.
Report window
Composing an engineering report is now easier than ever before using Stress-
Check’s report editor. The report editor may be activated from the Views Toolbar
by selecting the “Display Report” icon. Once the report window is visible, you may
insert a capture of the Model window, a capture of an extraction Chart, or an extrac-
tion Table using the “Capture Window Contents to Report” icon illustrated in FIG-
URE 70.
Display Report icon

5
Capture Window Contents to Report icon/menu
FIGURE 70 Report Composition Icons
FIGURE 71 illustrates a sample report window after a snapshot of the model win-
dow, extraction chart and extraction table have been captured and inserted using the
“Capture Window Contents to Report” icon menu options.
After capture, the inserted graphic images may be manually resized or repositioned.
You may also enter new text manually, insert new text or graphics from the Win-
dows clipboard, and apply a variety of formatting operations to the document using
the various options provided by the menu options associated with the Report Win-
dow. The Report Window’s menu options will replace the standard StressCheck
menu options whenever the Report Window has input focus. You may change
fonts, margins, spacing, etc. using the various options provided. You may also pre-

Report window
Model Capture
Chart Capture
5 Table Capture
FIGURE 71 Sample StressCheck Report
view the report, and dispatch it to a printer when you are satisfied with the for-
mat.
When you exit StressCheck, you will be prompted to save or discard your
changes to the report. You may also save the contents of the report to a Rich
Text Format file at any time using the File > Save or File > Save As options. If
the report is saved with the same name as your working database, it will be
automatically reopened for additional composition operations the next time
you open your StressCheck database.

Report window
Report Template A StressCheck report may be composed from a “clean” sheet of paper, or may be
based on a customized report template. A template may contain default header and
footer text as well as “tagged” fields which may be filled in automatically by a
VBA script, or by StressCheck’s autofill feature, based on a user specific template
information file. FIGURE 72 illustrates the content of the default StressCheck
report template file which is provided as part of the standard StressCheck installa-
tion. This file may be customized for a specific site, or a user may define a template
Tag Fields
Report Body Report Header
FIGURE 72 Default Report Template
for his/her own personal use. The following sections describe the construction of a
report template file. Once we have described how to compose a template file, we

Report window
will discuss how the tagged fields will be filled in by StressCheck or by a VB

script.
Template file As an example of how to construct a template file, we will create one starting
from an empty report window, and proceeding through the following steps:
- Access the header section of the report page
- Insert a header-table into the header
- Insert sub-tables into each cell of the header table
- Change the border attributes of table cells
- Change the shading attributes of the table cells
- Insert tagged fields into the table cells
- Insert default text into the table cells
- Insert special tags for date and page numbers
Starting from an empty report window as shown in FIGURE 73 the first step is
to access the header section of the report page. The header section of the tem-
plate contains information which will be repeated on each page of the report. In
a typical engineering report, the header will contain the company name, the
name of the engineer who is performing the analysis, and various other infor-
mation identifying the part number, report number, page number, date, etc.
FIGURE 73 Empty Report Template

Report window
To access the header section of the page in order to insert the necessary template
fields, activate the “Display Report” icon and then select the Edit > Edit Page
Header/Footer option from the main menu bar. Dotted lines will appear near the top
and bottom of the page that separate the header and footer sections of the page from
the main body of the page.
Next, place the cursor into the header area and select Table > Insert Table from the
main menu. Create a table which is 2 rows by 3 columns. A dialog window will
appear as shown in FIGURE 73, that accepts the table dimension information.
When you click OK, a table will appear, outlined with dotted lines. If you position
the mouse cursor over the vertical dotted line separating the first and second col-
umns of the table, a special cursor will appear that allows you to reposition the
dividing line. Drag the dividing line to the left until the first column is approxi-
mately 1.5 inches wide. Repeat for the dividing line separating the second and third
columns, dragging the dividing line to the right.
Next, change the cell borders for this master table by first selecting all cells of the
table. Position the cursor in the upper left cell, and drag to the lower right cell so
that all cells are highlighted. Choose the Table > Cell Border... menu option to acti-
vate the dialog window shown in FIGURE 74. If all cells are highlighted, you may
set the Range option to Selected Cells, or simply set the All Cells option. Enter a
value of 1 into each of the border width input fields, and click OK. The table bor-
ders will change from dotted lines to solid lines.
5
Next, insert the mouse cursor into each cell of the table and insert a sub-table into
each cell. From the main menu select Table > Insert Table and enter a single col-
umn and the number of rows shown in FIGURE 75, skipping the top center cell. If
you accidently create a table with the wrong number of cells, simple select the cells
of the table and press the Delete key on your keyboard to remove the erroneous
table. After inserting each sub-table, position the cursor into the top center cell of
the main table, and select Insert > Data Field... from the main menu. The dialog
windows shown in FIGURE 75 will appear so that you may define a tagged field
which will provide a place for your company name to appear. Enter a field name
“CompanyName” into the first input field, and type the company name in the sec-
ond input field as shown. After clicking OK to define the data field, click the Text
Centering icon to center the data field in the cell.
Next, place the cursor in the top cell and select Table > Cell Shading... and set the
shading attribute to 10% as shown in FIGURE 76. After shading the indicated cells,
place the cursor in each shaded cell and type the label shown in the figure. After
entering each label, you may center the labels by highlighting the entire column,
then clicking the centering icon in the toolbar of the report window.

Report window
FIGURE 74 Set Cell Borders of Template Table
Next, define the data fields in the cells below each shaded field as shown in
FIGURE 77. From the main menu bar select Insert > Data Field... and com-
plete the Field Name and the Field Data fields under Designer and Checker.
The Field Data represents the default value for the cell. Note that you will see
the default value for each data field appear if the View > Field Names option is
disabled. If the View > Field Names option is enabled, you will see the data
field names appear in curly brackets {}. To insert the date field, just do Insert >
Date and select the format of your choice.
A label may also be assigned to a cell. In those cases where a cell contains both
a label and a data field, simply type the label into the cell, then insert the data
field. For example, to enter the page fields, first insert a page number field
(Insert > Page Number), then type a text label for the word “of”, then insert a
page count field (Insert > Page Count).

Report window
Place cursor in top center cell for Company Name Data Field
2 rows 2 rows
4 rows 4 rows
2 rows
Text centering icon
Data field definition dialog
FIGURE 75 Define a Template Data Field

5
Referring to FIGURE 77, note that there are three reserved data field names: Mod-
elTitle, Description, and Model. In the central cell, type “Model Title:”, then select
Insert > Data Field... and enter “ModelTitle” under Field Name and click on OK.
You can place the cursor at the end of that cell, hit the “enter’ key and select Insert
> Data Field... again to enter “Description” under Field Name. When running a
standard StressCheck Handbook Analysis, these tags will automatically be filled in
by StressCheck with the title assigned by the user, the model description, and the
model name. In this sample template illustration, we did not use the reserved name
“Model”, but instead used “ModelName”. This means that a user defined script, or
the user defined model information file must provide the data that will be inserted
into these tagged fields. More on this later.
If you wish, you may set other cell attributes to control other cell borders, shading,
justification, etc. When you are finished, save your template file using the File >
Save As menu option, and save the file in SSE Format (*.sse). If your template is
intended for use by your entire company, place this file in the StressCheck installa-

Report window
Select cells for shading
FIGURE 76 Template Cell Shading
5
tion sub-directory /UI, with the name Template.sctm. If you prefer a different
name or location, you must perform an additional step to set a Windows regis-
try key to indentify the location and name of your default template file.
To do this, set your input focus back to your StressCheck model window, then
choose the menu option Tools > Edit Template.... A dialog window will appear,
as illustrated in FIGURE 78. Select the tab labelled Template File. At the bot-
tom of the dialog window, click the Browse button to locate the path to the
template file you have defined. After identifying the template file, click the
“Set Env. Var.” button. This will set a registry key for the “Current User” called
“ESRD6_TEMPLATE_FILE” to point to this file. If you wish to define a tem-
plate file for all users of your machine, you must use the Windows Regedit
utility to set a registry key named “ESRD6_TEMPLATE_FILE” in the “Cur-
rent Machine” registry folder. If neither of these registry keys are defined,
StressCheck will use the file named “template.sctm” in the /UI StressCheck
installation directory.

Report window
Page Number & Page Count
Date field
FIGURE 77 Template Data Field Definitions
5
Report Template Data The data fields defined in a report template file may be filled in by StressCheck in
the case of the reserved fields (“ModelTitle”, “Description”, and “Model”), using
data supplied in a Template Information file, or by a Visual Basic program using a
COM function call. This section describes the procedure for defining a template
information file containing the data which will automatically be inserted into
tagged data fields of a template based report. Please refer to the Programming
Guide for a description of the COM functions which may be used to deposit data
into a data field.
To define the data values that will automatically be inserted into template data files,
select the Tools > Edit Template option from the Main Menu Bar. The dialog win-
dow illustrated in FIGURE 79 will appear.
The data field names in the currently defined template file will appear in the dialog,
along with the default values defined in the template file. If you wish to override
the default values simply replace each string shown in the right-hand column.
When you have finished editing the data values, save the data to a template tag

Report window
FIGURE 78 Defining a Template File Path
FIGURE 79 Defining Template Data Values

Checking the overall quality of the solution
information file (extension .tti). In order to tell StressCheck to read values for a
template based report from your data file, you can simply save your data file in the
current working directory with the same name as your template file, but with the
extension .tti. If you prefer to save the file elsewhere, it must still have the same
name as the corresponding template file and the .tti extension. If you save the file
somewhere other than the current working directory, you must set a registry key by
clicking the “Set Env. Var.” button in the Edit Template > Info File dialog window
as illustrated in FIGURE 79. This sets a registry key in the “Current User” folder of
the Windows registry. If there is no template information file found in the current
working directory, and no registry key is set in the “Current User” registry, Stress-
Check will look for a key in the “Local Machine” registry. If this registry key is not
found, StressCheck will look for a file in the StressCheck /UI installation directory.

Having obtained a hierarchic sequence of at least three finite element solutions, it is
possible to estimate the overall quality of the solution. This quality is measured in
terms of the energy norm. The energy norm is closely related to the root-mean-
square measure of stress in elasticity when the ratio of the largest eigenvalue of the
material stiffness matrix to the smallest is not too large.
5
Specifically, StressCheck estimates the relative error in energy norm from:
2 C
π ( u FE ) – π ( u EX ) = u EX – u FE ≤ ----------
E 2β
N
where π(uFE) is the potential energy of the finite element solution, C and β are pos-
itive constants and N is the number of degrees of freedom. The rate of convergence
is characterized by β. This relationship is such that for sufficiently large N the “less
than or equal” sign can be replaced by “approximately equal”.
Knowing π(uFE) and N for three levels of extension (usually p-levels), in most
cases it is possible to solve for the unknowns π(uEX), C and β. A detailed descrip-
tion of this procedure is available in Ref. [8].
Care must be exercised in interpreting this error measure. In general, it is not suffi-
cient to ensure that the energy norm measure of error is small. It is also necessary to
check that the data of interest are converging to their limit values.

Estimating the relative error in energy norm

Having performed a p-extension, select the Error tab in the Results dialog win-
dow and the Results interface will come out together with the Graph window
as shown in FIGURE 80. To obtain an error estimate, there must be at least
(b)
(a)
FIGURE 80 Results interface for

Error Estimation.
(c)

three runs in the sequence. Typically, a sequence of solutions is computed as p

ranges from 1 to 8 on a fixed mesh (either downward or upward). The solution
name(s) and run numbers and associated degrees of freedom are displayed in the
combo listbox. Click on the last run and the run numbers will be entered automati-
cally by StressCheck. By selecting the Table tab, the results are tabulated as shown
in FIGURE 80c.
To obtain the results, select All elements, click on the Estimate button and then
click on the Accept button.
Note: In some cases the estimated limit cannot be computed. This happens when
the error associated with the available finite element solutions cannot be approxi-
mated by a function of the form C/N2β. Usually there is an error in the input data, or
some special symmetry or antisymmetry is present.
Computing the strain energy
To compute the strain energy for an element or group of elements, select the Error
tab from the Results interface, set the A/O/M to Select > Any Element > Selection,
pick the element(s) from the display area, select the Estimate button and then click
on Accept. The program will compute the strain energy of the selected elements
instead of the potential energy for all the elements when the Object: Any Element is
selected. The strain energy (U) is computed from the following expression:
5
1
U = ---
2 ∫ ε { σ } dV
Ω
where {ε} is the mechanical strain tensor, {σ} is the stress tensor and the integral is
performed over each element (Ω). For the sequence of solutions selected, the
results will include the run #, the unconstrained degrees of freedom (U_DOF), the
strain energy, convergence rate and % error associated with each run #, and the esti-
mated limit of the strain energy as shown in FIGURE 81.
Error indicator
An option is available in the Error interface to compute the error indicator for an
element or groups of elements. This option can be used for a particular linear solu-
tion corresponding to an elasticity problem only. The error indicator for the kth ele-

FIGURE 81 Strain energy: Tabular results.
ment (0 ≤ ηk ≤ 1) provides an indication of the relative influence of the element

on the global error of the solution.
To compute the error indicator, the program first determines the jumps in the
stress resultants across any shared edges (2D) or faces (3D). By definition, the
jump (∆) is the sum of the absolute values of the differences in stress result-
ants. An element (k) has as many ∆s as there are shared edges/faces. The larg-
5 est ∆ of the element (∆k) is then divided by the sum of all the jumps over all the
elements (∆T) to obtain a relative jump for that element (Φk=∆k/∆T). The error
indicator is the ratio between Φk of the element and the largest Φk for all the
elements in the mesh (ηk = Φk/Φkmax).
To use this option, select the solution of interest, click on the Indicate button,
and then click on accept (FIGURE 82). A report containing the error indicators
for each element will be available for that run #. You can enter a value between
0 and 1 in the Criteria input field, and the program will report only those ele-
ments with an error indicator exceeding that value.
Checking equilibrium and resultants

StressCheck provides information about the equilibrium of elements. The basic
idea is that in the displacement formulation of the finite element method, the
equilibrium conditions are satisfied approximately. As the number of degrees

FIGURE 82 Error indicators
5
of freedom increases, the displacement field of each element will satisfy equilib-
rium with increasing accuracy.
In the equilibrium test, the tractions are computed along the boundaries of elements
from the finite element solution and integrated along the perimeter. When no body
forces are applied, the contour integral of the tractions should be nearly zero. When
body forces are applied, the contour integral of the tractions should be nearly equal
and opposite to the applied body forces.
The equilibrium test is a rough measure of the quality of the mesh and should be
used for guidance in changing the mesh when the desired level of accuracy cannot
be reached by p-extension.
To perform an equilibrium test, select the Resultant tab in the Results dialog win-
dow and the Results interface will appear as shown in FIGURE 83. You may elect
to check equilibrium for any run number or a sequence of run numbers. In FIGURE
83, a sequence of run numbers from 1 to 8 is indicated.

Summation fields
5 FIGURE 83 Results interface for Resultants.
The extraction options depend on the object selected:

• Node: Computes force resultants and moments for a beam element.
• Edge: Computes the integral of the traction along an element edge for
2D problems.
• Face: Computes the integral of the traction over an element face for
3D problems.
• Element: Computes the contour integral of the traction over the ele-
ment (equilibrium check).
• Fastener: Computes the load transferred by a fastener.
The moment center refers to the coordinates of the point about which the
moment is to be computed. The default is the origin of the global system.

Plotting the data of interest
The available output functions will depend on the reference and analysis types. For
example, the appropriate functions for planar elasticity are Fx, Fy and Mz, that is
the force resultants in the global X and Y-directions and the moment about the Z-
axis.
The results of the equilibrium or resultant computation will appear in the Graph
window for the selected functions only, and also in the summation fields of the
Input area for all relevant functions. To select an output function, simply click on
the corresponding Force or Moment switch in the summation fields area of the win-
dow.
When performing equilibrium checks, click on the target element with the left
mouse button and then on the Accept button. The result will be displayed in the
Graph window. In the summation fields, only the results corresponding to the solu-
tion with the highest number of degrees of freedom (DOF) are displayed.
When computing resultants for a sequence of solutions, the graphical output will
show the convergence path, while the Input area will show the numerical values for
the run with the highest number of DOF. If the action: Check is selected and you
click on the Reset button, the previously obtained results will be cleared from the
summation fields. You are then ready to click the next element edge. If you do not
click on the Reset button then the values in the summation fields will be cumula-
tive.
5

In this section, the procedures for plotting the data of interest from finite element
solutions are described. StressCheck saves the solution vectors for the finite ele-
ment solutions obtained in the solution step. The data of interest can be computed
and displayed for each solution or sequence of solutions using the procedures
described in this section. StressCheck provides options for the computation and dis-
play of some frequently used data, such as displacement, strains, stresses, etc.
These data, called “Standard Functions”, are described in detail in the Analysis
Guide. Other data can be computed and displayed using the Formula or Calculator
options.
StressCheck provides a convenient means for displaying and printing computed

information in graphical form. To obtain a contour plot or deformed configuration,
select the Plot tab in the Results dialog window. The Results interface will appear

as shown in FIGURE 84. The procedures for plotting and printing graphical
information are described in the following.
FIGURE 84 Results interface for plotting.
Contour plot
Decide whether the desired data are to be plotted for all elements (this is the
default), for individual elements, for faces, or for boundaries. Click on the
Cancel button if you wish to cancel the selected object(s).
Select the solution name and run number for which you wish to obtain a con-
tour plot. This is accomplished simply by clicking on the appropriate line in the
combo listbox. Normally, you will use Plot: “Solution”. The “Formula” option
is used less frequently. This option is discussed at the end of this section.

To produce a color plot, select Contour: “Fringe”. Choose the shape of representa-
tion. The choices are: Undeformed shape, Deformed shape, Z-shape, and Principal
Directions. These options are described later.
Select the function from the list and the number of midsides. For plotting purposes,
each element is subdivided into a data extraction mesh which depends on the num-
ber of “Midsides”, that is, the number of grid points along each side not including
the end points. The desired data are then computed in the grid points of the data
mesh. The higher the number of midsides, the more accurate the display, and the
longer the required computer time. Generally, using 3 to 5 midsides is sufficient.
Do not specify a range for your first plot. StressCheck will automatically set the
range from the minimum to the maximum value of the function. Click on the Plot
button, and the fringes of the selected function will appear in the display window.
The range of values of the plotted function will also be displayed. At this point, you
may wish to select a Range and contour Intervals. For example, for the case shown
in FIGURE 85, the equivalent stress values vary between 6.5 and 416 on a data
mesh corresponding to nine midsides. It is better to set the range from zero to 400
and select 8 intervals. The values will then be shown at 50-unit increments. To
change the range values, turn on the switch in front of Range and enter the new val-
ues. The number of contour intervals is entered below. You can display the function
either in colors or in gray scale (Gray button). The Blend button will blend adjacent
colors. The Invert button will assign colors to the display function in the opposite
order. 5
To plot a standard function, for example Seq which stands for “equivalent stress”,
select Seq from the Functions combo box. Select the number of midside points for
the data mesh, and then click on the Plot button.
Remark: Continuity of contour fringes
As described above, the desired data are evaluated at the grid points of the data
mesh separately for each element. Therefore, two sets of values are produced at
interelement boundaries, one from each element. Any discontinuity in contour
fringes will be clearly visible. Any discrepancy is an indication of error in the plot-
ted data. Generally the highest stress values are of interest and therefore attention
should be focussed on the corresponding contour values. Contours corresponding
to low stress values can “jump” across interelement boundaries considerably and
should not be a cause for concern. In the equivalent stress contour, fringes shown in
FIGURE 85, the continuity of contour fringes is a confirmation of the quality of the
solution and the data computed from the solution. Turn on the Average option if
you wish to produce a plot with no jumps.

FIGURE 85 Equivalent stress fringes.
Plotting the deformed shape

It is good practice to plot the deformed shape first. Possible misrepresentations
or errors in the boundary conditions can be quickly identified from the
deformed shape. To plot the deformed shape, select Shape: Deformed, and
click on the Plot button. The deformed shape will then appear. If you wish to
compare the deformed shape with the undeformed shape, then turn on the
5 Overlay option. You can enlarge the deformation by clicking on the up arrow
or reduce the deformation by clicking on the down arrow next to the Scale
input field. An example is shown in FIGURE 86.
FIGURE 86 Deformed shape and overlay.

Graphic file output

To create a Graphic output file of the image in the main window, select File >
Graphic File Output from the Main Menu Bar, and the box shown in FIGURE 87
will appear. Enter the name of the file to be created (with or without extension),
select the desired option (JPEG, Postscript, HOOPS Metafile, Pict, or HPGL), and
click on the Save button. The box will disappear, and the graphic file will be cre-
ated.
FIGURE 87 Graphic output dialog box.
Displaying and plotting a reference function

In some cases you may wish to plot a function over the mesh defined by a formula
which is not related with the solution. In that case, select Plot: Formula, define the
formula, and supply the formula name and the corresponding coordinate system
name in the input area of the Results window. The default is, as always, the global
system. FIGURE 88 shows an example.
Note that formula-specified data can be displayed in two ways: the first, just
described, is to specify a formula which is independent of the solution. This option
is useful to visualize a formula which may be used as a boundary condition. The
other option is to use a formula to perform some mathematical combination of the
standard functions. In this case, select Fmla from the Function combo box and then
supply the name of a formula which is a function of some of the standard engineer-
ing functions; such as Sx, Sy, etc.

FIGURE 88 Plot formula.

5
Z-shape function
An alternative way to display information is illustrated in FIGURE 89. It is
available only for 2D analysis and 3D-beams. To produce such a display, pro-
ceed as follows: select Shape: Z-shape, the function of interest, the number of
midsides, and then click on the Plot button. In FIGURE 89, the number of mid-
sides is 5. To obtain a perspective view, select the Rotate icon from the Views
Toolbar and rotate the figure until a satisfactory view appears. Click on the up
and down arrows next to Scale to scale the displayed function. To lock in the
scale, turn the toggle switch on in front of Scale.
Animation
The Results > Plot interface provides a facility for animating the deformed
shape of a model or a solution sequence. The solution sequence may be a

5
FIGURE 89 Displaying information by the option “Z-shape”.
sequence of linear solutions for which you wish to animate the convergence of a
particular engineering function. The sequence may be a parametric solution (from a
design study) for which you wish to animate an engineering function as it changes
due to the change of a design variable. The sequence may also be a series of load
steps from a nonlinear analysis for which you wish to monitor the shape or engi-
neering function as the nonlinear solution converges.
The Animation toolbar (FIGURE 90) is activated from the Results > Plot interface
when you click the Animate button. Prior to activating the animation, you should
select whether you wish to animate a single solution or a sequence of solutions, and
choose the number of steps in the animation sequence.
If you are animating a deformed shape, the number of steps will be used to divide
the maximum range of the deformation scale into equal increments so that the first

Rewind/Forward
Single/MultiSolution Sequence
Pause
Steps in Animation
Play
Repeat Factor
Stop
Record
FIGURE 90 Animation toolbar.
step of the animation will have the minimum deformation (zero), and the final
step will have maximum deformation.
If you are animating a solution sequence, the run number supplied in the
Results > Plot interface will be the first solution in the sequence. After each
5 image is generated, the run number will be incremented to identify the next
solution in the sequence until the step limit is reached. For this reason, the run
number and number of steps should be chosen carefully to correspond to the
proper range of run numbers to be animated.
Animation will begin when you click the Plot button in the Results > Plot inter-
face. While the animation is active, it is possible to change the content of the
model by enabling or disabling the undeformed shape overlay, geometric and
mesh objects, boundary conditions, etc., by clicking the corresponding buttons
or toolbar icons. It is also possible to change the view of the model by rotating,
zooming, translating, etc., during the animation.
It is possible to control the speed of the animation by incrementing/decrement-

ing the speed control in the animation interface. The speed value is a “repeat”
value. It will control how many times the current image is repeated before
advancing to the next image. This has the effect of slowing down the animation
as you increase the speed value, and speeding up the animation as you decrease
the speed value. You may also pause the animation and then step backward or

Computing minimal and maximal values
forward one image at a time, by first clicking the Pause button, then clicking the
backward/forward arrows. To leave Pause mode, click the Pause button again.
When you wish to stop the animation, you may either press the escape (Esc) key
while the model window has focus, or click the OFF button in the Animation tool-
bar.
While the animation is progressing, you may capture the animation sequence as a
series of images, then play them back in rapid succession. This is useful if the
model is very complex, and the dynamic animation proceeds too slowly. Click the
Record button to start capturing the sequence of images. Once the sequence is cap-
tured, you may click the Stop button, and then click Play to playback the recorded
images. You may need to control the speed if the playback is too fast. Of course,
you cannot change the view or content of the model window during playback. You
must first stop the playback by clicking Stop or Play. The playback may also be
paused and single stepped as described above. If you wish to record a new
sequence, just click Record again.
Image frames captured during the “record” operation may be saved to JPEG files
automatically during the animated display sequence. Each time the “record” button
is depressed, the program will ask whether the animation frames should be saved. If
yes, the program will automatically save each frame to a .jpeg file named the same
as the current database but with a suffix identifying the frame. The quality (amount
of compression) of the .jpeg files can be controlled in the Tools/Preferences inter- 5
face.
To remove the animation toolbar, disable the Animate button in the Results > Plot
interface.

An important feature of StressCheck is that the location and magnitude of minimal
and maximal values of stress, strain and displacement data can be computed for a
solution or a sequence of solutions. Generally, the solution can be considered to be
of good quality if the error measured in energy norm is small and the location and
magnitude of the minimal or maximal values of the data of interest clearly converge
to a limiting value.

To compute minimal, maximal or average values of displacement, stress and

strain data, in the Results window select the Min/Max tab (FIGURE 91). Click
on the solution name in the scrolling listbox and enter the range of run numbers
for which you wish to compute minimal and maximal values. For example,
you may enter 8-8 if only run number 8 is of interest.
multi-solution button
FIGURE 91 Results interface for the extraction of minimal and maximal values.
Note: It is possible to compute the minimum/maximum values for different

solution ID’s. This is useful if you want to compare linear with nonlinear, or
solutions having different boundary conditions. To activate multiple solution
computation, select the multi-solution button near the upper left corner of the
Results interface. The solution selection controls will change so that you may
choose up to four different solution ID’s and corresponding run numbers.

Specify whether you wish to compute the data on all elements, on a group of ele-
ments, a set, a face, an edge, or along selected boundaries. The option ‘boundary’
makes sense only if an edge or a face is attached to it. Select the Method of search.
The options are:
• Block: the search will be performed on a three-dimensional grid. This
option is applicable only to solid elements.
• Grid: the search will be performed on a two-dimensional grid. This option
can be used for 2D elements or on surfaces or faces of 3D elements.
• Line: the search will be performed along element edges, along 2D bound-
aries or along the outline of a surface.
• Wet Surface: the search will be performed on the free faces of the elements
(excluding shared faces). This option is applicable only to solid elements.
The output will be displayed in the Graph window. Choose the Table tab for a tabu-
lar display or the Chart tab for an X-Y plot.
You can specify the format of the results to be displayed by entering it into the For-
mat field. The default format is in scientific notation and any C language format
specification can be used. For example, the number Pi (3.141592654...) will be dis-
played as: 3.141592654e+00 in format %16.9e or 3.14 in format %5.2f.
The Display Points button can be turned on to display the sample points in the
model. If you wish to create a point at the location of the maximum function value, 5
turn on the Locate Maximum button. If the Average button is turned on, the pro-
gram first computes the average of the selected function, extracted from each ele-
ment sharing an edge/face, for each point along a shared edge/face. The averaged
values are used to select the min/max function from all the points. The Et/Em
switch can be toggled when extracting min/max strains to indicate whether total or
mechanical strains should be computed.
Specify the function by selecting the appropriate function from the Function combo
box. If you wish to define a formula then choose Function: Fmla, and specify the
formula name and the coordinate system in which the formula is defined. Define
the formula in the Formula(e) dialog box.
Specify the data mesh, which is characterized by the number of “Midsides”. The
specified function will be computed in the grid points of the data mesh. The data
mesh determines the set on which the minimum and maximum will be sought. For a
typical p-version mesh, between 8 and 12 midsides should be used.

If an element is selected and the analysis reference is Extrusion, then it is pos-

sible to perform the extraction on any plane parallel to the XY-plane. The loca-
tion of the plane is controlled by the input in the Z-plane field. The value of the
Z-plane can be any number between –1.0 and 1.0, with Z-plane=0 representing
the mid-plane of the extrusion.
If an element edge is selected and the analysis reference is Extrusion, then it is

possible to interpret the extraction to indicate the position of the edge in any
plane parallel to the XY-plane. This feature allows defining the extraction pro-
cedure in 2D, and then interpreting the extraction as if the edge is located in an
arbitrary plane (defined by the Z-plane input field) when the model is extruded.
Specify whether you wish to compute Minimum, Maximum and/or Mean val-
ues by turning on the appropriate buttons.
Click on the Accept button. The points where the maximal (respectively mini-
mal) values occur will be identified for the run number with the largest number
of degrees of freedom, with a positive (resp. negative) sign in the display
screen. To clear these labels, click the right mouse button while the display
window has focus.
Min/Max/Mean values can be computed for one function at a time. When com-
puting minimal and/or maximal values from a sequence of at least 3 solutions,
StressCheck performs an estimation of the true value of the selected function
5 by projecting the results from the finite element solutions to an infinite number
of degrees of freedom. The result of this projection is reported as the “Esti-
mated Limit” together with the percent deviation from the value corresponding
to the solution chosen having the highest number of degrees of freedom. FIG-
URE 92 shows a typical graph output where the estimated limits computed by
StressCheck are displayed for the maximum and minimum values of Sx using
9 midsides.
Note: The computation of the estimated limit is performed as follows: If the

last three data points are non-monotonic with respect to the number of degrees
of freedom (N) then a weighted average is computed, the weight being the
number of degrees of freedom. When the last three data points are monotonic
with respect to the number of degrees of freedom then it is assumed that the
error decays in proportion to N-β where β is a positive number. These should
be understood as heuristic procedures which tend to work well when the errors
in the data are small, but may give a false indication of the limit when the
errors are large. The procedure may compute a limit even when the limit does
not exist or fail to compute a limit even when one does exist.

Computing data at points
FIGURE 92 Typical graph output.
Computing data at points 5

StressCheck provides convenient means for the computation of data in user-
selected points, along user-specified lines, element edges, faces or boundary seg-
ments. Select the Points tab in the Results dialog window and the input area will
appear as shown in FIGURE 93. Select the solution for which you wish to compute
data. Select the run number or the range of runs for which data are to be computed.
You can specify the format of the results to be displayed by entering it into the For-
mat field. The default format is in scientific notation and any C language format
specification can be used. For example, the number Pi (3.141592654...) will be dis-
played as: 3.141592654e+00 in format %16.9e or 3.14 in format %5.2f.
Note: It is possible to compute data in user-selected points for different solution

ID’s. This is useful if you want to compare linear with nonlinear, or solutions hav-
ing different boundary conditions. To select multiple solution computation, select
the multi-solution button near the upper left corner of the Results interface. The
solution selection controls will change so that you may choose up to four different
solution ID’s and corresponding run numbers.

multi-solution button
FIGURE 93 Results interface for

extraction of data at points.
The number of points must be filled in only if you wish to extract data at multi-
ple points along a boundary, edge or a line which you may select graphically. If
you wish to evaluate a formula, enter the formula name and define the formula.
The output will be displayed in the Graph window. If you wish it displayed in
tabular form then select the Table tab, if you wish an XY-plot select the Chart
tab.
Note: For the Point extraction interface. If an element edge is selected and the
analysis reference is Extrusion, then it is possible to interpret the extraction to
indicate the position of the edge in any plane parallel to the XY-plane. This fea-
ture allows defining the extraction procedure in 2D, and then interpreting the
extraction as if the edge is located in an arbitrary plane (defined by the Z-plane
input field) when the model is extruded.

The Method: Selection It is used to evaluate data at or along a pre-defined object such as a boundary, point,
node, edge, location, system, or set. Identify the particular object(s) by clicking on
the object(s) or by dragging the cursor until a desired group is completely enclosed
in a rectangular box. If the group includes objects which were not to be selected
then individual objects can be removed from the group by holding the Ctrl key
down while clicking on them. If the object selected is a boundary, then you may
specify an offset range.
For example, if the selected boundary is a line, then offset 0 means the starting
point of the line, offset 1 means the endpoint. An offset of 0 to 0.5 is understood to
mean that the specified function(s) will be evaluated in the starting point, the mid-
point of the line plus the number of points specified in the dialog box (# of pts.) in-
between (FIGURE 94). In the case of circles and ellipses the offset range is mea-
sured in the range 0 to 1 as well.
The second parameter range is ignored for other than 3D objects. If the selected
boundary is a surface, the initial and final range of each parameter will define an
extraction grid in the parameter space of the surface. The third parameter range
only applies to object system. When system is selected the three parameter ranges
apply to the x, y and z-directions extraction ranges.
Turn on the Display points button if you want to see the sample points being dis-
played on the screen as they are being evaluated. This option is useful to visualize
the density and location of the points along the selected object.
The Method: Path It is used for computing data along any arbitrary path on the mesh. The type of path 5
will depend on the object selected. To compute data along an arbitrary path that is
specified directly on the screen, select Object: Location, left-click on the starting
location, release the mouse button, and move the cursor to define the end location.
When the end location is reached, then double-click. StressCheck will compute the
data at the end-points and at evenly spaced intermediate points, the number of
which is specified in the dialog box (next to “# of pts.”). The results will include the
coordinates of the points, the data values and the minimal and maximal values
within the set.
If the Object: Node, Point, Any Boundary, System, or Edge is selected, the Method:
Path will cause the first click on the screen to snap to the closest node, point,
boundary, system, or edge to the mouse pick. Then, drag the mouse and the second
click will snap to the nearest object to the mouse pick thus creating a line connect-
ing the two objects. If you double-click on the second location, then StressCheck
will compute the data along the line. If you click for a third time at a different loca-
tion, a new line is created between the second and third object. You can create any

Extraction points
FIGURE 94 Evaluation of Seq at selected points along a line.
arbitrary path until you double-click on the last pick. Along each line segment
the requested function(s) will be computed in the number of specified points.
To use the option Object: Sets, give the name of a previously defined set. A set
can be defined by selecting the Sets tab in the Input dialog window.
One or more functions can be selected from the Function(s) box. To select
more than one function hold the Ctrl key while you click on the function name.
Average extraction A capability is available to compute the average of functions along element
edges, element faces, element volumes and arbitrary curves. The average is

understood in the integral sense, and when appropriate, the user controls the num-
ber of points used for the numerical integration. If the average is computed along an
element edge, face or volume, then the integration is performed using Gauss
quadrature, and the number of integration points is selected internally by the pro-
gram. If the average is performed along an arbitrary path which runs inside one or
more elements, then the integration is performed using the trapezoidal rule and the
user controls the number of integration points.
For example, if the average stress σy is requested along an element edge of length
a0, the program performs the following numerical integration:
a0
1
σ̃ y = -----
a0 ∫ 0
σ y ds
The arc length a0 is determined automatically based on the screen selection. If the
selected object is an element face, the integral is performed over the area of the
face. If the selected object is an element, then a volume integral is performed. The
length, area or volume of the selected object is also reported with the corresponding
average quantity.
To utilize this option, simply turn on the Average button. If a boundary is selected,
the number of points used for the trapezoidal rule of integration is entered in the #
of points field. 5
The quantities for which the average can be computed include the displacement
components in the global system, all directional strain and stress components, prin-
cipal strains and stresses. Additionally, the average can be computed for quantities
extracted in a local coordinate system and any other function that can be described
using the formula option in StressCheck. For example, an expression for “margin of
safety” (MS) for the interlaminar out-of-plane failure of composite adherents can be
written as:
σ̃ τ̃ xz
2 2
MS =  -----z  +  -----
- –1
 σ 0  τ0 
In this case, the average of the normal ( σ̃ z ) and shear ( τ̃ z ) stresses are computed
first before the expression of MS is evaluated.

Extraction in Local System You may extract engineering results from a grid of points relative to one or
more selected local coordinate systems. To extract from a local system follow
the steps below (see FIGURE 95):
Select Local System Extraction locations (+)
X range
5 Y range
Z range
FIGURE 95 Point Result Extraction from a System
• Points tab > Select > System > Selection. Pick the desired local system
symbols in the Model window. Enter the range of local coordinates
which describes the boundaries of the grid of points to be evaluated.
The local X, Y, Z limits are supplied in the fields labeled P1-Min, P1-
Max (X limits), P2-Min, P2-Max (Y limits), and P3-Min, P3-Max (Z
limits). Supply the number of midpoints to be in each direction. Click
on the Accept button.

Each point in the grid will be mapped into the corresponding element in the model,
the engineering results will be computed, and the results will be reported. Points
which lie outside the model domain will be ignored.
Note: The system combo box in the Input interface is used for deciding whether the
stress/strain/displacement selected must be interpreted in the global, local or mate-
rial direction.
Extraction on a Boundary You may extract engineering results from a sequence or grid of points relative to
one or more selected boundaries (curves or surfaces). To extract on a boundary, do
the following (see FIGURE 96):
• Points tab > Select > Any Boundary > Selection. Pick the desired curve/sur-
face(s) with the mouse cursor by clicking left mouse button. Enter the range
of the U,V parameters (P1 Min, P1 Max), (P2 Min, P2 Max) from which
you wish to extract by generating points with uniform spacing. Click on the
Accept button. Each point will be computed, and if a point lies within a
finite element domain, the engineering function(s) selected will be evalu-
ated and reported.

Properties
Selected surface
Extraction locations
P1 range
P2 range
5
FIGURE 96 Point Result Extraction from a Surface
Properties
StressCheck provides convenient means for checking the distortion, and the
area or volume of the model after the finite element solution was obtained.
Select the Properties tab in the Results dialog window and the Input area will
appear as shown in FIGURE 97. Select the solution for which you wish to
compute data. Select the run number or the range of runs for which data are to
be computed. Select the property you are interested in, and click the Accept
button.

Properties
5
FIGURE 97 Results interface for computing Properties.
Turn on the Distortion toggle switch if you want to know the minimum and maxi-
mum angles of the element(s). This feature is specially useful when performing
geometric nonlinear analysis because it provides feedback on the degree of distor-
tion of the elements in the deformed configuration.
Turn on the Area/Volume toggle switch if you are interested in the area (or volume)
of the undeformed model and its centroidal location. By turning on the Deformed
Area/Volume toggle switch you will get the area (or volume) and the centroidal
location of the deformed body.

Fracture Mechanics
Fracture Mechanics
StressCheck has an advanced method implemented for the computation of
Mode 1 and Mode 2 stress intensity factors in linear elastic fracture mechanics
in two- and three-dimensions. It also has the capability of computing the J-inte-
gral along a circular path around the crack tip for two-dimensional problems.
The J-integral can be computed for linear and elastic-plastic solutions of iso-
tropic materials and for linear solutions of orthotropic materials. For elastic-
plastic solutions, the integration path should be selected in such a way that it
does not cut through the plastic zone around the crack tip.
StressCheck also has an advanced method implemented for the computation of

“generalized stress intensity factors” (GSIFs) for linear elastostatic problems
and “generalized flux intensity factors” (GFIFs) for steady state heat transfer
problems in two-dimensions. The GFIFs/GSIFs are computed as a post-solu-
tion operation on the finite element solution. The method is applicable to iso-
tropic materials, anisotropic materials and multi-material interfaces, and for
cases where the singularities are characterized by complex eigenpairs. The
method is also available in three-dimensions for edge singularities in heat
transfer and elasticity.
To compute any of the fracture parameters, select the Fracture tab (FIGURE
98) in the Results window. Click on the solution name in the listbox and enter
5 the range of run numbers for which you wish to compute the values.
To compute the stress intensity factors (SIF) or the J-integral, select the corre-
sponding method in the combo box and then specify the radius of the circle
representing the integration path. To do this, turn on the Radius toggle switch
and specify a value such that the circle remains inside the solution domain. For
planar problems, the integration path must only intersect interelement bound-
aries which are radial with respect to the crack tip. Then click on the crack tip
node. Alternatively, you may turn off the switch in front of the radius, bring the
cursor to the node on the crack tip (edge of crack front in 3D), click the left
button of the mouse, and drag the cursor. A circle will appear which will grow
in size as you are dragging the cursor. Once the circle is of the desired size,
click the left button of the mouse to perform the computation.
To extract the GFIFs/GSIFs, enter the radius of the extraction circle, the num-
ber of terms to be computed (# terms), and, if the solution involves a thermal
loading, also provide the first eigenvalue of the heat transfer problem in the
EV-th field. Then select the singular point (edge) by pointing to it with the cur-

Fracture Mechanics
5
FIGURE 98 Fracture Mechanics interface.
sor and clicking the left button of the mouse. The results are displayed in tabular
and/or graphical format depending on the display selection.
The influence of the size of the extraction circle on the results is generally very
small if proper meshing is utilized. For details refer to the Advanced Topics Guide.

Extraction using the Formula and Calculator options
The Formula option

User-specified formulas can be used to combine the StressCheck functions into
a mathematical expression. The formula can be used for plotting, computing
resultants, computing data points or computing minimal or maximal values. To
illustrate the use of a formula, let us consider the linkage arm shown in FIG-
URE 86. Suppose we are interested in computing an expression based on the
tangential stress distributions around the inner hole which has been loaded by a
normal traction:
σ
MS = 1 – -----t
σ0
where σt is the tangential stress and σ0 is some constant reference value

defined by the parameter S0. Select the Points tab from the Results dialog win-
dow and complete the Points extraction input. Select the Function: Fmla, the
name of the formula (MS as shown in FIGURE 99), and the name of the local
system of the circle.
5 Note: The name of the coordinate system for the inner circle can be found
selecting Display/ Objects from the Main Menu Bar. The list of system is also
available in the system combo box of the point extraction interface. When a
system is selected from the list, it is highlighted in the model window.
To define the formula specify the formula name and the coordinate system
option in which the formula is defined and complete as shown below.
1 - Sy/S0
Then click on the Accept button of the Formula Box, and then click on the
Accept button in the Results interface. Note that the solution variable name
used to indicate the tangential stress is Sy, because a local cylindrical system
was selected. The results of the computation of the formula MS for 120 equally
spaced points (not counting the end points) along the circle will be displayed in
the Graph window as shown in FIGURE 99.
You can extract results from only one formula at a time.

5
FIGURE 99 St stress distribution along inner hole.
The Calculator option

The function calculator can be used to combine solutions in an arbitrary way. When
you select Calc. from the Function list, the dialog box shown in FIGURE 100 will
be displayed. This box allows you to combine the results from any of the available
solutions in the database. This option is especially useful for combining solutions
corresponding to different load cases, linear and nonlinear solutions, etc. To com-
plete the calculator box, simply supply the names of solutions ID’s you want to
combine, the corresponding run numbers, and an algebraic expression you wish to
evaluate. The formula field uses the StressCheck functions with the operand num-
ber of the corresponding Solution ID.

FIGURE 100 Calculator input window interface.
The entries shown in FIGURE 100 indicate that we are interested in the differ-
ence between 2 times the X-displacement corresponding to SOLA, Run 8 and
5 the X-displacement corresponding to SOLB, Run 7.
When combining principal strains or stresses, a warning message will be

issued by the program. Combination of principal quantities may not be valid
because a direction is involved. Combination of directional quantities (dis-
placement, strains or stresses) are generally valid. The only exception is when
combining solutions with different constraints.
Expressions supplied in the results extraction calculator are limited to 80 char-

acters. If you require an expression longer than this limit, you may define your
expression in the Formula interface, then reference this formula from the cal-
culator expression field by prefixing the formula name with the pipe symbol
(“|”). The expression defined in the formula interface should follow the same
rules as if it were defined directly in the calculator interface. For example, you
should refer to engineering quantities in the format “sx_1”. The formula inter-
face will issue an error message indicating that “Parameter "SX_1" not defined
for formula.” But this error message should be ignored.

The Failure Criteria option
Basic equations To analyze damage initiation in the polymer of a composite material and in bonded
composite joints, strain invariant failure criteria have been proposed in the litera-
ture. It has been shown that damage initiation can be correlated with the volumetric
and equivalent strains of the lamina which are functions of the strain invariants.
The strain invariants depend on the three principal strains and include the effects of
mechanical loads and environmental contributions.
The volumetric strain, the equivalent strain, and related quantities are defined as
follows:
Volumetric Strain = J 1 + J 2 + J 3
2 2 2
ε eq = 0.5 [ ( ε 1 – ε 2 ) + ( ε 2 – ε 3 ) + ( ε 3 – ε 1 ) ]
(EQ 5)
where εeq is the equivalent strain, ε1, ε2, ε3 are the principal strains, and Ji are the
invariants of the strain tensor:
J1 = ε1 + ε2 + ε3
J2 = ε1 ε2 + ε2 ε3 + ε3 ε1
5
J3 = ε1 ε2 ε3
(EQ 6)
Often the strains due to the combined effect of mechanical and thermal loads must
be computed, but these loads cannot be applied simultaneously to the component
because different constraint conditions are required for each type of load. A proce-
dure to combine these solutions is then required to compute the quantities of inter-
est.
The principal strains and strain invariants are computed from the directional strains
obtained from the finite element solution. To properly account for the effects of
mechanical and thermal loads, separate finite element solutions are typically
required because the constraint conditions for the mechanical loads may be differ-
ent from those for the thermal loads. Therefore, the procedure implemented in
StressCheck allows the user to compute strains by combining any two solutions
with the same or different number of degrees of freedom. Additionally, a facility
was added for incorporating arbitrary multiplying or additive factors for each solu-
tion and a global additive factor.

Let {ε}a represent the strain tensor at a point obtained from the finite element
solution corresponding to the applied mechanical loads, and let {ε}b be the
strain tensor at the same point obtained from the finite element solution corre-
sponding to the thermal loads. One of four combined strain tensor {ε} can be
extracted, using user-provided factors as follows:
{ε} = α { ε }a + β { ε }b + { δ }
{ ε } = { α } + { ε }a + { β } + { ε }b + { δ }
{ε} = α { ε }a + { β } + { ε }b + { δ }
{ ε } = { α } + { ε }a + β { ε }b + { δ }
(EQ 7)
where αi, βi, δi, i=1, 2,..., 6 are the user-provided factors:
T
 α xx 
 
 α yy 
 
T  α zz 
α = {α} =   (EQ 8)
 α xy 
 
5  α yz 
 
 α zx 
Other functions can be computed based on the combined strain tensor, such as
the equivalent strain, the strain invariants, and formula-based expressions uti-
lizing the combined strain tensor components. The combined strain tensor con-
tains six directional components according to the following convention:
 εx   εr   ε 11 
     
 εy   εθ   ε 22 
     
 εz   εz   ε 33 
{ε} =  , or  , or   (EQ 9)
 γ xy   γ rθ   γ 12 
     
 γ yz   γ θz   γ 23 
     
 γ zx   γ zr   γ 31 

where x, y, z, represent the strain directions in a global or local Cartesian coordinate

system; r, θ, z are the strain directions in a cylindrical coordinate system; and 1-2-3
refer to the strain directions in the material coordinate system (applicable to ortho-
tropic and laminate materials). Which one of these strain tensors is computed
depends on the coordinate system selection in the Results interface.
Interface The Failure Criteria option of the Calculator function is accessible from the Plot,
Points and Min/Max extraction as shown in FIGURE 101. This interface is acti-
Solution activation Extraction functions

Solution ID selection
Operation selectors Input factors

System selection Input subtab
FIGURE 101 Point Extraction and Failure Criteria Interfaces (Input

subtab).

vated whenever the Calc. function is selected from any of these Results tabs.
The Failure Criteria tab under the Input subtab, contains the following fields:
Solution ID activation buttons: They can be turned on and off to activate or

deactivate the contribution from a particular solution to the requested strain
function. At least one button must be on.
Solution ID and Run #: Selection of the solution ID and run number from the
set of existing finite element solutions that will be used to compute the com-
bined strain-related quantities. If zero is entered as a run number, then it is pos-
sible to get convergence results based on the run settings in the corresponding
Results tab (Points or Min/Max).
Extraction functions: List of strains and strain-related functions that can be

computed by combining the selected solutions with the user-provided factors.
They include all six directional strain components (εx, εy, εz, γxy, γyz, γzx), the
three principal strains (ε1, ε2, ε3), the equivalent strain (εeq), the strain invari-
ants (J1, J2, J3) and a formula option (Fmla).
When the formula option is selected, then the formula field is enabled and the
user can enter an expression in the corresponding field. Any expression con-
taining intrinsic functions and logical relations among the listed functions can
be entered. The formula expression can be up to 119 characters long. For a
complete list of intrinsic functions supported by StressCheck, refer to Chapter
5 2.
When computing directional strains, they can be in the global system, in a local
Cartesian or cylindrical system, or in the material directions, depending on the
system selection in the Results interface (see FIGURE 101). It is understood
that if a listed strain component is requested in a local system, the contributions
from SOL1 and SOL2 are computed in the same system.
If the Average button is turned on in the Points Results tab, then the selected
functions will be computed as average (in the integral sense) over the selected
edge, face or element.
By toggling the Et/Em button in any of the Results interfaces, the selected
strain function can be computed based on total (Et) or mechanical (Em) strains.
The number of functions that can be selected depends on the Results tab. From
the Plot and Min/Max tabs, only one function at the time can be selected for

extraction. From the Points tab any number of functions can be selected at the same
time.
Factors and operation selectors: Input fields to enter the factors to be used for
each solution in the context of Eq. 7. The factors can be constants or previously
defined parameter names. The toggle switch under each column controls whether
the factors are multiplied or added to the strains computed from each solution. Add
or Multiply applies for α and β factors only; the δ factors are always additive. The
sample expression located under the Add/Multiply buttons is adjusted based on the
setting of the operation selectors to help the user understand the use of the factors as
indicated by Eq. 7.
A total of 18 factors can be input in three columns. The first and second columns
are associated with SOL1 and SOL2, while the third column is always additive.
Each factor is added or multiplied to the strain component listed on the left.
Settings subtab: The Settings subtab located at the bottom of the form is used for
saving/retrieving the input factors α, and β and δ in the current StressCheck session
(database) and in the StressCheck input file. The interface is shown in FIGURE
102. To save a set of input factors, simply enter the name in the corresponding field
and click on the Save button. To retrieve a previously saved setting, simply click on
the down arrow key and select from the list of available names. The Delete button
can be used to remove the saved input associated with the selected name. The Purge
button will remove all the saved settings from the database. 5
File subtab: The File subtab is used for saving/retrieving the input factors α, β and
δ in a neutral (ASCII) file with the extension FAC. The interface is shown in FIG-
URE 103. To save a set of input factors to a file, type the name of the file in the
input field and click on the Write button. The file will be placed in the same direc-
tory as the current StressCheck session. To select a different directory, click on the
Browse button first to select the location for the file. To retrieve a previously saved
file, click on the Browse button to select the file and then click on the Read button.
Results The combined strains are reported in Chart/Table form when the extraction is per-
formed from the Points and Min/Max Results tab or in contour fringes when
extracted from the Plot tab. Each function extracted from the Failure Criteria tab of
the Calculator interface is identified by the prefix FC as shown in FIGURE 104.
The figure shows the Chart and Table output from the Points extraction interface in
which the Failure Criteria functions Ex (εx), Eq (εeq) and J1 (J1) were requested at
several points. Note that each function has the prefix FC- before the name in the
legend of the graph and in the column heading of the table. The solution ID and run
# included in the titles are the ones selected in the Points interface.

XY Graphing
FIGURE 102 Failure Criteria Interface (Settings subtab).
5
The same prefix is used to identify the selected function from the Plot Results
tab as indicated in FIGURE 105 In this case, the Failure Criteria function Ex is
identified in the legend as FC-Ex. The solution ID and run # included in the
legend are the ones selected in the Plot interface.
XY Graphing
StressCheck provides an integrated XY graphing function for viewing com-
puted data. A graph of computed data is automatically produced by all Results
classes except Plot Solution. A graph may also be displayed to illustrate non-
linear material properties and nonlinear radiation boundary condition coeffi-
cients. The graphic display is automatically formatted for each type of data, but
it is sometimes desirable to change the format from the default.

XY Graphing
FIGURE 103 Failure Criteria Interface (File subtab).
A graph will appear in its own Multiple Document Interface (MDI) child window
as illustrated in FIGURE 106. When the graph window has focus, the Main Menu
Bar and pulldown menus corresponding to the graph will replace the default menus
of the Main Model Window. Each time you change focus between the Main Model
window and the graph window, the Main Menu will become that associated with
the window currently having focus.
At the top of the graph window is a table containing one column for each data vari-
able represented, and a row for each value of each column. The check boxes at the
top of each column will enable or disable the visibility of each data column in the

XY Graphing
FIGURE 104 Graph and Table results for the failure criteria functions.

XY Graphing
FIGURE 105 Plot results for a failure criteria function.

XY Graphing
Tabular data
Independent variable
FIGURE 106 Graph window.
5
graph. The column which is highlighted is the independent variable represent-
ing the X-axis of the graph. The independent variable may be changed in the
View > Indep. Variable... menu option.
Data shown in the table may be edited by double-clicking in the table cell and
entering a new value. The graph will automatically be updated.
You may read the data value for any point on the graph by positioning the
mouse cursor over the data point of interest. A cross-hair will appear over the
data point and the X and Y values of the data point will be displayed at the bot-
tom of the graph.
Some extraction operations, especially those performed from the Handbook

interface, may generate more than one graph. When more than one graph is
available in the graph window, a series of tabs will appear between the table
and the graph. Selecting a tab will automatically change the graph to reflect
the data associated with the selected tab.

XY Graphing
The graph area of the window has two tabs. The Chart tab displays the graphic
depiction of the tabular data. The graphic depiction may be copied to the clipboard
using Ctrl-C, and pasted to a compatible application using Ctrl-V. The Table tab
displays a textual representation of the tabular data. The textual representation may
be copied using the mouse cursor to highlight text, and the Ctrl-C command to copy
and Ctrl-V to paste the text to a StressCheck Report window or other compatible
Windows application. The right mouse button will also activate a pop-up menu in
the Table tab to obtain access to the Copy/Paste functions.
When in the Chart tab, clicking the right mouse button will activate the property
sheet shown in FIGURE 107, which contains many advanced chart formatting
options. On-line documentation for the advanced property sheet is available by
clicking the HELP button.
FIGURE 107 Chart Control Properties.
Graph File menu

The File menu contains options for transferring data to and from the graph window.
Open...: Use Open to open an existing data file (.scg) containing data saved using
the Save option.

XY Graphing
Save...: Use this option to save the current graph data to an external ASCII file
with the extension .scg. This file may be loaded back into the graph window
using the File > Open option.
Save Image...: Use this option to save an image of the current graph in a JPEG
format file (.jpg). JPEG is a standard format used by many graphic utilities and
word processing programs.
Print...: Use this option to send an image of the current graph directly to a
printer. A standard Windows print dialog will appear so that you may select a
printer.
Graph Edit menu

The Edit pulldown menu contains options to copy an image of the graph to the
Windows clipboard, to copy the tabular data to the Windows clipboard, and to
clone the graph in another graph window.
Once the graph/table is deposited on the clipboard, it may be pasted into a

StressCheck report window or into any other Windows application which
accepts graphic/tabular data from the clipboard, using the Paste command
(ctrl-v).
5 The clone operation will create another graph window containing a graph iden-
tical to the original. The cloned graph can be used to preserve a particular set of
data so that the original graph window may be used to display a new set of data
from a subsequent extraction operation.
Graph View menu

The View pulldown menu contains options to customize the appearance of the
graph. These options are described below.
Symbols: Use this option to enable and disable the visibility of symbols for
each datapoint in the graph.
Legend: Use this option to enable and disable the visibility of the legend.
Grid: Use this option to enable and disable the visibility of the data table
appearing at the top of the graph window.

Setting convergence criteria
Indep. Variable: Use this option to display the dialog window in FIGURE 108.
Check the variable name corresponding to the table column that you wish to assign
to the independent variable of the graph.
FIGURE 108 Changing the independent

variable of the graph.
Properties...: Use this option to display the property sheet shown in FIGURE 109
which controls graph labeling and axis specific label and scale attributes.
(a) Labels tab (b) Axis tab
FIGURE 109 Graph property: Labels tab.

The Settings tab in the Results dialog window is used for defining convergence cri-
teria records that will be used in the context of the SOLVE! interface. These con-
vergence criteria records identify extraction operations which will be performed at
the completion of each finite element solution in order to test whether the solution
is of sufficient quality to be used as the basis for engineering decisions. Each indi-

vidual convergence record is composed of the name of one or more extraction

records and a corresponding numeric value which represents the level of error
considered acceptable for the given extraction.
Example For example, you might choose to perform a global energy error estimate at the
completion of each solution in a sequence of increasing p-level solutions.
Before you can define your convergence criteria in the Results > Settings Tab,
you must first define an extraction operation and save the definition of the
extraction record in the Settings tab of the specific extraction class as shown
for example in FIGURE 110. Note that the brief/verbose button is used in
conection with the handbook to produce a brief or extended report of the corre-
sponding extraction setting.
Supply solution ID and run limits
5 Choose error Estimate
Supply name for extraction record

Settings tab
FIGURE 110 Definition of Error Estimate Settings Record

After having defined extraction records in the various extraction class interfaces,
you may go to the Results > Settings tab to define your convergence criteria
records. FIGURE 111 illustrates the definition of a convergence criteria record.
List of convergence criteria already defined.
Supply Name and Title
Available extraction records
Enter convergence criteria (%) for each extraction
Click Name button to remove extraction from definition
FIGURE 111 Definition of Solution Convergence

Criteria
To define a convergence criteria record, first supply a name and title. Next, select
from the list of available extraction records such as the one described in the previ-
ous paragraph. If you make a mistake in the selection of an extraction record, just
click the button which appears in the table at the bottom of the interface which con-
tains the name of the extraction record, and it will be removed from the list.
If you wish to edit a previously defined convergence criteria record, just select the
corresponding record in the list box at the top of the interface, and the definition
will appear in the table at the bottom of the interface. Make your changes, then
click the Replace button.

You can now make use of this convergence record when you prepare to per-
form a solution. FIGURE 112 illustrates the selection of the convergence
record name in the SOLVE! interface.
Choose convergence criteria record
FIGURE 112 Selection of Convergence Criteria Record
The solution sequence will be controlled by the convergence criteria. For

5 example, if you had chosen to solve your problem for a sequence of polyno-
mial levels from 1 to 8, with the chosen convergence criteria, StressCheck will
solve the problem for plevels 1, 2 and 3, then perform each extraction defined
in the convergence record. Next, the estimated error for each extraction will be
compared to the criterion defined for that extraction. If the criterion is
exceeded, the solution sequence will continue for the next polynomial order. If
the criterion is met, the next extraction will be performed. If the criteria for all
extractions defined in the convergence record are satisfied, the solution proce-
dure will terminate.

6 The Handbook
Framework
The Handbook Framework in StressCheck is a simple, yet powerful environment for solving analysis
problems encountered in routine and variant design. The user interface to the Handbook Library makes
it easy to create new entries in the library and to access and use handbook models once they have been
constructed.
Handbook Library
Handbook Facilities 6
The Handbook Framework provides a Model Information interface used during

model construction to supply descriptive information for new handbook entries
and to capture and edit a model icon; a Handbook Browser to assist in the search
for available models; a handbook Analysis interface used for analyzing a user
selected model with user specified design dimensions, using pre-defined solution
methods and post-processing procedures; a Design Study interface used for
performing quick "What-if?" studies; a Results interface used for performing basic
post-processing operations such as error estimation, contour plotting, and point
function extraction in a simplified setting; a Material interface to modify existing
User’s Guide Chapter 6: The Handbook Framework 193

Handbook Library
material properties; and a Constraint interface to modify existing constraints

conditions.
Handbook Folders
The Handbook Library is essentially a directory structure that organizes
StressCheck models. Each folder in the directory structure represents a section
of the library. Each section can be further divided into folders to provide easy
organization of model files. The StressCheck installation contains a library of
sample handbook models which are intended to illustrate the kinds of
problems that can be constructed and placed in a handbook to be solved by a
typical design engineer. Most problems found in the sample library have been
defined in parametric form, though this is not a requirement. Handbook
models may be used in a production environment where dimensions will be
modified, load magnitudes adjusted, or material coefficients changed in order
to evaluate the engineering characteristics of a particular design. Also, models
may be entered into a handbook library simply to capture a static component
design. In this way, the handbook library serves as a repository for design
knowledge for future reference.
Note on handbook problems

The Handbook library has been designed to serve more than one purpose.
Besides combining handbook-type problems with the flexibility of finite ele-
ment analysis, it also serves as a verification manual and as an educational
tool.
The verification manual aspect is realized when solving problems for which
6 exact solutions are available or there is convincing experimental or numerical
information available in the literature. One of the key characteristics of some
problems in the Handbook library is that the mesh and the selection of the
extension process (p-discretization) were carefully designed to provide nearly
optimal rate of convergence in energy norm for the default value of the param-
eters. The following handbook problems fall in that group:
• Planar Elasticity: Plate with circular hole, Plate with U-notch in bend-
ing, Central crack in tension panel.
• Axisymmetric Elasticity: Externally cracked bar in tension, Internally
cracked bar in tension.
• Plate Elasticity: Bending rectangular plate: 4-simple.
194 Chapter 6: The Handbook Framework User’s Guide

Handbook Interface
• Plate Modal: Rectangular plates, all boundary conditions, first six natural
frequencies.
For parameter values other than the default values, the standard quality assessment
tools available in the program are essential for guaranteeing the quality of the solu-
tion, so that proper correlation with theoretical or experimental results can be made.
The educational aspect is twofold: First, for new users of finite element methods,
the handbook problems provide a simple way of exploring the essentials without
being distracted by details such as mesh generation, specification of material prop-
erties, loading and constraint assignments, etc. Once a problem is selected from the
list, it is ready for execution and extraction of the engineering data from the finite
element solutions, as will be demonstrated with an example. Second, the handbook
problems are excellent sources of well-defined problems, with correctly designed
meshes and properly specified boundary conditions, so that they can serve as exam-
ples for data preparation for new problems. Once a problem is loaded, it can be
modified to accommodate differences between the actual problem at hand and the
handbook problem and then saved as a input file for future use, or incorporated into
a Handbook as a new entry.
Note on the accuracy of handbook solutions

Most engineering handbooks are collections of solutions obtained before computers
became available. Some simple problems have closed-form solutions. For these
problems the exact solutions are available. The majority of the problems do not
have closed form solutions however, and therefore had to be solved by numerical
means, using Taylor or Fourier series approximations, for example. The solutions
for these problems in the engineering handbooks are approximations. In many
cases the p-version will yield more accurate results, substantiated by convergence 6
information. Such information is not available in handbooks. Some handbooks give
widely varying answers for the same problem (see, for example, Ref. [36]). Also,
the possibility of typographical errors exist, in fact some have been found when
comparisons were made with finite element results supported by convergence data.
Handbook Interface
The user interface to the Handbook library has been divided into two separate
components: the Model Information interface, and the Handbook Analysis

Handbook Interface
interface. The Model Information interface is used primarily during model

construction in order to define model specific information such as a title,
description, browser keywords, parameter definitions and rules, and model
icon. The Handbook Analysis interface is where a specific handbook model is
analyzed, using pre-defined model specific solution and post-processing
procedures. The following sections describe the entry of model information
which make it possible to use a StressCheck model as a Handbook model.
Later sections will describe how to use the Handbook Analysis interface to
solve and post-process a handbook model.
Model Information
FIGURE 113 illustrates the three tabs of the Model Information interface:
Model Info, Parameters, and Rules. The Model Info tab contains basic
information describing the current model.
Browser icon
FIGURE 113 Model Info interface.

6
Model Title You may supply a model title that will appear in the graph and report
documents produced by StressCheck during post-processing.
Model Comments The model comments is intended to provide instructions to a future handbook
model user. You may supply information describing default material
properties, a description of modeling assumptions, or perhaps a reference to
the source for analytical solutions to the model problem. To perform a
carriage-return (i.e. drop down to next line) click “Enter” key while holding
down the “Ctrl” key.

Handbook Interface
Info Button The handbook interface provides a facility for displaying documentation pertaining
to each individual handbook model via the Info button. The Info button will appear
in the handbook interface anytime a handbook model has an accompanying file
with a .pdf extension. This file might contain instructions for executing the model,
an explanation of modeling assumptions, sample graphic images, or any other
information which may be helpful to the handbook model user. For example, the
Parts Handbook contains two models (bracket and pipe_int3) that have
accompanying .pdf files. When clicking the Info button, the Adobe Acrobat reader
will automatically be loaded and the corresponding .pdf file will be displayed. A
.pdf file may be created using Microsoft Word, or any other document preparation
software with .pdf export capability.
Browser Keywords The browser keywords can help identify the current handbook model for future
users of the Handbook library. You may supply any number of keywords.
Model Browser The large icon button in the upper left corner of the Model Info tab activates the
StressCheck handbook browser. The browser shown in FIGURE 114 and FIGURE
View Model List, Icons, Details Refresh List
Directory
6
structure
FIGURE 114 Handbook Browser file view.

Handbook Interface
115 is an Explorer-like file browser that helps a user locate and preview Stress-
Check handbook models in a file list or icon view.
FIGURE 115 Handbook Browser icon view.
Browser Tree At the left side of the browser is a directory tree that assists in the navigation
through the Handbook library directory structure. To expand the directory
tree, click the plus (+) sign next to a folder icon. To collapse a branch of the
6 directory tree, click the minus (-) sign next to a folder icon. To display the
contents of a particular handbook folder, simply click the folder icon in the
directory tree to open it. The viewer to the right of the browser interface will
display either a file list or an array of model icons corresponding to each
StressCheck input file found in the currently selected handbook folder.
Browser Views The handbook browser viewer can be displayed in six different formats: large
icons, small icons, file list, file details, models (default), or model image array.
FIGURE 114 shows the interface for the model view option, while FIGURE
115 shows the interface for the model image array option. The Views arrow at
the top of the viewer provides the switch between the display formats.

Handbook Interface
When in model image array format, by right clicking on the desired entry in the
viewer and choosing Properties, a model dialog window containing additional
information for the currently highlighted model will be displayed (FIGURE 116).
FIGURE 116 Model Details window.
6
Model Selection To select a handbook model for use in StressCheck, double click the
corresponding entry in the browser viewer. This will load the model into
StressCheck, and dismiss the browser.
Model Icon The three buttons to the right of the Browser Icon in the Model Info interface may
be used for capturing, editing, and displaying an icon for the current StressCheck
model. When you click Capture Icon, a window will appear asking whether you
want to capture a standard 200x200 pixel icon. If you answer YES, a rectangle will
appear that you should position over the model window, otherwise a cross-hair
will appear so that you may outline the area of the screen to be captured for use as

Handbook Interface
an icon. The captured image is automatically deposited on the Windows

clipboard, and may be transferred to any other Windows application using the
clipboard. The icon image is also saved as a JPEG file with the same name as
the current database, with the extension .jpg. This image becomes the icon
associated with the current model and will appear if you click the Show Icon
button.
When you click Edit Icon, a bitmap editing interface will appear and the
currently captured model icon will automatically be loaded into the editor.
(See FIGURE 117). The icon editor provides standard bitmap editing
FIGURE 117 Model Icon editor.
operations such as line drawing, text labeling, erasure, etc. You may save your
new icon in either bitmap (.bmp) or JPEG (.jpg) format. Both formats are
recognized by the Handbook Browser. JPEG is the default, and is usually the
preferred format since it is more compact and thus requires less space for

Handbook Interface
image storage. If using SaveAs to save in bitmap format, be sure to use the Resize
Icon option to save with 256 colors or less. Images modified in the editor may also
be saved to the Windows clipboard for transfer to other applications.
Model Parameters The Parameters tab of the Model Information interface shown in FIGURE 118
provides for the definition of parameters used in the construction of a StressCheck
model. The parameter is assigned a name, a description, a default value, a limit
test, and a class. Optionally, instead of a default value, the parameter may be
assigned an expression which references other parameters in order to obtain a
value. To modify a parameter value, simply enter the new value in the
corresponding input field and press the TAB key to advance to the next field.
Information entered into the Parameters table is not saved in the current database
until the Accept button is clicked, or unless the Auto Step check box is checked
and an increment/decrement arrow is clicked. At this time, the limits and
parameter rules are checked for validity before the new information is actually
saved in the database.
Input tab
6
FIGURE 118 Model Parameters.
The parameter name may be a string of up to 15 characters, must begin with an

alphabetic character, and must not contain any arithmetic or special symbols.
Parameters which begin with an underscore (_) are considered special parameters
to be used in StressCheck to activate certain advanced features. The following
rules apply to parameter names:

Handbook Interface
• Parameter name should not be a coordinate name (X, Y, Z, R, T, P).

• Parameter name should not be E followed by a number, e.g.: E1, E2, ...
• Parameter name should not be C followed by a number, e.g.: C1, C2,
..., C10.
• Parameter name must not begin with a digit, or contain a special sym-
bol (! # $ % & ; : , . ? + - * / ‘ @ { } ( ) [ ] = ~ ^ | \ ”).
• Parameter name must be distinct from intrinsic functions (SIN, COS,
TAN, ATAN, DEG, RAD, SQRT, SINH, COSH, TANH, ASIN, ACOS,
ATAN, LOG10, LOG, EXP, ABS, ATAN2, MOD, MIN, MAX, IF,
PFIT).
The parameter description may be a string of up to 39 characters. This

description will appear in the Handbook analysis interface to assist the
handbook user in understanding the use of the parameter, and usually some
information about the valid range of values or the relationship with other
parameters.
The parameter expression may be up to 119 characters in length, and may be

any valid StressCheck expression as defined in Chapter 2, for example "sin(h)
+ cos(s) - pi".
The parameter value is a double precision number, and may be incremented

by the value of the step by clicking the up or down arrow next to the value.
The limit expression may be up to 39 characters in length. It may be a simple

relation as shown in FIGURE 118, or may be a series of simple expressions
such as ">0, <1.0".
A parameter can have a class assigned to it. The options are: General
6 (default), Property and Boundary Condition. The classification is used in the
Handbook Analysis interface to control the list of parameters presented to the
handbook user.
Parameter Settings The current parameter values may be saved as a “configuration” record in the
database by switching to the SETTINGS tab, and clicking the Save button
shown in FIGURE 119. This mechanism is also available in the Handbook
analysis interface and is useful for preserving various configurations of a
model that have been evaluated and may need to be recalled later for further
study. The FILE tab has a similar function, except that it preserves the
parameter information in an ASCII file, providing for the transfer of
parameter information between StressCheck databases.

Handbook Interface
Settings tab
FIGURE 119 Parameter Settings.
Parameter Rules The Rules interface provides a mechanism for enforcing valid relationships among
parameters. FIGURE 120 illustrates the format of a parameter rule and its
corresponding error message.
FIGURE 120 Model Rules.
The combined length of the relational expression and the corresponding error
message must be less than 119 characters. If the relational expression evaluates to
a FALSE result, then the defined error message will be deposited in the error log

Handbook Interface
file and an error dialog window will appear on the screen. When errors occur,
the most recent valid parameter values will be restored in the interface, and
the model will be preserved in its most recent valid state.
Handbook Model Preparation

When constructing a new model which is to be inserted into a handbook, it is
important to plan ahead. A good handbook problem should provide
parameters which will control the critical dimensions of the model problem,
and if appropriate, parameters should be provided to give the user control over
other critical design variables such as thickness, material coefficients, and
load magnitude. It is important to design the finite element mesh so that the
elements will not be overly distorted over a wide range of parameter settings.
This is usually the most important part of designing handbook models. It is
also important to determine the maximum and minimum value of each
parameter so that an appropriate parameter range may be assigned which will
maintain the integrity of the mesh. Also, plan to provide rules which describe
relationships among parameters which should be maintained to protect mesh
integrity. The following guidelines should help you construct useful handbook
problems:
• Provide parameters for critical dimensions and design variables.
• Design the mesh to be valid over a wide range of parameter settings.
• Determine minimum and maximum values for each parameter.
• Provide rules which describe relationships among critical design
parameters.
With these guidelines in mind, the usual sequence in the preparation of a

handbook model is as follows:
6
1. Enter Model Information including Title, Description, and Keywords.
2. Define Parameters to be used during the construction of the model.
3. Construct the model geometry, mesh, material, boundary condition, and
solution records, using parameters previously defined.
4. Define parameter rules that will enforce valid parameter value and
relationships.
5. Test model for proper simulation of the design problem and modify if
necessary.

Handbook Analysis
6. Capture and/or compose model icon.

7. Preserve solution settings for use in Handbook Analysis (SOLVE! button).
8. Perform post-processing operations and preserve input settings for use in
Handbook Analysis.
9. Save model input file(s) in desired handbook folder. When performing a File >
Save Input operation, the input file (.sci) and icon file (.jpg or .bmp) are both
automatically saved to the designated directory folder.
Many of these steps are described in detail in other chapters of this manual and so
will not be repeated here.
Handbook Analysis
Once a handbook model has been prepared as described in the previous section, and
saved in a handbook library file folder, it may be retrieved for use in the Handbook
Analysis interface shown in FIGURE 121. The Model Info tab contains the same
information as in the Model Information interface, but the title, comments and key-
word fields are read-only. As in the Model Information interface, the large icon in
the upper left corner will activate the model browser described in the previous sec-
tion. The Handbook user can browse the library of available models to locate the
model of interest. The model is loaded into StressCheck by double clicking the icon
or view listing. Once loaded, the model appears in the main display window with a
graphic depiction of the default boundary conditions. Other attributes such as mate-
rial, thickness, etc. may be enabled using the various attribute icons found in the
Main Toolbar. The icon capture buttons operate in a similar fashion to the corre-
sponding buttons in the Model Information interface. This capture facility is the
6
same as the one activated by the “camera” icon in the main toolbar. You may cap-
ture any portion of the screen, and this image will automatically be deposited on the
Windows clipboard. The image may also be edited using the bitmap edit facility,
and saved in bitmap or JPEG format.
Automatic Analysis
To perform a handbook analysis, switch to the Input sub-tab of the Analysis tab as
shown in FIGURE 122. The Input interface is intended to be used by a typical

Handbook Analysis
Library Icon (main toolbar)
Operation Tabs
Browser Icon
(Model Info tab)
FIGURE 121 Handbook Analysis model information.
design engineer who wishes to evaluate some variation of the basic part model
in order to determine the performance of the modified component.
Parameter This interface provides access to the model parameters so that the values may
6 be revised, and the model updated to reflect the new settings. These model
parameters (or design variables) may represent model dimensions, material
coefficients, boundary conditions, or any other aspect of the model that the
author of the entry deemed appropriate. To modify a parameter value, simply
enter the new value in the corresponding input field. You may also use the “=”
key to enter an expression that will be evaluated immediately and the result
deposited in the field. Refer to the model icon provided for a visual indication
of the meaning of each parameter.
Update The Update button will validate the input using the parameter limits and rules
before applying the new design values to the model. The Auto Step checkbox
may be activated so that parameter changes will be immediately applied to the

Handbook Analysis
FIGURE 122 Handbook Analysis AutoRun interface.
model when the increment/decrement arrows are clicked. If any parameter values
violate their predefined limits, an error message appears and all parameter values
will be returned to their previous valid settings. The Solve and Plot buttons are
then used to activate pre-defined model specific procedures to automatically solve
and post process the model. These pre-defined procedures must be provided by the
author of the handbook model and are stored as part of the model input.
Solve The Solve button activates a predefined model specific solution procedure which
may include any combination of linear, nonlinear, modal, buckling, coldworking
or limit load analyses. A message will appear when all solutions are complete. The 6
arrow next to the Solve button provides the options of generating a: Report with
Plots, a Report without Plots, or No Report. The Report will produce tabular
information (and graph) of model specific extraction results pre-defined by the
handbook model author.
Plot When solutions are complete, the Plot button will produce one or more model
specific graphic display(s) selected specifically for the model, which might depict
a deformed shape, stress distribution, principal stress direction arrows, etc. When
more than one set of plot settings are provided, the graphic results are display by
repeatedly clicking the Plot button. An specific graphic display can be chosen from
the selection given when clicking on the arrow next to the Plot button.

Handbook Analysis
Example problem
After opening a database, activate the Handbook Analysis interface by choos-
ing the Handbook Library icon from the Main Toolbar. When the Handbook
interface appears, click on the Browser icon of the Model Info tab to access the
Model Browser. Choose the 2D-Basic handbook, and double click on the
peter02.sci file. The problem, “Plate with central hole”, will be loaded into
StressCheck. See FIGURE 123.
FIGURE 123 Analysis tab, model icon and mesh for a handbook problem.
This problem represents a panel of length L, width W and thickness h, subject

to uniform traction S and with a central hole of diameter a. FIGURE 123
shows the Analysis tab of the Handbook interface with the icon and parameters
for the selected problem and the finite element mesh consisting of 5 quadrilat-
eral elements, with the loading and constraint symbols. The objective is to

Handbook Analysis
compute the stress concentration factor Ktg for the following value of the parame-
ters: a=1.0, h=1.0, L=15.0, S=1.0 and W=5.0.
Taking advantage of symmetry, only one-fourth of the domain was discretized. The
material properties assigned to this problem are those of an ASTM A-36 carbon
steel (E=29.0x106, v=0.295). The stresses and hence the stress concentration factor,
are independent of the material properties, however. Plane stress was assumed.
The execution parameters have been assigned so that a downward p-extension

(from p=8 to p=1) is initiated in the automatic mode after clicking on the Solve but-
ton.
After the solution is completed, a Report summarizing the results of the analysis
will come out. For this particular problem, it contains a summary of the parametric
values, an estimate of the global error of the solution (FIGURE 124) and a sum-
mary of the location and value of the maximum first principal stress in the model.
FIGURE 124 Example Handbook report: Estimate of global error of solution.

6
To compute the stress concentration factor Ktg, we simply extract the maximum
principal stress S1 and since the applied remote stress (S) is unity, we have:
S1
K tg = ------ = S 1
S
A reference value for this problem can be found in Stress Concentration Factors by
R. E. Peterson, John Wiley and Sons, Inc., 1974, Fig. 86 (for a/W=0.2, S1=3.14,
taken from graph).

Handbook Analysis
The relative difference between Peterson’s and the finite element result is:
3.155 – 3.14
e = ------------------------------ × 100 = 0.475 %.
3.155
Note that most handbook data were obtained by approximate methods which
are less accurate than the procedures implemented in StressCheck.
Exercise Consider the problem shown in FIGURE 125 which corresponds to a bolt head
in tension with a washer support. The bolt is made of malleable cast iron
ASTM A-47 (E=24x106 psi, v=0.29), and the washer material is an aluminum
alloy with Ew=10.1x106 psi. The support provided by the washer can be mod-
eled as a spring constraint with the spring coefficient given by the ratio
between the modulus of elasticity and the thickness of the washer. Determine
the magnitude and location of maximum first principal stress.
F=5000 lb
0.75
0.075 1.5
washer
6 0.125
0.5
1.5
FIGURE 125 Bolt with washer support. All dimensions in inches.

Handbook Analysis
You can load this problem from the Parts handbook by double clicking on the
bolt.sci file. The problem will be loaded into StressCheck. Modify the parameters
to match the provided information. (Answer: σ1max = 33940 psi)
Handbook Analysis Design Study

Choose the Design Study subtab of the Analysis tab if you wish to perform a “What
If?” type of analysis for the current handbook model. When performing a design
study, you control which parameters(s) will remain constant, and which parame-
ters(s) will vary during the analysis. You also control how many steps will be per-
formed during which the variable parameters will be increased from their minimum
to their maximum values.
The Design Study interface (FIGURE 126) provides access to the definition of each
parameter defined for the model. Each parameter may be either Constant or Vari-
able. When a parameter is constant, its value remains constant for each step of the
design study. The value of each variable parameter will change during the Design
study. To make a parameter variable, simply check the box at the left and then enter
the initial and final values in the corresponding fields. To make a parameter con-
stant, un-check the box.
FIGURE 126 Design Study interface.
# Steps You may supply the number of steps to perform during the parametric analysis. The
number supplied will be used to determine the value of the scale, which in turn is

Handbook Analysis
used to compute the value of each variable parameter. Step 1 corresponds to

the initial value of the variable parameters.
Scale The current value of each variable parameter is determined by the Scale value
(S) as show below:
a = a init + (a final - a init)S
You may preview the parametric configurations of the model by activating the
scale (enable the Scale check box), and clicking the up or down arrows to
increase or decrease the scale value. The Scale value will vary from 0.0 to 1.0
in increments of 1/(Steps-1).
p-level During a design study, the assignment of p-levels to the elements is held con-
stant. The p-level you enter will be assigned to all elements which have been
designated as variable in the definition of the model. All elements designated
as having fixed p-level will retain their assigned value.
Model Applicable only for Plate Bending, refers to the model order for the analysis.
Refer to the Advanced Guide for further details.
Solve button When you are ready to begin the design study, simply click the Solve button.
The model will be updated automatically and the resulting configuration will
appear in the model window. The solution for each design configuration will
be saved for subsequent post-processing.
Report button After a design study has been performed, a standard report is generated once
you click on this button.
6 Parameter Settings As described in the previous section which discussed the Model Information
interface, parameter settings maybe preserved for later recall. Select the
Settings tab, assign a Name to the parameter configuration, and click the Save
button to record the settings in the database. To recall a previously saved
configuration, just select it from the combo box. To save parameter settings to
an ASCII file, use the File tab.

Handbook Results
Handbook Results
When solving a handbook model, the default plot(s) and report(s) are produced.
However, you may require more information than is provided by these default pro-
cedures. In this case you may wish to use the Handbook Results interface to obtain
additional engineering results.
The Results tab in the Handbook interface provides a simplified version of the stan-
dard post-processing features found in the Results Class interface. Typical default
values have been assumed for input values which have been excluded from the
Handbook Results interface.
The Results tab (FIGURE 127) interface provides options for computing an esti-
mate of the error in energy norm, plotting standard engineering quantities, comput-
ing minimum and maximum engineering quantities, computing engineering data at
selected locations in the model, computing resultants, computing fracture mechan-
ics quantities, or computing various engineering properties such as deformed area/
volume or distortion. If you need to manipulate attributes of your engineering
results not provided for in the Handbook Results interface, you may use the stan-
dard View Results interface by selecting the View Results icon from the Main Tool-
bar.
FIGURE 127 Handbook Results interface.
All Handbook Results interfaces share some common controls. The left side of the
dialog window is the same in each Results interface. The Solution ID and Run

Handbook Results
number(s) will automatically be filled into the fields on the right side of the
window.
If appropriate, select model objects such as elements, nodes, points or bound-

aries where the computation is to be made. Use the combo-box controls at the
left of the interface to choose the Action > Object > Method to be used in per-
forming the computation. Use the mouse cursor to point to the object of inter-
est in the model window and click the left mouse button to select it.
For all options except PLOT, your results will be presented in the Graph win-
dow.
Error When computing an error estimate, the selections described above are suffi-
cient to perform the desired computation. The result is the estimated relative
error in energy norm for the sequence of solutions selected. At least three solu-
tions are required to perform an error estimate. Refer to FIGURE 127.
Plot When producing a plot of engineering data, you may choose from a variety of
options (FIGURE 128). First, using the Object combo-box control, you may
choose to plot the data for all elements or only a subset. For 3D models you
may plot only on selected faces or on faces which lie on a particular surface.
FIGURE 128 Handbook Results interface - Plot tab.
• You may choose whether or not to produce a fringe representation to

indicate the value of the engineering function.

Handbook Results
• You may select the engineering quantity of interest.

• You may choose whether to display the undeformed shape or the deformed
shape of the object, to use a special Z-shape display of engineering results
for 2D models, or to display arrows depicting the direction of a principal
stress or strain.
• You may control the resolution of the computation by specifying the num-
ber of element midsides used for constructing the data computation grid.
• You may choose whether or not to average the data at shared vertices, edges
and faces in order to smooth out the contours.
When you have made all of your selections, click on the Accept button to perform
the computation. Remember, one important indicator of solution quality is the
degree of continuity of engineering data across element boundaries. Averaging the
data can mask the quality of the engineering result.
Points In the Points tab (FIGURE 129), you may compute engineering quantities at
selected points or along selected boundaries or arbitrary paths through the model.
Use the Object and Method controls to choose objects used in performing the com-
putation. Next, select the engineering function you wish to compute at the chosen
locations in the model. Finally, if appropriate, indicate the number of evenly spaced
points on each object where you wish to perform the computation. Click on the
Accept button to perform the computation. Use the auxiliary variable input field if
FIGURE 129 Handbook Results interface - Points tab.
you want to include a variable parameter in the results. Turn on the Aux. toggle if
you want the auxiliary variable to be the independent variable of the graph.

Handbook Results
Min/Max In the Min/Max tab (FIGURE 130), you may search for minimum and maxi-
mum values of the selected engineering function. Search the entire model, or
only selected boundaries, edges, faces, or elements. Specify the resolution of
the data grid used for computing the selected engineering quantity. Click on the
Accept button to perform the computation.
FIGURE 130 Handbook Results interface - Min/Max tab.
Fracture In the Fracture tab (FIGURE 131),choose the appropriate fracture method,
FIGURE 131 Handbook Results interface - Fracture tab.
then supply the Radius of the integration circle to be used in the computation,

Handbook Results
point to the crack tip node or edge and click on the left mouse button. If you have
supplied a fixed radius of integration, the results will appear immediately other-
wise, drag the mouse cursor to display the radius of integration then click again to
obtain the results. Properties
In the Properties tab (FIGURE 132), after identifying elements to be included in the
computation, select the desired properties to be computed (Distortion, Area/Vol-
ume, etc.). Click on the Accept button to perform the computation.
FIGURE 132 Handbook Results interface - Properties tab.
Resultant In the Resultant tab (FIGURE 133) choose the appropriate output functions
depending on the reference and analysis type. For Planar Elasticity for example, the
appropriate functions are the resultant forces in the global X and Y-directions (Fx, 6
Fy), and the moment about the Z-axis (Mz). The results will appear in the Graph
window for the selected functions and also in the summation fields of the window
input area. To select an output function simply click on the corresponding force or
moment switch. To reset the summation fields of the window click on the Cancel
button.

Handbook Material
FIGURE 133 Handbook Results interface - Resultant tab.
Handbook Material
The Handbook interface contains a tab labelled Material shown in FIGURE
134. This interface provides access to material property definitions for isotro-
pic, orthotropic and laminate composites. A handbook user may modify exist-
ing material properties only. In order to add new materials or assign materials
differently, it is necessary to use the Material class of the Input interface.
FIGURE 134 Handbook Material interface.

Handbook Constraint
If laminate properties were assigned to the model, then it is possible to change the
ply stuck-up assignment by selecting the Laminate tab as shown in FIGURE 135.
This tab will appear in the Handbook interface only if laminate properties were
assigned.
FIGURE 135 Handbook Laminate interface.
Handbook Constraint
The Handbook interface contains a tab labelled Constraint shown in FIGURE 136.
This interface provides the handbook user a quick way to modify existing con-
straints of the type: Built-in, Symmetry, Anti-symmetry, Soft-simple and free only.
In order to add new constraints, or to modify general constraints, it is necessary to
use the Constraint class of the Input interface.
6
FIGURE 136 Handbook Constraint interface.

Constructing a handbook model

The main ideas for creating a handbook problem will be illustrated with an
example. This problem can be loaded from the Tutorial Handbook (file:
NotchedPlate2D.sci), but we will go through the steps of how it was created
first.
Problem description Consider the case of the notched cantilever plate in bending shown in FIGURE
137. This problem is characterized by seven parameters: The plate length (L),
width (W), and thickness (th), the support width (b) and height (h), the fillet
radius (ro) and the applied load per unit length (Ps).
Ps
th h
ro
L b
FIGURE 137 Notched plate geometry.
6
We will consider a two-dimensional case (plane-strain), and assume that the
line load is uniformly distributed along the edge of the plate. Of interest is to
compute the maximum first principal stress and to ascertain that the computed
value has converged within 5% relative error. The material is plain carbon steel
with a modulus of elasticity E=200GPa and Poisson’s ratio v=0.3. The steps to
create a parametric model (finite element mesh, material properties, loading,
constraints, and execution parameters) for this problems are described in the
following. Make sure that the analysis type is set to Planar Elasticity.

Model Information Make sure the analysis type is set to Planar Elasticity and the units are mm/N/sec/C.
From the Main Toolbar, select the Model Info icon and when the corresponding
dialog window appears (FIGURE 138) enter the Title you want to give to the hand-
book problem, the Comments describing characteristics of the model you think
would be important to a user, and the Keywords.
Model Info icon
FIGURE 138 Model Information: Model Info tab.
Create Parameters Select the Parameters tab (FIGURE 139) from the Model Information dialog win-
dow to enter the values as follows:
• Name: b, Description: Support width, Value: 40.0, Limit: >0.0
• Name: h, Description: Support height, Value: 60.0, Limit: >0.0
• Name: L, Description: Plate length, Value: 100.0, Limit: >0.0
6
• Name: Ps, Description: Line load in N/mm, Value: 6.0 > Class: B. Cond.
• Name: ro, Description: Fillet radius, Value: 6.0, Limit: >0.0
• Name: th, Description: Plate thickness, Value: 20.0, Limit: >0.0
• Name: W, Description: Plate width, Value: 100.0, Limit: >0.0 > Accept
The Description field is optional when defining parameters. The Limit field when
given, indicates the limiting value(s) the parameter can take, for example the plate
thickness must be greater than zero. If there is no entry for the Limit it is assumed
that the parameter can take any value. The class general was used for all parameters
except Ps.

FIGURE 139 Model Information: Parameters tab.
To protect mesh integrity it is sometimes necessary to enforce relationships

among parameters. For this problem it is necessary to have a plate length
greater than its fillet radius. To incorporate this requirement select the Rules
tab, and type the Relational Expression (L > ro) and the Error Message (L must
be > ro) you want to display when the relationship is violated. FIGURE 140
shows the three rules defined for this particular problem.
FIGURE 140 Model Info: Rules tab.

Solution Domain Select the Create Model icon from the Main Toolbar and when the Input dialog box
appears select the Geometry tab. First, create two rectangles in parametric form
(see FIGURE 141a).
• Geometry tab > Create > Rectangle > Locate > Data tab > Input Switch on
> X: 0.0, Y: -th/2, Z: 0.0, Width: L, Height: th, Rot-Z: 0.0 > Accept.
• Create > Rectangle > Locate > Data tab > Input Switch on: X: L, Y: -h/2, Z:
0.0, Width: b, Height: h, Rot-Z: 0.0 > Accept > Click on the Center Model
icon.
The rectangles are created by first creating the corner points and then the connect-
ing lines. The next step is to create an auxiliary line to be used to locate nodes
inside the plate and two fillets:
• Create > Line > Locate > Input switch on: X: L + th/2, Y: -h/2, Z: 0.0,
Length: h, Angle: 90 > Accept.
• Create > Fillet > Selection > radius: ro. Move the mouse cursor to the dis-
play area and click on the two perpendicular lines where one of the fillets
should be located. Repeat at the other location.
Note that the program automatically locates points at the tangency points of the cir-
cle and lines. The display should look like the one shown in FIGURE 141a. The
description of the solution domain is now complete.
Mesh To create the nodes and elements for the solution domain defined above select the
Mesh tab in the Input dialog window and proceed as follows (see FIGURE 141b):
• Mesh tab > Create > Node > Point. Move the cursor to the display area and
click on the four corner points of rectangle 2, the two left points of rectan-
gle 1, and the four points at the ends of the fillets, and then click on Accept.
A total of 10 nodes should be created attached to the same number of 6
points.
• Create > Node > Intersection. Move the cursor to the display area and click
on the auxiliary line and the top line of the rectangle. Repeat for the bottom
line.
• Create > Node > Midpoint. Move the cursor to the display area and click on
the leftmost two nodes located over the vertical line defining the first rect-
angle. A node will be created between the two nodes selected. Repeat the
operation three more times by selecting the next two nodes to the right; the
two nodes on the auxiliary line and the two nodes to the right on the second
rectangle. A total of four nodes should be created by the midpoint method.

• Create > Node > Projection. Move the cursor to the display area and
click on the mid-side node created on the auxiliary line and then, while
holding the control and shift keys, click on one of the fillets and click
Accept. Repeat for the other fillet. Two nodes should be created by this
method.
click on the top node of the upper fillet and then, while holding the
control and shift keys, click on the auxiliary line and click Accept. A
node will be created on the auxiliary line as a projection of the fillet
node. Repeat the operation for the lower node of other fillet.
• Create > Quadrilateral > Selection. Move the cursor to the display area
and click on four nodes that define a quadrilateral element. A total of
10 quadrilateral elements should be created.
• Create > Triangle > Selection. Move the cursor to the display area and
click on three nodes that define a triangular element. Two triangular
elements should be created.
At this point the display should look like FIGURE 141b. The finite element
mesh for this problem is now complete. Various mesh configurations are possi-
ble for this problem. However one should keep in mind that the mesh should
perform properly for values of the parameters other than the default. Keeping
the elements around the fillet in a symmetric arrangement (note that the auxil-
iary line is positioned in such a way that the element edges bisecting the fillet
are always normal to the respective circles) minimizes the distortion of the
mesh as the values of the parameters are changed. For example, FIGURE 141c
shows the mesh when b=30.0 and th=30.0 instead of the default values of 40.0
and 20.0 respectively.
6 Thickness Select the Thickness tab in the Input dialog window and complete the follow-
ing information:
• Thickness tab > Select > All Elements > Selection > Thickness: W >
Accept
This will assign the current value of the parameter W to all elements in the
mesh. If unit thickness was already assigned to the elements, simply click on
the existing record, enter W in the thickness field and click on the Replace but-
ton. This will update the thickness definition created automatically by the pro-
gram.

rectangle 2
rectangle 1 fillets
auxiliary
line
(a) Geometry (b) Mesh for default

values of parameters
(c) Mesh for th=b=30.0
FIGURE 141 Geometry and mesh for example problem.
Material Select the Material tab in the Input dialog window and enter the material properties.
Assign the material to all the elements in the mesh. To define the material complete
the following information:
6
• Material tab > Define > Linear > Selection > Define tab > ID: MAT >
Material: Linear > Type: Isotropic > Units: SI > Case: Pl. Strain > E:
200000 > v: 0.3 > Click on the Accept button.
To assign the material ID “MAT” to all the elements in the mesh, proceed as fol-
lows:
• Material tab > Assign tab > Select > All Elements > Selection > ID: MAT >
Type: Homogeneous > Click on the Accept button.
Note that the modulus of elasticity was entered as 200000 MPa, and therefore the
stresses will be given in MPa.

Loads Select the Load tab in the Input dialog window and enter the following infor-
mation:
• Load tab > Select > Any Curve > Traction > ID: LOAD > Direction:
XY > Y:-Ps/th. Move the cursor to the display area and click on the
leftmost line of the model, then click on the Accept button. A distrib-
uted traction pointing downward will be displayed on the element
edges attached to the line.
The load line Ps was converted to a traction by dividing by the plate thickness
th. The total force applied to the plate will be the product of the line load mul-
tiplied by the width of the plate (F=Ps W=600 N). To check the total applied
load to the plate, use the “Check” action:
• Check > All Elements > Selection > ID: LOAD > Turn off the toggle
in front of Moment-X: and Moment-Y: > Accept.
The program will integrate the applied tractions and report the sum of the
forces in the global coordinate directions and the moment about the z-axis. The
result will be displayed in the report window as shown in FIGURE 142. This
feature is provided to allow users to verify the magnitude of the applied loads,
prior to performing the analysis.
FIGURE 142 Load Check report.
Constraints Select the Constraint tab from the Input dialog window and enter the following
information:
• Constraint tab > Select > Any Curve > Built-In > ID: CONST. Move
the cursor to the display area and click on the rightmost line of the
model then click on the Accept button. The symbols of the built-in

(clamped) constraint will be displayed on the element edges attached to the

line.
FIGURE 143 shows the load and constraint symbols on the mesh.
Load symbols
Constraint
symbols
FIGURE 143 Boundary conditions for the example problem.
Solution ID Select the Solution ID tab from the Input dialog window and enter the following
information:
• Solution ID tab > Define > Name > Selection > Solutions tab > Solution
ID: SOL > Constraint ID: CONST > Load ID: LOAD > Click on the
Accept button.
By this action, a unique solution name (SOL) has been assigned to the constraint 6
and load name pair created previously.
p-Discretization Select the p-Discretization tab and enter the following information:
• p-Discretization > Select > All Elements > Selection > Space: Trunk > p-
Discretization: Variable > Level: 1 > Accept.
Trunk space and variable polynomial order (p-level) starting at p=1 has been
assigned to all elements in the mesh. The data input is now complete.

Before adding the problem to the Handbook set, the analysis should be exe-
cuted. This will allow you to verify that the data have been properly input and
the quality of the solution is acceptable.
Model icon At this point you can capture or compose the model icon. Click the Capture
Icon in the Model Information dialog window, and when a window appears
asking whether you want to capture a standard 200x200 pixel icon, answer
YES. You may need to first scale down the model so that it will fit in the
capture rectangle. The icon image is saved as a JPEG file with the same name
as the current database, with the extension .jpg. This image becomes the
current icon associated with the current model and will appear if you click the
Show Icon button. You can select Edit Icon... to add labels to the capture as
shown.
Problem Solution From the Main Toolbar, select the Compute Solution icon. Select the Linear
tab and enter the following information:
• Extension: Upward-p > p-limits: 1 to 8.
6
Select the SOLVE! tab and choose the following options:
• Execute: Initialize > Run Mode: Automatic > Method: Iterative > Con-
vergence: None > Click on the Solve button.
To preserve the solution settings for future use by the handbook user, select the
Settings tab. When the dialog window appears, give a name to the solution (1-
Linear), press the “Enable for use in Handbook” icon, and click on the Save
button. See FIGURE 144.

Enable for use in Handbook
FIGURE 144 Solution settings for handbook user.
Note: When saving Solution or Results settings, the configuration will be saved in
alphanumeric order by the name assigned to the configuration. This is also the
order that the configurations will be retrieved and executed when activating the
Solve button. For this reason, it is recommended to use digits for the initial charac-
ters of the configuration name to control the order of execution.
Results After the solution is completed, perform the Error Estimation to assess the overall
quality of the solution, and verify the convergence of the maximum value of the
first principal stress. 6
Error Estimation: Select the View Results icon from the Main Toolbar, then
choose the Error tab and enter the following information:
• Error > Select > All Elements > Selection > Input tab > Solution: SOL >
Run:1 to 8 > Click on the Accept button.
The estimated relative error in energy norm for the default values of the parameters
is 2.13% at p=8 (see FIGURE 145).
Select the Settings Tab, assign a title and name to the post-processing operation,
press the “Enable for use in Handbook” icon and the “Enable to use report in Hand-
book” icon, and click on the Save button to record the settings in the database.

FIGURE 145 Error estimation for example problem.
Convergence of σ1: Select the Min/Max tab in the Results dialog window and
provide the following information:
• Min/Max > Select > All Elements > Line > Input tab > Solution: SOL
> Run: 1 to 8 > Function: S1 > Midsides: 20 > Maximum button on >
Click on Accept.
To record the settings in the database, select the Settings tab, assign a title and
name to the post-processing operation, press the “Enable for use in Handbook”
icon, the “Enable to use graph in Handbook” icon, and the “Enable to use
report in Handbook” icon, and click on the Save button (see FIGURE 146).
6
The maximum value of the first principal stress computed using 20 midsides is
S1=12.55 MPa at p=8 with a difference of 2.10% with respect to the estimated
limit value. FIGURE 147 shows the convergence of the magnitude and loca-
tion of the maximum principal stress as a function of the number of degrees of
freedom (DOF). As expected, the maximum principal stress is at the top fillet.
The contour plot of the first principal stress shown in FIGURE 147 was
obtained selecting the Plot tab and completing the required information as fol-
lows:

FIGURE 146 Min/Max interface: Settings tab.
• Plot > Select > All Elements > Selection > Solution: SOL > Run: 8 > Plot:
Solution > Contour: Fringe > Shape: Undef. > Func.: S1 > Midsides: 20 >
Click on the Plot button.
To record the settings in the database, select the Settings Tab, assign a title and 6
name to the post-processing operation, press the “Enable for use in Handbook”
icon, and click on the Save button.
To write the input file of the current model, select File from the Main Menu Bar and
when the File pulldown menu appears select Save Input. Enter a File name in the
corresponding input field. Remember that an icon file (*.jpg) is also saved in the
same pace as the StressCheck input file (*.sci). Place these files in a directory that
will serve as a handbook directory, so anyone can have access to this problem from
the Handbook interface.

FIGURE 147 Convergence of the maximum principal stress.

MODELING GUIDE

StressCheck
Volume 2
MODELING GUIDE
Release 7
August, 2005

Copyright 2005
COPYRIGHT NOTICE
U.S.A.

rights reserved.

DISCLAIMER

or change.
1
1 Introduction
This manual is intended to provide step by step instruction in the use of various construction techniques
to build the geometric components of a finite element model. It is recommended that before consulting
this guide, the reader should be thoroughly familiar with the material in the User’s Guide, especially
those sections pertaining to the User Interface Conventions.
For a discussion of topics of a more general nature such as the construction of a complete finite element
model, please consult the other manuals of the StressCheck documentation set.
Geometric Modeling in StressCheck

The construction of 2D and 3D geometry in StressCheck is based on the Parasolid
geometric modeling kernel from Unigraphics Solutions.
While there are a few new construction methods for point and curve objects, most
of the Parasolid functionality is related to the construction of solid geometry, and
the various operations used for creating three dimensional solid bodies using bool-
ean and blend operations. This functionality fits quite easily into the familiar
Action/Object/Method command paradigm, and all objects may be created in para-
metric form. The implementation supports the construction of:
• Solid Primitives: Box, Sphere, Cylinder, Cone, Torus
Modeling Guide Chapter 1: Introduction 1

Terminology
1
• Solid Booleans: Union, Subtraction, Intersection
• Clip Operations: Front and Back clip
• Blend Operations: Rolling ball, Chamfer, Variable radius
Release 7 automatically trims the surfaces affected by boolean operations, and

automatically constructs all intersection curves and points of intersection for
use in construction of the finite element mesh.
Not only can you construct solid geometry in StressCheck, it is also possible to
import solid models that were created by third party Computer Aided Design
tools such as CATIA. It is also possible to import files which were exported
from a CAD system using the IGES standard. Although imported geometry is
essentially a static snapshot of a part, it may be modified within StressCheck
through the addition of new solid objects, and the application of boolean or
blend operations. For more detailed discussion of CAD model importation,
refer to the StressCheck User’s Guide.
Terminology
There are a few important terms and concepts that should be understood before
reading the following sections.
• Point - A one dimensional geometric object defined primarily in terms
of an X,Y,Z location in 2D or 3D space. For example, a point, node, or
local coordinate system.
• Curve - A two dimensional geometric object defined in terms of a sin-
gle variable in 2D or 3D space. For example, a line, circle, ellipse,
spline curve, or surface curve.
• Surface - A three dimensional geometric object defined in terms of two
variables in 3D space. A surface object has no volume or thickness.
For example, a plane, sphere, cylinder, cone, torus, tabulated cylinder,
surface of revolution, spline surface, etc.
• Solid Body - A three dimensional topological object representing a
volume bounded by a collection of bounding faces (trimmed surfaces)
edges (trimmed curves) and vertices (points).
• Sheet Body - A three dimensional topological object composed of a
single face (trimmed surface), edges (trimmed curves) and vertices
(points). The sheet body does not represent a bounded volume. NOTE:
2 Chapter 1: Introduction Modeling Guide

Terminology
1
In StressCheck, when we talk about creating a surface, we really mean a
sheet body.
• Wire Body - A two dimensional topological object composed of one or
more trimmed curves. NOTE: In StressCheck, when we talk about creating
a curve, we really mean a wire body.
• General Body - A three dimensional topological object composed of one or
more faces (trimmed surfaces), edges (trimmed curves) and vertices
(points). The general body does not represent a bounded volume. A general
body is created by applying a boolean or face blending operation to one or
more sheet bodies.
• Boolean operation - The union, subtraction, or intersection of one or more
solid or sheet bodies. If all objects which are components of the boolean
operation are solids, the result is a solid topology. If one or more of the
components are sheet bodies, the result is considered to be a general body
(not necessarily a solid)
• Blend operation - The application of a fillet (rolling ball, chamfer, or vari-
able radius) to a solid body.
• Body - A topological entity which is either a primitive or the result of any
boolean, blend, or clipping operation. A body may be a solid, a sheet, or a
general topology.
• Target - In a boolean, clip or blend operation, the first object selected is
considered to be the target, and is generally the object which remains after
the operation is complete.
• Tool - In a boolean, clip or blend operation, the objects selected after the
target which do not survive after the operation is complete.
• Implicit associativity - The result of a boolean or blend operation is a col-
lection of faces (trimmed surfaces), edges (trimmed curves) and vertices
(points). The edges represent the curves of intersection between neighbor-
ing faces. The vertices represent the points of intersection between the
coincident edges. Since these relationships are automatically created by the
various modeling operations, their relationship (associativity) is considered
to be implicit. The user is free to make changes to solid objects without
regard for the resulting changes in implicit associative relationships.
• Explicit associativity - It is sometimes necessary to create new explicit
associative relationships in a model. For example, attaching a point as an
offset, a midoffset, or a projection to a surface or curve, or as an intersec-
tion of 2 or more curves. This explicit associativity will prevent the future
application of a boolean or blend operation to a solid body. As a conse-
quence, explicit associative relationships should not be added to a body
Modeling Guide Chapter 1: Introduction 3

Important Limitations
1
unless it is certain that no future boolean or blend operations will be
necessary for the corresponding body (i.e. its faces, edges or vertices)
Important Limitations
Some basic rules and limitations must be kept in mind while constructing sol-
ids-based finite element models in StressCheck.
• StressCheck is not intended to be used as a general purpose CAD sys-
tem. It is a solids-based analysis system.
• There is no solid modeling history tree maintained in StressCheck.
This means that you cannot edit a previous operation to change the
type of solid modeling operation that was performed. For example, if
you create a box, a cylinder and a sphere, then union the box with the
sphere, then subtract the cylinder, you may not edit the history to
change the union operation to a subtraction operation. To accomplish
this, you must delete all objects and operations that were performed
after the step that you wish to change, make the change, then manually
retrace your modeling steps. However, you may edit dimension infor-
mation for objects in your model, as long as the modification does not
change the topological structure of the model.
• You may not perform a boolean or blend operation on an object which
has explicit associative relationships. For example, you may not attach
a point as an offset to a surface which is part of a solid body, then
attempt to apply a blend to the same body. The blend operation effec-
tively deletes the original definition of the body and replaces it with the
blended representation. The deletion is not possible if there is an
explicitly associated object.
4 Chapter 1: Introduction Modeling Guide

2 Basic Geometry
2
Construction
This chapter contains a general overview of the features of StressCheck used for creating geometric
objects. For a discussion of creating individual geometric and mesh objects, please refer to the chapters
which follow this one.

With the adoption of Parasolid as the geometric modeling kernel, it is necessary to
consider the choice of units of measurement before creating any geometry in
StressCheck. Since Parasolid limits the definition of any geometric object to fit
within a cube that is 1000 by 1000 units, it is necessary to introduce a system of
units so that the objects represented in Parasolid will be properly scaled to fit within
this cube. You may think of this cube as being measured in meters, and that the
largest object that can be created in Parasolid as being one that is 1000 by 1000
meters.
Length units are now chosen as part of a standard system of units. The choices are
Other, inch/lbf/second/F, and millimeter/Newton/second/C. The "Other" option
indicates that the units are unknown. In all cases, it is the user's responsibility to
insure that units are given consistently throughout the program.
Modeling Guide Chapter 2: Basic Geometry Construction 5

Geometry Class
When StressCheck starts up, the default units are in/lbf/sec/F. The default units
may be chosen as part of the preference settings. When loading in an existing
StressCheck model file, the units will be assumed to be “Other” unless the .sci
file contains a specific record assigning specific units.
2 To manually select your system of units, use the Edit > Units menu option or
change the Units Selector combo box. You may change the system of units
only if no geometry has yet been defined. If you have an existing model that is
not Parasolid based (no CAD geometry was imported), you may insert a record
into the .sci file that will assign the units. The units selection will not affect the
values used to define the model. The record appears as shown below (see
"*13"). The other two options are: MM_N_SEC_C and OTHER.
!STRESSCHECK Version: V6.0.30b - Professional Edition
!---------------------< Batch File: aunits >----------------------
! Date: 01/06/08 Time: 14:48:39
!-------------------------------------------------------------------------------
!
*5
*13
INCH_LBF_SECOND_F
*10000
No comment
*6
0
*7
Dummy
Geometry Class
From the Class menu, select Geometry (FIGURE 1). The set of options under
Geometry provide for the definition of the solution domain using points, lines,
circles, ellipses, rectangles, etc. StressCheck lets you separate the definition of
boundaries from the definition of the finite element mesh. Separation of geo-
metric objects from mesh objects provides for a great deal of flexibility in
6 Chapter 2: Basic Geometry Construction Modeling Guide

Geometry Class
modeling. Geometric objects may be defined parametrically so that the domain

may be easily changed by adjusting individual design variables. Since boundary
conditions may be attached to geometric boundaries, the finite element mesh may
be easily changed without affecting boundary condition definitions.
In the following sections, various techniques will be described for the construction 2
and revision of basic geometric objects.
FIGURE 1 Geometry Input.
Simple object creation

Suppose we wish to construct a geometric point. The most straightforward method
would be to supply the X, Y, and Z coordinates manually, and press the Accept but-

Geometry Class
ton in the Input dialog window. The point will appear in the model window at
the appropriate location. Try this.
Class: Geometry
Action: Create
2 Object: Point
Method: Locate
X: 1.0
Y: 2.0
Z: 0.0
Buttons: Accept
You should now see a point represented in the display area by a mark similar to
+. Try changing the value in the X: input field and pressing Accept again. Now
change the Y: value and press Accept again. You should now see three points
in the display area. If you don’t, it may be because the coordinates you entered
were outside the current range of the display area. Click the Center Model
icon , this should bring all points into view.
Now, suppose you made an error in the coordinate input of the last point. The
simplest way to change this point would be to press the DeLast button to delete
the last object created, then revise the input coordinate which was incorrect and
press Accept again. Try this.
The DeLast button can be very useful to retrace several creation operations.
But, be aware that DeLast is NOT the same as an UNDO operation. It only
deletes the last object created. Once an object is deleted, it can be recovered
using Undo. Repeating DeLast will continue to delete objects in the reverse
order of creation until all objects have been deleted.
StressCheck also provides an Undo operation. Undo reverses the outcome of

the previous object or data operation. If an object was created, Undo will
remove the object. If an object was deleted, Undo will recover the deleted
object. If an object was moved, the object will be returned to its original loca-
tion. Undo does not have any effect on display operations such as rotation,
translation, selection, cancellation, etc. Note: If your database contains data
from a finite element solution, this data will be destroyed by an Undo opera-
tion.

Geometry Class
After an Undo operation, the Redo operation will reverse the effect of the Undo.
The Undo can be repeated to retrace all object or data related operations back to the
start of the current StressCheck session. Redo can be used to reverse all Undo oper-
ations. As soon as a new operation is requested other than an Undo, the possibility
of a Redo is eliminated.
2
Another way to remove all objects from your model is to use the File > Erase
Database option in the Main Menu Bar. There are other methods for revising
objects, but these will be discussed later on.
FIGURE 2 Screen and direct input.
It is often useful to construct objects using a mixture of screen and manual input.
FIGURE 2 illustrates the use of graphic and text input. To begin, you must first
select the Create action, the Point object, and the Locate method. By turning off the
X: input toggle, the X coordinate of the point will be taken from the location of the
graphic cursor pick, while the Y and Z coordinates will be taken directly from the
specified keystroke input. If you move the mouse and place the graphic cursor in
the display area, then click the left mouse button, the X-coordinate corresponding
to the location of the screen will be computed by the program, and the Y and Z
coordinates will be taken from the manual input area to locate the new point. If you
move the mouse to a new location and click again, the same Y and Z coordinates
will be used, with a new X coordinate computed from the new location of the
mouse cursor. The result will be a series of points along a horizontal line at Y = 2.0

Geometry Class
and Z = 0.0. To display the point labels, select Display > Objects from the
Main Menu and enable the corresponding Points Label toggle button.
Similarly, if you now turn on the X: toggle and turn off the Y: toggle, then click
at various locations in the display area, you will see that a series of points are
2 created along a vertical line which has the same X coordinate as the last point
created earlier. The last X coordinate was preserved in the X: input field when
you turned the toggle on. You may note that each time you click the mouse but-
ton in a new location on the screen, a new Y coordinate appears in the Y: input
field. This feedback mechanism is an important concept to understand. It can
be used to obtain coordinate information from existing objects, and subse-
quently be used to create a new object.
The X:, Y: and Z: input toggles may be simultaneously activated and deacti-
vated by toggling the Input check box. When the Input toggle is on, the pro-
gram will automatically manage the control of the input toggles when you
change actions, objects or methods. When the Input toggle is turned off, the
label changes to Screen and all input buttons are turned off, indicating that
input will be taken from graphic cursor input. You are free to change individual
input toggles manually at any time to obtain the desired combination of
graphic and manual input.
Repeated object creation

When objects are to be created at regular intervals, it can be helpful to make
use of the Repeat function in the Input dialog window (Repeat # =). FIGURE 3
illustrates the input data settings required to create the three circles shown in
the display at the left. In this case it is possible to invoke the creation operation
by clicking the Accept button or by clicking the left mouse button in the dis-
play area. Since all coordinate toggle buttons are turned on, the coordinates
will be taken from the manual input provided for X, Y, and Z. Try this yourself
(Note: The Repeat toggle is automatically switched off when the Accept button
is clicked, but not when the left mouse button is clicked in the display area).
Also try turning off one of the coordinate input toggle buttons, then move the
mouse cursor to the display area and click the left mouse button. You will see
that the same pattern of circles will be created, but the location of the first cir-
cle in the series will depend on which input toggle button(s) are turned off. Try
this. Remember that you can use the DeLast button or Undo icon to delete the
last object(s) created. When objects are created in groups with the Repeat
option, DeLast or Undo will delete all objects which were created with a single
repeat operation. Try this too.

Geometry Class
FIGURE 3 Input by repetition.
Repeated object creation can be very useful with other creation methods also. For
example, if you wish to create points at evenly-spaced intervals around a circle, you
can use the Offset method with the Repeat option, and specify the offset interval
you wish to use (e.g. 45 degree spacing).
Parametric object creation

In some situations it is important to be able to create objects which are based on the
value of some dynamic design variable. For example, we might wish to create a cir-
cle based on a design variable for the radius. FIGURE 4 illustrates a parametric cir-
cle creation operation.
First you must create the design variable parameter d. Select Edit > Model Info
from the Main Menu Bar or select the Model Info icon from the Main Toolbar.
Click on the Parameters tab and fill in the fields as shown in the figure to create a
parameter called d, with a Value equal to 1.0. Click on the Accept button of the
window and the parameter will be created. If you want to increase or decrease it,
turn the Auto Step toggle switch on and click on the arrows next to the value. This

Geometry Class
FIGURE 4 Parametric input.
record indicates the name of each parameter currently defined and its current
value. If you wish to change any of the parameters click on its field, enter the
change and click on the Accept button. The new value will be immediately
updated as long as it does not violate the user defined rules. Alternatively, you
may delete the selected parameter by clicking on the first field or its name, then
pushing the Delete key of the keyboard. Note that you cannot delete a parame-
ter if it is currently being used.
Once you have defined the parameter d, complete the Geometry section of the
Input dialog window as shown in FIGURE 4 and press the Accept button to
create the circle. Now if you want to increase the value of d to be 2.0, return to
the Parameters tab in the Model Info dialog window, turn on the AutoStep tog-
gle switch, enter a value of 1.0 in the Auto Step field, and click on the Value
upward arrow. The value of the parameter d will be increased after each step.

Geometry Class
You can change the AutoStep value to whatever value you want. You should see on
the screen the radius of the circle change automatically.
Repetitive parametric object creation

2
It is possible to create objects which are parametric and which follow a repetitive
pattern. For example, you may wish to create a series of circles which are evenly
spaced according to the value of a parameter. FIGURE 5 illustrates the input
required to create parametric circles repetitively.
FIGURE 5 Repetitive Parametric Input.
We will again use the parameter d defined earlier to represent the distance between
their origins. This time, when we define the parametric expression, we must intro-
duce the # symbol to represent an index variable. This index variable begins with
the value 0 and is incremented by one for the number of times the object creation is
repeated. Turn on the Repeat# function and enter the number 3. In this case the
index variable will have the values 0, 1, and 2. The X coordinate of each circle will
be computed from the expression 1.0+d*#, where # has the values 0, 1, and 2. The
resulting X coordinates will thus depend on the current value of parameter d. Circle
1 will have X= 1.0, Circle 2 will have X= 1.0+d, and Circle 3 will have X=

Geometry Class
1.0+2*d. Now if you click the upward arrow in the Parameters box, the spacing
between points will be adjusted automatically.
As before, you may invoke object creation by clicking on Accept or by moving

the mouse cursor to the display area and clicking the left button. Experiment by
2 turning off the Y: input toggle and then moving the mouse cursor to the display
area and clicking the mouse button. The Y coordinate will depend on the cur-
sor location, but the X coordinate will be based on the expression entered man-
ually. If you turn on the Input Grid under the View pulldown menu, the Y
coordinate will snap to the nearest grid point.
Now that you have mastered the basic object creation mechanisms, try to cre-
ate some other objects.
General procedures for object creation

The graphical creation of some objects requires multiple cursor picks. The
number of picks required depends on the object and the input toggles which are
turned on for manual data entry. For example, if all input toggles are turned off,
a line requires two picks. Try this.
Select Create > Line > Locate and set the input mode to Screen. Position the
cursor for the first endpoint and click. Position the cursor for the second end-
point. Notice that the line is dynamically updated as you move the cursor
around on the screen. As soon as you click the left mouse button a second time,
the line is created.
Select Create > Circle > Locate and again set the input mode to Screen. Posi-
tion the cursor for the center of the circle, click, and hold. Position the cursor to
represent the circular radius. Notice that the circle is dynamically updated as
you drag the cursor on the screen. As soon as you release the left mouse button,
the circle is created. This same procedure is used for rectangles and ellipses.
Try creating lines and circles with some of the input toggle buttons turned on.
The construction technique should be intuitive. For example, try selecting Cre-
ate > Circle > Locate and turn only the radius button on. Supply a value for the
circular radius, and move the mouse cursor to the display area. You should see
a circle appear and move around on the display as you move the cursor. As
soon as you click the mouse button, the circle with the fixed radius given will
be constructed. Try other variations of input toggle buttons on and off to verify
that the construction methods work as you expect.

Geometry Class
The construction of polylines and splines requires N+1 cursor picks, where the first
N picks identify the locations of the control points, and the N+1 pick is actually a
double click on the last control point to indicate the end of the input sequence.
There are many methods for creating the various objects. One important distinction
should be made to clarify the interpretation of the methods. The Location method 2
uses only the manual input and/or the location of the cursor pick to identify coordi-
nates in the display area. All other methods require other objects on the screen to be
selected either as reference locations or to indicate an associative attachment. Asso-
ciativity will be discussed in the next section.
There are so many ways to create geometric objects that it would be impossible to
describe all possible variations. With the experience gained so far, you should be
able to follow your intuition and discover many of the techniques on your own.
Associative object creation

Associativity is an important feature of any design tool which is to be used for para-
metric analysis. Associativity is simply a geometric relationship among objects.
This relationship is maintained as the underlying objects are modified. For exam-
ple, lines may be defined associatively with respect to 2 endpoints. If one or both
endpoint locations are changed, the definition of the associated line is automatically
updated. Similarly, a circle may be associated with either a local coordinate system,
or with 3 points. If the locations of these underlying objects are changed, the defini-
tion of the circle is updated automatically.
All associative relationships are based on the order of creation of the associated
objects. It is possible to create points which are associatively attached to bound-
aries, lines parallel to lines which are normal to other lines, etc. The associative
relationships are evaluated strictly in the order of object creation. Therefore, if you
wish to change an existing object definition in such a way that it depends on
another object which was created later than itself, it may not be possible. The pro-
gram will attempt to revise the order of evaluation to accommodate the desired
change, but it may not be possible. In this case, you must delete the object you wish
to change, and recreate it. To determine object dependencies, first select the objects
of interest. Then click the “Assoc.” button. All objects that depend on the initially
selected objects will also be selected.

Geometry Class
Creation methods
The following is a summary of creation methods provided in StressCheck:
Locate - Create an object in global cartesian coordinate space.

2
Work Plane - Create an object in the local xy coordinate system of a work
plane. A work plane can be any plane surface created. Objects created using
the Work Plane method will be associatively attached to the selected plane.
The coordinate location of the new object in the work plane may be entered
FIGURE 6 Object Creation in a Working Plane.
manually in the input area, or by using the mouse to digitize points on the sur-
face of the plane. To improve visualization, select the Surface Grid icon
from the Display Options Toolbar. This will cause all surfaces to be displayed
with a surface grid. To create objects in more than one working plane at a time,
select the desired planes then define the object dimensions in the input area and
click the Accept button. If your model has several plane objects, it is some-
times helpful to “blank” the unwanted planes temporarily so that it is easier to
locate objects in the desired working plane.

Geometry Class
Local - Create an attachment between an object and a local coordinate system.

This method is useful for orienting subassemblies of geometry.
Point - Use an existing geometric point as a reference point for constructing a new
object. This method is associative only for creating a line or a node.
2
2-Point - Select 2 points to create a line.
3-Pt. Circle - Select 3 points to create a circle.
3-Pt. Arc - Select 3 points to create a circular arc.

3-Pt. Plane - Select 3 points to create a plane.
Midpoint - Select two objects (points, systems, nodes, or one of each). The new
object will be created midway between the two selected objects. If the selected
objects are moved, the location of the midpoint object will be automatically recal-
culated.
Offset - Select a boundary object and provide an offset value to locate a new object
(such as node or point) along the boundary object selected. If the boundary is a
conic, the offset is measured in degrees. If the object is not conic, or the dimension
is a linear dimension, the offset is a fraction of the linear dimension.
Mid-Offset - Select two nodes, two points or two systems that have been created
as offsets. A new object will be created midway between the two selected objects.
If the selected objects are moved, the location of the Mid-Offset object will be auto-
matically recalculated.
Projection - Select an object (point, system or node) to be projected and a bound-

ary to project the object to. The projection is the shortest distance from the first
object to the boundary. If the first object is moved or the boundary is changed, the
projection will be recomputed automatically.
Sample - Create an object by sampling existing (usually spline based) geometry.

When the user selects an object on the screen, the program will automatically sam-
ple points from each selected object and produce an “equivalent” geometric primi-
tive which may be more easily parameterized. Sampling consists of identifying the
starting and ending points, the radius, the height, etc. The new object created is not
associated with the original object in any way, so it is possible to delete the original
object if desired.

Geometry Class
The most frequent use of the Sample method of geometry creation is in con-
verting objects which were loaded into StressCheck by means of a CAD inter-
face such as IGES. Most CAD systems use spline curves and surfaces as a
standard representation of all basic geometric shapes. However, these basic
shapes cannot be easily parameterized. StressCheck provides the Sample
2 method so that a user may create an exact copy of the spline object in a stan-
dard parametric form. When creating a curve by sampling, if the base object is
an arc of a curve, the new curve will also be an arc. When creating a surface by
sampling, if the base objects are arcs or surface patches, the new surface cre-
ated will also be a surface patch.
Another use for Sample is if you wish to create a copy of an existing object, but
want to override one of its defining parameters. For example, you may have an
existing circular arc, and you want to create another circular arc with the same
center, same orientation, but with a different radius. You may create the new
circle by sampling, and override the radius input value in the input area with
the desired radius for the new circular arc. When you select the existing circu-
lar arc, a new arc with the desired radius will be created by the program. In a
similar fashion, you may override the arc range and/or the center location
(X,Y,Z).
The Sample Method may be used to create the following objects:

• Points - Select a curve or surface to create a pattern of points along a
curve or over the selected surface. The number of points created is
determined by the Repeat input value.
• Line - Select a curve. StressCheck will construct a line by sampling
the curve’s two end-points.
• Plane - Select a surface. StressCheck will construct a plane by sam-
pling three points.
• Sphere - Select one circle shaped curve. StressCheck will construct a
true sphere by sampling the center and radius.
• Cylinder - Select two circle shaped curves or one cylinder shaped sur-
face and sample the base, center, radius, and height to construct a true
cylinder or cylindrical patch.
• Cone - Select two circle shaped curves or one cone shaped surface.
StressCheck will form a true cone, frustrum, or conic patch by sam-
pling the base, center, radius, and height.
• Torus - Select three circle shaped curves that do not lie in the same
plane. StressCheck will construct a true torus or toroidal path by sam-

Geometry Class
pling the minor radius of the first circle selected, as well as the major radius
and torus center based on the circular path traced out by the three circles’
centers.
Delta - Use the selected point, system or node as a reference point and add the
specified X, Y, Z coordinate delta supplied to the base coordinate value. The result- 2
ing object definition will be associative.
Intersection - Select two boundaries. If both boundaries are curves, or if one is a

curve and one is a surface, then a point, system or node will be created at the inter-
section of the boundaries. The object created will be associative. By default, the
program assumes that two curves will be selected as the basis for determining the
location of the point of intersection. If you wish to select one surface and one curve,
hold the Shift and Ctrl keys simultaneously while picking the surface.
Note: It is possible to identify an existing object as an intersection of two other

objects even though it is not a “computed” intersection. This is sometimes useful
when a model contains all the necessary surface and curve geometry, but for some
reason an existing curve is not known internally to be an intersection. You may
want to attach nodes and elements to the curve and automatically inherit associativ-
ity with the intersecting surfaces. Instead of creating a boundary by the intersection
method, you may Move a Circle by the Intersection method. For example, a model
may contain a cylinder, a plane, and a circle which happens to lie exactly at the
intersection of the plane and the cylinder. You want the program to know that this is
the case, so you use the Move action, select the circle, then select the cylinder and
the plane. Now the program knows that any nodes or elements attached to the circle
should inherit associativity with the cylinder and the plane.
Intersect-Multi - The standard intersection method accepts only two boundary

objects as input. It is sometimes necessary to attach a point or node object to more
than two intersecting boundaries. To accomplish this, use the Intersect-Multi
method so select any number of intersecting curves. Use the mouse cursor to high-
light all desired intersecting curves, then click on the Accept button. StressCheck
will compute the location of the intersection using only the first two selected
curves, but will incorporate all selected curves into the definition of the object for
purposes of associativity evaluations. In this way, if a node is defined as an inter-
section attached to multiple curves, and if the any of the curves happen to represent
an intersection of surfaces, the node will inherit the associativity to the underlying
surfaces for purposes of element construction.
Offset Parallel - Select a line and a point to create a new line parallel (in the Z=0
plane only) to it at a given distance. If a positive distance is given the parallel line

Geometry Class
will be located at that distance on the same side as the point. If the distance is
negative it will be located on the other side of the point. The relationship is
associative.
Normal - Select a line and a point to create a new line normal (in the Z=0
2 plane only) to it. A positive or negative length will determine which side of the
original line to create the new line. The relationship is associative.
Tangent - Select a point and a circle to attach a line which lies tangent to the
circle and passes through the point. A line may also be attached tangent to two
circles. The curves must be coplanar. The relationship is associative. A node is
automatically created at the point of tangency, since this will be needed later
during mesh creation.
Reverse - Edit a boundary. The parametric range(s) for the boundary will be
reversed. This is useful when creating a ruled surface or a spline surface when
one of the 2D boundaries is parameterized in the wrong direction. The reverse
operation may be applied before or after the surface is created.
Object delete, blank, unblank

To delete an object, use the Select action to identify the object(s) to be deleted.
Next click on the Delete button at the bottom of the Geometry input dialog
window. If you delete an object by mistake, select Edit > Undo from the Main
Menu Bar or use the Undo button in the Edit Toolbar to retrieve the deleted
object.
You may also delete objects of a certain type by setting the object icon for the
type of object you wish to delete, for example circles. If any other type of
object happens to be selected, only circles will be deleted or blanked.
To blank objects (remove them from the display temporarily) use the Select
action to identify the object(s) to be blanked and then use the Blank Objects
icon in the Edit Toolbar or the Display > Selection > Blank operation. To
restore blanked objects, click on the Unblank Objects icon in the Edit Toolbar.
When attempting to delete objects from a model, it may be necessary to first

delete objects that have associative relationships that depend on the object that
you wish to delete. To determine object dependencies, first select the objects
of interest. Then click the “Assoc.” button. All objects that depend on the ini-
tially selected objects will also be selected. At this point you may perform a
Delete, Copy, or Blank operation.

Geometry Class
Geometry revision
So far, we have described how to create geometric objects. It is very important to be
able to change objects after they have been created. The Action section of the
Geometry input class contains several mechanisms for changing aspects of the
finite element model. These actions are summarized as follows: 2
Create - Use Create to construct a new object by the selected method.
Select - Use Select to identify an object for manipulation by some other command.
Each time you select an object, the location and dimension information will appear
automatically in the Geometry Input fields which have been set for screen input. If
a field has been set for manual data entry, the current manual input data will be
retained. Note that the coordinate location is always given in global coordinates
when using Select. Use the Edit action to obtain actual input definitions of an
object. See the discussion below on the use of Replace and Move with objects
which have been selected.
Note: By default, each selection operation automatically cancels the selection of

other objects. Pressing the Shift key while selecting an object will enable the selec-
tion of multiple objects. Pressing the Ctrl key will allow you to cancel a particular
selected object without affecting other selected objects. The Cancel Highlighted
Objects icon in the Edit Toolbar may be used to cancel all selected objects,
regardless of type. The Cancel Specific Object Type icon in the Edit toolbar may be
used to cancel the selection of all objects of the current object type.
Edit - Use Edit object to identify an object so that you may examine and/or revise
its definition. If the object is associative then those objects on which it depends are
also highlighted in the display area. For example, if you Edit a point defined as an
intersection, the two intersecting boundaries will be highlighted to indicate the
associativity condition. See the discussion below on the use of Replace and Move.
Note that when you select the object on the screen, the method by which the object
was originally defined will appear in the Input dialog window. You should not
change the method when using Edit. Use Move To for this purpose.
Move To - Use the Move To action to change the definition of an existing object.
This action is intended for direct manipulation of an object with the mouse cursor.
When you edit the object, it will be highlighted, and the method by which the object
was created will be highlighted. To revise the definition of the object, simply define
it as though it were a new object. You may change the method and/or supply coor-
dinate or dimension information manually in the input fields. When you are ready
to apply your changes, click the left mouse button in the Model Window, or click

Geometry Class
the Replace button. Using the Replace button will preserve any coordinate or
dimension information from the original object definition corresponding to
input fields which are turned off. If you use the cursor to activate the revision,
the appropriate coordinate, dimension, or associativity information will be
determined by the proximity of your cursor pick.
2
Check - Use Check to perform various query operations on the geometry or
mesh. Check may be used to obtain object definition information or to compute
the distance between objects. When used with coplanar lines in particular, the
distance will be computed between parallel lines, or the angle between non-
parallel lines. Check is also used in the Mesh Class to identify distorted ele-
ments and free edges.
The Replace or Move buttons can be used to change the definition of all
selected objects. If you have identified an object using the Select, Edit, or
Move To actions, you may manually change a coordinate or dimension value
in the input area using the Replace or Move buttons. Replace will change the
coordinate or dimension of the selected object to the one given. Move will
increment the coordinate or dimension by the magnitude given in the input
field. Only the coordinate or dimension values selected for manual entry will
be affected during the Replace or Move operation. For example, if you select 5
different points, then turn on only the X: coordinate input toggle and enter a
new value, a Replace operation will assign the same X coordinate to all points
whereas a Move operation will translate all selected points by the magnitude of
the value in the X: coordinate input field.
The Copy button at the bottom of the Input dialog window may be used to cre-
ate copies of objects currently selected in the graphic display area, and will
only be activated if the object is set to “Any Object”. Before actually perform-
ing the copy, the program will check to see that all objects referenced by asso-
ciative relationships are also selected. If any object referenced associatively is
not selected, the copy operation will be aborted.
All objects which are selected and which are defined in global coordinates will
be copied and attached to a new local system. All of their associative objects
will be copied and the associative relationship will be transferred to the new
copies of the related objects. The new local system created by the Copy com-
mand may be used to translate or rotate the copied objects as a group. The
coordinates and rotation angles of the new local system may be specified in the
input area prior to activating the copy operation.

Geometry Class
FIGURE 7 illustrates the use of the copy operation. In this example, a two dimen-
sional profile of a slot inside a rectangle is to be copied to a different z-plane (say
z=5). This particular geometry was created as a rectangle object, two circular arcs,
and two tangent lines. The nodes were automatically created as intersections of the
tangent lines with the circular arcs.
2
To copy an object or a group of objects, first select the objects to be copied. Select
the desired group of objects by drawing a box around them. This will cause the
objects to be highlighted. Next, enter the distance the copy is to be translated and/or
the angles to be rotated, in the corresponding input fields as shown in the figure. In
this case the translation is Z=5. Finally, click the Copy button. The copied objects
will appear non-highlighted. If you wish to delete the copy, simply click on the
DeLast button. Note that the copy has been associated with a local coordinate sys-
tem so that it is easy to orient the copied objects as a group.
Once the geometry is copied, it is easy to create a ruled surface or spline surface
from geometric curves in each profile. Note that if the original group of objects had
contained elements, the resulting copy would also have elements, which could in
turn be used to construct solid elements using the Face to Face creation method dis-
cussed later. Finally, it should be noted that it is possible to use the Repeat toggle to
create several copies of the selected group of objects in one operation. Simply enter
the number of copies to be made, and supply the translation/rotation offset in the
auxiliary fields to the right of the standard input fields.

Geometry Class
FIGURE 7 Copy operation.

3 2D Geometry
Creation
3
This chapter contains a general overview of the features of StressCheck used for creating two dimen-
sional geometry for a finite element model. For a general description of the preparation of a complete
finite element model, please refer to the User’s Guide.
Geometric construction of 2D objects

The following sections describe the construction of two-dimensional geometric
objects in StressCheck. The fundamental building blocks of 2D models are point,
system, and curve objects. In 3D we add surface and solid objects to the tool kit of
available modeling objects. Three-dimensional geometry construction will be dis-
cussed in the next chapter.
Interface conventions
Keep in mind the data entry and graphic cursor construction techniques described
in the User’s Guide. The mouse cursor is used to locate control points in the graphic
display area, and the input fields in the Geometry dialog window override the cur-
sor location for supplying coordinates and object dimensions.
Modeling Guide Chapter 3: 2D Geometry Creation 25

Geometry construction also relies heavily on the Action > Object > Method
command paradigm. In geometry construction the most important Action is of
course Create. In 2D, the following objects may be constructed: system, point,
line, circle, fillet, ellipse, polyline, spline, formula curve, composite curve,
label, and rectangle. Once an object has been selected a method must be cho-
sen. While not all methods may be appropriate for every object, each method
works in a consistent way for every corresponding object.
Coordinate System Object

Local coordinate systems are fundamental geometric objects in StressCheck.
3 Local systems give you the ability to conveniently orient objects in 2D or 3D
space. You may also attach a local system to another local system in order to
build up a complex geometric assembly. For example, imagine a robot arm
with a local system at each joint: shoulder, elbow, wrist, finger, and knuckle. If
you rotate the system at the elbow of the arm, the objects defining the forearm,
hand and fingers will rotate simultaneously.
The basic information required to define a local system is its coordinate loca-
tion and its orientation. Some construction methods reduce the amount of
information required by the user to define a system, since the X,Y, Z coordi-
nate location can be computed from an associative relationship like an offset
(e.g. 45 degree offset on a circle), intersection, projection, or other associative
relationship. Defining an associative relationship always requires a graphic
mouse selection to determine the base object(s).
FIGURE 8 illustrates a simple local coordinate system definition and its visual
representation. Note that there is a graphic embellishment of each principal
axis of the local system. Each axis is displayed in a different color.
To create a system passing through three points select Action: Create > Object:
System > Method: 3-Pt. Plane. The system is located at the position of the first
pick point, the x-axis is in the direction of the line from point 1 to point 2, and
the z-axis is normal to the plane defined by the 3 points.
Point Object
The basic definition of a point is a geometric coordinate location X,Y, Z (R, Z
in Axisymmetric analysis). Points may be created by several methods, and may
be used as the basis for other associative objects such as lines, splines, circles,
26 Chapter 3: 2D Geometry Creation Modeling Guide

FIGURE 8 Local coordinate system.
etc. Points are to be distinguished from nodes, which are element topological
objects. Points cannot be used in the definition of an element, only nodes may be
used in this way.
Line Object
A line may be created in several ways. In parametric form, the X, Y, Z location of
its starting point, its length, and orientation angle measured in degrees from the
positive global X axis in the XY plane of the global or a local system are required.
FIGURE 9 illustrates the visual appearance of a line object. Note that a Point
appears at the starting point so that it is possible to determine at a glance the direc-
tion of the line. Note also that curves and surfaces will appear dashed by default
when element objects are enabled for display. This is so that if an element edge
overlaps a curve, it will be possible to see the element edge showing between the
dashes of the curve. If you wish to visualize curves as solid lines, you must disable
elements from the display using the View > Display Objects Toolbar > Display ele-
ments.
Lines are parameterized with a single parameter in the range 0.0 ≤ P1 ≤ 1.0. This
makes it easy to associate other objects with a line. For example, if you wish to
attach a point, system, or node to a line at a position 75% of the length from the
starting point, simply specify an offset value of 0.75. This associative relationship
will be maintained automatically. If the length or orientation of the line changes, the
associated object will always stay at a position corresponding to 75% of the current
length of the line. This associative relationship can be parametric.

FIGURE 9 Line object.
Methods for line creation 2-Point: Select Create > Line > 2-Point. Pick two points, the first point picked
will define the origin (P1=0.0) of the line.
Locate: Select Create > Line > Locate. Enter X, Y, Z coordinates which locate
the origin of the line. Enter the Length of the line. Enter the Orientation of the
line relative to the selected coordinate system (default is Global) rotating CCW
about the z-axis. The orientation angle is measured in degrees. Click on
Accept.
Normal: Creates a line normal to a selected line at a specified offset location on

the line. A reference point is required to create a construction plane and estab-
lish the positive direction of the line. Select Create > Line > Normal. Enter a
value for the Offset (origin location on the selected line) between 0 and 1.
Enter the Length of the new line. Pick a line to which the created line will be
normal. Pick a point (not on the selected line) that establishes the positive
direction of the line.
Off. Parallel: Creates a line parallel to a selected line at a given offset distance
from the line (FIGURE 10). The length of the new line is measured from the
origin (P1=0 position) of the line selected. Select Create > Line > Off. Parallel.

Enter the Distance of the parallel offset. Enter the Length of the new line. Pick the
reference line to which the new line will be created parallel. Pick a point (not on the
reference line) that establishes direction of offset.
reference point
3
FIGURE 10 Example for Off. Parallel method.
Offset: Creates a line whose origin lies on a selected boundary object and is ori-
ented relative to a specified coordinate system. Select Create > Line > Offset. Enter
Offset value. Enter the Length of the line. Enter the Angle (measured in degrees).
Pick a boundary object (line, surface curve, circle, spline, etc.) that is defined by a
single parameter range (P1).
Point: Creates a line with its origin at a specified point. This method is identical to
the Locate method except the X, Y, Z coordinates are obtained by selecting a point.
Select Create > Line > Point. Enter the Length of the line. Enter the Orientation of
the line relative to the selected coordinate system (default is Global) rotating about
the z-axis. The orientation angle is measured in degrees. Pick a point (origin of the
line).
Sample: Creates a line whose end points are the at the P1min and P1max locations
on a selected spline object (FIGURE 11). Select Create > Line > Sample. Pick a

spline. Note: This method will not work with a closed boundary object (e.g.
circle, fillet, ellipse, etc.) because P1min and P1max are the same point. You
can use the X, Y, Z inputs to force the end points to be at a specified coordinate
location. Values in any of these input boxes override the coordinates of the
spline end points.
FIGURE 11 Example for Sample method.
Tan to 1 Pt.: Creates a line tangent to a circle passing thorough a point. Select
Create > Line > Tan to 1 Pt. Pick the point. Pick the circle in the area to which
the line is tangent.
Tan to 2 Cir.: Creates a line tangent to two circles. Select Create > Line > Tan
to 2 Cir. Pick the first circle. Pick the second circle.
Work Plane: Creates a line which lies in a selected plane and its origin is
located relative to the plane reference (local) system (FIGURE 12). Select Cre-
ate > Line > Work Plane. Enter X. Enter Y. Enter the Length of the line. Enter
Angle (measured in degrees). Pick the plane.

FIGURE 12 Example for Work Plane method.
Circle Object
The basic definition of a circle is a coordinate location of its center and a radius. By
default, a circle is automatically created with a local system at its center so that it is
easy to move the circle and to introduce an orientation in 2D or 3D space. The cir-
cle object is parameterized as follows: 0.0 ≤ P1 ≤ 360.0 degrees. It is possible to
create a circular arc by specifying a subset of the parameter range as illustrated in
FIGURE 13. The location, radius, and parameter range of the circle may be para-
metric. A circle may be defined by the 3-Pt Circle method in which the circle will
pass through all three points.
A circle may also be defined as by the 3-Pt CRP (center, radius, point) method. In
this case, the first point represents the center of the circle, the second point estab-
lished the radius of the circle and the direction of the local x axis, and the third
point establishes the plane in which the circle lies. The circle is a complete 360
degree circle.

A circular arc can be defined using the 3-Pt. Arc method. The three points
together determine the orientation and radius of the circle. The first and last
points selected define the range of the arc. The second point determines which
side of the circle to represent with the arc.
FIGURE 13 Circle object.
Ellipse Object
The basic definition of an ellipse is a coordinate location of its center, a major
axis radius and a minor axis radius. By default, an ellipse is automatically cre-
ated with a local system at its center so that it is easy to move the ellipse and to
introduce an orientation in 2D or 3D space. The ellipse object (FIGURE 14) is
parameterized as follows: 0.0 ≤ P1 ≤ 360.0 degrees. It is possible to create an
elliptical arc by specifying a subset of the parameter range.
Fillet Object
Like a circle object, the underlying definition of a fillet is a coordinate location
of its center and a radius. However, the fillet is an associative object which is

FIGURE 14 Ellipse object.
tied to two boundary curves. You need only select the two curves and supply a fillet
radius. The program will automatically compute the location of the center and the
arc range of the fillet, and will also locate two points at the tangency locations. The
fillet object is parameterized as follows: 0.0 ≤ P1 ≤ 360.0 degrees. See FIGURE 15
for an illustration of some fillet definitions. At present, fillets can be constructed
between line objects and circle objects (i.e. two lines or one line and one circle),
only.
Polyline Object
The polyline object is not a true object. It is actually composed of line objects
which are associated with point objects. The polyline object is provided as a conve-
nient tool for constructing a series of straight lines connecting points. Try to con-
struct a polyline as the one shown in FIGURE 16. First, create 6 arbitrary points,
then select:
• Geometry tab > Action: Create > Object: Polyline > Method: Point > Now,
click once on each of the points, in the desired order, but double-click on
the last point to indicate the end of your selection.

FIGURE 15 Fillet object.
FIGURE 16 Polyline object.

Spline Object
The spline object is a rational spline curve which may be constructed by four differ-
ent methods: Locate, Point, Offset, and Mid-Offset. Use the “Locate” method to
construct the spline curve by locating a series of points at new locations in model
space. Use the “Point” method to construct the spline curve by selecting a series of
existing points. In both cases, click twice on the last location or point to complete
the construction. The polynomial order of the spline is assigned automatically. The
spline will pass through all points. The spline object is parameterized as follows:
0<= P1 <= 1.0. See FIGURE 17 for an illustration of a spline curve constructed by
the Locate or Point methods.
3
FIGURE 17 Spline definition.
Use the “Offset” method to construct a spline curve which is at a fixed distance
from a reference curve. A series of sample points, the number of which is specified
as an input value (res), is computed on the reference curve. The normal to the curve
at each point is computed and a construction point is located on the normal vector at

a fixed distance from the reference curve as specified by the input “offset”
value as follows:
y
n1 2
1 r2 = r1 + n1 × offset
r2
r1 offset
3
x
The computed construction points are then used to construct the new spline
curve. The resulting spline curve maintains an associative relationship with the
reference curve. The offset value may be parametric. See FIGURE 18 for an
illustration of a spline curve constructed by the “Offset” method.
FIGURE 18 Spline constructed by Offset method.

Use the “Mid-Offset” method to construct a spline curve which is positioned rela-
tive to two reference curves. A series of sample points, the number of which is
specified as an input value (res), is computed based on sample points taken from
the reference curves. Moving along each curve from the initial offset to the final
offset, a sample point is taken and new construction point is computed as follows:
2
1 3
3
y r1 r3 r2
r3 = r1 + ( r2 – r1 ) × offset
x
If the offset value provided is 0.5, each point on the resulting spline curve will be
equidistant from the reference curves. The resulting spline curve maintains an asso-
ciative relationship with the reference curves. The offset value may be parametric.
To construct the mid-offset curve, click once on each of two curves. If the resulting
curve is twisted, it may be necessary to use the Select/Boundary/Reverse technique
to reverse one of the curves. See FIGURE 19 for an illustration of a spline curve
constructed by the “Mid-Offset” method.
NOTE: The repeat feature may be used to construct several spline curves with one
construction operation. The Offset method may be used to construct a spline associ-
ated with all selected curves by entering the appropriate input and clicking on the
Accept button.
Formula Object
The formula object is a curve defined by two formulas which determine the value
of the dependent and the independent variables of the curve. The formula object
will be associated automatically with a local system in which the independent and
dependent variables will be evaluated. This also makes it possible to easily orient

FIGURE 19 Spline constructed by Mid-Offset method.
the curve after it is created. The formula object is parameterized as follows:

P1-Min ≤ P ≤ P1-Max. See FIGURE 20 for an illustration of a formula defini-
tion.
Rectangle Object
The rectangle object is not a true object. It is actually composed of four line
objects which are associated with four point objects, which are attached to a
local system object. The rectangle object is provided as a convenient tool for
constructing a rectangular shape. It may be defined as constant or parametric,
and once created, may be rotated by rotating the underlying local coordinate
system. See FIGURE 21 for an illustration of a rectangle object definition.
Label Object
The label object is used to display a text label on the screen at a location identi-
fied by the mouse cursor. Enter the desired text in the label field in the input

FIGURE 20 Formula definition.
FIGURE 21 Rectangle object.

area, then trace a line on the screen to indicate the location of the label. See
FIGURE 22 for an illustration of a Label definition.
FIGURE 22 Label object.
Composite Curve A composite curve is defined by connecting some number of 2D curves into a
single composite curve. The resulting curve may subsequently be used to con-
struct a tabulated cylinder, surface of revolution, ruled surface, or spline sur-
face. The curve coordinates are defined in the range 0 ≤ P1 ≤ 1. The curve
coordinates may be understood to be a fraction of the defined curve length.
To construct a composite curve in StressCheck requires that you first construct

the 2D curve segments as illustrated in FIGURE 23a, then trim the curves by
choosing Select > Any Curve > Trim to Point, clicking on each curve to be
trimmed, and clicking on the point where that curve intersects the fillet.
Finally, choose Create > Comp. Curve > Selection, and select the curves with
the mouse cursor in the desired order. Double-click on the final curve to com-
plete the definition. Be sure to select the segment of the curve you wish to
include in the definition. The composite curve will be constructed by identify-
ing the intersection points between each curve segment and attaching the seg-

ments end to end at the intersection points (FIGURE 23b). The points of selection
(see pick points labeled 1-7 in FIGURE 23a) on each curve segment should be in
the range of the curve which is intended to become part of the composite curve.
Segment Curves 7
Composite Curve
6
3
1 3
4 5
2
(a) (b)
FIGURE 23 Definition of a composite curve.


4 3D Geometry
Creation
4
This chapter contains a general overview of the features of StressCheck used for creating three dimen-
sional geometry for a finite element model. For a general description of the preparation of a complete
finite element model, please refer to the User’s Guide.
Construction of 3D primitive objects

The following sections describe the construction of three dimensional solids and
surfaces in StressCheck. Solids and surfaces are used in 3D in much the same way
that curves are used in 2D, as a basis for describing the finite element mesh in an
associative and sometimes parametric way. Before creating 3D solids or surfaces,
you must select the 3D reference system in the Reference/ Theory Toolbar and then
select Class > Geometry.
Note that the same rules apply when creating 3D geometric objects as in 2D. The
mouse cursor is used for locating control points in the graphic area, and the input
fields in the geometry dialog box override the cursor location. For example, if you
wish to create a point, you may supply the Z-coordinate in the input area by
enabling the Z: toggle button and entering a specific Z coordinate. Now when you
move the mouse cursor to the screen and click the left mouse button, a point will be
created using the X and Y coordinates from the screen, and the specified Z coordi-

nate from the input area. As in 2D input, the objects may be created parametri-
cally or in a repeating fashion.
Finally, it should be noted that when dimensions are explicitly entered in the
input area, for example the radius of a sphere, a sphere with the given radius
will be created at the location specified when you click the mouse button. If
you happen to specify the X, Y, or Z coordinate of the location of the center of
the sphere, in the input area, the coordinate(s) specified will be used to define
the location of the sphere.
Solid vs. Surface - What’s the difference?

In StressCheck, there are five types of three dimensional objects, a solid body,
a sheet body, a wire body, a surface and a curve. Solid, sheet and wire bodies
are considered to be topological objects, while surfaces and curves are consid-
ered to be geometric objects. Both solid bodies and sheet bodies may be com-
posed of point, curve and surface geometric objects. Wire bodies are composed
4 of curve and point geometric objects.
Solid bodies are a representation of a three dimensional bounded volume. A

solid body may be a simple primitive such as a box, sphere, cylinder, cone or
torus. A solid may also be a complex object formed by a boolean, blend or
clipping operation involving two or more solid bodies, and represented by a
collection of faces (trimmed surfaces), edges (trimmed curves) and vertices
(points). As such a “parent” solid body is composed of “child” point, curve and
surface geometric objects. These child objects may not be modified directly by
the user, rather they are controlled by the definition of their parent solid. On the
other hand, these child points, curves and surfaces may be referenced by other
objects in explicit associative relationships, or for the purpose of defining
boundary conditions in the finite element model. Solid bodies do not have
parametric limits in the same way that sheet bodies do. Solids are composed of
child surfaces, which are trimmed by the intersection curves that define the
connection between neighboring surfaces in the solid object. However, the sur-
faces which compose a solid have the same 2 parameter (U,V) reference sys-
tem as sheet bodies have for purposes of associating objects or defining
boundary conditions.
Sheet bodies are three dimensional surface objects that do not represent a vol-
ume and do not have thickness. A surface object may be a simple primitive
such as a plane, sphere, cylinder, cone, torus, surface of revolution, tabulated
cylinder, extruded body, or spline surface. In general, a sheet body is a

bounded rectangular patch defined in a 2 parameter system. You may control the
extent of the rectangular patch by supplying limits in the U,V (p1,p2) parameter
system. For example, you may define a cylindrical sheet body with radius=1,
height=2, and parameter limits p1min,p1max = (0,90) and p2min,p2max =
(0.25,0.75). Similarly, a point on the surface of a sphere may be identified by a pair
of surface coordinates representing the longitude angle (0 degrees ≤ P1 ≤ 360
degrees) and latitude angle (-90 degrees ≤ P2 ≤ 90 degrees).
In general, we refer to sheet bodies as surface objects, and solid bodies as solid
objects. As mentioned above, both may be composed of point, curve and surface
geometry, however the user does not directly control the definition of geometric
objects. The user is in control only of topological objects.
Parametric Surface/Solid Description

Most 3D surface and solid objects in StressCheck are defined by a location and one
or more basic dimensions of the object. For example, the sphere is described by the 4
X, Y, Z location of its center, and a radius. Each of these may be defined parametri-
cally. To create a sphere, you may enter the X, Y, Z coordinates, the radius, and, if
defining a surface, the P1 (P1-Min: 0, P1-Max: 360) and P2 (P2-Min: -90, P2-Max:
90) values into the input fields of the Input Window. If creating a solid, depress the
SOLID button. If creating a surface, do not depress the SOLID button. Click on the
Accept button or move the mouse cursor to the graphic area and click the left
mouse button. Remember that it is necessary to enable each input field in order to
enter location and dimension information.
Once a 3D surface or solid primitive has been created, it may be edited graphically.
Simply choose Action: Edit, Object: Cylinder, and use the graphic cursor to select
the cylinder you wish to edit. The dimension information will appear in the input
area (only in fields which are disabled for manual data entry). You may now enter
new dimension information and click the Replace button to change the object defi-
nition. Only those dimensions associated with enabled input fields will be changed
for the selected object. For example, if the cylinder is originally defined with
radius=10.0 and height=30.0, and you enable only the radius input field, enter any
nonzero value and click Replace, only the radius of the cylinder will be changed to
the value specified. Note: You may also use a generic object type such as Any
Body, Any Boundary, or Any Object to edit a solid or surface.

Surface/Solid Orientation
Another common editing operation performed on 3D surfaces is to change the
orientation of the surface. By default all system based surfaces and solids, such
as planes, spheres, cylinders, cones, and tori, are created such that the xy-plane
of the local system associated with the surface is parallel to the XY-plane of the
global coordinate system when the XYZ rotation angles are zero. This condi-
tion is illustrated in FIGURE 24
z
y
Z x
Y Local
4
Global X
FIGURE 24 Default orientation of system based 3D surfaces.
When a system based surface is constructed graphically the XZ plane of the

local system is created in the plane of the screen. To change the orientation of
an object, you must edit the local system associated with the object. This can
be accomplished as follows:
1) choose Action: Edit,

2) choose Object: System,
3) make sure that all input fields are disabled for manual input,
4) use the mouse cursor to select the local system you wish to modify,
5) change the local system rotation values (rot-X, rot-Y, rot-Z),
6) click the Replace button to update the definition of the local system and
associated 3D object.

Solid/Surface Bounding Curves
Sheet and solid bodies are composed of geometric point, curve and surface primi-
tives. In some cases, a solid or sheet body may have been constructed from pre-
existing curve or surface primitives. For example, a ruled surface or a spline surface
may be constructed by joining geometric curves with a sheet body composed of
new point, curve and surface primitives. In this case, the curves represent the
bounding curves that define the extent of the sheet body. Visually, the original con-
struction curves and the new bounding curves may be easily confused. It is impor-
tant to distinguish them when constructing a finite element mesh by hand.
In most cases, nodes should be attached to bounding curves, and not to the underly-
ing construction curves, in order for the elements to be properly associated with the
parent sheet body surfaces. Before defining any nodes that are to be attached to
bounding curves, it is a good practice to first blank the underlying construction
curves, so that they will not be selected by mistake when attaching nodes. One tech-
nique for doing this is to select all surface curves with a “marquee pick”, then use
the “Invert Selection” icon to automatically cancel the selection of the surface 4
curves and select the non-surface curves. Then blank the currently selected curves.
When you pick curves during subsequent node creation operations, the nodes will
only be attached to the desired bounding curves.
Surface Patches
It is sometimes desirable to limit the portion of a surface which is displayed to some
range of the P1, P2 parameters. The default parameters for the Sphere for example,
are 0 degrees ≤ P1 ≤ 360 degrees and -90 degrees ≤ P2 ≤ 90 degrees. This may be
accomplished by enabling the range parameters and entering the desired ranges
when the surface is created, or changing them afterwards.
Note: the parameter range of a solid primitive may not be modified in this manner.
FIGURE 25 illustrates this procedure. For the sphere on the left, the range for the
first parameter was set to 250 ≤ P1 ≤ 325, and for the second 0 ≤ P2 ≤ 45.
Solid/Surface Types and Definitions

The following sections describe basic construction techniques for each of the 3D
surface types (and corresponding solid primitives if appropriate) currently sup-
ported in StressCheck. It is important to pay particular attention to the definition of

the surface coordinate ranges since node definitions are often associated with
surfaces using the surface coordinates.
surface patch
sphere
4
FIGURE 25 Definition of a 3D surface patch.
Sphere A sphere is defined by the coordinates of its center and a radius (FIGURE 26).
The surface coordinates are defined in the range 0 ≤ P1 ≤ 360 (longitude), and
-90 ≤ P2 ≤ 90 (latitude). A sphere can be created either as a surface or a solid
object.
P2
P1
FIGURE 26 Definition of a sphere.
radius

Cylinder A cylinder is defined by the coordinates of the center of its base, a radius and a
height (FIGURE 27). The surface coordinates are defined in the range 0 ≤ P1 ≤ 360
(circumferential direction), and 0 ≤ P2 ≤ 1.0 (along the cylinder axis). The P2 coor-
dinate is understood to be a fraction of the defined height of the cylinder. For exam-
ple, to place a node on the surface at half the height of the cylinder, the P2 offset
parameter should be given as 0.5. In this way, a node will remain at the same rela-
tive location on the surface, even when the height of the cylinder is changed. A cyl-
inder can be created either as a surface or a solid object.
P2
height
FIGURE 27 Definition of a cylinder.
4
P1
radius
Cone/Frustrum A cone is defined by the coordinates of the center of its base, a base radius (r1), a
top radius (r2), and a height (FIGURE 28). The surface coordinates are defined in
the range 0 ≤ P1 ≤ 360, and 0 ≤ P2 ≤ 1.0. A cone can be created either as a surface
or a solid object.
Torus A torus is defined by the coordinates of the center of its base, a major radius (r1),
and a minor radius (r2). The surface coordinates are defined in the range 0 ≤ P1 ≤
360 (along the circumference corresponding to r1), and 0 ≤ P2 ≤ 360 (along the cir-
cumference corresponding to r2). These dimensions are illustrated in FIGURE 29.
A torus can be created either as a surface or a solid object.
Plane A plane is defined by the coordinates of its local origin, a width and a height. The
surface coordinates are defined in the range -0.5 ≤ P1 ≤ 0.5, and -0.5 ≤ P2 ≤ 0.5.
The surface coordinates may be understood to be a fraction of the actual dimension
of the plane in each direction (FIGURE 30). When using the 3-Point method, the
local origin is located at the first point picked.

r2=0
P2 r2
height
r1 P1 r1
FIGURE 28 Definition of a cone or frustrum.
P1
4
r1 P2
r2
FIGURE 29 Definition of a torus.
local origin
z (Locate method)
y x
P2 P1
width
height
FIGURE 30 Definition of a plane.

Box A box is defined by the coordinates of the center of its base, the width (x dimen-
sion), height (y dimension), and depth (z dimension). These dimensions are illus-
trated in FIGURE 31. A box can be created as either a surface or a solid. A surface
Base local system

z
Depth x
y
Width
Height
FIGURE 31 Definition of a box.

4
box is not a true object, since it may not be edited as a single entity. Once created, it
exists only as a collection of points lines and planes. However, it may be created
with parametric dimensions, and the underlying objects will respond to changes in
the corresponding dimensional parameters. Since the surface box is not a true
object, it may not be used as a tool or target in a subsequent boolean or blend oper-
ation. If a surface box is defined parametrically, the local system is located at one
corner of it. Otherwise, the system is at the same location as for the solid box.
A solid box is a true entity, and may be used as a primitive tool or target in a subse-
quent boolean, blend or clipping operation. A solid box may be created with para-
metric dimensions, and may be edited as a single entity.
Spline Surface A spline surface is a form of fitted surface constructed from a set of control points
obtained from an underlying set of 2D curves oriented in 3D space (FIGURE 32).
The surface coordinates are defined in the range 0 ≤ P1 ≤ 1, and 0 ≤ P2 ≤ 1.
To construct a spline surface in StressCheck requires that you first have a set of at
least two 2D curves. These curves may be of any type, and may be defined as open
or closed. In the example shown in FIGURE 32 the spline surface is associated with
a spline curve, two circular arcs, and an elliptical arc.

FIGURE 32 Definition of a spline surface.

4
P1
P2
To construct the spline surface, simply use the mouse cursor to select each
curve in the order you wish to use them to represent the surface, and double
click on the final curve to indicate that you are finished selecting curves.
StressCheck will automatically choose the polynomial order of the spline fit-
ting. The surface remains associative, so that if you change the definition of
any of the underlying curves, the surface will automatically be re-evaluated. If
you make a mistake while selecting curves, use the Backspace key to de-select
the last curve selected.
A spline surface must have at least two sampling curves. If the spline surface
appears to be twisted, it is possible that one or more of the 2D boundary curves
have a reversed parameter direction with respect to the other curves used. Use
the Reverse method (Select > Any Boundary > Reverse) to change the sam-
pling direction of the erroneous curve.
Spline Surface Closure Conditions: Note that the input fields labeled P2-Bot
and P2-Top provide additional control over the shape of the surface in the P2
parameter direction. To use this advanced feature, the inputs should only be

assigned non-zero values if the surface will be closed in the P2 parameter direction.
If a value of zero (0) is assigned, or if the toggle is turned off, the surface will
appear open in the P2 parameter direction. If a value of one (1) is assigned, the sur-
face will be closed like a cup at the corresponding edge of the surface (either where
parameter P2 is minimum or where it is maximum). The closure condition is
obtained by averaging the sample points from the first profile (P2-Min) or last pro-
file (P2-Max) and adding an additional profile to the surface definition consisting
of a small circle at the computed average location. If a value of two (2) is assigned,
the closure condition will be obtained by averaging the sample points from the first
or last profile and replacing that profile with the new computed points.
Use the “Offset” method (Create > Spline > Offset) to construct a spline surface
which is at a fixed distance from a reference surface. A series of sample points, the
number of which is specified as an input value (res), is computed on the reference
surface. The computed construction points are then used to construct the new spline
surface. The resulting spline surface maintains an associative relationship with the
reference surface. The offset value may be parametric. See FIGURE 33 for an illus-
tration of a spline surface constructed by the “Offset” method. 4
FIGURE 33 Create spline surface by the Offset method.
Use the “Mid-Offset” method to construct a spline surface which is positioned rela-
tive to two reference surfaces. A series of sample points, the number of which is
specified as an input value (res), is computed based on sample points taken from
the reference surfaces. Moving along each surface from the initial offset to the final

offset in each u,v direction, a sample point is taken and new construction point
is computed.
If the offset value provided is 0.5, each point on the resulting spline surface
will be equidistant from the reference surfaces. The resulting spline surface
maintains an associative relationship with the reference surfaces. The offset
value may be parametric. To construct the mid-offset surface, click once on
each of two surfaces. If the resulting surface is twisted, it may be necessary to
use the Select > Boundary > Reverse technique to reverse one of the surfaces.
See FIGURE 34 for an illustration of a spline surface constructed by the “Mid-
Offset” method.
FIGURE 34 Create spline surface by the Mid-Offset method.
Note: The Repeat feature may be used to construct several spline surfaces with
one construction operation. The Offset method may be used to construct a
spline associated with all selected surfaces by entering the appropriate input
and clicking on the Accept button.
Surface of Revolution A surface of revolution is defined by an arbitrary 2D curve and a line which
represents the axis of revolution. The surface coordinates are defined in the
range 0 ≤ P1 ≤ 360, and 0 ≤ P2 ≤ 1. The surface coordinate in the direction of
the basis curve (P2) is understood to be a fraction of the curve length.

P2
Line 2D-curve
(a) (b) P1 4
FIGURE 35 Definition of a surface of revolution.
To construct a surface of revolution in StressCheck requires that you first construct

2D curves as illustrated in FIGURE 35a. After constructing the 2D curves, select
the curve to be rotated first, then the line which is to be used as the axis of revolu-
tion. By default, the surface will revolve about the axis the full 360 degree range
(FIGURE 35b). You may supply the start angle and end angle in the input area if
you wish to restrict the angle of revolution.
Tabulated Cylinder A tabulated cylinder is defined by an arbitrary 2D curve and a point which does not
lie on the curve, (or a line) representing the direction of extrusion from the curve at
P1 = 0. The surface coordinates are defined in the range 0 ≤ P1 ≤ 1, and 0 ≤ P2 ≤ 1.
The surface coordinates in each direction are be understood to be a fraction of the
defined surface dimensions.
To construct a tabulated cylinder StressCheck requires that you first construct a 2D

curve and a point (or line), as illustrated in FIGURE 36a. Then, simply select the
curve and the point (or line) with the mouse cursor (FIGURE 36b). Use method
“Point” to create a Tabulated Cylinder using a reference point (Create > Tab. Cyl >

2D-curve
Point
(a)
P1
(b)
P2
4
FIGURE 36 Definition of a tabulated cylinder.
Point). Use method “Selection” to create a Tabulated Cylinder using a refer-

ence line (Create > Tab. Cyl > Selection).
Ruled Surface A ruled surface is defined by connecting two arbitrary curves. The surface
coordinates are defined in the range 0 ≤ P1 ≤ 1, and 0 ≤ P2 ≤ 1. The surface
coordinates in each direction are understood to be a fraction of the defined sur-
face dimensions.
To construct a ruled surface StressCheck requires that you first construct the
curves as illustrated in FIGURE 37a, then simply select the two curves with the
mouse cursor. Note that it is necessary that the two curves be oriented so that
the positive parameter direction is the same.
Note: If the spline surface appears to be twisted, it is possible that one of the
boundary curves has a reversed parameter direction with respect to the other
curve. Use the Reverse method (Select > Any Curve > Reverse) to change the
sampling direction of one of the curves.

positive (a)
parameter
direction
P1
(b)
P2
4
FIGURE 37 Definition of a ruled surface.
Offset Surface An offset surface is defined as a replica of a 3D surface which has been offset from
the original surface. Each point on the new offset surface is located at the specified
distance from a corresponding point on the base surface, in the direction of the nor-
mal vector at each point. An example is shown in FIGURE 38. To construct an off-
set surface StressCheck requires that you first construct the 3D surface as illustrated
in FIGURE 38, then supply the offset distance in the input area and select the sur-
face with the mouse cursor. Be sure that the offset surface is not self-intersecting, or
an error message will appear.
To locate points or nodes on the offset surface, use the same surface coordinate sys-
tem as the originating surface.
Fillet Surface A fillet surface represents the shape of the path traced between two surfaces by a
rolling ball. The surface coordinates are defined in the range 0 ≤ P1 ≤ 1, and 0 ≤ P2
≤ 1. The surface coordinates in each direction may be understood to be a fraction of
the defined surface dimensions.

Offset Surface
Original Surface Offset Distance
4 FIGURE 38 Definition of an offset surface.
To construct a fillet surface in StressCheck requires that you first construct the
two surfaces to be filleted as illustrated in FIGURE 39, supply the fillet radius
in the input area, and select the two surfaces with the mouse cursor. The prox-
imity of your cursor picks will determine on which side of each surface the fil-
let surface will appear.
When creating elements attached to a fillet surface, it is important to place

nodes exactly on the edge of the fillet surface (p1=0.0 and p1=1.0, or to attach
them to the intersection curve which is constructed automatically to represent
the point of tangency between the fillet surface an the parent surfaces. This will
ensure that the associativity with the secondary surfaces will be recognized
when one edge of the element lies along the edge of the fillet and the other
edges lie one of the parent surfaces.
Surface Curves
A surface curve is a curve attached to one or two surfaces which has an asso-
ciative relationship with those surfaces. The surface curve may be created by
one of three methods. It may be created as an intersection of two surfaces, as a

P1
P2
FIGURE 39 Definition of a fillet surface.
projection of a secondary curve to a surface, or as a trace along the surface between

two points which have been projected to the surface.
A surface curve representing the intersection between two arbitrary surfaces may
be constructed by selecting Create > Surface Curve > Intersection. Use the mouse
to select the two surfaces to be intersected. The location of the mouse pick should
be in close proximity of the desired intersection curve so that if there is more than
one intersection between the surfaces, the program will construct the desired curve.
See FIGURE 40.
A surface curve representing the projection of a secondary curve to a surface may

be constructed by selecting Create > Surface Curve > Projection. Use the mouse to
select first the surface, then the curve to be projected.
A surface curve representing the trace between two points which have been pro-
jected to a surface may be constructed by selecting Create > Surface Curve > 2-
Point. Use the mouse to select first the surface, then the two points to be projected.

Performing 3D Boolean Operations
4
Surface Curve
FIGURE 40 Definition of an intersection curve.
The surface curve retains its associativity with the parent points, curves and
surfaces, so that if the dimension or location of the objects are changed, the
surface curve will be recomputed automatically. If a node is attached to the sur-
face curve, the underlying associativity with the objects is inherited by the
node so that when an element is attached to the node, element edges and/or
faces will be properly associated with the objects for the purposes of geometry
and boundary conditions.

The following paragraphs describe the use of boolean operations for construct-
ing complex three dimensional solids in StressCheck. A boolean operation
requires as input two or more solid bodies, one of which is designated as the
target body, and the remaining bodies are designated as tool bodies. The target
body is generally the object which survives after the boolean operation is per-

formed. When performing a boolean operation, you are designating existing solid
bodies as input to a boolean operator, and the result is one or more solid bodies. The
definitions of each target body, tool body and boolean operator are preserved in
StressCheck’s data structure. The result body (or bodies) may then be used as input
to a subsequent boolean, blend, or clipping operation. Although a tool body may be
suppressed from display, its definition remains in the StressCheck data structure,
and its dimensions may be modified either through an edit operation or through a
parametric change.
Create/Body/Bool-Subtract
For example, we can create a simple box and cylinder, then subtract the cylinder
from the box to form a hole. Create the box as follows:
• Geometry tab > Create > Box > Locate > Solid button on > Input toggle
switch on, X= 0, Y= 0, Z= 0, Width= 5, Height= 5, Depth= 1, Rot-X= 0,
Rot-Y= 0, Rot-Z= 0 > Accept button.
4
Create the cylinder as follows:
• Create > Cylinder > Locate > Solid button on > Input toggle switch on, X=
0, Y= 0, Z= -1, Radius= 1, Height= 3, Rot-X= 0, Rot-Y= 0, Rot-Z= 0 >
Accept button.
To perform the boolean subtraction,

• Create > Body > Bool-Subtract. Pick the box, then pick the cylinder, then
click on the Accept button. As you pick each target and tool object, it will
be highlighted. To cancel a selection, hold the Ctrl key while you pick it. To
cancel the whole operation prior to completion, click the right mouse but-
ton. As noted above, the first object picked is assumed to be the target of
the boolean operation.
Note that there can be only one target object for a boolean operation, but there can
be multiple tool objects. FIGURE 41 illustrates the simple boolean operation
described above. After completion of the boolean subtraction operation, examine
the object index by selecting the tab labelled “Index”. As illustrated in FIGURE 43,
the index contains an entry for the original box and cylinder objects, one entry for
the boolean operation, and an entry for the result of the boolean operation. As you
select each entry in the index, the corresponding objects will be highlighted in the
model window. This facility gives you the ability to review the construction history
for your model. This is also one way for you to modify the dimensions of objects
that may have been used as targets or tools in a boolean operation. For example, if

Input
Result
FIGURE 41 Definition of a Boolean Subtraction
you choose Edit > Any Body >..., then click the Index entry for the cylinder,
you will see the cylinder highlighted as shown in FIGURE 42. You may imme-
diately switch back to the Data tab, where you will see the input dimensions for
the original cylinder object displayed in the input interface. If you were to
change the radius of this cylinder from 1.0 to 0.5, then click on the Replace
button, the radius of the hole produced by subtracting the cylinder from the box
would be reduced to 0.5 as illustrated in FIGURE 43.
The boolean union and boolean intersection operations may be performed in a

similar fashion.

FIGURE 42 Editing cylinder for the

example problem.
4
Note: Since StressCheck does not allow you to manage the history of the modeling
operations, you are limited to relatively simple dimensional changes in the defini-
tion of your model. In order to perform more sophisticated editing operations, you
must work backwards to the point of modification, make the modification, then
retrace your modeling steps manually. This is usually performed by either clicking
DeLast for each step you wish to remove, by using the Undo command, or by
selecting the Index entries for the each boolean operation in the reverse order of
construction, and clicking Delete.
Create/Body/Stitch
The stitch method makes it possible to construct a single body by sewing multiple
patches together. The resulting body will automatically become a solid body if pos-
sible. For example, we can construct a solid cylinder by sewing circular surfaces at
the ends of a cylindrical surface.
Create a cylinder surface with the following input:

• Create > Cylinder > Locate > Solid button off > Input toggle switch on, X=
0, Y= 0, Z= 0, Radius= 1, Height= 2, P1-Min= 0, P1-Max= 360, P2-Min=
0, P2-Max= 1, Rot-X= 0, Rot-Y= 0, Rot-Z= 0 > Accept button.
To create a circular surface the geometric icon should be set to surfaces . If you
wish to create a circular curve, set the geometric icon for curves .

Before Edit
pick
After Edit
FIGURE 43 Editing a Boolean Body
• Create > Circle > Locate > Input toggle switch on, X= 0, Y= 0, Z= 0,
Radius= 1, Rot-X: 0, Rot-Y: 0, Rot-Z: 0 > Accept button.
• Create > Circle > Locate > Input toggle switch on, X= 0, Y= 0, Z= 2,
Radius= 1, Rot-X: 0, Rot-Y: 0, Rot-Z: 0 > Accept button.
Now stitch the three surfaces together to form a solid cylinder by doing the fol-
lowing:
• Create > Body > Stitch. Select each surface and click on Accept.
You can verify whether the resulting body is a solid, or a general body by
examining the entry in the Index tab. The record corresponding to the new

body will be labelled as “Body-Surf” if the result was a general (surface) body, and
“Body-Solid” if the result is a true solid.
Create/Body-Imprint/...
Body-Imprint methods are provided to make it possible to create new edges and
faces in an existing target body, usually for the purpose of performing subsequent
solid body construction operations using the new faces. The methods provided are:
Both Bodies, Curve Normal, Curve Vector, Faces, Plane, and Target Body.
Both Bodies – Select a target body and a tool body. The curve(s) representing the
intersection(s) between the selected tool body and the target body will be imprinted
on both the target body and the tool body.
Curve Normal – Select a target body and one or more curves. Imprint the selected
curve(s) on the selected body by projecting them locally down face normal. You
must click on the Accept button to initiate the operation after picking the final
4
curve.
Curve Vector – Select a target body and one or more curves. Imprint the selected
curve(s) on the selected body by projecting them in the given direction. If no local
system is supplied, the direction of projection corresponds to the direction of the
positive global Z axis. If a local system is selected (hold CTRL-SHIFT when pick-
ing a local system), the direction of projection corresponds to the direction of the
positive local Z axis. You must click on the Accept button to initiate the operation
after picking the final curve.
Faces – Select a target body and one or more faces of neighboring bodies. The
curve(s) representing the intersection(s) between the selected faces and the faces of
the target body will be imprinted on the target body. You must click on the Accept
button to initiate the operation after picking the final face.
Plane – Select a target body and one planar sheet body. The curve(s) representing
the intersection(s) between the extended plane of the surface and the faces of the
target body will be imprinted on the target body. The sheet body will be suppressed
from view.
Target Body – Select a target body and a tool body. The curve(s) representing the
intersection(s) between the selected tool body and the target body will be imprinted
only on the target body.

Create/Body-Sheet/...
The following Body-Sheet methods are provided: Extrude, Fill Hole, Spin, and
Sweep.
Extrude – Select a curve that will be extruded to form a sheet body surface, and
supply the distance to the first edge of the new surface and the distance to the
second edge. Select a local coordinate system (while holding the CTRL-SHIFT
keys) to determine the direction of the extrusion (along positive Z-axis). If no
system is selected, the extrusion will be relative to the positive global Z-axis.
Fill Hole – Fill a hole in a surface by selecting one or more edges which bound
the hole to be filled. After the operation, the face(s) formed by filling the hole
will become part of the body that contained the selected edge(s). The body that
results from a fill operation will not be a solid. To convert a body to a solid,
you must use the Create/Body-Copy operation.
Spin – Select a curve to be spun about the Z-axis of a local system to form a
4 surface. The local system must be selected while holding the CTRL-SHIFT
keys. If no local system is supplied, the spin will be performed about the global
Z axis. Supply the angle (in degrees) that the curve will be spun. The local sys-
tem must be selected before selecting the curve.
Sweep – Construct a sheet (surface) body by sweeping a profile curve along a

path curve. First select the profile curve, then the path curve.
Create/Body-Solid/...
The following Body-Solid methods are provided: Extrude, Extrude Face, Spin,
Spin Face, Sweep, and Thicken.
Extrude Select a sheet (surface) body that will be extruded to form a solid body, by
specifying distances or selecting objects as limits for the extrusion. Select a
local coordinate system (while holding the CTRL-SHIFT keys) to determine
the direction of the extrusion (along the positive Z-axis). If no system is
selected, the extrusion will be relative to the positive global Z-axis. The local
system should be selected before selecting the profile surface or limiting
objects. There are several methods for extruding a surface to form a solid,
which relate to the location of the front and back faces of the extruded solid.
The methods for defining the front and back faces may be used in any combi-
nation. The distance method requires as input the distance from the profile sur-

face where the front and back faces should be located. The other methods require as
input the number of limiting faces which occur between the profile surface and the
limiting surface of the extrusion. If the limiting faces are located in a negative
direction relative to the local system Z axis (or global system Z axis if no local sys-
tem is selected), the input limit should be provided as a negative value. These meth-
ods for specifying the limits of an extruded solid are described below.
Distance – specify the distance from the selected surface to the front/back
face of the solid. “Dist1” refers to the front face, “Dist2” refers to the back
face. Negative distance is measured relative to the positive normal of the
surface.
Surface – Select a surface that will be used as the front or back bounding
face of the extruded solid.
Face – Select the face of a body that will be used as the front or back
bounding face of the extruded solid.
Sheet – Select a sheet body that will be used as the front or back bounding
face of the extruded solid. If the extrusion is successful, the sheet body will
be trimmed and become part of the extruded solid.
4
Body – Select a body that will limit the extruded solid. The extruded solid
will conform to the shape of the selected solid body face or faces where the
extrusion intersects the body. If the extruded solid intersects the solid body
in more than one location, the limit value will identify the interval through
which the solid will be extruded.
Example: Create two solid boxes and a plane as follows:

switch on, X= -2, Y= 0, Z= 0, Width= 0.3, Height= 2.0, Depth= 2.0, Rot-
X= 0, Rot-Y= 0, Rot-Z= 0 > Accept button. This will create Box 1.
switch on, X= 2, Y= 0, Z= 0, Width= 0.3, Height= 2.0, Depth= 2.0, Rot-X=
0, Rot-Y= 0, Rot-Z= 0 > Accept button. This will create Box 2.
• Geometry tab > Create > Plane > Locate > Input toggle switch on, X= 0,
Y= 0, Z= 1, Width= 0.5, Height= 0.5, Rot-X= 0, Rot-Y= 90, Rot-Z= 0 >
Accept button. This will create the plane.
Create your solid body by setting the following options in the Input dialog box:
• Geometry tab > Create > Body-Solid > Extrude > Data tab > Limit1: Face,
Limit2: Face > Input toggle switch on, Limit1: 1, Limit2: 2. Select the local
coordinate system (while holding the CTRL-SHIFT keys) of the plane to
determine the direction of the extrusion (along the positive Z-axis of the

local system). Click on the plane, then on the front face of Box 2 and
then on the back face. A body with the dimensions of the plane will be
created inside Box 2.
The order of selection of faces should be consistent with the Z-axis of the local
coordinate system. To define the Limit values, you can choose any positive
integer number for the face facing the positive Z-axis of the plane, and any
negative integer number for the face facing the negative Z-axis of the plane.
Plane
Box 1
Box 2
4
To create a solid body in Box 1 similar to the one already created in Box 2, set:
• Geometry tab > Create > Body-Solid > Extrude > Data tab > Limit1:
Face, Limit2: Face > Input toggle switch on, Limit1: -1, Limit2: -1.
Select the local coordinate system (while holding the CTRL-SHIFT
keys) of the plane to determine the direction of the extrusion (along the
positive Z-axis). Click on the plane, then on the Box 1 face farther
from the local system and then on closest face. A solid body will be
created inside Box 1.
•

To create a solid body that goes from one box to the other, set the following
options:
• Geometry tab > Create > Body-Solid > Extrude > Data tab > Limit1: Face,
Limit2: Face > Input toggle switch on, Limit1: -1, Limit2: 2. Select the
local coordinate system (while holding the CTRL-SHIFT keys) of the plane
to determine the direction of the extrusion (along the positive Z-axis) and
then click on the plane. Click on the most distant faces of the boxes relative
to the local coordinate system of the plane, starting by the Box 1 face.
Box 1
Box 2
4
Extrude Face Select the face of a solid body that you wish to extrude. By default, the extrusion is
performed in the direction of the positive Z-axis of the global system. To select a
local system, hold the Ctrl-Shift keys while picking a local system object in the
model window. A typical usage of this function is to imprint a curve onto an exist-
ing face of a solid body, then extrude into the body to form a slot, or away from the
body to form a tab. Supply as input the length of the extrusion.
Example: Using the geometry given in the example above, select the following
options:

• Geometry tab > Create > Body-Solid > Extrude Face > Data tab > Dist:
1. Click on face Z=2 of Box 2 and then on Accept. The face will be
extruded as shown below.
Box 1
Box 2
4
Spin Select a sheet (surface) body to be spun about the Z-axis of a local system to
form a solid. The local system must be selected while holding the CTRL-
SHIFT keys. If no local system is supplied, the spin will be performed about
the global Z axis. Supply the angle (in degrees) that the surface will be spun.
The local system must be selected before selecting the surface.
Example: Create a plane and a system as indicated below:

• Geometry tab > Create > Plane > Locate > Input toggle switch on, X=
2.0, Y= -0.5, Z= 2.0, Width= 1.5, Height= 1.0, Rot-X= 0, Rot-Y= 0,
Rot-Z= 0 > Accept button.
• Geometry tab > Create > System > Locate > Input toggle switch on,
X=-1.0, Y=0.0, Z=1.0, Rot-X= 90, Rot-Y= 0, Rot-Z= 0 > Accept but-
ton.

• Geometry tab > Create > Body-Solid > Spin > Data tab > Input toggle
switch on, angle: 45. Select the plane and then click on the local coordinate
system (while holding the CTRL-SHIFT keys).
Spin Face Select the face of a solid body that you wish to spin. By default the spin is per-
formed about the positive Z-axis of the global system. To select a local system, hold
the Ctrl-Shift keys while picking a local system object in the model window. Sup-
ply as input the angle to spin the face about the corresponding Z-axis.
Example: Create a solid cylinder and a system as indicated below: 4

• Create > Cylinder > Locate > Solid button on > Input toggle switch on, X=
0, Y= 0, Z= 0, Radius= 2, Height= 2, Rot-X= 0, Rot-Y= 0, Rot-Z= 0 >
Accept button.
• Geometry tab > Create > System > Locate > Input toggle switch on, X= -
4.0, Y= 2.0, Z= 3.0, Rot-X= 90, Rot-Y= 0, Rot-Z= 0 > Accept button.
• Geometry tab > Create > Body-Solid > Spin Face > Data tab > Input toggle
switch on, angle: 60. Select one cross section of the cylinder and then click
on the local coordinate system (while holding the CTRL-SHIFT keys).
Click on Accept.

Sweep Construct a solid body by sweeping a profile surface along a path curve. First
select the profile surface, then the path curve.
Example: Create a circle and a line as indicated below:

• Geometry tab > Create > Surface Icon > Circle > Locate > Data tab >
Input toggle switch on > X= 0, Y= 0, Z= 0, Radius= 2, Rot-X= -90,
Rot-Y= 0, Rot-Z= 0 > Accept.
• Geometry tab > Create > Line > Locate > Data tab > Input toggle
switch on > X= 3, Y= 0, Z= 0, Length= 4, Angle= 75 > Accept.
• Geometry tab > Create > Body-Solid > Sweep > Data tab. Select the
circle and then click on the line.
Thicken The “thicken” operation will convert a sheet body (a surface with no thickness)
into a solid body with the supplied thickness. The existing surface will be
treated as a mid-surface, to which material will be added in the positive and/or
negative directions with respect to the normal at each point on the surface.
Select a sheet body or general body that will be thickened, and supply the dis-
tance in the positive and/or negative direction(s) to extend the volume of the
solid body.
Example:
• Geometry tab > Create > Plane > Locate > Input toggle switch on, X=
0.0, Y= 0.0, Z= 0.0, Width= 1.0, Height= 1.0, P1-Min=-0.5, P1-
Max=0.5, P2-Min=-0.5, P2-Max=0.5, Rot-X= 0, Rot-Y= 0, Rot-Z= 0 >
Accept button.
• Geometry tab > Create > Body-Solid > Thicken > Data tab > Input tog-
gle switch on, Above= 0.5, Below= 1.5. Click on the plane and hit
Accept to create the solid body.

Create/Body-Copy/Heal…
This feature is currently under development and it is only available in a Beta-form.

A brief description of its functionality follows: 4
When creating a Body-Copy object, the goal is to construct a new body from an
existing body, where some of the details have been removed. Usually this kind of
operation is useful when one wishes to remove a fillet or a hole from an imported
solid body in order to simplify the model, or when you wish to convert an existing
static geometric definition to a parametric definition. When removing a detail, there
are several different methods for filling the hole/gap that is left by the missing
detail. These methods are used to “heal” the body. Depending on the situation, each
healing method may produce a different result. NOTE: The new body will lose all
parametric and associative attributes of the original body.
The input to the Body-Copy creation operation is the X,Y,Z translation and X,Y,Z
rotation to be applied to the new body. If no translation or rotation input is pro-
vided, the original body will simply be replaced by the new body with the appropri-
ate face removal healing operations performed.
The six possible healing methods are as follows:
No Heal – Faces are removed from the body, and the hole is left with no attempt
made to fill it. This option is used most often with surface bodies in order to cut a
hole in the surface after imprinting a closed curve.
Heal Cap - Find a surface in which all edges of a hole lie and attach this to a face
covering the hole.

Heal Parent - Grow the faces the parent had around the hole until they cover it.
Heal Child - Extend faces around the hole until they cover the hole.
Heal Shrink - If extending faces does not yield a solution, then shrinking the
faces is tried.
Heal Any - Use any of the available heal methods.
To select faces for removal and subsequent healing, simply click each face to
be removed then click on the Accept button. Alternatively, you may designate
faces to be preserved by holding the CTRL-SHIFT while selecting the faces to
retain. If you wish to make a copy of the complete part, you must CTRL-
SHIFT and select all faces of the part.
Create/Body-Trim/Heal…
4 This feature is currently under development and it is only available in a Beta-

form. A brief description of its functionality follows:
The Body-Trim operations are similar to the Body-Copy. Use the same selec-
tion techniques to identify faces to be removed from the body. The most com-
mon use of the Body-Trim functions are to remove faces from a sheet body
which has had curves imprinted. The objective is to trim faces in order to con-
struct a profile which may be extruded, thickened, swept or spun. The differ-
ence between the Body-Copy and the Body-Trim operations are as follows:
1. The Body-Trim operations modify the original body and retain the paramet-
ric and associative relationships of the original body.
2. The trim operation is sensitive to the type of body the operation is performed
on (solid, sheet, general). If the target body is not a solid, selected faces will
always be removed from the body and no healing will be performed (i.e. The
healing method chosen is irrelevant, and consequently is ignored.)
To select faces for removal, simply click each face to be removed, then click on
the Accept button. Alternatively, you may designate faces to be preserved by
holding the CTRL-SHIFT while selecting the faces to retain.

Performing 3D Clipping Operations
Performing 3D Clipping Operations

The following paragraphs describe the use of clipping operations in StressCheck.
The purpose of a clipping operation is to divide a body into two sections, and then
subtract the portion of the object which lies on one side or the other of the clipping
surface. Note: The clipping surface may be any type of surface, not necessarily a
plane.
Using the example described in the last section [Create>Body>Bool-Subtract], we

can perform a clipping operation to create a body which represents one half of the
block with the hole. To perform the clipping operation, first construct a clipping
plane in such a way that it divides the body to be clipped as desired. To create a
clipping plane, do the following:
• Geometry tab > Create > Plane > Locate > Input switch on, X= 0, Y= 0,
Z= 0.5, Width= 3, Height= 6, Rot-X= 0, Rot-Y= -90, Rot-Z= 0. Click on
Accept.
4
The direction of the normal vector for the clipping surface determines whether to
use a Clip-Back or a Clip-Front operation. You can see the vector triad by placing
the cursor on the clipping surface. A Clip-Back operation will remove the solid
geometry on the back (negative) side of the clipping surface. In a similar fashion,
the Clip-Front operation will remove the target body that lies on the positive side of
the clipping surface.
To remove the portion of the body which lies on the negative Z side of the clipping
plane, do the following:
• Create > Body > Clip-Back. Pick the body to be clipped, then pick the clip-
ping surface. The result will be as shown in FIGURE 44.
Note: If your model is parametric, make sure that the clipping surface is also para-
metric so that it will encompass the entire object to be clipped, regardless of the
dimensions corresponding to any given set of parameter values.
Performing 3D Edge Blending Operations

The following paragraphs describe edge blending operations that may be used to
create rolling ball or variable radius fillets and chamfers in a solid body. Note that
the edges to be selected for application of the edge blend must all belong to a single

Create Clip Plane
Pick body then clip plane
Result
FIGURE 44 Definition of a Clipped Body
solid body. Using the simple object from the previous section as an example,
we can apply a rolling ball fillet to the edges of the hole as follows:
• Create > Blend Edge > Rolling Ball > Input switch on > Radius=0.2.
Select the top and bottom edge of the hole, then click on the Accept
button. The resulting body is shown in FIGURE 45.
A chamfer is a flat blend which replaces the selected edge(s) with a flat face,
where the distance from the original edge on the selected face is distance “R1”,

Select edges
Result
FIGURE 45 Definition of a Rolling Ball Edge Blend
and the distance from the original edge on the neighboring face is distance “R2”. To
create a chamfer edge blend, do the following:
• Create > Blend Edge > Chamfer > Input switch on > R1=0.1, R2=0.2. First,
select each edge to be blended. Then, while holding both the Shift and Ctrl
keys down simultaneously, select each face which corresponds to where the
distance “R1” will be measured. Click on the Accept button to complete the
chamfer creation. NOTE: Before picking the edges to be blended, it is usu-
ally necessary to turn off surface shading, in order to avoid a conflict
between selecting faces and edges. FIGURE 46 illustrates the construction

of a chamfer edge blend. A chamfer may also be used to construct a
First pick edges
0.2
0.1
Next pick “R1” faces
Result
Disable surface shading before picking edges
FIGURE 46 Definition of a Chamfer Edge Blend
countersink hole as shown in FIGURE 47.
A variable radius blend is characterized by 6 parameters: two radii at each end

(R1-a, R1-b, R2-a, R2-b) and a shape factor at each end of the blend (Rho1,
Rho2). The following situations are possible depending on the selection of
these parameters:

Pick “r1” face

r2
r1
Pick chamfer edge
Result
FIGURE 47 Definition of a Counter Sink Hole Using a Chamfer Edge Blend.
• if all radii are equal and Rho1=Rho2=0, then the blend has a circular cross
section
• if different radius values are specified on each side and the Rho values are
zero, then the blend has an elliptical cross section.
• further control of the shape of the cross section is obtained by setting Rho
to non-zero values:
-Rho between 0.0 and 0.5 for elliptical cross section
-Rho= 0.5 for a parabolic cross section
-Rho between 0.5 and 1.0 for hyperbolic cross section. As rho approaches
1.0, the blend becomes more and more L-shaped. Rho should be always
strictly less than 1.0.
To create the variable radius edge blend illustrated in FIGURE 48, proceed as fol-
lows:

• Create > Blend Edge > Variable Radius, R1-a= 0.2, R1-b= 0.2, R2-a=
0.4, R2-b= 0.4, Rho1= 0.0, Rho2= 0.0, pick the edge, click on the
Accept button. The “R1” input corresponds to the radius of the blend at
the start of the edge, “R2” corresponds to the radius of the blend at the
end of the edge. The start and end of the edge is based on the edge
direction, and can currently be determined only by trial and error. It
should be noted that you may apply a variable radius blend to only one
edge at a time.
Pick Edge
R2
R1
4
Result
FIGURE 48 Definition of a Variable

Radius Edge Blend

Performing 3D Face Blending Operations
Performing 3D Face Blending Operations

A rolling ball face blend operation is essentially a hybrid between the surface fillet
and the edge blend operations. The surface fillet takes as input two distinct sheet
bodies, produces a new sheet body that represents the fillet surface, and retains the
original sheet bodies. The face blend is more general in that it can take as input two
or more sheet bodies and produces a single sheet body as a result which contains
portions of the original sheet bodies as well as a new face representing the fillet
joining the original sheet bodies. In the case of the face blend, you must distinguish
the “right” wall of the blend from the “left” wall of the blend. For the face blend,
there are two cases which are valid:
- All left and right wall faces belong to the same solid
- All left wall faces must belong to one general body and all right wall faces
must belong to a different general body
To create a face blend, do the following:

4
• Create > Blend Face > Rolling Ball > Radius= 0.2. Next pick the faces
which represent the left wall of the blend. To pick the faces which represent
the “right” wall, hold the Shift and Ctrl keys simultaneously while you pick
the right wall faces. You will see that the left and right wall faces are high-
lighted in a different color. You may pick right and left wall faces in any
order, as long as you use the Shift/Ctrl keys to distinguish them while pick-
ing. When all faces have been selected, click on the Accept button. If the
input was a solid body the result is a solid body. If the input faces came
from two distinct general bodies, the result will either be a general body, or
a solid body if the result satisfies the appropriate requirements of a solid
body. The face blend is illustrated in FIGURE 49.
As a more general example of the face blend, suppose we have the result of the pre-
vious example, and a plane which cuts across the general body as shown in FIG-
URE 50. Select each face of the general body as the right wall, and select the plane
as the left wall. The result of this face blend for a radius of 0.2 is shown in FIGURE
50.

Example: Construction of a Simple Bracket
Right wall
Left wall
4
Result
FIGURE 49 Definition of a Face Blend

The following paragraphs describe the construction of a simple bracket model
in parametric form using solid modeling operations. The result that will be
achieved at the end of this exercise is shown in FIGURE 51.
Begin by defining the parameters that will be needed to describe the various
dimensions of the bracket. FIGURE 52 illustrates the Model Info interface
where parameters are defined. Shown are the parameter names, descriptions

Pick 3 right wall faces
Pick 1 left wall face
Result
FIGURE 50 Definition of Another Face Blend
and default values assigned for the bracket model. Parameter definition and the
Model Info interface is discussed in detail in the User’s Guide.
After defining the parameters, create the basic model geometry as follows:
switch on, X= 0, Y= 0, Z= 0, Width= width, Height= height, Depth= thick,
Rot-X= 0, Rot-Y= 0, Rot-Z= 0 > Accept button.
• Create > Box > Locate > Solid button on > Input toggle switch on, X= 0,
Y= (height-fthick)/2, Z= 0, Width= width, Height= fthick, Depth= fheight,
Create the cylinders as follows:

• a - bholedist
• b - bholeradius
• c - csdepth
• d - cswidth
f • e - fheight
• f - fholeradius
• g - filletradius
• h - fthick
• i - height
h g • j - thick
e
• k - width
c b
d
4
k
j
FIGURE 51 Simple Bracket Model
• Create > Cylinder > Locate > Solid button on > Input toggle switch
on, X=0, Y= height/2, Z= fheight, Radius= width/2, Height= fthick,
on, X=0, Y= height/2, Z= fheight, Radius= fholeradius, Height=
fthick, Rot-X= 90, Rot-Y= 0, Rot-Z= 0 > Accept button.
on, X=0, Y= -bholedist, Z= 0, Radius= bholeradius, Height= thick,
on, X=-width/2, Y= -height/2, Z= 0, Radius= thick, Height= width,

FIGURE 52 Example Bracket Parameter Definitions

4
To create a clipping plane, do the following:

• Geometry tab > Create > Plane > Locate > Input switch on, X= 0, Y=0, Z=
0, Width= width, Height= height*2, P1-Min= -0.5, P1-Max= 0.5, P2-Min=
-0.5, P2-Max= 0.5, Rot-X= 0, Rot-Y= 0, Rot-Z= 0.
At this stage, your model should look like FIGURE 53.
Before applying boolean operations, perform a clip operation (see FIGURE 54) to
divide the transverse cylinder in half as follows:
• Create > Body > Clip-Back. Pick the transverse cylinder as the target, then
pick the clipping plane.
You are now ready to begin applying boolean operations to your model. First, (as
shown in FIGURE 55) join the Base Box, Flange Box, and Flange Cylinder by
doing the following:
• Create > Body > Bool-Union. Pick the clipped Transverse Cylinder as the
target, pick the Base Box, pick the Flange Box, and pick the Flange Cylin-
der. Click on the Accept button.
Next, (see FIGURE 56) create the two holes by subtracting the base hole cylinder
and the flange hole cylinder from the result body of the previous operation as fol-
lows:

Flange Cylinder
Flange Hole Cylinder
Flange Box
Base Hole Cylinder

Fillet edge Base Top Face
Transverse cylinder
Base Box
Countersink edge
Clipping plane
FIGURE 53 Bracket Construction Primitive Object Illustration
• Create > Body > Bool-Subtract. Pick the result body from the previous
operation as the target, pick the Base Hole Cylinder, and pick the
Flange Hole Cylinder. Click on Accept
Next, create the fillet (see FIGURE 57) as follows:
• Create > Blend Edge > Rolling Ball, Radius= filletradius. Pick the fil-
let edge. Click on Accept
Finally, create the countersink (see FIGURE 58) on at the base hole as follows:
• Create > Blend Edge > Chamfer, R1= cswidth, R2= csdepth. Pick the
countersink edge (you may have to disable surface display). Hold Ctrl/
Shift keys and pick the base top face. Click on Accept.
Save an input file for this model, as it will be used in the next chapter to illus-
trate meshing techniques. You are done

FIGURE 54 Bracket Construction

Clip Operation
Pick transverse cylinder first
Pick clipping plane second
Flange Cylinder
Flange Box
Transverse Cylinder

Base Box Boolean Union Operation

Flange Hole Cylinder
Base Hole Cylinder

Boolean Subtraction Operation
Result Body Target
Fillet edge
FIGURE 57 Bracket
Construction Fillet Creation

Pick chamfer edge first
Pick top face
FIGURE 58 Bracket Construction Countersink Creation


5 Mesh Creation
This chapter contains a general overview of the features of StressCheck used for creating finite element
meshes for two and three dimensional models. For a general description of the preparation of a com-
plete finite element model, please refer to the User’s Guide.
Introduction to Meshing
From the Class menu, select Mesh. The set of options under Mesh provide for the
specification of the solution domain using nodes, fasteners and elements. In the fol-
lowing sections, various techniques will be described for constructing and revising
basic mesh objects.
Mesh Objects
Mesh objects in StressCheck are distinct from geometry objects. This is because
mesh objects are usually attached to geometry objects in order that the model will
be parametrically associative. Nodes are similar in definition to geometric points, in
that they may be constructed by a variety of methods (global, local, offset, projec-
tion, etc.). However, only nodes may be referenced in the construction of elements.
Modeling Guide Chapter 5: Mesh Creation 91

Node A node is a location in space which will be referenced in the definition of an

element. A Node may be defined using a variety of methods including: global
location, attached to a local system, as an offset to a curve or surface, as a mid-
point between two point-like objects (point, node, system), as a midoffset
between two point-like objects that are attached to the same curve or surface,
as a projection of a point-like object to a curve or surface, as an intersection of
two curves, or attached to a point. Nodes which have been attached to a point
will automatically inherit the associativity of the underlying point and pass this
information to elements which reference the node.
Element Elements are constructed by selecting previously defined nodes. If the nodes
used to construct an element are attached to geometry, the elements will inherit
the characteristics of the underlying geometry. If the two nodes of an element
edge are attached to the same curve or surface, the edge will be mapped
directly from the associated curve or surface. If all the nodes of an element face
are all attached to the same surface, that face will be mapped directly from that
surface. If one or more edges of a face are curved, but not all edges of the face
are attached to the same surface, the face will be curved by performing a blend-
ing operation to produce a Coons patch representation for the face.
Elements are simply topological constructions used to represent the shape of a

part. All other characteristics of the element are determined from the current
reference system and theory, and from the user assigned attributes. In this man-
5 ner, a single element definition may be used to solve a problem of elasticity or
heat transfer, and may be used to perform a plane strain, plane stress, axisym-
metric, or plate bending analysis.
All elements constructed in StressCheck are geometrically associative, i.e. they

inherit shape characteristics from underlying geometry. StressCheck also sup-
ports midside elements (isoparametric). Isoparametric elements can be created
manually by identifying the vertex nodes, then drag individual edges to mid-
side nodes to utilize isoparametric mapping (Create>Edge>Midpoint). This
new functionality makes it possible to automatically generate a mesh that has
isoparametric elements, and to later add or modify elements manually. When
adding a new element, each edge of the new element that is shared with an
existing element will automatically inherit the midside node from its neighbor.
Beam, triangle, quadrilateral, tetrahedron, pentahedron, and hexahedron now
support manual creation of isoparametrically mapped elements. This enhance-
ment also supports the connection of geometrically associative elements to iso-
parametrically mapped elements.
92 Chapter 5: Mesh Creation Modeling Guide

Isoparametric elements may be imported into StressCheck using the NASTRAN

Bulk Data interface. Isoparametric elements include the 6-noded triangle, 8-noded
quadrilateral, 10-noded tetrahedron, 15-noded pentahedron, and 20-noded hexahe-
dron. These elements perform well in StressCheck as long as the polynomial level
does not exceed 4 or 5.
Edge An Edge is defined as the path between two nodes in the definition of an element.
An element edge may be referenced in the definition of boundary conditions, and
for post processing operations such as resultants, point extraction, 3D fracture
mechanics, etc. Using the MoveTo action, a straight edge may be attached to a
curve or surface by selecting the edge and dragging it to the desired curve or sur-
face. Similarly, a curved edge may be converted to a straight edge by simply select-
ing it when the action is MoveTo.
Edge Curve An Edge Curve is a sequence of edges which are slope continuous. The tolerance
used to determine continuity is 0.1 degrees by default. When the EdgeCurve option
is selected, a tolerance field will appear so that you may specify the degree of con-
tinuity. The EdgeCurve option is convenient for specifying boundary conditions or
performing post processing operations in order to avoid having to select each edge
individually.
Face A Face is defined as the surface formed by neighboring nodes of an element. An

element face may be referenced in the definition of boundary conditions, and for 5
post processing operations such as resultants, point extraction, etc. Using the
MoveTo action, a straight face may be attached to a surface by selecting the face
and dragging it to the desired surface. Similarly, a curved face may be converted to
a straight face by simply selecting it when the action is MoveTo.
Face Surface A Face Surface is a sequence of faces which are slope continuous. The tolerance
used to determine continuity is 0.1 degrees by default. When the FaceSurface
option is selected, a tolerance field will appear so that you may specify the degree
of continuity. The FaceSurface option is convenient for specifying boundary condi-
tions or performing post processing operations in order to avoid having to select
each face individually.
Triangle A Triangle is an element formed by connecting three nodes which may be used to
represent a membrane, an axisymmetric solid or a plate. The nodes may be selected
in any order. Three dimensional shell elements are not currently supported in
StressCheck, however a triangular element may be constructed in the 3D reference
system in order to construct a pentahedral element by the face to face construction

method. If used for this purpose, triangular elements must be deleted from the
model before performing an analysis.
Quadrilateral A Quadrilateral is defined as an element formed by connecting 4 nodes which

may be used to represent a membrane, an axisymmetric solid or a plate. The
nodes may be selected in any order. Three dimensional shell elements are not
currently supported in StressCheck, however a quadrilateral element may be
constructed in the 3D reference system in order to construct a hexahedral ele-
ment by the face to face construction method. If used for this purpose, quadri-
lateral elements must be deleted from the model before performing an analysis.
Fastener A Fastener is a special element that can be attached to circular boundaries. It

consists of a rigid core with two degrees of freedom connected to the planar
body by a distributed radial (normal) spring. The rigid core can be connected to
the rigid core of other fasteners directly or using a link element, it can be fixed
in one or both directions, it can be loaded by a force or an imposed displace-
ment.
Link A Link is a special element to connect fastener elements allowing the simula-
tion of the shear/bending stiffness of the fastener.
Beam A beam is defined by selecting two nodes in any order. The order in which the
nodes are picked when creating the elements define the principal directions of
5 the beam. The beam axis is defined in the direction from the first to the second
node. A beam element cannot be used in conjunction with other element type.
Beam elements can be used in Planar and 3D Elasticity.
Tetrahedron A Tetrahedron is defined as an element formed by connecting 4 nodes in the

three dimensional reference system. The nodes may be selected in any order.
The tetrahedron element may also be constructed by the Face to Face method
by connecting an existing triangular element face and one node.
Pentahedron A Pentahedron is an element formed by connecting 6 nodes in the three dimen-

sional reference system. When constructing the element, you should select
them in order by first selecting the nodes which will form one triangular face
and then the opposite triangular face. The pentahedral element may also be
constructed by the Face to Face method by connecting two existing triangular
element faces, or by connecting one quadrilateral face and one edge.

Hexahedron A Hexahedron is defined as an element formed by connecting 8 nodes in the three

dimensional reference system. The nodes may be selected in any order. The hexa-
hedral element may be constructed by the Face to Face method by connecting two
existing quadrilateral faces.
Editing Nodes A node definition may be modified either graphically, or manually. To edit a node
graphically, choose the MoveTo action and select the desired node. The interface
will automatically update to indicate the method by which the node was originally
defined. If you wish to preserve the original method, simply redefine the node
graphically as though you were creating it from scratch. If you wish to change the
method of creation, simply change the method and redefine the node.
To change a node definition manually, choose the Edit action and select the node
with the mouse cursor. The original definition will appear in the input interface.
You may then enter new information (x,y,z location, offset, etc.) and click the
Replace button. The Edit action is not appropriate for nodes create by intersection,
midpoint, midoffset, or projection methods, since these methods require graphic
selections of the associated objects.
If you wish to change the definition of a node by incrementing/decrementing the

global/local coordinate or offset value, you may enter the delta value in the corre-
sponding input field and click the Move button. Move operations are only appropri-
ate for objects that have purely constant (not parametric) attributes (coordinate or
offset). 5
Editing Elements There are several ways to change the definition of an element in StressCheck. An
element may be modified graphically using the MoveTo action. Simply select the
element to be modified, and then reselect the nodes which will be used to represent
the element. If you wish to change the type of element (e.g. from hexahedron to
pentahedron), simply specify the object you wish to create, select the element to
modify, and select the appropriate nodes.
It is also possible to change an element definition by entering the node numbers.

Use the Edit action and select the desired element. The corresponding node num-
bers will appear in the user interface. Simply enable the input check boxes and
enter the node numbers for the new element definition. Click the Replace button
when you are finished entering the new node numbering.
The Edit action may be used to change an element definition using a specific node
ordering. When constructing an element using the Create action, the nodes ordering
is always determined automatically in such a way that the volume of the element is

maximized. This automatic ordering may not produce the desired element for
your particular model. If this is the case, use the Edit action and select the
desired element. Select the nodes in the order you want to preserve in the ele-
ment definition.
If you make a mistake while selecting nodes during the construction of an ele-
ment, you may use the Backspace key to cancel the last node selected.
You may obtain a list of nodes each element is attached to by switching to the
Index tab in the Mesh Class interface. Selecting the element listed in the index
will feed back a list of nodes referenced by the element. Selecting a node will
feed back the underlying objects that the node is attached to.
Basic 2D Mesh Construction

Once the underlying geometry has been defined, you may create the finite ele-
ment mesh attached to the geometry. StressCheck provides both manual and
automatic mesh construction tools.
To construct a finite element manually, you must first have nodes. Nodes and
points are distinct objects and should not be confused. Points are geometric
objects. Nodes should be thought of as topological objects that define connec-
tions between elements. They may be defined by a variety of methods: by glo-
5 bal coordinates, by local coordinates, attached to a geometric point, as an offset
on a geometric boundary, or as an intersection between two geometric bound-
aries, etc. These are all associative relationships. When you attach a node to a
circle, the node definition “remembers” this attachment. If you later attach an
element to a node which is attached to a circle, this associativity will be inher-
ited by the element.
Note that if you attach a point to a boundary and then attach a node to this
point, the node will inherit the association with the boundary, and the element
attached to this node will automatically obtain a curved edge or face.
Try the following example. Set the Reference selector to Planar, and create a
circle as follows:
• Geometry tab > Create > Circle > Locate > Input switch on, X: 0,
Y: 0, Z: 0, Radius: 1, P1-Min: 0, P1-Max: 360, Rot-Z: 0 > Accept but-
ton.
Create two global nodes as follows:

• Mesh tab > Create > Node > Locate > Input switch on > X: 2, Y: -1, Z: 0 >
Accept button
and
• Mesh tab > Create > Node > Locate > Input switch on > X: 2, Y: 1, Z: 0 >
Accept button
Next create two Offset nodes as follows:
• Mesh tab > Create > Node > Offset > Input switch on > Offset: -45 (Click
on the circle).
and
• Mesh tab > Create > Node > Offset > Input switch on > Offset: 45 (Pick the
circle again).
Now your are ready to create a quadrilateral element as follows:
• Mesh tab > Create > Quadrilateral > Selection. Select each node with the
mouse cursor (select them in any order).
After selecting the fourth node you should get an object which looks like FIGURE
59. If you make a mistake and select the same node twice, nothing bad will happen.
FIGURE 59 Element construction.

No element will be created until you identify 4 distinct nodes. If you choose
the wrong node while defining an element, you may click the right mouse but-
ton to abort the generation. Alternatively, you may use the Backspace key on
the keyboard to un-select the nodes you have selected by mistake. In this way
you do not have to select all nodes again, only continue from where you left
off.
Note: It is also possible to define an element by selecting all desired nodes in

one operation using what is called a marquee pick. Simply drag a box around
the desired nodes. You may select nodes one at a time, two at a time, or several
at a time in order to create elements. Similarly, you may cancel them in the
same fashion by pressing the CTRL key while you perform a marquee pick.
Notice that the element is automatically assigned one circular edge. This was
possible because of the associativity of nodes 3 and 4. The program will
always look for associativity and create curved edges when possible. This is
usually what users intend. If not, you may replace a curved edge with a straight
edge by using the Mesh tab > MoveTo > Edge > Selection command and then
clicking on the curved edge. The curved edge will become straight. Try this. If
you select this straight edge again, the edge will become dynamic and will float
until you select the circular boundary. Then the edge will become circular,
again.
In general, you may make an edge curved by using the “Create” or “Move To”
5 actions with the “Edge” object.
If the nodes on the straight edge which you wish to attach to a geometric
boundary are not already associated with the selected boundary, the nodes will
be projected onto the selected boundary automatically, converted to associative
boundary offset node definitions, and the curved edge will be incorporated into
the element definition.
Try the example illustrated in FIGURE 60. First erase all objects by using the
File > Erase Database option from the Main Menu Bar. Next, create the circle
as follows:
• Geometry tab > Create > Circle > Locate > Input switch OFF (Screen)
Move mouse cursor to the display area and construct a circle by clicking the
location of the center and moving the mouse to display the dynamic circle.
Click again to fix the radius and construct the circle (size is arbitrary for this
example). Now create four arbitrary nodes in a pattern similar to the example
shown in FIGURE 60.

Before After
FIGURE 60 Projection of element edge to a boundary.
• Mesh tab > Create > Node > Locate > Input switch OFF (Screen)
Next, construct an arbitrary element as follows:

5
• Mesh tab > Create > Quadrilateral > Selection. Click on the four nodes.
Now you are ready to project the element edge closest to the circle onto the circle.
Do the following:
• Mesh tab > Create > Edge > Selection
Click near the element edge closest to the circle. Move the cursor until it is near the
circle and click again. The element will be attached to the circle.
Semi-automatic Element Creation in Two Dimensions

In addition to standard element creation by individual node selection, and marquee
selection, StressCheck provides a facility for creating individual elements automat-
ically. If you press both the CTRL and SHIFT keys as you move the cursor across
the model window, the program will automatically sense the location of the cursor,
and its proximity to various nodes that have been defined in your model. Stress-
Check will display a candidate element connecting the nodes within the proximity

of the mouse cursor that will form a valid element of the chosen type (beam,
triangle, quadrilateral). If the candidate element is the one you want, simply
click the left mouse button, and the element will be created. It is sometimes
necessary to rotate or translate the model to obtain an orientation in which the
candidate element can be properly identified. You may manipulate the view
using the right mouse button, without leaving the element creation operation.
FIGURE 61 illustrates the display of a candidate element.
Press CTRL & SHIFT keys while

moving mouse cursor
FIGURE 61 Candidate 2D Element Construction
To create a quadrilateral element automatically, do the following:

• Mesh tab > Create > Quadrilateral > Selection. Press both the CTRL
and SHIFT keys. Move mouse cursor over model window until the
desired element appears. Rotate or translate model if necessary. Click
left mouse button to create the candidate element.
Sometimes it is difficult to position the mouse cursor in such a way that the
candidate element is the one that you are seeking. In this case, it is possible to
indicate to StressCheck that you wish to identify the first edge of a 2D element,

then move the cursor to complete the element. To operate in this two step mode,
simply hold the SHIFT key while you move the cursor over the model window. You
will see a candidate edge appear. When you see the desired edge, click the left
mouse cursor. When you click the mouse, the nodes associated with the chosen
edge will be highlighted. As you move the cursor, the selected edge will be pre-
served, and any subsequent candidate element will include this edge. When you
click again, the candidate element will be created. FIGURE 62 illustrates the dis-
play of a candidate edge.
Press SHIFT key while

moving mouse cursor
FIGURE 62 Candidate Edge Construction
Sometimes even with these candidate element construction procedures, it is diffi-

cult to identify the desired element. In this case, you may wish to select the desired
nodes individually, then let the automatic procedure complete the element for you.
To select nodes individually, you may use either the marquee pick or individual
node selection. You may marquee pick nodes at any time by simply pressing the left
mouse button and dragging the selection rectangle until it surrounds the desired
nodes. When you release the mouse button, press the CTRL & SHIFT keys and the
automatic selection procedure will take over and display the candidate edge if you
have identified fewer than half the necessary nodes, or the complete candidate ele-
ment if you have identified at least half of the necessary nodes. If you wish to select
individual nodes just release the CTRL and SHIFT keys, and make your individual
node selections.

To summarize the semi-automatic element construction procedures:
NO key depressed - select individual nodes (suppress candidate element

evaluation and display)
SHIFT key depressed - select nodes in proximity of cursor to form first
edge/face of candidate element
CTRL and SHIFT keys depressed - select nodes in proximity of cursor
to produce complete candidate
element.
Note: As with any multi-step construction technique, you may abort the cur-
rent operation by clicking the right mouse button.
Mapped Mesh Generation in Two Dimensions

StressCheck provides a simple way to generate quadrilateral elements using
the “Quad-Mesh” object and “Edge to Edge” method. Each curve has its own
reference parameter U (also referred to as P1). A mesh of quadrilateral ele-
ments will fill the space between two reference curves, by placing nodes along
each curve equally spaced in the parameter space of each curve, in the range
specified in the input. P1-Min and P1-Max refer to the first curve selected
while P2-Min and P2-Max refer to the second curve selected. N-P1 is the num-
ber of elements to be constructed. Use the mouse cursor to select each refer-
5
ence curve. Click DeLast to remove the elements if desired. See FIGURE 63
for an example of mapped meshing in two dimensions.
Mesh Revision
So far, we have described how to create mesh objects. It is very important to be
able to change objects after they have been created. The Action section of the
Mesh Input Class contains several mechanisms for changing the contents of
your finite element model. The various actions available in the Mesh Class
may be used in exactly the same way as for Geometry, to Create, Select, Move
To, and Edit a mesh object. In the Mesh Class, the Check action provides addi-
tional functionality as described below.
Use Check to find the distance between nodes. If you check element edges
(Check > Edge > FreeEdge), all free edges will be selected and highlighted
automatically. This is very useful for determining whether there are any cracks
or improperly connected regions in the model.

FIGURE 63 Example of mapped meshing in

two-dimensions.
Use Check on elements to compute element properties or to check element distor- 5

tion. FIGURE 64 illustrates the use of the element cross-sectional property calcula-
tion using the Check action. Simply select Mesh tab > Check > Any Element >
Properties and the desired element with the mouse cursor, then click on the Accept
button. The cross-sectional properties of the selected element will be reported in the
Edit Window. This report may be saved to a file if desired.
Element distortion may be checked in a similar manner. FIGURE 65 illustrates a

distortion check and the report produced in the edit window. Note that the triangu-
lar element has two vertex angles outside limits. If you wish to override the default
valid angle range, simply enter the values in the MinAng and MaxAng input fields.
The smallest vertex angle accepted for an element is zero degrees and the maxi-
mum is 179 degrees. An element with a vertex angle greater than 179 degrees will
be rejected, and the execution will not be performed until the element is modified.
Note on vanishing angles Elements with vanishing angles (zero vertex angle) are acceptable for linear and
material nonlinear analysis but cannot be used for general nonlinear analysis. They
require a special post processing procedure (which is performed automatically by
the program) and their main advantage is that they facilitate meshing in filleted

FIGURE 64 Element Property Check.
FIGURE 65 Element Distortion Check.
regions. Quadrilateral, triangular, hexahedral, pentahedral and tetrahedral ele-

ments can have one or more vanishing angle. An example is shown in FIGURE
66. Note that the contour fringes in the region of the fillet are not affected by
the vanishing angle element.

FIGURE 66 Model problem with vanishing angle elements.

5
Merging nodes The Merge button at the bottom of the Input dialog window may be used to remove
redundant nodes from the model. The program searches for nodes which are less
than 10-6 units apart. All but one of the coincident nodes are discarded, and all ele-
ment definitions are updated appropriately. The tolerance for determining redun-
dancy may be adjusted by supplying a parameter called “_merge_tol” with the
desired tolerance value. Note that using Merge when the model has an intentional
crack will cause the crack to be removed.
If any nodes are selected prior to the merge, the operation will be limited to those
selected nodes. Otherwise the merge operation will apply to all nodes.
Note: Refer to Automeshing at the end of this chapter for an explanation about the
available options for automatic mesh creation.

Basic 3D Mesh Construction

With the introduction of solids-based geometry construction, three dimensional
model construction has become much simpler and more robust. Solid models
automatically provide many of the associative relationships that are needed for
proper meshing. In this section a few basic guidelines are given that should be
followed when preparing a three dimensional model for meshing.
When you are preparing to mesh a solid body, the first step is usually to attach
nodes to all the point objects that were automatically created as part of the
solid model representation. These points represent the locations of curve inter-
sections, and as such, must be used to locate the vertices of finite elements. As
noted earlier, nodes which are attached to points will automatically inherit the
underlying associativity of the corresponding point. StressCheck makes it easy
to attach nodes to points in a model as follows:
1. Select all points you wish to attach nodes to. Use a marquee pick, or use the
invert selection icon to select all points not currently selected.
2. Choose Create > Node > Point
3. Click on the Accept button. Nodes will automatically be attached to all

selected points.
5 If you wish to attach nodes to every point there is no need to select them, sim-
ply click on the Accept button. There are several additional issues to be consid-
ered while constructing your finite element model. Your meshing strategy will
depend to some degree on whether your model will be parametric or not. Static
(non-parametric) models are much easier to mesh because you do not need to
be concerned about whether elements will become distorted under various
parametric configurations. Another important consideration of parametric
models is that the underlying solid modeling operations can be significantly
affected by changes to the dimensions or changes in the relative positions of
target and tool objects that were used to construct the resulting solid body that
you wish to mesh.
One type of problem that can arise when parameter values are changed is that
the topological structure of the solid model can change. For example, if a rela-
tively small change is made to the size or position of an interior cylindrical tool
representing a hole in a target body, the overall topology of the resulting body
remains the same. However, a relatively large change in size or position of the
tool may cause the tool to intersect the surface of the target body in such a way

that the topology of the resulting body is changed. Any topological change that
alters the number, type, or orientation of the curves or surfaces used to represent the
resulting body may cause dependent geometry and mesh definitions to become
invalid. The only protection against such changes is to provide parameter limits and
relational rules that will prevent the assignment of values to parameters which will
cause the model to become invalid. Rule definition is discussed in detail in the
User’s Guide.
It is possible that a simple parametric change will result in a subtle change to the
representation of the body that is not visually perceptible, but which will have a sig-
nificant impact on dependent object definitions. The most common change of this
nature occurs when a surface or curve which forms a face or edge of a solid body
changes orientation as a result of a parametric change. For example, a circle repre-
senting the edge of a cylindrical hole may change direction from clock-wise to
counter-clock-wise. Because this situation is not uncommon, it is important to take
some basic precautions that will protect your model from being adversely affected
by such changes.
One important precaution that will help to maximize the robustness of a parametric
model is to minimize the use of “offset” definitions. In other words, do not create
points, nodes, or systems attached to the curves or surfaces of a solid model using
the offset method. If the parameter space of the underlying curve or surface would
happen to reverse as a result of a parametric change, the relative position of the
associated point-type objects would reverse, thus having many potential and undes- 5
ired consequences. The alternative to creating offset definitions is to use intersec-
tion, mid-offset and projection methods where ever possible. These definitions are
more resilient to parametric model changes. There are usually plenty of points gen-
erated automatically as part of a solid model that may be used as reference points
for mid-offset and projection methods, and it is fairly rare that an offset definition is
necessary. Furthermore, it is usually possible to define new reference objects in
some parametric fashion that can then be used to create the necessary associative
objects needed for mesh construction.
Note: If you do create auxiliary curves or surfaces for this purpose, it is not a prob-
lem to attach point-type objects as offsets on the auxiliary boundaries, since their
orientation is not subject to the same reversal conditions as the edges and faces of a
solid body.

Semi-automatic Element Creation in Three Dimensions

As described in the earlier section, StressCheck provides a facility for creating
individual 2D elements automatically. This same facility is provided for con-
structing 3D elements. While creating elements with the Selection method, if
you press the CTRL and SHIFT keys as you move the mouse cursor across the
model window, the program will automatically sense the location of the cursor
and display a candidate 3D element based on the proximity to various nodes
that have been defined in your model. If the candidate element is the one you
want, simply click the left mouse button, and the element will be created. FIG-
URE 67 illustrates the display of a candidate 3D element.
Press CTRL & SHIFT keys

while moving mouse cursor
FIGURE 67 Candidate 3D Element Construction
To create a hexahedral element automatically, do the following:

• Mesh tab > Create > Hexahedron > Selection. Press and hold both the
CTRL and SHIFT keys. Move mouse cursor over model window until

the desired element appears. Rotate or translate model if necessary. Click

left mouse button to create the candidate element.
As in 2D, it is sometimes difficult to position the mouse cursor in such a way that
the candidate element is the one that you are seeking. In this case, it is possible to
indicate to StressCheck that you wish to identify the first face of a 3D element, then
move the cursor to complete the element. To operate in this two step mode, simply
hold the SHIFT key while you move the mouse cursor over the model window. You
will see a candidate face appear. When you see the desired face, click the left mouse
button. When you click the mouse, the nodes associated with the chosen face will
be highlighted. As you move the cursor, the selected face will be preserved, and any
subsequent candidate element will include this face. When you click again, the can-
didate element will be created. FIGURE 68 illustrates the display of a candidate
face.
Press SHIFT key while

moving mouse cursor
FIGURE 68 Candidate Face Construction

As in 2D, you may select nodes individually or using the marquee pick. If you
subsequently hold the CTRL and SHIFT keys down, the automatic element
construction will take over and display the candidate face if you have identi-
fied fewer than half the necessary nodes, or the complete candidate element if
you have identified at least half of the necessary nodes.
When constructing a tetrahedral or pentahedral element using the candidate

face technique, the first candidate face is triangular.
Note: When constructing a 3D element automatically, the candidate element

may not look valid. This is because candidate node selection is based on con-
structing an element with straight edges and faces. Before the candidate ele-
ment is displayed however, the associativity is evaluated and the actual
element using the candidate nodes is then displayed. It is not possible to actu-
ally create an invalid element, since it will first be evaluated, and then rejected
if it is not valid. This technique is more efficient than evaluating the associativ-
ity for all possible candidate elements.
Example Mesh Generation in Three Dimensions

Using the solid model constructed in the previous chapter and shown in FIG-
URE 69, we will now construct a finite element mesh for a simple three dimen-
sional solid body. The objective will be to create nodes attached to the curves
5
and surfaces of the solid body such that we can create a mesh which is well-
behaved within reasonable limits of the parameter values. We begin by creat-
ing a surface curve as a projection of the countersink edge from the top surface
of the base of the lug. This curve will be needed later to create a ring of hexa-
hedral elements around the countersunk hole. To do this, do the following:
• Geometry tab > Create > Surface Curve > Projection. Pick the bottom
surface of the base of the lug. Pick the outer ring of the countersink
(you may need to turn off Display Surfaces to do this).
The next step is to attach nodes to all the points which were automatically cre-
ated as part of the solid part representation. To do this, do the following:
• Mesh tab > Create > Node > Point > Accept. A message will appear
asking if you want to attach a node to every point. Click Yes. A node
will be created at the location of each point.
Next, create mid-offset nodes between the nodes just created as shown in FIG-
URE 70.

Existing points to attach nodes
Countersink outer ring
Bottom surface of lug base

(hidden)
FIGURE 69 Sample bracket geometry
• Mesh tab > Create > Node > Mid-Offset. Disable the Display Points icon to
suppress displaying point objects. Pick each pair of nodes that were
attached to points (each mid-offset node will automatically be created after
the second selection is made). At the center of the fillet, and at the top of
the flange, use the new mid-offset nodes to create a secondary layer of mid-
offset nodes.
Now create one midpoint node between the two mid-offset fillet nodes.
• Mesh tab > Create > Node > Midpoint. Pick the two mid-offset nodes of the
fillet.
Next, create projection nodes using the previously created nodes as a reference (See
FIGURE 71). The steps to do this are as follows:

Midoffset node
Secondary midoffset node
Nodes attached to points
Midoffset node
Midpoint node
5
FIGURE 70 Construction of mid-offset nodes
Disable the Display Surfaces icon to suppress displaying surface objects. This
is important to avoid projection of nodes to surfaces by mistake. Unless other-
wise indicated, all nodes should be projected to curves in order to insure that
the nodes inherit the multiple associativities implied by the intersection curves
representing the edges of the model.
• Mesh tab > Create > Node > Projection. Select a node and then push
Control-Shift and select the curve that it will be projected to.
You may automatically project all of the newly-created nodes on the counter-
sink ring to the other rings. To do this, do the following:
• Create > Node > Projection. Select all the nodes on the countersink
ring that were created in the previous step. Then, while holding Con-

Reference node
Projection node
Reference node
Reference curve
Projection node
5
FIGURE 71 Construction of projection nodes
trol-Shift, pick the inner countersink ring, the lower countersink ring, and
the projected countersink curve on the bottom surface. Click Accept.
Next, create three nodes as projections to the opposing surfaces as shown in FIG-
URE 72.
• Create > Node > Projection. Enable Display Surfaces icon (surfaces must
be visible to project nodes). Pick the reference node. Pick the reference sur-
face while holding Control-Shift (you must rotate model to view surface).
Repeat for each reference node.

Back flange surface
Reference nodes
Bottom surface
Projected nodes
5
FIGURE 72 Projection of nodes to surfaces
At this point, all the nodes that are necessary for meshing are available. The
next step is to begin creating elements. You have several methods at your dis-
posal for creating 3D element: manual node selection, marquee node selection,
face to face construction based on existing elements, and semi-automatic ele-
ment construction.
To mesh the sample bracket model, first create the two pentahedral elements as
shown in FIGURE 73.
• Create > Pentahedron > Selection. Pick nodes individually or using a

marquee pick, by first choosing three nodes representing one triangular
face, then the three nodes representing the other triangular face.

Auto method candidate face nodes 5
FIGURE 73 Construction of penta elements
Repeat for the second pentahedral element. Alternatively, use the automatic method
as follows:
• Create > Pentahedron > Selection. Press and hold the SHIFT key. Move the
mouse cursor to the proximity of the three nodes which will form the first
face of one pentahedral element. Click the left mouse button to select the
first 3 nodes of the element. Move the mouse cursor to the nodes represent-
ing the opposite face. Click the left mouse button to create the candidate
element.
Next create the ring of hexahedral elements around the countersink hole as illus-
trated in FIGURE 74.

5
FIGURE 74 Construction of elements around countersink
• Create > Hexahedron > Selection. Pick nodes individually, use a mar-
quee pick, or use the automatic methods to create each hexahedral ele-
ment.
Next, create the hexahedral elements in the fillet region as shown in FIGURE
75.
• Create > Hexahedron > Selection. Pick nodes individually, use a mar-
quee pick, or use the automatic methods to create each hexahedral ele-
ment.

FIGURE 75 Construction of elements in the fillet region
Next, create the hexahedral elements to fill the region around the countersink hole
and the flange hole as shown in FIGURE 76.
• Create > Hexahedron > Selection. Pick nodes individually, use a marquee
pick, or use the automatic methods to create each hexahedral element.
This mesh is quite robust for a wide variety of parameter combinations, as shown in
FIGURE 77. Before deployment as a handbook model, appropriate rules should be
defined so that inappropriate combinations of parameters will be prevented.

5
FIGURE 76 Completion of bracket mesh
Meshing Face to Face

When there is a need to convert a two dimensional model to a three dimen-
sional model, one possible procedure is to make a copy of the original 2D
model which is offset in the Z direction. In this case, there is a one to one cor-
respondence between the elements in the original model and the elements in
the copy. One option for constructing 3D elements (hexa or penta) between the
two planar sections is to use a face to face element construction method to join
an element in one section with a corresponding element in the other section.
This can be a very tedious process, especially when there may be tens or hun-
dreds of elements in each section.

FIGURE 77 Parametric variation of the bracket model.
An alternative technique is to use the Create > Mesh > Face To Face method. This
meshing method will allow you to join a set of elements in one section with a set of
elements in another section in one operation. This technique works best when one
of the sections being joined is a copy of the other. The procedure is to simply select
the elements from the first section with a marquee pick. Then select the elements in
the second section with a marquee pick while holding both the SHIFT and CTRL
keys. The two sections will be distinguished by different highlight colors. Click on
the Accept button to complete the face to face meshing operation. After all desired
3D elements have been created, the user must delete the 2D elements. If you are

creating elements between only two sections, you may immediately click the
Delete button to remove the currently selected 2D elements. If you have more
than two profiles, you will have to wait until all sections have been meshed
before deleting the 2D elements.
Note: The underlying assumption of this meshing method is that there is a one
to one correspondence between elements in the first section and those in the
second section, based on the ordering of the elements. For example, suppose
there are three elements in each section. Elements in the first section are num-
bered 1, 2, 3, and elements in the second section are numbered 8, 9, 10. The
nodes from element 1 will be joined with the nodes from element 8 to form a
new 3D element. Similarly, elements 2 and 9 will form a new element, and 3
and 10 will form a new element. Naturally, each pair of elements must be of
the same type (quad/quad or tri/tri). If any of the restrictions are violated, an
error message will appear and no elements will be created. This technique can
be used in other cases only if some care has been taken to construct the ele-
ments in each section such that the order of elements in the first section corre-
sponds properly with the order of the elements in the second section.
Mapped Mesh Generation in Three Dimensions

StressCheck provides a simple way to generate shell elements in three dimen-
sions, using the “Quad-Mesh” and “Tri-Mesh” objects in the Mesh Input Class.
5 These objects may be used to perform “mapped” meshing of surface objects.
“Quad-Mesh” will create a quadrilateral shell mesh and “Tri-Mesh” will create
a triangular shell mesh. NOTE: Shell elements are not currently supported for
analysis.
Each surface has its own (U,V) coordinate system (also referred to as P1,P2).
The surface will be meshed uniformly by locating nodes at equal distances in
the (U,V) coordinate space. For mapped meshing purposes, the range of the
U,V coordinate space is from 0 to 1 in each direction for all surface types. Sim-
ply select the surfaces to be meshed, and enter the following information in the
input area:
N-P1: The number of elements to create in the U direction
N-P2: The number of elements to create in the V direction
P1-Min: the U coordinate value where the first node should be placed.

Resulting 3D mesh
Standard Select
Shift/Ctrl Select
5
Delete 2D elements on front and back face
FIGURE 78 Mesh Face to Face.
P1-Max: the U coordinate value where the last node should be placed.
P2-Min: the V coordinate value where the first node should be placed.
P2-Max: the V coordinate value where the last node should be placed.
When you click on the Accept button, the elements will be created. Click on the
DeLast button to remove the elements created by the previous Accept operation.
See FIGURE 79 below for an example of mapped meshing. In this example, P1
refers to the circumferential direction of the cylinder, and P2 refers to the height
direction of the cylinder.

Mesh Refinement
FIGURE 79 Example mapped meshing.
Since StressCheck does not yet support shell elements for analysis purposes,
these elements may only be used to create solid elements by the Face-to-Face
method. You may select two quadrilateral shell elements, or two triangular
shell elements to create a hexahedral or pentahedral element respectively. The
shell elements must be deleted before using the solid elements for an analysis.
StressCheck also provides a simple way to construct hexahedral elements

using the “Hexa-Mesh” object and “Face to Face” method. Use the mouse cur-
5
sor to select two reference surfaces. The parameter range P1-Min to P1-Max
and P2-Min to P2-Max will be applied to each surface. A grid of nodes will be
attached to each surface and the space between the two surfaces will be filled
with hexahedral elements. See FIGURE 80 for an example of mapped hexahe-
dral meshing.
Note: Refer to the section on Automeshing at the end of this chapter for an
explanation about available options for automatic mesh generation in 3D.
Mesh Refinement
In the Input dialog window, select the h-Discretization tab and the Input Area
will appear as shown in FIGURE 81.

Mesh Refinement
FIGURE 80 Example of mapped meshing in 3D.
StressCheck allows manual and automatic h-refinement. The refinement is per- 5

formed by subdividing each element using a user-specified parameter, the number
of “midsides”. For example, if the number of midsides selected is 1 then each quad-
rilateral element will be subdivided into 4 elements by locating a node in the mid-
point of each side and connecting the nodes on opposite sides. StressCheck
currently supports uniform h-refinement, and simple graded refinement.
Uniform Refinement To perform a uniform mesh refinement select Refine > Any Element > Uniform
Div. > Midsides:1 and click on the element you want to refine. Alternatively,
choose Select > Any Element > Uniform Division > Midsides: 1, click on all ele-
ments to which the refinement should be performed, and then click on the Accept
button. If you select All Elements instead of Any Element, the refine operation will
be performed on all elements of the mesh.
FIGURE 82 illustrates the effect of a uniform refinement operation applied to a
straight sided hexahedral element.

Mesh Refinement
5 FIGURE 81 h-Discretization interface.
Graded Refinement To perform a simple graded refinement with one layer of elements select
Refine > Node > Simple Graded > Midsides: 1 > Grading factor: 0.15 and click
on the node to refine. You can select a group of nodes first by using the Select
option/ Accept button combination as explained above.
Graded refinement may be applied to a node, edge, or face (3D only), and may
be performed using either Action: Refine or Action: Select as described above.
FIGURE 83, FIGURE 84, and FIGURE 85 illustrate the affect of applying a
grading factor of 0.15 to a node, an edge, or a face of a hexahedral element.
Note: When performing edge or face refinement, you may choose whether to
apply the refinement only to the parent element of the selected edge/face, or to
all elements which share the selected edge/face.
When the parent element has either curved edges or faces, the resulting ele-
ments will inherit the curvature only on the edges or faces of the original par-

Mesh Refinement
FIGURE 82 Sample uniform element refinement.
FIGURE 83 Sample graded node refinement.
ent element. New edges and faces which are internal to the original parent element
will always be straight.
Mesh refinement is not supported for tetrahedral elements.

Mesh Refinement
FIGURE 84 Sample graded edge refinement.
FIGURE 85 Sample graded face refinement.
Use the DeLast or Undo button to delete the last set of refined elements cre-
ated.

Automeshing
Automeshing
StressCheck provides two general purpose automeshers that may be used to mesh
either 2D sheet bodies or 3D solid bodies. The default automesher utilizes the
MeshSim automesher from Simmetrix, Inc. An optional automesher is the PAM
automesher from ADA, Inc. Each automesher has unique capabilities and advan-
tages. Both automeshers are accessed from the Mesh Input Class. To use the Mesh-
Sim automesher, choose Create/Mesh. To use the PAM automesher, choose Create/
Mesh PAM. Each automesher has its own methods for generating a mesh.
In order to automesh a sheet or solid body in StressCheck, you must perform two
distinct steps. First, create one or more mesh records that contain the instructions
for performing the meshing operations. Next, click the Automesh button to execute
the instructions contained in the automesh records. The automeshing procedure
consists of first deleting all existing elements from a previous automeshing opera-
tion (if any), followed by the new automeshing operation. Any time a parametric
change is applied to the model, these automeshing records will be used to re-mesh
all bodies in the model.
Each body may be referred to only once by the meshing records. Mesh records are
processed in numerical order. The first mesh record found for each body will be
used. Redundant mesh records will be ignored. If a mesh record does not refer to a
specific sheet or solid body, it will be assumed that it applies to all bodies. In order 5
for a mesh record to refer to a body, the body must be selected during creation of
the mesh record.
Both meshers introduce midside nodes into the mesh in order to control element
curvature. StressCheck can interpret these midside nodes in two ways. The midside
nodes may be used as the basis for isoparametric mapping, or may be used to deter-
mine associativity with the geometric entities of the solid body to perform geomet-
ric mapping. Geometric mapping produces a more accurate representation of the
geometry than isoparametric mapping, which uses a parabola to approximate all
shapes. Geometric mapping is more expensive than isoparametric mapping since it
must obtain at least 6 sample points from each edge, and 36 sample points from
each surface, whereas isoparametric mapping uses at most one midside node along
each element edge. After meshing is completed, the mapping method used may be
changed by the user by selecting the Main Menu option: Tools/Convert Element
Mapping.
NOTE: The MeshSim automesher performs all element quality tests using isopara-
metric mapping. Therefore, it is possible that changing to geometric based mapping

MeshSim automesher
will cause some elements to become illegal. Always check mesh quality
(Check/Mesh/Distortion) after changing the mapping method. The PAM
automesher performs all element quality tests using high order approximation
methods, so it is less likely that geometric mapping will introduce unexpected
distortion.
MeshSim automesher
The MeshSim automesher is a two-step procedure. In the first step, the
attributes of the desired mesh are described, and in the second step, the
automesher is invoked. The first step is accomplished in the usual way by
selecting an Action/ Object/ Method and entering the attributes in the input
fields, then clicking the Accept button to create a record. When all attribute
records are entered, clicking the Automesh button performs the second step. If
you wish to change the attributes of the resulting mesh, simply edit the
automesh records, replace them, and invoke the automesher again. To edit
existing automesh records, you must switch to the Index tab in the Mesh Class
interface, select the record of interest, then switch back to the Data tab to make
your changes. Each time you click the Automesh button, the existing mesh will
be deleted, and the automesher will regenerate the mesh using the currently
5 defined attributes. There are several different types of automesh records that
may be created, each with a corresponding automesh Method. These methods
may be divided into two groups, global and local methods. Only one global
meshing record should be provided for each distinct solid body. The inputs
associated with each method are shown for each method. If an input value is
not provided (check box is turned off), the automesher will perform according
to its own internal default rules.
It should be noted that the operation of the automatic mesher depends on the
currently selected reference system. In 3D, the mesher will automatically mesh
only solid bodies. If you wish to automesh a subset of the available solid bod-
ies, or if you wish to automesh a sheet body, you must select the desired body
explicitly before activating the automesher. In all other reference systems (Pla-
nar, Plates and Extrusion), the mesher will automatically mesh only sheet bod-
ies. If you wish to automesh a subset of the available sheet bodies, you must
explicitly select them before activating the automesher.
The global meshing methods are:

MeshSim automesher
* Auto - attributes that control the automesher everywhere in the domain where
local meshing attributes are not provided. This record may be used for either solid
or sheet bodies. Select the solid or sheet bodies, supply the input values, then click
Accept to save the record. The input values are:
Ratio: The user may specify the ratio of the length of the largest edge relative
to the length of the smallest coordinate aligned bounding box which surrounds
the part. Default value: 0.5, valid range: 0.1 to 1.0.
D/H: The user may specify the ratio of the distance from the chord formed by
an element edge and its associated curve (D) or surface, relative to the length
of the element edge (H). Default value: 0.4, valid range: 0.01 to 0.4. If the
check box is disabled, the mesher will not check the D/H ratio of the elements
it produces. If the check box is enabled, the ratio is tested and the elements
produced will satisfy the limit supplied. In addition, when the D/H checkbox
is enabled, the automesher will utilize additional rules that take into consider-
ation the individual characteristics of geometric surfaces found in the model.
For example, the mesher will attempt to mesh a cylindrical surface with fewer
elements in the length direction than in the circumferential direction.
MinLen: The minimum length of any element edge in the mesh (in model
units). Default value: 0.05, valid range: 0.001 to 0.4
Isopar: If the check box is on, midside nodes and isoparametrically mapped 5
elements will be generated.
Trans: Rate of transition from small elements in a local region to larger ele-
ments in a global region. A value of zero (0.0) indicates that elements will
remain small in both the local and global region. A value of one (1.0) indi-
cates that elements will transition rapidly from a small size in the local region
to a larger size in the global region. A value in between will indicate a more
gradual transition between local and global mesh regions.
* Extrude - attributes for generating an extrusion mesh for a solid body. An extru-
sion mesh is only applicable when the front face of the solid domain is identical
with the back face. In this case, it is possible to instruct the automesher to first gen-
erate a mesh of planar (2D) elements on the front face, and then extrude these ele-
ments through the depth direction of the body to form solid elements. Select the
front faces, select the back faces in the same order that the corresponding front
faces were selected, supply the input values, then click Accept to save the record.
The input values are:

MeshSim automesher
D
Curve/Surface
Element chord
H
FIGURE 86 Automatic meshing controls.
5
Ratio: The user may specify the ratio of the length of the largest edge rel-
ative to the length of the smallest coordinate aligned bounding box which
surrounds the part. Default value: 0.5, valid range: 0.1 to 1.0.
D/H: The user may specify the ratio of the distance from the chord
formed by an element edge and its associated curve (D) or surface, rela-
tive to the length of the element edge (H). Default value: 0.4, valid range:
0.01 to 0.4. If the check box is disabled, the mesher will not check the D/
H ratio of the elements it produces. If the check box is enabled, the ratio
is tested and the elements produced will satisfy the limit supplied. In
addition, when the D/H checkbox is enabled, the automesher will utilize
additional rules that take into consideration the individual characteristics
of geometric surfaces found in the model. For example, the mesher will
attempt to mesh a cylindrical surface with fewer elements in the length
direction than in the circumferential direction.
units).

MeshSim automesher
Isopar: If the check box is on, midside nodes and isoparametrically mapped
elements will be generated.
Mirror: If this check box is on, the extruded mesh will be mirrored about the
center of the domain.
Layers: The total number of layers in the extruded mesh (including mirrored
layers if any).
Depth: Thickness of the first layer of elements measured from the front face.
Layer thickness grows geometrically.
Trans: Rate of transition from small elements in a local region to larger ele-
ments in a global region of the surface mesh generated for the front face. A
value of zero (0.0) indicates that elements will remain small in both the local
and global region. A value of one (1.0) indicates that elements will transition
rapidly from a small size in the local region to a larger size in the global
region. A value in between will indicate a more gradual transition between
local and global mesh regions.
Angle: Smallest vertex angle allowed. Default value: 0.0
The local meshing methods are:
* Boundary Layer - A boundary layer mesh record describes a local mesh refine-
ment region around one or more selected edges of a sheet or solid body, or one or
more selected faces of a solid body. The boundary layers will be constructed in a 5
geometric progression, based on the ratio of the thickness of the first layer (To) rel-
ative to the total thickness of the refinement region (T-Total). Given the thickness
of the ith layer to be ti, the total thickness of the refinement region to be T, and the
number of layers to be n, the following relation holds for geometric growth: t i+1 =
rti and
T = t0 + t0r + t0r2 + t0r3 + … + t0rn-1
= t0 (rn-1)/(r-1)
The automesher calculates the growth factor r from the input values. Isotropic layer
heights (r=1) can be generated by setting the total height equal to the number of
layers times the first layer height, T = n*t0. Select the edges and/or faces, supply
the input values, then click Accept to save the record. The input values are:
Ratio: The ratio of the length of the longest attached element edge to the
length of the geometric edge or face.
Layers: Number of layers of elements generated near the selected edge/face.

MeshSim automesher
To: Thickness of the first layer of elements attached to the selected edge/
face.
T-Total: Total thickness of the local mesh refinement region.
Side: (For face boundary layers only) If negative, the mesh refinement
will be generated on the side of the face corresponding to its negative
normal. If positive, the refinement will be generated on the side of the
face corresponding to its positive normal. If zero (0.0), the refinement
will be generated on both sides of the face.
* Crack Face - Identify a face of a 3D solid body that will be treated as a crack
path. After a mesh is automatically generated, the element connectivity on one
side of the crack will be modified to produce a crack in the domain. Select the
plane representing the crack, and then click Accept to save the record.
* Crack Edge - Identify an edge of a 2D sheet body that will be treated as a

crack path. After a mesh is automatically generated, the element connectivity
on one side of the crack will be modified to produce a crack in the domain.
Select the curve representing the crack path and click Accept to save the
record. Then, click on the Automesh button.
* Curve - Identify an edge of a sheet or solid body where you wish to control
the local meshing attributes. Elements will be associatively attached to the
selected edge. The attributes assigned to the edge will override the global
5 meshing attributes only for this edge. Select the edge, supply the inputs, and
click Accept to save the record. The input values are:
Ratio: The ratio of the longest element edge to the length of the selected
curve.
D/H: The user may specify the ratio of the distance from the chord
formed by an element edge and its associated curve (D) or surface, rela-
tive to the length of the element edge (H). Default value: 0.4, valid range:
0.01 to 0.4. If the check box is disabled, the mesher will not check the D/
H ratio of the elements it produces. If the check box is enabled, the ratio
is tested and the elements produced will satisfy the limit supplied. In
addition, when the D/H checkbox is enabled, the automesher will utilize
additional rules that take into consideration the individual characteristics
of geometric surfaces found in the model. For example, the mesher will
attempt to mesh a cylindrical surface with fewer elements in the length
direction than in the circumferential direction.
units).

MeshSim automesher
* Local - Identify a point, edge or face of a sheet or solid body where mesh refine-
ment is desired. Select the points, edges and/or faces, supply the input value, and
click Accept to save the record. The input value is:
Size: The length of the element edges in the neighborhood of the selected
object(s).
* Simple Graded - (For use in combination with Extrusion meshing only). Identify
an edge in the front face of a solid body where a mesh refinement is to be inserted.
All faces attached to the edge must be quadrilaterals. The elements attached to the
edge will be hexahedral. Select the edge(s), supply the input values, and click
Accept to save the record. The input values are:
Layers: Number of layers of elements around the selected edge.
Radius: Distance from the edge to the outer face of the mesh refinement layer.
Since the automatic mesher capability is relatively new, we expect that some time
will be required to improve the effectiveness for general classes of problems. We
also expect that as the automesher improves, a newer release may produce a differ-
ent mesh for the same problem, than was produced by an earlier release. Conse-
quently, the meshes that you obtain may be different from the examples shown in
the documentation.
* Face to Face - This is not a true automeshing method. It is a method for filling the
space between two sheet bodies with a single layer of elements. Consequently the 5
elements are constructed immediately and no automesh record is produced. This
technique for constructing elements should not be used with other automeshing
methods. Refer to “Meshing Face to Face”.
Limitations
The limitations of the mesher are as follows:
1. In 3D, the mesher generates only 10 noded isoparametric tetrahedral elements. In

2D, the mesher will produce triangular elements by default, but the mapping can be
chosen to be either isoparametric or geometric. A 2D mesh can be constructed
which uses quadrilateral elements (if possible) using the Create > Quad-Mesh
option. In this case, only geometric mapping is supported.
2. In 3D, the mesher works best on solids that do not have thin sections. Models
with thin sections will cause the mesher to generate too many elements to be solved
in a reasonable time by StressCheck. The number of elements that can be solved

MeshSim automesher
will depend on the amount of memory your system has. A system with 256MB
of RAM can be expected to handle 3D mesh sizes up to 2000 elements.
Mesh Improvements
Finding a combination of the four input parameters, Ratio, D/H, MinLen and
Isopar, that gives the most satisfactory mesh, may require some experimenta-
tion. To begin, enable only the Isoparametric check box. After generating the
mesh, use the Check > Mesh > Distortion function to test the validity of the
elements contained in the mesh. If all elements pass the distortion test, you
may proceed to assign material properties and boundary conditions. If some
elements are identified as invalid, it will be necessary to delete the current
mesh by clicking the DeLast button, and try a new set of input parameters.
In general, reducing the value of any of the three parameters will produce a
finer mesh. The valid range for “Ratio” is from 0.1 to 1.0, but reducing the
value of “Ratio” below 0.25 should be considered a last resort, since it controls
global mesh density. The valid range for “D/H” is from 0.01 to 0.4, but in most
cases should not be reduced below 0.2. The valid range for “MinLen” is from
0.001 to 0.4.
Checking mesh quality

5
When using the automatic mesher and performing the mesh distortion check,
you may find that several elements are identified as distorted (outside the range
of valid vertex angles of 10 to 150 degrees) or invalid. Elements that are
invalid require remeshing with different settings. Elements identified as dis-
torted are acceptable to perform an analysis, but in some cases the stiffness
matrix generation of these elements may fail as the polynomial order increases
during a p-extension.
During the distortion check, not only the compliance with the vertex angles is
verified, but also that the jacobian determinant is greater than zero in a number
of interior points (27 for 10-noded tetrahedral elements). During the computa-
tion of the stiffness matrix, the number of integration (internal) points depends
on the p-level for the element. If an element is too distorted, it may happen that
for a given p-level (larger than 1), a negative jacobian is found in some integra-
tion points and the stiffness matrix computation will not proceed. An error
message will be issued by the program identifying the corresponding element.

MeshSim automesher
If you encounter a situation like this, one possible solution is to fix the p-level of
the element to the value for which the stiffness matrix was successfully computed.
This should only be done in those cases where the distorted elements are outside the
region of primary interest.
Automatic Mesh Generation in Two Dimensions

StressCheck provides five automated mesh generation facilities for two-dimen-
sional geometry constructions: Auto, Boundary Layer, Crack Edge, Curve and
Local. StressCheck also provides a simple method for trimming two dimensional
boundaries which is accessible from the Geometry Class under the Select action.
Choose any two dimensional curve object, and the trim method. Point the mouse
cursor to the section of the curve to be removed and click the left mouse button.
The indicated section of the 2D boundary will be removed.
Auto To explore the use of the 2D automatic mesh generator using the method Auto, start
constructing the following geometric objects (see FIGURE 87):
• Create a rectangle at (0.0, 0.0, 0.0) of size 2.66 x 4.58.
• Create two circles at (4.54, 2.52, 0.0), one of radius 0.875 from 0 to 360,
and the second one of radius 2.42 from
-90 to 90.
• Create one horizontal line at (0.0, 0.46, 0.0) of length 7 and a second one 5
starting at the top-right point of the rectangle of length 4.
• Create a point at the intersection of the first line created in the previous step
and the large circle.
• Create a line by the 2-Point method using the last point and the bottom-
right point of the rectangle.
We first trim the lines and circle as shown in FIGURE 88a

• Select > Any Curve > Trim to Curve. With the mouse cursor click on each
segment you want to trim out of the model.
Next, toggle the Surface/Curve selector to Surface and create a plane as fol-
lows (FIGURE 88b):
• Create > Plane > Locate > X: -1.0 > Y: -1.0 > Z: 0.0 > Width: 10.0 >
Height: 6.0 > P1-Min: 0 > P1-Max: 1 > P2-Min: 0 > P2-Max: 1 > Accept.
To use the 2D automesher we need to create a sheet body.

MeshSim automesher
FIGURE 87 Geometric description for the attachment lug problem.
(a)
(b)
5
FIGURE 88 (a) Trimmed geometry and (b) plane for sheet body creation.
We now have to imprint the geometry into the plane and then create a sheet
body as follows (FIGURE 89):
• Create > Body-Imprint > Curve Normal. Select the plane (will be high-
lighted in green) and then all the lines and circles (will be highlighted
in red) except the horizontal line that runs across the lug, and then click
on Accept (FIGURE 89a).
• Create > Body-Trim > No Heal. Select the regions of the plane that
will be removed (will be highlighted in red), and then click on Accept
(FIGURE 89b).

MeshSim automesher
FIGURE 89 (a) Imprint geometry into plane, and (b) create body-trim.
As a result of the last operation, the geometric description will look like the one
shown in FIGURE 90a. Next, select the Mesh tab from the dialog window and cre-
ate the mesh as follows:
• Create > Mesh > Auto > Ratio: 1, D/H: 0.4, MinLen: 0.05, Isopar switch
off, Trans: 0 >Accept button > Automesh button.
The mesh generated using the indicated settings is shown in FIGURE 90b. 5
FIGURE 90 (a) Trim sheet body and (b) mesh from the 2D automatic mesh generator.
Crack Edge To explore the 2D automatic mesh generator using the method Crack Edge, start by
creating the sheet body shown in FIGURE 89b. Draw a line representing a crack
path as follows:

MeshSim automesher
• Geometry tab > Create > Point > Sample > Click on the internal edge
of the circle. Make sure the Display Surfaces is off.
• Geometry tab > Create > Line > Point > Length: 0.7, Angle: -30 >
Click on the point created in previous step.
• Geometry tab > Create > Body-Imprint > Curve Normal. Select the
plane and then the line. Click on Accept. See FIGURE 91.
FIGURE 91 Creating the internal crack.
Make a global and then a local meshing. After the mesh is automatically gener-
5 ated, the element connectivity on one side of the crack will be modified to pro-
duce a crack in the domain.
• Mesh tab > Create > Mesh > Auto > D/H: 0.4, Isopar switch on >
Accept button
• Mesh tab > Create > Mesh > Crack Edge. Now, select the line repre-
senting the crack path and click Accept to save the record. Then, click
on the Automesh button.
Curve To explore the 2D automatic mesh generator using the method Curve, use the
global meshing created when working with the Auto method. Select the circle
as the place where you want to control the local meshing and supply the fol-
lowing input:
• Create > Mesh > Curve > Ratio: 0.1, D/H: 0.4 >Accept button >
Automesh button.
The mesh generated using the indicated settings is shown in FIGURE 93.

MeshSim automesher
FIGURE 92 Crack Edge meshing method.
FIGURE 93 Curve meshing method.
Local To explore the 2D automatic mesh generator using the method Local, identify a
point or edge of a sheet body where mesh refinement is desired. Select the points,
and/or edges, supply the length of the element edges in the neighborhood of the
selected object(s), and click Accept to save the record. When all records are
entered, click on the Automesh button. FIGURE 94 shows an example.

MeshSim automesher
FIGURE 94 Local meshing method.
Automatic Mesh Generation in Three Dimensions

StressCheck provides a three dimensional automeshing capability based on
5 solid geometry. The meshing is accessible from the Mesh Class by choosing
Create > Mesh > Auto. The “Isoparametric” input field provides control over
the quality of the element mapping used for the elements generated by the
automesher. In the default position, the “Isoparametric” checkbox is turned on.
Therefore, all elements will be mapped according to the standard quadratic
form for midside (10 noded tetrahedra) elements. If the checkbox is turned off,
StressCheck will test to see whether element edges/faces lie on geometric
curves/surfaces. If so, the mapping for the element will be upgraded for those
geometrically associative edges/faces to produce the standard geometrically
associative mapping used by manually created elements. The benefit of the
geometric mapping is that the geometry will be better represented by the ele-
ments, and the solution produced by these elements will be of higher quality.
The cost of the improved mapping is extra computational time. Furthermore,
the geometric mapping will sometimes cause a very thin element to become
invalid. Invalid elements will automatically be flagged at solution time, or can
be identified by using the Check/Mesh/Distortion function. If any geometri-
cally mapped elements are invalid, the mesh should be regenerated with the
“Isoparametric” checkbox enabled.

MeshSim automesher
Ratio should be determined based on the size of the largest element that is to be cre-
ated, relative to the overall dimension of the part to be meshed. If the “Ratio” input
field is not enabled, StressCheck will automatically compute the size of each face in
the model and provide a value for the ratio based on the equation: ratio=(maximum
face dimension)/(2*maximum part dimension). This evaluation ignores the fact that
surfaces may be penetrated by one or more holes, which would affect the actual
maximum element size possible. You may wish to override this automatic computa-
tion of “Ratio” to produce a better mesh. If so, try to take into consideration the
location of holes that may exist in the various surfaces of the part to be meshed in
order to estimate the maximum element size that it will actually be possible for the
automesher to create.
StressCheck provides eight automated mesh generation facilities for three-dimen-

sional geometry constructions: Auto, Boundary Layer, Crack Face, Curve, Extrude,
Face to Face, Local and Simple Graded.
To explore the use of these meshing tools, consider the solid model of a linkage arm
detail shown in FIGURE 95. The steps for creating the solid model are described in
the following.
To specify the domain, select the Geometry tab in the StressCheck Input dialog
box, and define two cylinders by the commands:
• Geometry tab > Create > Cylinder > Locate > Solid button on > Input: X:
100.0, Y: 0.0, Z: 0.0, Radius: 10, Height: 10, Rot-X: 0.0, Rot-Y: 0.0, Rot-Z: 5
0.0 > Accept. Similarly, create the outer cylinder with a radius of 20
inches.
Then, construct a rectangular domain using the following steps:

• Create > Box > Locate > Solid button on > Input: X: 50.0, Y: 0.0, Z: 0.0,
Width: 100, Height: 20, Depth: 10, Rot-X: 0.0, Rot-Y: 0.0, Rot-Z: 0.0 >
Accept.
You are now ready to begin applying boolean operations to your model. First, join
the Box and the larger cylinder by doing the following:
• Create > Body > Bool-Union. Pick the Cylinder as the target, and pick the
Box. Click on the Accept button.
Next, create the hole by subtracting the smaller cylinder from the result body of the
previous operation as follows:
• Create > Body > Bool-Subtract. Pick the result body from the previous
operation, and pick the Cylinder. Click on Accept

MeshSim automesher
Next, create the fillet as follows:

• Create > Blend Edge > Rolling Ball, Radius = 15. Pick the fillet edges.
Click on Accept. The solid model should look as shown in FIGURE
95.
FIGURE 95 Part of a linkage arm.
Save an input file of the model for future use.
Auto To create an automatic mesh using the method Auto, select the Mesh tab from
the dialog window and create a global meshing record. as follows:
5 • Create > Mesh > Auto > Ratio: 0.5, D/H: 0.2, MinLen switch off, Iso-
par switch off, Trans: 0.5 > Accept button > Automesh
FIGURE 96 Automesh Method: Auto
The mesh should look like the one shown in FIGURE 96. If you want to
explore other settings, simply switch to the index tab, select the existing

MeshSim automesher
record, switch back to the Data tab, update the input and click on Replace. After
that, click on Automesh and the previous mesh will be replaced by a new mesh.
Boundary Layer To explore the 3D automatic mesh generator using the method Boundary Layer,
start by using the global meshing record given above. A boundary layer mesh
record describes a local mesh refinement region around one or more selected faces
of a solid body. Select the Mesh tab from the dialog window, and create the mesh as
follows:
• Create > Mesh > Bndry Layer > Ratio: 0.5, Layers: 2, To: 3, T-Total: 5,
Side: -1.0 > Turn the Display Elements off and click on the cylindrical sur-
face of the link > Click on the Accept button > Turn the Display Elements
on and click on the Automesh button.
The mesh should look like the one shown in FIGURE 97.
cylindrical surface
FIGURE 97 Automesh Method: Boundary Layer.
Crack Face To explore the 3D automatic mesh generator using the method Crack Face, start
with the solid model shown in FIGURE 95 and add a plane representing a crack
face as follows:
• Geometry tab > Create > Plane > Locate > X: 20, Y: -10, Z: 10, Width: 10.
Height: 30, Rot-Y: 90 > Accept.
• Create > Body > Bool-Union > Click on the Plane and the Body Solid.
Click on the Accept button. See FIGURE 98.

MeshSim automesher
FIGURE 98 Creating a plane

to locate a crack.
Blank the front surface of the link in order to access to the plane just created by
selecting the face and then selecting Blank Object from the Edit toolbar. Then,
select the Mesh tab and create the following meshing records:
• Create > Mesh > Crack Face. Select the plane representing the crack,
turn off all surfaces, and select the line where the plane intersects the
bottom of the linkeage arm. Click Accept to save the record.
‘crack’ plane
After setting the local meshing record proceed to define the global meshing
record.
• Create > Mesh > Auto > D/H: 0.2, Isopar: 0.0 > Accept button >
Automesh button.
The mesh is shown in FIGURE 99.

MeshSim automesher
After the mesh is automatically generated, the element connectivity on one side of
the crack will be modified to produce a crack in the domain. To check the existence
of the crack, make sure the Wetted Faces toggle is off and then:
• Mesh tab > Check > Face > Free Face
FIGURE 99 Automesh Method: Crack Face
5
You need to shrink the elements and turn on Element Handles to see the free faces
associated with the crack faces.
Curve To explore the 3D automatic mesh generation using the method Curve, start with
the solid model shown in FIGURE 95 and proceed as follows: select one of the
outer circles of the link for controlling the local meshing and create local and global
mesh records.
• Create > Mesh > Curve > Ratio: 0.1, D/H: 0.1 > Accept button.
• Create > Mesh > Auto > Ratio: 0.5, D/H: 0.4, Isopar: 0.0, Trans: 0.5 >
Accept button > Automesh button.
The mesh is shown in FIGURE 100. Note that the addition of curve refinement pro-
duced more elements in the region of the hole than for the case shown in FIGURE
96, while in the rest of the link the meshes are similar.

MeshSim automesher
FIGURE 100 Automesh Method:

Curve
Extrude To explore the 3D automatic mesh generator using the method Extrude, the
front face of the solid domain has to be identical with the back face. For the
problem shown in FIGURE 95, we select the front surface of the linkage arm,
select the back surface, and supply the input values.
• Create > Mesh > Extrude > Ratio: 0.5, D/H: 0.4, MinLen: 0.05, Mirror:
0.0, Layers: 1, Depth: 0, Trans: 0.95 > Click on Accept to save the
record > Click on Automesh. The mesh is shown in FIGURE 101.
5

Extrude
Local To explore the 3D automatic mesh generator using the method Local, identify a
surface of the solid body where mesh refinement is desired. Select the faces,

MeshSim automesher
supply the input value, and click Accept to save the record. For the problem shown
in FIGURE 95, perform the following actions:
• Create > Mesh > Local > Size: 4 > Select the cylindrical surface of the hole
> Click on Accept to save the record
Make a global meshing.

• Mesh tab > Create > Mesh > Auto > Ratio: 1, D/H: 0.4, MinLen: 0.05, Iso-
par switch off, Trans: 0 >Accept button > Automesh button. The mesh is
shown in FIGURE 103.

Local
5
Simple Graded To explore the 3D automatic mesh generator using the method Simple Graded,
identify an edge in the front face of a solid body where a mesh refinement is to be
inserted. All faces attached to the edge must be quadrilaterals. Select the edge, sup-
ply the input values, and click Accept to save the record. Start with the solid model
shown in FIGURE 95, and create two planes as follows.
Click on the Accept button
Click on the Accept button
Create the mesh by supplying the following input values.

PAM automesher
• Mesh tab > Create > Mesh > Simple Graded > Layers: 2 > Radius: 3 >
Click on a side edge of Plane 1 > Accept
• Create > Mesh > Extrude > Ratio: 0.5, D/H: 0.4, Isopar: 0.0, Layers: 2,
Depth: 2, Trans: 0.5 > Click on the whole front face of the arm and in
the same order on the back face of the arm. Click on Accept to save the
record > Click on Automesh. The mesh should look like the one
FIGURE 104 Automesh

Method: Simple Graded.
5
PAM automesher
PAM is a p-version automeshing software designed to create tetrahedral
meshes as coarse as possible, while, mesh qualities, such as element edge
angles, element face angles, curvilinear edge curvature, are ensured according
to user’s definitions. All sample models described in this section are available
in the Tutorial Handbook of StressCheck.
The PAM automesher currently supports 7 methods. Briefly, these are:
1. Auto – global membrane/shell and solid meshing. This method will produce
triangular elements for sheet bodies and tetrahedral elements for solid bodies.

PAM automesher
2. Extrude – global pentahedral element mesher. This method requires a sheet body
that will be meshed with triangular elements that will be extruded into pentahedral
elements in the direction of the positive Z axis of a chosen local coordinate system.
The distance and number of layers is user provided.
3. Extrude Face – global pentahedral element mesher. This method requires a sheet
body that will be meshed with triangular elements that will be extruded into penta-
hedral elements in the direction of a the positive Z axis of a chosen local coordinate
system until a chosen geometric surface is encountered. The distance and number
of layers is user provided.
4. Spin - global pentahedral element mesher. This method requires a sheet body that
will be meshed with triangular elements that will be extruded into pentahedral ele-
ments as they elements are spun around the central Z-axis of a chosen local coordi-
nate system. The spin angle and number of layers is user provided.
5. Thicken – global pentahedral element mesher. This method requires a sheet body
that will be meshed with triangular elements that will be extruded into pentahedral
elements as they are extruded above and/or below the sheet body a given distance.
The distance and number of layers is user provided.
6. Boundary layer – global tetrahedral/pentahedral mesher. This method requires a

solid body of which faces have been selected for application of the boundary layer
mesh. The solid is meshed with tetrahedral elements. The triangular faces lying on 5
the chosen geometric faces will be extruded normal to the surface into pentahedral
elements a given distance and number of layers.
7. Crack face – global tetrahedral mesher. This method requires a general solid
body with one or more internal geometric surface that represents a crack or inter-
face between contacting bodies. The body will be meshed as usual with tetrahedral
elements, but those faces that lie on selected interior surfaces will not be connected
with their neighbors on the opposite side of the surface.
All PAM extrusion meshing methods (i.e. Extrude, Extrude Face, Spin, Thicken,
and Boundary Layer) will produce mesh layers of equal thickness by default. In
order to specify the thickness of each individual layer, it is necessary to define a
table, using the AOM settings Select/Extruded Body/Constant. Supply the table
name, layer number and layer thickness for each layer. When creating the automesh
record, supply the table name instead of the total thickness (or Angle for Spin
method) of the mesh extrusion.
To define the table records needed by the PAM automesher, do the following:

PAM automesher
1. Choose the Action/Object/Method as Select/Extruded Body/Constant.
2. Supply a table name that will be referenced in the PAM automesh record.
3. The “Type” combo box will automatically be set to “Thickness”.
4. Specify the layer number. The first layer next to the extrusion surface will be
layer 1.
5. Supply the thickness value of the layer being defined.
6. Repeat steps 4) and 5) for each mesh extrusion layer.
For more details about using each of these methods, please review the explana-
tions and examples found in the following sections.
Method Auto This is a general purpose meshing method for sheet bodies and solid bodies.
The only meshing attributes required are the minimum and maximum vertex
angles that are to be allowed for each element. To create the automesh record,
choose Create/Mesh PAM/Auto and select one or more sheet or solid bodies,
then click Accept. Click the Automesh button to complete the operation.
Method Extrude This is a special purpose meshing method to produce a three dimensional pen-
tahedral mesh by extruding an automatically generated triangular mesh on a
5 sheet body. The inputs are the minimum and maximum vertex angles, the num-
ber of layers, and the total thickness (T-Total) of the extruded body. The extru-
sion will be constructed in the direction of the Z-axis of a selected local
system. If no local system is selected, the extrusion will be constructed in the
global Z-axis direction. To create the automesh record, choose Create/Mesh
PAM/Extrude and select one or more sheet bodies. To select a local system,
hold the Ctrl-Shift while selecting one local coordinate system. Click Accept to
create the automesh record. Click the Automesh button to complete the opera-
tion.
Method Extrude Face This is a special purpose meshing method to produce a three dimensional pen-
tahedral mesh by extruding an automatically generated triangular mesh on a
sheet body. The inputs are the minimum and maximum vertex angles, and the
number of layers. An input is also provided for total thickness (T-Total), but
this should only be provided if a table is to be used to specify individual layer
thickness information. The extrusion will be constructed in the direction of the
Z-axis of a selected local system. If no local system is selected, the extrusion
will be constructed in the global Z-axis direction. To create the automesh

PAM automesher
record, choose Create/Mesh PAM/Extrude Face and select one or more sheet bod-
ies. To select a local system, hold the Ctrl-Shift while selecting one local coordinate
system. Click Accept to create the automesh record. Click the Automesh button to
complete the operation.
Method Spin This is a special purpose meshing method to produce a three dimensional pentahe-
dral mesh by extruding an automatically generated triangular mesh on a sheet body.
The inputs are the minimum and maximum vertex angles, the number of layers, and
the total angle swept by the extruded body. The extrusion will be swept about the
local Z-axis of a selected local system. If no local system is selected, the extrusion
will be swept about the global Z-axis. To create the automesh record, choose Cre-
ate/Mesh PAM/Spin and select one or more sheet bodies. To select a local system,
hold the Ctrl-Shift while selecting one local coordinate system. Click Accept to cre-
ate the automesh record. Click the Automesh button to complete the operation.
Method Thicken This is a special purpose meshing method to produce a three dimensional pentahe-
dral mesh by extruding an automatically generated triangular mesh on a sheet body.
The inputs are the minimum and maximum vertex angles, the number of layers, the
total thickness (T-Total) of the extruded body, and the side of the sheet body to be
extruded. The extrusion will be constructed in the direction normal to the sheet
body. If the input value supplied for “Side” is negative, the extrusion mesh will be
constructed on the side of the sheet body corresponding to the direction of the neg-
ative normal vector by the right handed rule. If the value for “Side” is positive, the
extrusion mesh will be constructed on the side of the positive normal. If the value 5
for “Side” is zero, the extrusion will be repeated on both sides of the sheet body. To
create the automesh record, choose Create/Mesh PAM/Thicken and select one or
more sheet bodies. Click Accept to create the automesh record. Click the Automesh
button to complete the operation.
Method Bndry. Layer This is a special purpose meshing method to produce a three dimensional pentahe-
dral mesh by extruding the triangular faces of tetrahedral elements on the selected
face(s) of a solid body. The inputs are the minimum and maximum vertex angles,
the number of layers, and the total thickness (T-Total) of the extruded mesh layers.
The extrusion will be constructed normal to the selected face(s) of the solid body.
To create the automesh record, choose Create/Mesh PAM/Bndry. Layer and select
one or more faces of a solid body. Click Accept to create the automesh record.
Click the Automesh button to complete the operation.
Method Crack Face This is a special purpose meshing method the purpose of which is to disconnect
mesh faces that lie on a common internal surface, the result of which is to create a
crack in a three dimensional mesh. This method requires that a solid body be identi-

PAM automesher
fied for a volume mesh by the method AUTO. To create the automesh record,
choose Create/Mesh PAM/Crack face and select one or more internal faces of a
solid body. Click Accept to create the automesh record. Click the Automesh
button to complete the operation.
PAM Quality Control
PAM requires only two parameters in the current release to control mesh qual-
ity: MinAng and MaxAng. These input values will be used as convergence cri-
teria during meshing. PAM will optimize the mesh until all element face angles
and edge angles satisfy these input criteria. There is no method to force face
angles or edge angles to satisfy the input MinAng criteria if the CAD model
contains geometry constrained angles which are smaller than the input
MinAng (see FIGURE 106). In this case, PAM treats such elements as valid.
Other mesh qualities such as element edge curvature, element aspect ratio, etc.
are controlled by PAM automatically.
PAM Operation
PAM input Since the current PAM release can create a tetrahedral mesh for solid or 3D
sheet bodies, it works only in 3D mode. You can import the Halbgabe.sci
5 model from the Parts Handbook to practice the following operations.
When AOM is set to Create > Mesh-PAM > Auto under the Mesh class, an
input dialog is shown as in FIGURE 107. By default, MinAng=5.0, Max-
Ang=175.0, where the input values are in degrees. The valid range of MinAng
is 1.0~15.0, and the valid range of MaxAng is 165.0~179.0. Although both
MinAng and MaxAng affect PAM performance, MinAng generally has a
greater effect than MaxAng. This is due to the fact that PAM requires a large
computational effort to optimize the mesh for a large MinAng.
A meshing record is required to run PAM. To define it, simply input the
MinAng and MaxAng values and click on the Accept button. You can check
under the index tab to confirm the meshing record definition (see FIGURE
108a). If the Accept button is clicked multiple times, more than one meshing
record will be defined. Redundant records should be deleted by selecting them
in the list box and clicking the Delete button..

PAM automesher
Geometry constrained angle smaller

than input criteria.
Element edge angles and face angles are

less than MinAng criterion.
Two elements with geometry constrained angles are 5

treated as valid elements.
FIGURE 106 PAM convention for min/max angle conditions.
After a meshing record is defined, you can run PAM by clicking on the Automesh
button. A dialog with a progress bar is shown (see FIGURE 108b). You can termi-
nate PAM at any time by clicking on the Cancel button or by pressing the <Esc>
key (FIGURE 108c). Note that once the Automesh button is clicked, StressCheck
calls PAM to mesh all solid bodies in the current database by default. If you want to
mesh selected bodies, you must select the bodies at the time of creation of the
meshing record. FIGURE 109 shows a CAD model and the mesh created by PAM
for the defualt values of the min and max angles.

PAM automesher
User input angle criteria
FIGURE 107 Input dialog

window for Mesh-PAM
AOM setting creation.
PAM meshing progress
(b)
Defined meshing record.
(c)
(a)
When <Esc> Key is pressed or

Cancel button is clicked.
FIGURE 108 Mesh-PAM record, progress bar dialog, and cancel

warning

PAM automesher
FIGURE 109 A CAD model and created mesh
Checking mesh quality After PAM finishes the meshing process, it is important to check the mesh quality
by selecting Check > Mesh > Distortion, and entering the values of the MinAng and
MaxAng for warning.
After all the elements are evaluated, a text window will be displayed with the qual-
ity control information. This includes the number of acceptable elements, the num-
ber of distorted elements, the number of rejected elements, as shown in FIGURE 5
110. Elements identified as being distorted are in general acceptable to perform an
analysis, but in some cases the stiffness matrix generation of these elements may
fail as the polynomial order increases during p-extension. If elemetns are identified
as rejected, then the existing mesh cannot be used in analysis. To create another
mesh using PAM, modify the existing meshing record and create a new mesh.
FIGURE 110 Mesh check text window

PAM automesher
Creating a surface mesh for sheet models
PAM can create a surface mesh for a sheet model. The operations are the same
as those for meshing a solid model. FIGURE 111 shows examples of sheet
model surface meshes. You can load models such as PAM_022_a.sci and
PAM_Sheet_TjoinGen2.sci from the Tutorial Handbook to practice sheet
meshing.
FIGURE 111 Sheet model surface mesh examples

PAM automesher
Creating a solid mesh by extrusion
PAM has several methods to convert a surface mesh into a solid mesh. These
include: Thicken, Spin, Extrude, and Extrude Face. A Boundary Layer method can
also be used to create a boundary layer mesh on user specified faces.
Thicken The thicken function extrudes a surface mesh (triangles) in the direction normal to
the surface to form pentahedral elements. To create a mesh by thicken consider the
sheet body shown in FIGURE 112. This model can be loaded from the Tutorial
Handbook: PAM_022_a.sci. Next, create the following meshing record:
• Create > Mesh-PAM > Thicken, and select the sheet body to be meshed.
The selected sheet model will be highlighted. Next, input the five parame-
ters as shown in FIGURE 112 and click on Accept.
FIGURE 112 Input settings for thicken mesh example
The meaning of each input is as follows:

MinAng --- Minimum edge angle criteria.
MaxAng --- Maximum edge angle criteria
Layers --- Number of layers. If Side=0 then the number of layers is under-
stood to be at each side of the surface.

PAM automesher
T-Total--- Total thickness or ID of defined table.

Side --- Side to be extruded
Side= 0-- thicken both sides
Side= 1-- thicken in the direction of the positive normal
Side= -1-- thicken in the direction of the negative normal
To begin the PAM meshing process, click on the Automesh button. FIGURE
113 shows the mesh with three layers of elements on each side of the surface.
CAD surface
FIGURE 113 Thicken mesh for Layers=3, T-Total =0.001, Side=0.

PAM automesher
To try both-side-thickening with a single layer of elements through the thickness,

load the PAM_035_a_ExtrudeDelta.sci input file from the Tutorial Handbook and
delete the existing Mesh-PAM extrude record. Then complete as follows:
• Create > Mesh-PAM > Thicken, and input Layers =0, T-Total=0.001, and
Side=0. Click Accept and then Automesh.
PAM creates a one layer mesh with the CAD surface located in the middle of the
layer (See FIGURE 114). The input file with the meshing record described above
can be found in the Tutorial handbook:
PAM_035_a_OneLayer_ExtrudeBothSide.sci. FIGURE 115 shows some exam-
5
T-Total=0.001, Layers=0, Side=0 CAD surface
FIGURE 114 Mesh for Layers=0, T-Total=0.001, and Side=0
ples of meshes constructed by the thicken method.
Note: if the sheet model is not planar, the T-Total should be given carefully. The
total thickness is constrained by the smallest radius of curvature concave to the
extrude direction. PAM checks the input thickness, and if it exceeds the limit com-
puted by PAM based on the curvature, then meshing cannot be performed.
If you would like to try more examples, you can load PAM_pipe_7__sheet2.sci and
PAM_040_a.sci from the Tutorial Handbook. They already have defined mesh
records, so you can see the resulting mesh by simply clicking on the Automesh but-
ton (see FIGURE 115).

PAM automesher
FIGURE 115 Thicken mesh examples.

PAM automesher
Extrude By using the extrude method, you can extrude a surface mesh in the X- Y- or Z-axis
direction of a user defined local system to form pentahedral elements. To create an
extruded mesh of a sheet model, load the file PAM_thin_sheet.sci from the Tutorial
Handbook, set AOM to Create > Mesh-PAM > Extrude under the Mesh tab.
To define the meshing record, you must pick the sheet body to be meshed. The
sheet model will be highlighted red. Then, while holding Ctrl-Shift, click on the
local system that will be used for the extrusion direction. The system will be high-
lighted green. Next, you must input the four parameters and choose the axis of
extrusion as shown in FIGURE 116.
FIGURE 116 Input settings for extrude mesh example.
After you input the parameters, click on the Accept button to define the meshing
record. Then click on the Automesh button to begin the PAM meshing process.
Extrude Face The extrude face method allows you to extrude a surface mesh to another specified
face along the X- Y- or Z-axis direction of a user defined local system to form pen-
tahedral elements. To create an extrude mesh of a sheet model to a face, load the
model PAM_Extrude_ToSurf_test01.sci from the Tutorial Handbook.
The file already contains a meshing record that was created by selecting:

PAM automesher
• Create > Mesh-PAM > Extrude Face > pick objects in a specific order.
First, pick the sheet body to be meshed (the smaller surface), and it will be
highlighted in red. Second, while holding Ctrl-Shift, pick the surface to which
the mesh will be extruded (the larger surface), and it will be highlighted in
green. Third, while holding Ctrl-Shift, pick the local system for the extrusion
direction (it will be highlighted green). After that, you must input the four
parameters and choose the axis of extrusion as shown in FIGURE 117.
FIGURE 117 Input settings for extrude face mesh example problem
After the parameters are input, click on the Accept button to define the mesh-
ing record. Then, click on the Automesh button to start the PAM meshing pro-
cess. FIGURE 118 shows the mesh produced by PAM for this example.
Spin This method spins a surface mesh around the X- Y- or Z-axis of a user defined
local system to form pentahedral elements. To create a spin mesh from a sheet
body, load the file PAM_surf03.sci from the Tutorial Handbook and select the
Mesh tab:
• Create > Mesh-PAM > Spin. To create the meshing record, first pick
the sheet body and it will be highlighted red. Then, while holding Ctrl-

PAM automesher
Second sheet body
First sheet body
Local system
FIGURE 118 Resulting mesh for extrude face example
Shift, pick the local system which will be highlighted green. Next, input the
four parameters and the spin axis as shown in FIGURE 119, and click on
the Accept button to define the meshing record. Click on the Automesh but-
ton to start PAM meshing process..
The meaning of the input fields ia as fllows:

MinAng --- Minimum edge angle criteria.
MaxAng --- Maximum edge angle criteria
Layers --- Number of layers to be spun.

PAM automesher
FIGURE 119 Input settings for spin mesh example
Angle--- Total spin angle or ID of defined table (in degrees )

5
The valid range of the Angle parameter is: -360.0 <= Angle <= 360.0
Either planar or curvilinear sheet bodies can be spun. However, when you cre-
ate the local spin system, you must be sure that all bodies are located on the
same side of the Y-Z plane of the local system.
Additional examples of spin meshing are available in the Tutorial Handbook:

PAM_Plug2_5mm_3Ch_Plug_Sheet_Gen_spin.sci and
PAM_spline_sheet_spin.sci. These models already have the proper meshing
records.
Note: When the given angle is very close to 360.0 or -360.0 within a tolerance
(1.0e-5), PAM will connect the last spun layer to the first spun layer.
Spinning is possible even in the special case where some linear edge of the
sheet model coincides with the spin Y axis. When this occurs, PAM will create
a special pentahedral mesh around the spin axis (see FIGURE 120).

PAM automesher
FIGURE 120 Spin mesh when some linear edge of the sheet model coincides
with the spin Y axis
However, when other edges connecting with the rotation axis edge are not perpen-
dicular to axis of rotation (see FIGURE 121), the resulting mesh will have a re-
entrant corner that may not be adequate for analysis. Care should be exercised
when using the spin method for these cases.
Spin axis
FIGURE 121 Spin mesh when other edges connecting with the edge of
rotation are not perpendicular to it.

PAM automesher
Additional examples of meshes created by spin are available in the Tutorial

Handbook: PAM_surf03_SpinAxis.sci and
PAM_surf03_CutCorner_SpinAxis.sci.
Boundary Layer The Boundary Layer method creates a layer of elements after a tetrahedral
mesh of a solid body has been created. PAM will collect all tetrahedral element
faces on user specified face(s) and use the Thicken function to form a layer of
pentahedral elements. This function is useful when the user wishes to solve a
model with very thin layers of differing material properties on solid models.
FIGURE 122 shows a tetrahedral and a pentahedral boundary layer mesh.
Boundary layer faces
Solid model
Tetrahedral mesh with pentahedral boundary
layer mesh
pentahedral boundary layer mesh

Tetrahedral mesh
FIGURE 122 A tetrahedral and a pentahedral boundary layer mesh
You can practice boundary layer mesh creation by loading the input file
PAM_2HOLEMODEL_Split.sci from the Tutorial Handbook. To make a solid

PAM automesher
mesh with a boundary layer mesh, you must define two meshing records: the
boundary layer meshing record and the solid meshing record.
To define the boundary layer meshing record, set AOM to Create > Mesh-PAM >
Bndry Layer in the Mesh class, pick the face(s) of the solid on which the boundary
layer mesh is to be formed, input the parameters (Layers=3, T-Total=0.003 for the
example), and click on the Accept button. To define the solid meshing record, set
AOM to Create > Mesh-PAM > Auto, pick the solid body, input the parameters
(MinAng=5, MaxAng=175 for the example), and click on the Accept button. After
these two meshing records have been defined, click on the AutoMesh button to ini-
tiate the PAM meshing process.
Note: In the current release of PAM, the boundary layer mesh thickens the mesh in
the positive normal direction, outside from the solid model face. Therefore, the
meshed model is slightly larger than the solid model. Keep this in mind when solv-
ing the model.
Since all defined meshing records will be processed by StressCheck, it is permissi-

ble to define multiple meshing records and then run PAM by clicking on the
AutoMesh button once. The four meshing records shown in FIGURE 123, for
example, will all be executed and the meshes created once the Automesh button is
clicked. The model PAM_Many_MeshRecords.sci available in the Tutorial Hand-
book illustrates this.
5
Creating a disconnected mesh on a crack face
The Crack Face method allows for the creation of disconnected meshes on user
specified crack face(s). To create a mesh that is disconnected on crack faces, two
meshing records are necessary: the crack face meshing record and the solid mesh-
ing record. In order to visualize the partition faces you wish to specify as crack
faces, you may first need to blank some faces of the model.
Next, set AOM to Create > Mesh-PAM > Crack Face in the Mesh class, select the
crack faces, and click on the Accept button to define the crack face meshing record.
To define the solid meshing record, set AOM to Create > Mesh-PAM > Auto, select
the entire solid body, input the angle constraints, and click the Accept button. After
the two meshing records have been defined, click the AutoMesh button to begin the
PAM meshing process. FIGURE 124 shows general cases of disconnected meshes
that PAM can create.

PAM automesher
4 meshing records
Spin, extrude, thicken, and solid meshing were performed after

clicking on AutoMesh button.
FIGURE 123 Illustration of processing all defined meshing records simultaneously

5
You can practice the creation of disconnected meshes by loading the following
files from the Tutorial Handbook:
PAM_Crack_Face_Cyl_test.sci
PAM_Crack_InBody_Ell_test.sci
PAM_Crack_Face_InterfaceSpl_2Areas_test.sci
These models already have the proper meshing records.
Currently, PAM can only create a disconnected mesh on a separate object,

which includes the following objects:
Separate Face ---Outer face of a solid model or user specified crack partition face
Separate Edge---Edge only used by Separate Face
Separate Point---Point only used by Separate Edge
FIGURE 125 shows some crack face mesh examples.

PAM automesher
Crack faces cutting the solid model

Crack faces inside the solid model
Solid body
Partition face
Solid general body
Crack faces created by imprinting closed

curve(s) onto the partition face.
Solid general body
FIGURE 124 General cases of possible disconnected meshes
Defining a table to use with PAM meshing 5
When creating meshing records for Thicken, Extrude, Spin, etc., the thickness of
each layer or angle of spin of each layer is calculated by dividing T-Total or Angle
by the number of layers. We will now introduce another method for controlling the
thickness of each layer.
You can define data tables and use their IDs as input parameters when creating the
meshing records. To define a table, click on the Tables tab in the Model Input Dia-
log, set AOM to Select > Extrude Body > Constant, and enter the table ID, Layer
number, and corresponding Value.
For example, suppose you wish to define a table that will be used to control differ-
ent layer thicknesses while extruding a surface mesh. First, input the table ID
(TEST in FIGURE 126a). Second, input the layer number (1 in this case) and the
layer thickness value (0.1 in this case). Finally, click on the Accept button to define
the first table value. To define the second table value, simply input layer number 2
and the corresponding value, and click on the Accept button. Repeat this operation
until the nth layer and value are defined. To confirm the defined tabular data, click

PAM automesher
Crack face cutting the solid body
Crack face inside the solid model
Crack faces on the interface of different

materials bonded together.
5
FIGURE 125 Crack face mesh examples

the combo box (see FIGURE 126b). In our example, there are five values
defined in the table named TEST.
Once the data table is defined, its ID can be used in the following PAM mesh-
ing operations:
1. Mesh > Mesh-PAM > Thicken: Use the defined table ID in T-Total
input field.
2. Mesh > Mesh-PAM > Extrude: Use the defined table ID in T-Total input
field.

PAM automesher
(b)
FIGURE 126 Example of tabular data
(a)
3. Mesh > Mesh-PAM > Extrude Face: Use the defined table ID in T-Total
input field.
4. Mesh > Mesh-PAM > Bndry Layer: Use the defined table ID in T-Total 5
input field.
5. Mesh > Mesh-PAM > Spin: Use the defined table ID in Angle input field.
To see an example, load PAM_035_a_ExtrudeDelta_Table.sci from the Tutorial

Handbook. When you mesh the model, notice that the layer thickness is controlled
by the tabular values.
Since the values in the defined table will be used to define the thickness or spin
angle, it is not permissible to define a table with a zero value or with a mixture of
positive and negative values. If the values in the table will be used to define the spin
angle, you must ensure that the sum of those values is <= 360.0 and >= -360.0.
If you wish to define a table to be used with Create > Mesh-PAM > Extrude Face,
keep in mind that the tabular values do not represent the thickness of each layer.
Since the surface mesh and the face on which the mesh will be extruded may not be
planar, each layer does not have fixed thickness (see FIGURE 127). Node n is
extruded to k at the opposite face, the layer thickness at n is the distance d between
p and m, and the total thickness at node n is the distance D between n and k. The

PAM automesher
ratio of d to D corresponds to the value in the table. Therefore, you must ensure
the sum of those values does not exceed 1.0.
a layer
レイヤー
押し出される
Mesh to be extruded メッシュが押し出され
Face to which mesh
メッシュ
るフェース
is extruded
p m
n k
d
D
FIGURE 127 Using tabular data with Extrude Face meshing
Using PAM Element Sets and Element Face Sets

5
When meshing a general body solid model or a sheet model, PAM can create
element sets and element face sets. This makes it more convenient to assign
materials and to define loads and constraints.
In the case of a Parasolid general body solid model, there may be several solid
regions. PAM can create an element set for each solid region to make assigning
different materials more convenient. In case of a Parasolid sheet model, if an
extrusion mesh (thicken, extrude, spin, extrude face) is created, PAM will cre-
ate an element set for each extrusion layer. Additionally, if the sheet model has
more than one surface, PAM creates an element set for the elements extruded

PAM automesher
from each surface. FIGURE 128 and FIGURE 129 show the element sets of a Para-
solid solid model and sheet model.
FIGURE 128 Element sets created for parasolid solid regions (shown by a unique color)
Element sets created for layers
Element sets created for surfaces
FIGURE 129 Element sets created for extrusion layers and surfaces of a sheet model

PAM automesher
Using element sets to assign materials
To practice material assignment, you may load the file

PAM_GenBody_PLT_CYL_Box.sci from the Tutorial Handbook and mesh it
by clicking on the Automesh button under the Mesh Tab, Create > Mesh-PAM.
Once the mesh is created, you may assign material(s) to some elements.
To do this, you must define material(s), select the elements to which you wish
to assign certain materials, and click on the Accept button. While this may
seem simple, a complicated model shape can make it difficult to select the
desired elements. In this case, the element set created by PAM will assist you
in assigning materials. The model shown in FIGURE 131 is a Parasolid general
solid body with three solid regions. Suppose the mesh was already created, you
now want to assign a different material to each region, and you already have
defined three materials: 2014-T6, ASTM-A242, ASTM-A36 (see FIGURE
130).
5 FIGURE 130 Defined materials
Now, you can click on the Assign tab to begin material assignment. First, set
AOM to Select > Any Element > Selection under the Material class. An ele-
ment set can be selected by holding Ctrl-Shift and clicking on any element to
which you want to assign the material. When an element is selected using this
method, a set browser appears that lists the sets to which the currently selected
element belongs. In the set browser, all candidate sets are shown in the list box
at the right with one of them highlighted. Use the mouse to select the desired
set and click on the “Move candidate to selection list” button to add the candi-
date set to the selection sets box (one element set has been added to the selec-
tion sets box in FIGURE 132). You can use Ctrl-Shift to select as many
element sets as you want to add to the selection sets box. Note that during your
Ctrl-Shift selecting, element set highlighting status may be changed in the
model window whenever you want.

PAM automesher
Element set 1
Element set 3
Element set 2
FIGURE 131 Element sets of Parasolid solid region.
To assign a material to a collection of element sets, be sure that they are all in the
selection sets box, click on the “Select all sets” button, select the appropriate mate-
5
rial ID and material color, and click Accept.
By repeating the above operation, you can use the Ctrl-Shift selection method to
select element sets of other solid regions and assign a material to them until all ele-
ment sets are assigned a material. FIGURE 133 shows a sample model that has
been assigned three different materials.

PAM automesher
Ctrl-Shift-Pick here
Select all sets Delete selected sets
5 Purge selection Cancel selection Move candidate to selection list
FIGURE 132When using Ctrl-Shift to select an element, candidate element sets can be shown in a set
browser. You can add any element set to the selection list.

PAM automesher
Use a regular mouse selection or a marquee pick to select elements that do not
belong to any element set.
FIGURE 133 Elements in three element sets are assigned different materials, which are
represented by different colors.
FIGURE 134 shows three material assignments, which can be confirmed by click-
ing on the record in the pulldown menu. When a material assignment record is
selected, all elements corresponding to the selected assignment will be highlighted
in the model.
5
You can load PAM_GenBody_PLT_CYL_Box_MeshWithMat.sci from the Tuto-
rial Handbook to see this material assignment..
FIGURE 134 Created material assignments

PAM automesher
Using element face sets to assign loads and constraints
FIGURE 135 shows a sheet model with three surfaces and the model’s spin
mesh without any assigned materials. You can load this model from the Tuto-
rial Handbook (PAM_surf03_Hole_Gen2_WithSysRec.sci). After it is loaded,
mesh the model by clicking on the Automesh button, and assign materials
using the method described in the previous section. FIGURE 136 shows the
mesh with three materials assigned. .
Surface regions on sheet model
FIGURE 135 A general sheet model with three surfaces and its spin mesh.
5
FIGURE 136 Mesh created by spinning a sheet model with three

surfaces. Three different materials are assigned to three element sets.

PAM automesher
Suppose you want to define a built-in constraint on the model. To do this, go to the
Constraint class interface, and set AOM to Select > Face > Built In. Then, hold the
Ctrl-Shift keys and use the mouse cursor to pick the two desired element faces as
shown in FIGURE 137. When you click the Accept button, the constraint will be
defined.
Ctrl-Shift-Pick here Built-in constraint
FIGURE 137 Use Ctrl-Shift to assign a built-in constraint to two element face sets.
An element face set can also be used to define a load. Suppose you want to assign a
pressure load to the element faces forming the tubular shape (see FIGURE 138).
You will find it very difficult to select all of the element faces of the pipe, because
5
many element faces are not visible, many element faces must be indivudually
picked, and the marquee pick does not work on curved surfaces. Fortunately, since
PAM created element face sets during the meshing process, you can select the ele-
ment faces you want by a single Ctrl-Shift selection.
Element faces forming a tubular shape.
FIGURE 138 Tubular shapes in the mesh.

PAM automesher
PAM can create element face sets for all possible “surfaces”, which include
CAD surfaces and “quasi-surfaces” formed by spinning (extruding) CAD
curves on CAD surfaces (see FIGURE 139). An element face set of quasi-sur-
faces is very useful when assigning a load on a spin (extrude) mesh.
Quasi-surfaces
FIGURE 139 CAD surfaces and Quasi-

CAD surfaces Surfaces.
If you use Ctrl-Shift to select an element face of the first tubular shape and then
use Ctrl-Shift to select an element face of the second tubular shape, you will
see that all element faces forming the two tubular shapes are highlighted.
Now, you can assign a normal load on those element faces to create the the
5 pressure load. To visually confirm the defined load, you must be in wire-frame
mode so that the load marks can be seen. FIGURE 140 shows the defined inner
pressure load on the tubular shapes, where the arrows indicate the direction of
the load.

PAM automesher
To view this model, load PAM_surf03_Hole_Gen2_WithLoadConstr.sci from the

Tutorial Handbook.
FIGURE 140 Defined inner pressure load on

tubular shape.

PAM automesher

ANALYSIS GUIDE

StressCheck
Volume 3
ANALYSIS GUIDE
Release 7
August, 2005

Copyright 2005
COPYRIGHT NOTICE
U.S.A.

rights reserved.

DISCLAIMER

or change.
1
1 Planar Elasticity
This chapter contains guidelines for the preparation of input data, for obtaining a linear solution and for
performing post solution operations for problems in Planar Elasticity. It is recommended that you
become thoroughly familiar with this chapter prior to attempting other analysis types, such as axisym-
metric, plate bending or 3D analyses.
Setting the analysis type and choice of units

After opening a new database, adjust the reference and theory selectors to Planar
Elasticity. Under this analysis mode, all the input forms will contain the appropriate
fields and functionality supported for the reference and theory.
Analysis Guide Chapter 1: Planar Elasticity 1

Entering geometric data
1
The choice of units Any length unit can be chosen, the only requirement is that the units must be
consistent throughout the data set. It is good practice to work consistently
either with standard U.S. units or SI units. By standard U.S. units we under-
stand the following system of units: pound force (lbf), inches, seconds and
degrees Fahrenheit. SI units are mass (kg), meters, seconds and degrees
Kelvin. When using SI units in mechanical engineering practice then it is often
convenient to use millimeters for length units, Newtons for force units and
MPa for stress units.

From the Main Toolbar select the Create Model icon and then select the Geom-
etry tab in the Input dialog window (FIGURE 1). Geometry provides for the
FIGURE 1 Geometry input.
2 Chapter 1: Planar Elasticity Analysis Guide

1
specification of the solution domain using points, lines, circles, ellipses, rectangles,
etc. StressCheck lets you separate the definition of boundaries from the definition
of the finite element mesh. You will find that this feature gives you a great deal of
flexibility and convenience. You will be able to change the mesh and the new ele-
ments will be assigned the correct boundary conditions by StressCheck automati-
cally. Refer to the Modeling Guide for a detailed description of geometry
construction in StressCheck.
Select the Mesh tab when you are ready to define nodes and elements. Nodes may
be associated with previously defined points, specified as intersections of two
boundary curves, assigned as offsets on boundaries, defined directly, etc. The pro-
cess of entering geometric data is demonstrated through an example in the follow-
ing. Additional examples and exercises are presented to highlight important
features.
Example StressCheck makes it very convenient to specify solution domains bounded by

lines, circles, ellipses, and curves created by data fitting. Consider, for example, the
domain shown in FIGURE 2, representing part of a linkage arm.
Circular fillet, radius = 15
y
B
10
20
x
A 20
100
FIGURE 2 Part of a linkage arm. The length units are in millimeters.
Before creating any geometry make sure the units selector is set to mm/N/sec/C. To
specify the domain, select the Geometry tab in the StressCheck Input dialog box,
and then construct a rectangular domain using the following steps:

1
• Class tab: Geometry > Action: Create > Object: Rectangle >
Method: Locate > Input: (Make sure the toggle switch is ON) X: 0.0,
Y: -10.0, Z: 0.0, Width: 80, Height: 20, Rot-Z: 0.0 > Button: Accept. If
the rectangle is not visible, click on the Center Model icon.
Note that the logical sequence was to select the Class (GEOMETRY), an
Action (CREATE), an Object (RECTANGLE), and the Method by which the
object is to be created (LOCATE, that is, specify the data which define the rect-
angle: the coordinates of a vertex point, the width and the height, measured
from the vertex point). The resulting rectangle consists of four lines and four
points.
Define next the inner circle by the commands:

• Create > Circle > Locate > Input: X: 100.0, Y: 0.0, Z: 0.0, Radius: 10,
P1-Min: 0.0, P1-Max: 360.0, Rot-Z: 0.0 > Accept
Similarly, create the outer circle with a radius of 20 units. Finally, create the fil-
lets using the commands:
• Create > Fillet > Selection > Input ON > Radius: 15. Then using the
mouse, click on the upper horizontal line and the outer circle (in either
order) close to the points where the points of tangency are expected.
A circular fillet will appear on the screen. Points were created at the fillet end
points. The radius of the fillet will be exactly 15 units, the center of the circle
and tangent points (nodes) are computed by StressCheck automatically. Repeat
for the lower horizontal line and the outer circle. It is possible that the fillet will
not turn out as you expected. This is because fitting a circle of a fixed radius
tangent to a circle and a line can have as many as five solutions. If you encoun-
ter such a problem then click on the DeLast button (delete the last command)
and try again, clicking on the circle and the line close to where you expect the
tangent points.
This completes the specification of the solution domain.

Entering mesh data
1
Entering mesh data
Elements types
Five types of elements are available in Planar Elasticity: quadrilateral, triangular,
beam, fastener and link elements.
Quadrilateral elements: They are constructed by selecting 4 nodes in any order.
Triangular elements: They are constructed by selecting 3 nodes in any order.
Fastener elements: Special elements that can be attached to circular boundaries.
Link: Special elements to connect fastener elements.
Beam elements: They are constructed by selecting 2 nodes in any order.
If neighboring nodal points lie on a boundary, for example a circle, then that bound-
ary is automatically associated with the element. This means that the mathematical
description of the boundary curve will be used for computing the stiffness matrix
and load vector for the element.
When the elements are displayed, an inscribed 3x3 mesh represented by dotted
lines, appears on the screen. These lines are there to help you see the element
clearly. If you do not wish these lines to be displayed then turn off the Element
Handles icon, or select View/ Display Options Toolbar/ Element Handles and the
inscribed mesh will disappear.
Designing the mesh

A general rule is that finite element meshes should be so constructed so that the ver-
tex angles of triangular elements are as close to 60 degrees as possible, and the ver-
tex angles of quadrilateral elements are as close to 90 degrees as possible. The p-
version is much more ‘forgiving’ with respect to deviation from the optimal vertex
angles than the h-version; nevertheless it is recommended to keep vertex angles
between 10 and 150 degrees.
There are simple rules for designing finite element meshes for problems in Catego-
ries A and B (Refer to the Advanced Guide, Theoretical Background chapter, for

Entering mesh data
1
the classification of problems in Categories A and B). These rules are outlined
in the following.
Consider, for example, the linkage arm detail shown in FIGURE 2. Assuming
that the applied loading is smooth and the constraints do not introduce singular
points, the solution of this problem will be very nearly in Category A. Strictly
speaking, the solution will not be perfectly in Category A because the curva-
ture of the boundary changes abruptly at the endpoints of the fillets. The solu-
tion is not analytic at these points which therefore must be nodal points. Note
that StressCheck automatically inserts nodes at the endpoints of fillets. The
abrupt changes in curvature will not introduce high stresses, nor will they
cause the computed stresses to oscillate. Therefore no mesh refinement is nec-
essary around these points.
Node numbers
Elements numbers
FIGURE 3 Portion of a linkage arm. Mesh details.

Entering mesh data
1
Example Let us now construct the mesh shown in FIGURE 3. The first step is to define the
nodes. Create nodes 1 to 6 by placing a node at the specified points.
• Class tab: Mesh > Action: Create > Object: Node > Method: Point. Click
on the four fillet end points and the two leftmost corner points on the rect-
angle. Click Accept to create the nodes.
StressCheck will indicate the node by a small square. The numbering sequence for
the nodes is unimportant. Nodes 1 and 7 shown in FIGURE 3, should be on a radial
line, or nearly so, to ensure that the angles of the neighboring quadrilateral elements
are nearly 90 degrees. To create nodes 7 and 8:
• Create > Node > Projection. Click on node 1, hold Control-Shift, click on
the inner circle, and click Accept. Node 7 will be created over the circle as
a projection (shortest distance) from node 1. Repeat by choosing node 3,
holding Control-Shift, clicking on the inner circle, and clicking Accept to
create node 8.
The rest of the nodes can be created as offsets on the given circles. To create nodes
9 through 12:
• Create > Node > Offset > Repeat # = 4 > Input ON > Offset: 0 + 90. Then
click on the inner circle.
To create nodes 13 through 15:

• Create > Node > Offset > Repeat # = 3 > Input ON > Offset: -90 + 90. Then
click on the outer circle. The result should be the set of nodes shown in
FIGURE 3.
Note: Turn the Repeat switch off before going to the next step.
Now we are ready to create the elements shown in Figure 3. To create a quadrilat-
eral element:
• Create > Quadrilateral > Selection. Then click on the four nodes which
define the element in any order.
To create a triangular element:

• Create > Triangle > Selection. Click on the nodes at the vertices which
define the triangular element in any order.

Entering mesh data
1
Checking the mesh

In order to ensure that all elements are properly connected, that is, there are no
unintended free edges, select:
• Mesh tab > Check > Edge > Free Edge.
Those element boundaries which are not connected to other elements will be
highlighted.
Note on vanishing angles

Elements with vanishing angles (zero vertex angle) are acceptable for linear
analysis. They require a special post processing procedure (which is performed
automatically by the program) and their main advantage is to facilitate mesh-
ing in filleted regions or when different materials are used. Quadrilateral and
triangular elements can have one or more vanishing angle. Consider a square
plate subjected to axial traction as shown in FIGURE 4. The central quadrilat-
eral element has all four corners with vanishing angles, however the stress dis-
tribution is practically not affected by this element. For a unit traction the stress
should be unity everywhere.
FIGURE 4 Quadrilateral element with all four vertex angles equal to zero.

Assigning thickness
1
Assigning thickness
For problems of Planar Elasticity it is necessary to associate some thickness with
the elements. To assign thickness, select Class: Thickness or click on the Thickness
tab in the StressCheck Input box (FIGURE 5).
FIGURE 5 Thickness input.
To assign a constant thickness to all elements, select All Elements and enter the
value of the constant. Click on Accept and StressCheck will confirm your entry in
the scrolling list. If you wish to assign different thickness to different elements,
select Any Element and click on the elements to which the thickness is to be
assigned. The elements will be highlighted. Alternatively, drag the cursor on the
screen until the desired group of elements is identified. If you wish to remove one
or more elements from the selected group, click on the elements to be removed
while holding the Ctrl key. To add element(s) to the set, click on the element(s)
while holding down the Shift key. Enter the desired thickness and click on Accept.

Entering material properties
1
Example For the problem shown in FIGURE 2 assign constant thickness (5.0 mm) to all
elements.
• Class tab: Thickness > Select > All Elements > Selection > Thickness:
5.0 > Accept.

To enter the material properties you must provide two types of information:
definition of material properties and assignment of material properties. Both
activities are performed by selecting the Material tab in the StressCheck Input
box, or Class: Material from the Main Menu Bar. FIGURE 6a shows the mate-
rial interface displayed on the screen when the Define tab is used for providing
the material coefficients. FIGURE 6b shows the interface when the Assign tab
is used for assigning the defined properties to the elements in the mesh.
To define the material properties select the Define tab from the Material inter-
face.
1. Decide whether you wish to enter your own material properties, or select the
material properties from a Standard list provided by StressCheck. To use mate-
rial properties from the standard list, click on the Browser button located at the
side of the material ID and when a list of material designations appears in the
window of the Material Library form [FIGURE 7], click on the material name
and the property data will be entered automatically in the Material input win-
dow. Finally, click on Accept.
2. Specify whether you wish to define a Material: Linear elastic, Elastic-Plastic

or Nonlinear-Elastic.
3. Specify the material Type. For linear materials the choices are Isotropic,
Transversely Isotropic, Orthotropic, Laminated Isotropic, Laminated Orthotro-
pic, Anisotropic, or Stiffness. Anisotropic in two-dimensions designates mate-
rials with only one plane of material symmetry. The plane of material
symmetry must be the xy-plane. Materials with one plane of symmetry are
called monoclinic materials. After you select the material type, the scroll win-
dow automatically adjusts to provide the appropriate number of fields into
which the material properties are to be entered. For the types of elastic-plastic
and nonlinear-elastic materials, refer to the Advanced Guide.

1
(a) (b)
FIGURE 6 Material properties input.
4. Name a material by entering the name into the field next to ID: (e.g. mat, steel,
alum, etc.). If the toggle switch in front of material ID is turned on then the name of
the material will be shown for each element once you completed the assignment
process.
5. Specify whether the analysis is to be Plane Stress or Plane Strain.
6. Provide the material properties if not using a material from the list. All values of
the material coefficients can be constant or parametric. Entering a parameter name
in the input field is all it takes to associate any coefficient to a previously defined
parameter.
7. Click on Accept to associate the material name (entered into the material ID box)
with the specified material properties.

1
FIGURE 7 Material Library Form.
The material properties on the Standard list of StressCheck are average values
taken from various reference books. Actual material properties may vary sub-
stantially from these generally accepted average values. It is the user’s responsi-
bility to ensure that the material data are properly defined for the specific
application.
8. Assign the material ID to the elements. When only one material is defined,
the material assignment is not required. The program will utilize that material
for all the elements in the mesh. If more than one material is defined, or if the
material is other than isotropic, then an assignment is needed.
Important notes:
• For orthotropic and transversely isotropic materials, a coordinate sys-
tem other than the global can be specified. In such cases the material 1-
axis is aligned with the x-axis of the selected local coordinate system
while the material 2-axis is aligned with the y-axis. If the selected
coordinate system is cylindrical (r,θ,z), then the material 1-axis is
aligned with the θ-direction and the material 2-axis is aligned with the
r-direction. Such local coordinate systems can have an arbitrary rota-
tion about the global Z-axis only. Rotations of the local system about
any other axis will be ignored during the analysis. To account for arbi-
trary orientation of local systems, laminate property assignment should
be used as described below.
• Stiffness, in force per unit length, can be defined to be assigned to link
elements only.

1
Example Let us define and assign the materials properties of a 2014-T6 aluminum alloy
(Plane stress, SI units) to the linkage arm. Select the following options:
• Class tab: Material > Action: Define > Object: Linear > Method: Selection
> Define tab > Click on Browser button and when the Material Library
form appears, click on the aluminum alloy 2014-T6 record. When you
select a material from the list, the name is automatically entered in the input
fields. > ID: ALUM > Type: Isotropic > Case: Pl. Stress > Click on Accept.
You can replace the name before clicking on the Accept button.
To assign materials to elements:

• Assign tab > Select > All Elements > Selection > ID: ALUM > Accept
Laminate properties
A capability is available to facilitate the input of orthotropic material properties for

individual plies and for sub-laminate properties for 2D plane-strain analysis.
• Individual plies: When the ply-angles are not contained in the standard 2D
working plane (the plane-strain XY plane), the material matrix needed for
the 2D analysis is extracted from the 3D material matrix. The material coef-
ficients are entered in the material directions, and then assigned to the ele-
ments together with the ply angle information. The program performs all
necessary transformations to compute the equivalent 2D properties in the
XY work plane. The appropriate 3D properties are applied when the planar
model is extruded into 3D as well.
• Sub-laminates: When a set of plies needs to be combined in a single ele-
ment (sub-laminate), the properties of the sub-laminate are obtained by
homogenization. Again, the 3D material coefficients in the material axes of
the composite are entered together with the stacking sequence of the sub-
laminate, and the program computes the equivalent 2D properties in the
Stress Check XY work plane for the stack. The appropriate 3D properties
are applied when the planar model is extruded into 3D as well.
The procedures implemented in Stress Check to account for these two cases include
the following:
• The material coefficients of a single orthotropic ply are entered in the direc-
tion of the material axes (FIGURE 8a) by selecting the material type:
Lamin.Ortho under the Define tab. Nine engineering coefficients (E11, E22,
E33, G12, G23, G31, ν12, ν23, ν13) in the direction of the material axes and
the ply thickness must be provided, since plane strain is the default setting

1
for this material. The three coefficients of thermal expansion (α11, α22,
α33) and the mass density can also be entered, but they are not required
unless there is thermal loading or a modal analysis is needed.
• After the material properties for a single ply are entered, the ply group
information must be provided. This includes the angular orientation of
each ply in the group, in accordance with the following convention: A
positive ply angle (θ) is measured as a counterclockwise rotation about
the z-axis of a local coordinate system (xyz). The z-axis of the local is
aligned with the material 3-axis as shown in FIGURE 8b.
z≡3 1
θ x
y
3 Y
1 X 2
2
‘plane-strain’
plane
Z
(a) Material axes (b) Rotation of material axes
FIGURE 8 Material coordinate systems for orthotropic plies.
• The ply groups are assembled into a stack by providing the total num-
ber of layers in the laminate, defining whether the stack is symmetric
or not and entering the stacking sequence based on the ply group
names. This procedure provides great flexibility, since various ply
groups of different material properties can be assembled together to
represent a single sub-lamina.
• Finally, the stack is assigned to the elements in the mesh identifying
the local coordinate system, the z-axis of which is assumed to coincide
with the material 3-axis and must be perpendicular to the plane of the
ply (FIGURE 8b). The case of a single ply assigned to an element is
treated as a particular case of a sublaminate with one orthotropic layer.

1
• It is also possible to assign laminated plies directly instead of having to

assign a stack, which in turn references a ply definition. This feature is par-
ticularly useful when the saved material properties are used for all the plies
in the stack.
The two-dimensional material stiffness matrix needed for the plane-strain analysis
is then extracted from the three-dimensional material matrix in global coordinates
(XYZ). When a group of laminae is assigned to a single element, the three-dimen-
sional effective properties are computed by homogenization (refer to the Advanced
Guide for additional information).
Example 1 Consider a 16-ply [4(0/90)]S simply supported laminated composite strip under
plane strain conditions subjected to a sinusoidal traction as shown in FIGURE 9.
This problem can be loaded from the Tutorial Handbook: BeamLaminate.sci. Each
q=cos(πx/a)
h=1.0
a=20
z
Local System
SYS2
x
a/2 Uy=0
Ux=0
(symmetry)
FIGURE 9 16 ply [4(0/90)]S simply-supported laminated composite

under sinusoidal load. Notation and mesh.

1
ply is of the same material and thickness (t=0.0625 mm) with the following
properties:
EL = 1.38x105 MPa, ET = 9.3x103MPa
GLT = 4.6x103 MPa, GTT = 3.1x103 MPa
νLT = 0.3, νTT = 0.5
where L is in the direction of the fibers and T is the transverse direction.

Because of symmetry, only half of the beam was considered for the analysis.
The mesh shown in FIGURE 9 consists of 16 quadrilateral elements, one for
each ply. The material properties for a each ply are entered in the material def-
inition form shown in FIGURE 10a. All nine engineering coefficients (E11,
(b) Ply group definition
(a) Material definition

(c) Stacking definition
FIGURE 10 Input forms for laminate properties definition. Example 1

1
E22, E33, G12, G23, G31, ν12, ν23, ν13) in the direction of the material axis and the
ply thickness must be provided in this form. The relation between the L-T and the
1-2-3 notation is: E11=EL, E22=E33=ET, G12=G31=GLT, G23=GTT, ν12=ν13=νLT
ν23=νTT. The three coefficients of thermal expansion (α11, α22, α33) and the mass
density are optional. After the material properties for a single ply are entered, the
ply group information must be provided as shown in FIGURE 10b: The ply group
name, the lamina material name and the ply layout angles. To do this, go to the
Assign tab, and select Type: Laminate-Stack > Ply. The stack form shown in FIG-
URE 10c can be accessed under the Assign tab > Type: Laminate-Stack > Stack,
and is used to assemble the ply groups by providing the stack sequence name, the
number of layers in the laminate, whether the stack is symmetric or not, and the
stacking sequence based on the ply group names. As shown in FIGURE 10, the
material property SINGLEPLY, defined in the material definition form, is used for
defining the ply group PLY90 with a ply layout of 90 degrees with respect to the z-
axis of a local system and the stack LAM90 is created with a single ply using the
plygroup PLY90. The last step is to assign the material stacks to the elements and
select the local coordinate system, the z-axis of which is assumed to be aligned with
the material 3-axis. The material assignment form is shown in FIGURE 11. The ori-
Stack name
Local system
FIGURE 11 Input form for

laminate properties assignment
to elements.

1
entation of the local coordinate system relative to the global system is used for
determining the material matrix in the global system as explained in the previ-
ous section.
Example 2 Consider the same 16-ply [4(0/90)]S simply supported laminated composite
strip of example 1, but with the eight central layers replaced by a single ele-
ment with equivalent properties. The finite element mesh is shown in FIGURE
12. The ply group definition and stacking sequence for the 8 individual layers
sub-laminate
h
a/2
FIGURE 12 Mesh for a 16 ply [4(0/90)]S simply-supported laminated composite

under sinusoidal load with the eight central layers replaced by a single element.
(4 above and 4 below the sub-laminate) are the same as shown in FIGURE 10
for example 1. The corresponding input records for the sub-laminate are shown
in FIGURE 13. The central 8-plies [2(0/90]s can be described in more than one
way in the input field of FIGURE 13a. For example, 0/90/0/90/90/0/90/0, or
(a) Ply group definition (b) Stacking definition

FIGURE 13 Input forms for sub-laminate properties definition. Example 2

Entering load data
1
2[0/90];2[90/0] as shown. Alternatively, define only half of the sub-laminate (4-

plies) in the ply definition form (that is, 2[0/90]) and then use the Symmetric option
in the stacking sequence. This problem can be loaded from the Tutorial Handbook:
BeamSublaminate.sci.
Entering load data

To enter load data select the Load tab in the StressCheck Input box or Class: Load
from the Main Menu Bar. The input area will appear as shown in FIGURE 14.
Specify a unique name which identifies the loading case you are about to enter. In
engineering practice often multiple load cases must be investigated. Each load case
must be given an unique name in the ID field.
FIGURE 14 Input forms for Load and Formula definition.

Entering load data
1
Types of loading
Several types of loading such as traction, spring, body forces or point loads are
available for Planar Elasticity. The definitions are as follows:
Traction loading: means that a distributed load (in force per unit area) is
imposed on a boundary or edge. Traction is a vector quantity. Thus, two vector
components must be given. These may be in the normal-tangent reference
frame, in the global system, or an arbitrary local system. Traction loads can be
applied to geometric boundaries of element edges, including beam elements.
Spring loading: means that load is applied through a displacement imposed on

a distributed spring which has been specified on an edge or boundary. Specifi-
cally, the traction vector components in the normal-tangent reference frame,
for example, are dependent on the imposed spring displacements:
=
where Tn, Tt are traction vector components in the normal and tangent direc-
tions, respectively, ann, ant, att are spring coefficients (in units of force per
length cubed), δ , δ are imposed displacement components in the normal-
n t
tangent directions. These are the loading terms. The symbols un, ut represent
the normal-tangent displacement vector components.
Spring loading can be applied in global or local x-y coordinates as well. The
only restriction is that the spring coefficients and the imposed spring displace-
ments must be in the same reference frame. For further information on spring
loading please refer to pp. 75-76 of Ref. [3]. Spring loading is not available for
beam elements.
Body force loading: The body is loaded by some distributed load, such as by
gravity or inertia due to acceleration. The body force is in terms of force per
unit volume. If the acceleration vector a = {a x, a y} and the mass density ( ρ )
are known then enter the products:
Fx = ax ρ
Fy = ay ρ

Entering load data
1
Body forces can be applied to quadrilateral, triangular and beam elements.
Thermal loading: means that a temperature distribution (in degrees) will be speci-
fied on elements. Refer to Chapter 6: Conduction Heat Transfer for specification of
temperature distribution obtained from a heat transfer analysis. Thermal loading
can be applied to quadrilateral, triangular and beam elements.
Point force: The elastic body is loaded by a concentrated force F = {F x, F y} . You

may specify Fx, Fy directly or the absolute value of F and the direction of F in
terms of the angle measured counterclockwise from the positive x-axis. Refer to the
Advanced Topics Guide for additional information about the use of concentrated
forces. Point loads can be applied to points, nodes or fasteners. The points can be
located inside a quadrilateral or triangular element, but only on nodes for beam ele-
ments.
Bearing load: Normal sinusoidal traction applied to a circular hole by specifying

the magnitude and angular direction of the force or the two vector components Fx
and Fy.
Moment: Nodal moment applied to beam elements only.
Interference loading: Magnitude of the diametral interference that can be speci-

fied to a fastener element. If the interference is positive, means that the diameter of
the fastener is larger than that of the hole. If the interference is negative (gap), the
diameter of the fastener is smaller than that of the hole.
Example Specifying normal traction by formula. Let us apply a sinusoidally distributed nor-
mal traction to the inner bore of the linkage arm detail shown in FIGURE 2. We
will use the mesh shown in FIGURE 3. Let the thickness be 5.0mm, and the mate-
rial the aluminum alloy 2014-T6, E=7.52 x 104 MPa, v=0.397. We shall assume
plane stress conditions. A load of 104 N is applied in the axial direction (i.e., the
positive x-direction) such that the corresponding normal traction is sinusoidally dis-
tributed compressive traction over half of the bore. Therefore the normal traction Tn
is of the form:

 C cos ( t ) for t ≤ π
---
 2
Tn = 
 π
 0 for t > ---
2


Entering load data
1
where t is the angular polar coordinate measured from the x-axis of a local
coordinate system located in the center of the circle. The coefficient C is to be
determined such that the resultant of the traction Tn is the desired axial force.
Referring to FIGURE 15 and noting that the positive normal traction is in the
direction of the positive (outward) normal, we have:
π π
--- ---
2 2
2 π
Fx = – ∫ Tn cos ( t )brdt = – Cbr ∫ cos ( t )dt = – --- Cbr
2
π π
– --- – ---
2 2
where Fx is the desired axial force (Fx = 104 N), b is the thickness (b = 5.0 mm)
and r is the radius (r = 10 mm). From this we have:
2F x 4
2 × 10
C = – --------- = ------------------------ = – 127.3MPa.
πbr π × 5 × 10
y1
Tn
t
x1
FIGURE 15 Positive normal traction for a circular hole.
Select the Formulae icon and in the Formula entry form enter the following
information:

Entering load data
1
• Name: TN > System: Cylin./Polar > Angle: -pi to pi > Use Constants, C1:-
127.3 > Formula: if (abs(t):le: pi/2; C1*cos(t); 0). That is if the absolute
value of t is less or equal π/2 then TN = C1*cos(t), else TN = 0. An equiva-
lent expression for TN is: if ((t:ge: -pi/2) :and: (t:le: pi/2); C1*cos(t); 0).
Click on the Accept button.
In the Load input area

• Class Tab: Load > Action: Select > Object: Any Curve > Method: Traction
> ID: LOAD > Direction: Norm./Tan. (Normal/Tangent) > Normal: TN >
System: select the local coordinate system of the inner circle from the pull-
down menu. The selected system will be highlighted on the screen. Make
sure that the tangent toggle switch is off. Select the inner circle and click on
the Accept button. To display the name of the local system select the dis-
play Objects icon from the Views Toolbar.
The applied traction will be displayed as shown in FIGURE 16. The up and down
buttons located next to the load Scale input field can be used to change the size of
the arrows representing the specified traction.
Bearing load The type of loading illustrated in FIGURE 16 frequently occurs in professional
practice because it approximates the interaction of a neat fit pin or fastener with a
plate. StressCheck lets you specify this type of loading very conveniently. In the
method combo box simply select Bearing, enter the magnitude and direction of the
load (in force units) then point to a circular boundary and click on the Accept but-
ton. The program will compute the sinusoidal traction corresponding to the speci-
fied bearing load.
Checking applied loads StressCheck makes it very convenient to check the magnitude of the applied
mechanical loads.
• Load tab > Check > All Elements > Selection > System: Global > Moment-
X: 0.0 > Moment-Y: 0.0 > Button: Accept.
The moment center, the load vector components FX, FY, and the moment Mz will
be displayed in an Edit window. You will find the FX value to be virtually the same
as the applied load of 104 N.
Alternatively you may check the size of the mechanical load applied to any part of
the domain by selecting Action: Check, Object: Any Element and while holding
down the Shift key clicking with the left button of the mouse on the desired ele-
ments. You may exclude selected elements by holding down the Ctrl key while
pointing to the ones to be excluded.

Entering constraint data
1
FIGURE 16 Display of the applied sinusoidal traction.

To enter constraint data select the Constraint tab in the StressCheck Input box
or Class: Constraint from the Main Menu Bar. The Input Area will appear as
shown in FIGURE 17. Specify an unique name for the constraint data you are
about to enter. This is necessary because StressCheck allows more than one
constraint case to be entered. Each case must be identified by a unique name.
Types of constraints
Select the type of constraints, such as General, Rigid Body, or Spring Coeffi-
cient as explained in the following:
General: means that you will prescribe one or two displacement components
to one or more boundaries or edges. For example, if one of the boundaries of
the solution domain lies on a plane of symmetry, which in 2D elasticity appears
as a line of symmetry then you will prescribe the normal displacement to be
zero.

1
FIGURE 17 Input area for constraint.
Built-In: Both displacement components are set to zero.
Symmetry: The normal displacement component is set to zero. Symmetry con-

straints are applicable only to straight edges in 2D and flat faces in 3D.
Antisymmetry: The tangential displacement component is set to zero. Antisymme-

try constraints are applicable only to straight edges.
The advantage of using these particular cases is that they can be interpreted auto-
matically in 3D when the extrusion option is used. Refer to Chapter 4 for details
about extrusion.

1
Spring Coefficients: When an elastic body is constrained by means of distrib-

uted springs, or loaded by displacements imposed by distributed springs then
the spring coefficients modify the stiffness matrix. For this reason the spring
coefficients are entered as constraints. The relationship among tractions, spring
coefficients and displacements is described in page 20 under the subheading
“Spring loading”. Spring coefficients can only be applied to quadrilateral and
triangular elements.
Rigid Body: StressCheck provides a convenient way for specifying rigid body
constraints when the object node is selected. In the case of two-dimensional
elasticity the user selects two nodes. StressCheck will then create a right-
handed local coordinate system, the x-axis of which is directed from the first
node to the second. In this local coordinate system both displacement compo-
nents are constrained for the first node; only the displacement component in
the direction of the local y-axis is constrained for the second node.
Single Node: Individual nodes can be constrained or their displacement pre-

scribed in one or more directions. In the case of beams, the nodal rotation can
also be prescribed.
Hinge: Releases the rotation degree of freedom associated with a beam ele-
ment at that node. A hinge is specified by selecting a node and one element
attached to that node. If more than two elements are hinged at a node, multiple
records can be created selecting the same node and a different element each
time.
Fastener Displacement: Prescribed displacement components for the center

of a fastener element. A fastener element has two degrees of freedom associ-
ated with its center.
Fastener Connection: Rigid connection between two or more fasteners that

share the same xy-coordinates.
Contact Zone: To especify the spring constant for a contact problem. For
additional details refer to the Advanced Guide, Multi-body Contact chapter.
Remark Using point constraint. Point constraints are admissible only for preventing
rigid body displacement and rotation. The body must be in equilibrium under
the action of external forces. Unlike the errors introduced by concentrated
forces, errors introduced by point constraints are not localized, except when
the point constraints are to prevent rigid body displacement only. In every
other case the reactions at point constraints are dependent on the mesh and the

1
polynomial degree of elements. For this reason multipoint constraints should not be
used in StressCheck.
To specify a general, built-in, symmetry, antisymmetry or spring coefficient con-

straint, the objects CURVE or EDGE must be selected first. To specify rigid body,
single node, or hinge constraint, the object NODE should be selected first. To spec-
ify a fastener constraint the object FASTENER must be selected first.
Example Enforcing symmetry and nodal constraints. Referring to FIGURE 2, we will

enforce symmetry conditions on boundary segment AB.
• Class Tab: Constraint > Action: Select > Object: Any Curve > Method:
Symmetry > ID: SYM > Click on the boundary segment AB (FIGURE 2).
The symmetry constraints will be indicated schematically, as shown in FIGURE 18.

The constraint record will also appear in the scrolling list.
At this point there is nothing to prevent rigid body motion in the Y-direction. To
impose the proper rigid body constraint, select:
• Constraint tab > Select > Node > Node > ID: SYM > Click on any of the
nodes, for example the one in the lower left corner of the domain > Direc-
tion: XY > Data Type: Fixed ON> X: turn this toggle off > Y: turn this tog-
gle on. Click on the Accept button.
The constraint imposed on the node will then be displayed symbolically. Note that
the same ID was entered for both constraints.
FIGURE 18 Symmetry and rigid body constraints enforced.

Defining the solution ID
1
By toggling the Element Handles icon to the “off” position you can display the
mesh and boundary conditions as shown in FIGURE 18.

Because StressCheck allows more than one load case and constraint case to be
defined, it is necessary to associate a unique solution name with each desired
constraint and load name pair. To do this, select Class: Solution ID, or select
the Solution ID tab from the StressCheck Input box. The constraint name(s)
and load name(s) previously defined are displayed on this form (FIGURE 19).
FIGURE 19 Solution ID input.

Setting discretization parameters
1
Example To complete the solution record for the linkage arm problem supply the following
information:
• Class Tab: Solution ID > Action: Define > Object: Name > Method: Selec-
tion > Solutions tab > Solution ID: SOL > Constraint ID: SYM (or click on
item in listbox) > Load ID: LOAD (or click on item in listbox). Click on the
Accept button.

Extension means a systematic increase in the number of degrees of freedom, as
explained in the User’s Guide. In StressCheck, typically uniform p-extensions are
used. It is possible, however, to select any group of elements and limit or fix the
polynomial degree for those elements. The options under the p-Discretization tab
are described in the User’s Guide.
If uniform p-extension, using the trunk space, is to be employed then nothing needs
to be done because this is the default option for StressCheck.
Example Let’s assign the product space to all the elements of the linkage arm problem. Select
the p-Discretization tab and the input area will appear as shown in FIGURE 20.
After selecting the Action, Object, and Method, you may select Space: Product, and
p-Discretization: Variable (meaning that the p-level is allowed to change). Enter the
starting p-level, and click on Accept. The entry will then appear in the scrolling list.
Writing an input file

It is good practice to periodically save the input data in an ascii file. To do this,
select File > Save Input or select the Save Input File icon from the Main Toolbar. A
form listing the current input file names will appear. Enter a file name which will
remind you of the job but which is different from the database name, then click on
the Save button. Typically, the database is deleted once an analysis is completed in
order to save disk space. The input file can be loaded into a new database by select-
ing File > Read Input.

Executing a linear analysis
1
FIGURE 20 p-Discretization input.

To execute a linear analysis select Edit > Solution from the Main Menu Bar, or
click on the Compute Solution icon from the Main Toolbar. When the Solution
dialog window appears (FIGURE 21) select the Linear tab and complete the
requested information. The available options are as follows:
Extension: Choosing “Upward-p” means that the solution will be computed

from the minimum to the maximum p-levels specified under p-limits. This
option requires more CPU time than the “Downward-p” but requires less disk
space. The highest possible p-level in StressCheck is 8.

1
Choosing “Downward-p” means that the solution will be computed from the maxi-
mum to the minimum p-level specified under p-limits. Generally, the downward
sequence is faster but requires more disk space than the upward. For the highest p-
level, the upward and downward options yield identical results. For the lower p-lev-
els there will be minor differences owing to the fact that the maximal number of
integration points are used when the downward option is selected. You may give
the p-limits in either order.
Choosing “Uniform-h” means that h-extensions (uniform mesh refinement) can be

performed when all boundary conditions were specified on “boundaries” rather
than element edges. When boundary conditions are prescribed on boundaries then
the loads and constraints are inherited by those elements which have an edge on the
boundary. StressCheck will generate finite elements by uniformly subdividing each
standard finite element. For example, for the mesh shown in FIGURE 3, the first h-
refinement (Run Limit: 2) will produce the finite element mesh shown in FIGURE
22.
If you select “Uniform-h” then you need to specify Run Limit, which is the number
of h-refinements to be performed. At each refinement level each finite element is
subdivided into four elements. Therefore the maximum number of elements gener-
ated is Mmax = 4r M1 where r is the specified run limit, and M1 is the number of ele-

1
FIGURE 22 Finite element mesh after one h-refinement.
ments at the start. You also need to specify the p-level which will remain the
same for all the runs.
Choosing “Adaptive-p” means that the polynomial order of the elements with a
variable or uniform p-discretization assignment will be determined based on
the values of the error indicator for each run until convergence is realized. The
Run limit input field refers to the maximum number of adaptive runs the pro-
gram will perform if convergence is not realized. The p input field is to provide
the initial p-level for the adaptive p-extension. The convergence options are
described below.
After completing the information in the Linear tab, select the SOLVE! tab and
choose from the available options:
Execute: Two options are available: “Initialize” and “Restart”. “Initialize”

means that the sequence will be computed from the initial p-level to the final
or, in the case of h-extensions, from the current mesh to the number of refine-
ments specified by the Run Limit entry.
“Restart” is used when a sequence of p-levels has been computed and you wish
to compute higher p-levels, retaining the computed data. In this case simply
increase the upper p-limit to its desired value and select “Restart”. In the case
of large problems it is good practice to run initially from p=1 to 4, check the
results, then if necessary, increase the p-limits until the desired level of preci-
sion is reached.

1
Run Mode: You may execute an “Automatic” sequence of solutions for the exten-
sion and limits selected. Or you may execute “Stepwise”, requiring you to click on
Solve following the solution for each p-level or h-refinement.
Method: It may be either “Iterative” or “Direct”. Method: Iterative means that the
Iterative Solver (default) is used. Method: Direct means that the Direct Solver is
used.
Converge: Two options are available: “None” means that the number of runs is
controlled by the settings in the linear tab. “Error=2%” means that the number of
runs will be controlled by the global error of the solution. When the estimated rela-
tive error in energy norm is less than or equal to 2%, the execution will stop. The
highest polynomial order cannot go beyond the limits entered in the linear tab.
Additional convergence criteria can be defined in the Settings tab of the Results
interface, as described in the User’s Guide.
Display: Two options are available:
- Sequence: The sequence switch is used to visually monitor the progress of the
solution on the mesh. The elements are removed from the screen as they are
included into the solver.
- Results: Engineering results may be monitored by using the layout switch to acti-
vate the Results layout. Select the desired Output controls in the Results class of
interest before activating the solution. Monitoring result functions is useful in the
case of large problems where sequential analysis executed stepwise is employed.
The monitored data provide information on the basis of which you can decide
whether to stop or continue the extension process. For small problems full p-exten-
sion should be performed.
Path to scratch directory: Prior to performing a solution, you may specify the
path to a scratch disk/directory where you wish to have your temporary files created
during the solution. The temporary files created during a finite element solution can
become quite large, so it is often desirable to have them created in a scratch disk
where there may be a large area (hundreds of megabytes) of free space to be used
only during the solution step. Files which are truly temporary are automatically
deleted by the program when they are no longer needed. Other files are left behind
because they may be used if you plan to restart a previous solution (downward p-
level or nonlinear analysis). If you do not plan to “restart”, then it is safe to remove
these special temporary files (.wrk).

Quality assessment and extraction procedures
1
Browse: The Browse button is used to identify a directory path where you wish
to have intermediate work files created. These files are disposable files.
Buttons: The Solve button will activate the currently specified solution. The
Check button will produce a report summarizing parameters, element distor-
tion, geometric and material properties, applied load, and the disk space
required to perform the currently specified solution.
Example To execute a linear analysis for the linkage arm problem (uniform p-extension,
p=1 to 8) enter the following information:
• Class Tab: Linear > Extension: Upward-p > p-limits: 1 to 8.
Next, choose the SOLVE! tab to get the solution. Complete the requested infor-
mation as shown below:
• Class Tab: SOLVE! > Execute: Initialize > Run Mode: Automatic >
Method: Iterative > Convergence: None > Display: Sequence > Solve
button. The status window will display the progress of the solution.

To perform post-processing operations you must select Edit > Results from the
Main Menu Bar, or select the View Results icon from the Main Toolbar. The
following describes the extraction procedures for Planar Elasticity.

To obtain a global error estimate, select the View Results icon from the Main
Toolbar, the Error dialog window will come out together with the Graph win-
dow. For the example problem, the interface settings and the graph of the error
estimate are shown in FIGURE 23.
By selecting the Table tab, the results are tabulated as shown in FIGURE 24.
The figure shows the degrees of freedom, the computed and extrapolated val-
ues of the potential energy, and the estimated relative error in energy norm.
The run number and the convergence rate can be displayed by checking the
corresponding check box at the top of each column as shown in FIGURE 24.

1
FIGURE 23 Relative error in energy norm for example

problem.
Small variations in the numerical results can be realized if the finite element mesh
is not created exactly as described.
Standard StressCheck functions

StressCheck is based on the displacement formulation of the finite element method.
Therefore the basic information generated by StressCheck is an approximation to
the displacement vector components. This approximation is characterized by the set
of standard shape functions and their coefficients. Thus, in the case of two-dimen-

1
FIGURE 24 Estimated relative error in energy norm. Tabular data.
sional elasticity, the displacement vector components on the kth quadrilateral

element are known in the form:
n n
(k) (k) (k) (k)
ux = ∑ a i N i (ξ,η) , uy = ∑ an + i Ni (ξ,η) , ξ ≤ 1, η ≤ 1.
i=1 i=1
where n is the number of shape functions. The value of n depends on the poly-
nomial degree p and whether the product or trunk space is used. N i (ξ,η)
(i=1,2,...,n) represent the standard shape functions defined in Ref. [3].
All user-specified functions are computed from the displacement vector com-
ponents. StressCheck computes a set of commonly used functions, such as
stresses, strains, etc. The available standard functions are listed in Table 1. In
addition, any combination of the Standard StressCheck functions can be com-
puted through user-specified formulas or through the use of the calculator.

1
Strains By definition, the normal strains are:
∂u x ∂u y ∂u z
εx = , εy = , εz =
∂x ∂y ∂z
and the shear strains are:
∂u x ∂u y ∂u y ∂u z ∂u z ∂u x
γ xy = + γ yz = + γ zx = + .
∂y ∂x ∂z ∂y ∂x ∂z
TABLE 1. Standard functions. Planar Elasticity.
Symbol Explanation and commonly used symbol

Ex Normal strain εx
Ey Normal strain εy
Ez Normal strain εz
Gxy Shear strain γxy

E1 Principal strain ε1
E2 Principal strain ε2
Eeq Equivalent strain εeq
Ux Displacement component in the x-direction ux
Uy Displacement component in the y-direction uy
Sx Normal stress σx
Sy Normal stress σy
Sz Normal stress σz
Txy Shear stress τxy
S1 Principal stress σ1
S2 Principal stress σ2

1
TABLE 1. Standard functions. Planar Elasticity.

Seq Equivalent stress σeq
Tmax Maximum shear stress τmax

Error Error indicator
Fmla Formula. Using this option, any mathematical expression
containing the standard functions can be computed for a
given solution.
Calc Calculator. Using this option, any mathematical expres-
sion containing standard functions can be computed for
any arbitrary combination of solutions.
The principal strains are the eigenvalues of the strain tensor. They are the strain
values ε which satisfy the condition
γ xy γ xz
ε x – ε ------- -------
2 2
γ xy γ yz
det ------- ε – ε ------- = 0.
2 y 2
γ xz γ yz
------- ------
- εz – ε
2 2
In two-dimensional problems, γyz = γxz = 0.
The three roots are the principal strains, denoted by ε1, ε2, ε3. In two-dimen-
sional problems the principal strains are ordered such that ε 1 ≥ ε 2 and
ε 3 = ε z . In three dimensions ε 1 ≥ ε 2 ≥ ε 3 . The normalized eigenvectors
are the unit vectors which define the directions of the principal strains.

1
The equivalent strain is related to the von Mises theory of yield. For linear elastic
isotropic materials
1 2 2 2
ε eq = ----------------------- ( ( ε 1 – ε 2 ) + ( ε 2 – ε 3 ) + ( ε 3 – ε 1 ) )
2
2(1 + ν)
for elasto-plastic materials
p 2 2 2 2
ε eq = --- ( ( ε 1 – ε 2 ) + ( ε 2 – ε 3 ) + ( ε 3 – ε 1 ) )
9
Stresses The sign convention for the stress tensor components is illustrated in FIGURE 25.
In two dimensions τxz = τzx = 0 and τyz = τzy = 0. The directional stresses, are
computed multiplying the material matrix [E] by the strain tensor (See Advanced
Topics Guide, Theoretical Background):
{σ} = [E]{ ε}
z
σz
τzx τzy σx
τyz
σy τxz σy
τyx
τxy y
σx
x
σz
FIGURE 25 Sign convention for stress components.

1
The principal stresses are the eigenvalues of the stress tensor. They are the
stress values σ which satisfy the condition
σx – σ τ xy τ xz
det τ xy σy – σ τ yz = 0.
τ xz τ yz σz – σ
The three roots of the equation above are the principal stresses, denoted by σ1,
σ2, σ3. In two-dimensional problems the principal stresses are ordered such
that σ ≥ σ and σ 3 = σ z . In three dimensions σ 1 ≥ σ 2 ≥ σ 3 .The corre-
1 normalized
sponding 2 eigenvectors are the unit vectors in the direction of the
principal stresses.
The equivalent stress σeq is by definition
1 2 2 2
σ eq = --- ( ( σ 1 – σ 2 ) + ( σ 2 – σ 3 ) + ( σ 3 – σ 1 ) )
2
σeq is related to the von Mises yield criterion.

The maximum shear stress is, by definition
σ1 – σ3
τ max = ---------------------
2
where σ1 is the largest and σ3 is the smallest principal stresses. τmax is related
to the Tresca yield criterion. In Planar Elasticity, the maximum shear stress is
computed from σ1 and σ2.
In the case of beams, the sign convention for member forces and rotations is
illustrated in FIGURE 26. The local ξ-axis is directed from node 1 to 2 and the
local η-axis is determined as the cross product of the global z-axis and the ξ-
axis.

1
Mz(1) Mz(2)
Na(1)
η
Na(2) θ(2)
1 2
ξ 1 2
θ(1)
y
V1(1) Forces V1(2) Rotations
x
z
FIGURE 26 Positive member forces and rotations for beam elements.

StressCheck provides convenient means for displaying and printing computed
information in graphical form. To obtain a contour plot or deformed configuration,
select the Plot tab from the Results window.
Example Deformed shape, and equivalent stress fringe plot for the linkage arm. To display
the deformed shape for run 8, complete the following information:
• Class Tab: Plot > Action: Select > Object: All Elements > Method: Selec-
tion > Input tab > Solution: SOL > Run: 8 > Plot: Solution > Shape:
Deform > Overlay toggle ON > Midsides: 10. Click on the Plot button
(FIGURE 27).
To plot the equivalent stress, Seq, on the undeformed shape, make the following
selection:
• Plot tab > Select > All Elements > Selection > Input tab > Solution: SOL >
Run: 8 > Plot: Solution > Contour: Fringe > Shape: Undef > Functions: Seq
> Midsides: 10 > Range min: 0, max: 400 > Intervals: 8. Click on Plot and
the fringes of the plotted function will appear in the display window (FIG-
URE 28).

1
FIGURE 27 Deformed shape for the example problem.
FIGURE 28 Equivalent stress fringe plot for the

example problem.

1

To compute minimal and maximal values of displacement, stress and strain data, in
the Results window select the Min/Max tab.
Example Maximum and minimum values of the directional stress σx. To compute the maxi-
mum and minimum values of the stress component σx (StressCheck name Sx, see
Table 1) for the linkage arm problem, complete the entries in the Results input area
as follows:
• Class Tab: Min/Max > Action: Select > Object: All Elements > Method:
Grid > Input tab > Solution: SOL > Run: 1 to 8 > Function: Sx > Midsides:
8 > Minimum and Maximum buttons ON. Click on Accept.
The convergence of the maximum and minimum values of Sx will be displayed as a

function the number of degrees of freedom as shown in FIGURE 29. The estimated
limits are also included. Refer to the User’s Guide for more details.
FIGURE 29 Convergence of the minimum and maximum

value of σx.

1
The location of the maximum and minimum values are in the Table window.
You will see that the coordinates of the location are x=100, y= -10 for the max-
imum and x=110, y =0 for the minimum, as it would be expected. Computing
stress resultants
The stress resultants are the force vector components (Fx, Fy) and the moment
about the z-axis (Mz) which are statically equivalent to the stresses acting
along an element edge or along a group of element edges. The stresses are
computed from finite element solutions. Specifically,
N2
Fx =
∫N T tds
1
x
N2
Fy =
∫ N T t ds
1
y
N2
Mz =
∫ ( ( x – xo )Ty – ( y – yo )Tx )t ds
N1
where N1 and N2 refer to the nodes (in counterclockwise order) which define
an element edge, Tx and Ty are the traction vectors computed from the stress
components and the direction cosines of the normal to the element edge (nx,
ny):
T x = σ x n x + τ xy n y
T y = σ y n y + τ xy n x
t is the thickness, ds represents a differential length along the element edge, xo,
yo are the coordinates of the point about which the moment is to be computed.
The default values are xo=yo=0.

1
In the case of fastener elements, the force resultants are simply the components of
the fastener load. In the case of beams, they are the projection of the axial and shear
forces in the direction of the global coordinate axes.
To compute stress resultants select Class: Resultant and the input area will appear
as shown in FIGURE 30. Enter the name of the solution for which you wish to
compute stress resultants and specify the desired run numbers. Enter the coordi-
nates (xo,yo) about which the moment is to be computed. Turn on the toggle
switches in front of the desired resultants then, using the cursor, click on the ele-
ment edge for which the resultant is to be computed.
FIGURE 30 Input area for Resultant/equilibrium check.
When performing equilibrium checks, click on the target element with the left
mouse button and then on the Accept button, the result will be displayed for all of
the selected outputs. In the summation fields, only the results corresponding to the
solution with the highest number of degrees of freedom (DOF) are displayed.

Model problem: Tension strip with a circular hole
1
When computing resultants for a sequence of solutions, the graphical output

will show the convergence path while the Input Area will show the numerical
values for the run with the highest number of DOF. If you click on the Cancel
button, the previously obtained results in the summation fields will be cleared.
You are then ready to select the next element edge. If you do not click on the
Cancel button then the values in the summation fields will be cumulative.
Example For the linkage arm shown in FIGURE 3, let us check that the resultant force in
the x-direction equilibrates the applied load of 104 N. Select:
• Class Tab: Resultant > Action: Select > Object: Edge > Method: Selec-
tion > Input tab > Solution: SOL > Run:8 to 8 > Forces: Fx and Fy,
Moments: Mz > Click on the left-most edge of element 1, and then
click on the Accept button. The resultant Fx force value will appear in
the text input area as shown in FIGURE 30. The value of Fx= -9992 is
very close to the applied load and Fy is very close to zero.
Problem description
A rectangular panel with a circular hole in the center (FIGURE 31) is loaded
by a constant traction Tx=σ0. It has unit thickness, a length to width ratio (L/
W) of 3 and Poisson’s ratio v=0.3. Assuming plane stress conditions, the goal
of the computation is to determine the nominal (Kt) and net (Kn) stress concen-
tration factors for (a/W) = 0.40.
By definition the cross-section stress concentration factor is:
σ max
K t = ------------ (1)
σ0
and the net-section stress concentration factor is
σ max ( W – a )
K n = ------------ ------------------ (2)
σ0 W

1
2a 2a 2W σο
2L
FIGURE 31 Tension strip with a central hole.
In many engineering problems the relationship between the maximum stress and
one or more design variables is of interest. This problem illustrates how Stress-
Check can be used for handling such problems very conveniently. We will exploit
symmetry and formulate the mathematical problem as shown in FIGURE 32.
D C
AB: un = Tt = 0.0 (symmetry)
BC: Tn = 1.0, Tt = 0.0

E W=1.0
CD: Tn = Tt = 0.0 (stress free)
a
DE: un = Tt = 0.0 (symmetry) y
x A B
L=3.0
FIGURE 32 The solution domain and boundary conditions.

1
Since the objective is to compute data which are normalized with respect to the
constant traction Tn = σο on boundary segment BC, it is convenient to select
σο = 1. Specification of units for σο is not important because the data of
interest, Kt and Kn are dimensionless.
Finite element discretization

FIGURE 33 shows the five quadrilateral element mesh used to perform the
computation. Elements 4 and 5 around the hole are included to account for the
stress gradient in that region.
Make sure the reference and theory selectors are set to Planar Elasticity and
select the Model Info icon. When the dialog window appears select the Param-
FIGURE 33 Finite element mesh.
eters tab and enter the following information:

• Name: a > Description: Hole radius > Value: 0.4 > Limit: >0.00 >
Class: General > Accept button.
Geometry and mesh To generate the geometry and mesh select the Create Model icon, then select
the Geometry tab and follow the steps indicated below:
• Create > Rectangle > Locate > Data tab > Input switch ON > X: 0.0, Y:
0.0, Z: 0.0, Width: 3.0, Height: 1.0, rot-Z:0.0 > Accept.
• Create > Circle > Locate > X: 0.0, Y: 0.0, Z: 0.0, Radius: a, P1-Min: 0,
P1-Max: 90 > Rot-Z: 0.0 > Accept.

1
• Create > Circle > Locate > X: 0.0, Y: 0.0, Z: 0.0, Radius: (1+a)/2, P1-Min:
0, P1-Max: 90 > Rot-z: 0.0 > Accept.
Now select the Mesh tab and provide the following information:
• Create > Node > Offset > Repeat # = 3 > Offset: 0 + 45. Use the mouse cur-
sor to select both circles.
Note: After completing the last step, turn the Repeat switch OFF.
• Create > Node > Intersection. Associate nodes to three corners of the rect-
angle by selecting the pair of lines at each intersection.
• Create > Node > Offset > Offset:=1/3. Use the mouse cursor to select the
lower side of the rectangle.
A node will be displayed on the lower side of the rectangle at L/3 from the origin.
Note the equal sign before 1/3: The equal sign placed in the first position of an
input field causes StressCheck to evaluate the expression that follows it (i.e., =1/3
is replaced by 3.333e-001). Any expression written in FORTRAN format will be
evaluated.
• Create > Node > Projection. Use the mouse cursor to click on the last node
created on the lower side of the rectangle. Then, while holding Control-
Shift, click on the upper side of the rectangle and choose Accept. A node
will be displayed on the upper side of the rectangle.
• Create > Quadrilateral > Selection. Define 5 elements by associating the
appropriate nodes (See FIGURE 33).
Thickness Assign unit thickness to all elements by selecting the Thickness tab and completing
the required information.
• Select > All Elements > Selection > Thickness:1.0 > Accept.
Material properties Assign the material properties by selecting the Material tab and complete the fol-
lowing information:
• Define tab > ID: prop > Material: Linear > Units: US > Type: Isotropic >
Case: Pl. Stress (Plane Stress) > E: 1000, v: 0.3 > Accept.
• Assign tab > Select > All Elements > ID: prop > Accept.
Load For the load application select the Load tab and complete the following informa-
tion:

1
• Select > Any Curve > Traction > ID: LOAD > Direction: Norm/Tan >
Normal: 1.0. Use the mouse cursor to select the right side of the rectan-
gle. Click on the Accept button (see FIGURE 34). Selecting a unit load
has the advantage that Kt is directly σ1max.
Constraint For the constraint application select the Constraint tab and assign zero normal
displacement along the boundaries where symmetry boundary conditions are
specified.
• Select > Any Curve > Symmetry > ID: CONST. Use the mouse cursor
to select the left side of the rectangle and then holding the Shift key
click on the lower side of it. Click on the Accept button (see FIGURE
34).
FIGURE 34 Specified boundary conditions.
Execution Before executing the problem, complete the Solution ID and the p-Discretiza-
tion (trunk space) inputs as described below:
• Solution ID tab > Solution ID: SOL > Constraint ID: CONST (or click
on item in listbox) > Load ID: LOAD (or click on item in listbox).
• p-Discretization tab > Action: Select > Object: All Elements >
Method: Selection > Space: Trunk > p-Discretization: Variable > p-
Level: 1. Click on the Accept button.
To execute the Linear Analysis, upward from p=1 to p=8 in automatic mode,
select the Compute Solution icon from the Main Toolbar and complete as fol-
lows:

1
• SOLVE! tab > Execute: Initialize > Run Mode: Automatic > Method: Itera-
tive > Converge: None > Display: Sequence > Solve button.
Results After the execution is completed we can extract results from the finite element solu-
tions. Select the View Results icon from the Main Toolbar to perform post-process-
ing operations.
Error Estimation
The converging sequence of finite element solutions for a/W ratio equal to 0.40
(FIGURE 35) can be obtained by selecting the Error tab from the Results window
and completing the following information:
• Error tab > Select > All Elements > Selection > Input tab > Solution: SOL
> Run: 1 to 8 > Click on the Accept button.
FIGURE 35 Estimated relative error in energy norm (a/W = 0.40)
Concentration factors
The nominal and net stress concentration factors for the p=8 solution are computed
by determining the maximum normal stress σmax=σx(0,a) and then using equations
(1) and (2) with σ0=1.0. Using the values of Sx=3.747 for p=8:
a
----- = 0.40, K = 3.747, K = 2.248
W t n

1
Note: The location and direction of the maximum normal stress, σmax, can be
extracted by selecting the Points tab from the Results window:
• Points tab > Select > Node > Selection > Input tab > Solution: SOL >
Run: 8 to 8 > Func(s): Sx. Use the mouse cursor to select the node at
the edge of the hole at the intersection with the vertical line. Click on
the Accept button.
To obtain the solution for other values of the hole radius, select the Model Info
icon from the Main Toolbar, select the Parameters tab to update the value of
“a”, and execute the linear analysis again.
Equilibrium check
Another assessment of the quality of the solution is obtained by verifying the

equilibrium condition of the elements in the finite element mesh. In particular,
the equilibrium check of the elements in the region of maximum stress is of
interest (elements 4 and 5). Select the Resultant tab and complete the following
information:
• Resultant tab > Select > Any Element > Selection > Input tab > Solu-
tion: SOL, Run: 1 to 8 > Fx: Make sure the button is on. > Fy: Make
sure the button is on. Then, move the cursor into the Model window
and point to the element of interest (say element 4). Click on Accept.
A plot of the resultants Fx and Fy will be displayed on the Graph 1 window.

Note in FIGURE 36 that both resultant force components converge to zero as
the number of degrees of freedom (DOF) increases. If you want to display the
plot of more than one element at the time, as shown in FIGURE 36, select Edit/
Clone from the main window and a duplicate of the Graph 1 will be created
labeled Graph 2. Next go to the input area and repeat the same procedure as
before with another element (say element 5). Graph 1 will now display the
information corresponding to the last element selected.
Remark The nominal and net stress concentration factors for all a/w values, compare
very well with the published data. In the manual “Stress Concentration Fac-
tors” by R. E. Peterson, the reported Kt and Kn are:
a-
---- = 0.40, K t = 3.75, K n = 2.24
W

Principal directions
1
FIGURE 36 Equilibrium check for elements 4 and 5.
Recommended exercise
Change the parameter to a=0.17. Compute Kt and Kn, compare the results with
published data. From R. E. Peterson for a/W=0.17, Kt=3.10 and Kn=2.57.
StressCheck makes it possible to visualize the principal direction vectors for the
principal stress and strains. The display of the principal directions is available for
triangular and quadrilateral elements with any material property assignment and for
linear and nonlinear analysis. The implementation also includes the display of the
principal directions for linear and nonlinear solutions in three-dimensions.

1
The principal direction vectors in two-dimension are computed from the eigen-
value problem for stresses or strains. In the case of the principal stresses, for
example, the eigenvalue problem can be written as:
σ x τ xy n
–σ 1 0 x = 0
τ yx σ y 0 1 ny
where σ is the eigenvalue and (nx, ny) are the direction cosines of the corre-
sponding eigenvector (components of the unit vector in the global coordinate
system). There are two eigenvalues associated with this equation (σ1, σ2) and
therefore there are two eigenvectors. In the implementation of the visualization
of the principal directions (those corresponding to σ1, σ2, ε1 and ε2), the eigen-
value problem is solved at each point of the plotting grid and the vectors are
displayed as double-headed arrows at each point. The control of the density of
points of the plotting grid and the size of the arrows is done under the Plot tab
of the Results interface.
Example
Consider the ladle hook shown in FIGURE 37, loaded by a sinusoidal traction
simulating the bearing load at the hook and pin locations. Isotropic material
properties were used (E=29x106psi, v=0.295) and the analysis was performed
by uniform upward p-extension for polynomial orders ranging from 1 to 8.
This problem is available in the Parts Handbook with the file name ladle-
hook.sci. The finite element mesh for the two-dimensional (plane-stress) anal-
ysis consists of 20 elements. The estimated relative error in energy norm is
under 1% for p=8 (1485 DOF).
The contour fringe of the first principal stress (S1) and the corresponding prin-
cipal direction vectors for p=8 are shown in FIGURE 38. The figure also
shows the settings of the plot interface and a detail around the highly stressed
region of the hook. To display the principal directions the option Shape must
be set to Princ. Direction as shown in FIGURE 38. Note that the Midsides
input field controls the density of the grid plot. In the example it was set to 5,
that means that a (5+1) × (5+1) grid is used over each quadrilateral element for
the contour plot and vector arrows. The Scale input allows controlling the size
of the arrows of the principal directions.

Using beam elements
1
FIGURE 37 Ladle hook. Error estimate.
FIGURE 39 shows the principal direction vectors corresponding to the first princi-
pal (S1) and second principal (S2) stresses over the model, without the color
fringes. Note that at every point, the directions of the two principal stresses are
mutually orthogonal.
The implementation allows the visualization of the principal direction vectors for
the planar elasticity problems (plane-stress, plane-strain and axisymmetric) for lin-
ear, material nonlinear and general nonlinear analyses. The same features are also
available for three-dimensional elasticity.
Using beam elements

In this section a simple example problem is worked out to present the beam capabil-
ities implemented in StressCheck. An example of the computation of natural fre-
quencies for beams is available in the Modal and Buckling Analyses Chapter.
Note: Beam elements cannot be used in conjunction with other element types.

Using beam elements
1
FIGURE 38 First principal stress for p=8. Contour and principal direction vectors.
Problem description
Consider the three-hinged semicircular arch acted upon by a transverse load as
shown in FIGURE 40. The arch has a mean radius of ro=100 inches, and the
distributed load is q=10 lb/in. The material properties are those of ASTM-A36
steel (E= 29x106 psi, ν= 0.295).
The cross section W 10x15 (AISC Database) is used. The mean radius is mea-
sured with respect to the centroid of the section. The objective of the analysis is
to find the axial load, shear force and bending moment diagrams for the arch.
The procedure to create a parametric model of the three-hinged arch is
described in the following.

Using beam elements
1
FIGURE 39 Principal direction vectors for S1 and S2.
hinge
q
na
d
θ
W 10x15
d= 10 in
A= 4.41 in2
I= 68.9 in4
2ro
FIGURE 40 Three-hinged semicircular arch acted upon by a transverse load.

Using beam elements
1
Geometry and mesh After opening a data base, and having set the analysis type to Planar Elasticity
and the units to in/lbf/sec/F, select the Model Info icon from the Main Toolbar
and then select the Parameters tab. Enter the following information:
• Name: q > Description: Distributed load > Value: 10 > Accept.
• Name: ro > Description: Arch radius > Value: 100 > Limit: >0.0 >
Accept.
Provide the geometric description of the beam by selecting the Create Model
icon and completing the following information:
• Geometry tab > Create > Circle > Locate > Input switch on > X: 0.0, Y:
0.0, Z: 0.0, Radius: ro, P1-Min: 0.0, P1-Max: 180.0 > Rot-Z: 0.0 >
Accept.
• Mesh tab > Create > Node > Offset > Repeat #= 5, Offset: 0.0 + 45.0 >
Click on the circular arc created in the previous step, and then turn off
the Repeat switch.
• Mesh tab > Create > Beam > Selection. Click on the first two nodes to
create the first beam element. A beam element is defined by selecting
two nodes in any order. After the fourth beam element has been cre-
ated, the finite element mesh should look as indicated in the FIGURE
41.
Note: The order in which the nodes are picked when creating the elements
define the principal directions of the beam. The beam axis is defined in the
direction from the first to the second node. The direction of the positive normal
to the beam axis is computed as the cross product between the positive z-axis
and the tangent to the beam axis. In this problem, element 1 was created by
connecting nodes 2 and 1, element 2 by connecting nodes 3 and 2, and so on.
Therefore the outward normal is positive for all the elements. If the elements
are not created following a consistent order in the node selection, the direction
of the positive normal to the beam AXIS will be different from element to ele-
ment.
Section properties To define the section properties select the Section Prop. tab from the Stress-
Check Input dialog window (FIGURE 42). Three options are available: User,
Library and Database. The option User means that the section properties will
be provided by the user. The required inputs are the cross sectional area (A),
the moment of inertia about the centroidal axis (I) and the third (M3) and
fourth (M4) moments of the area which are defined as follows:

Using beam elements
1
3 3 2
4 2
4 1
5 1
FIGURE 41 Finite element mesh for the beam problem.
3
dA
A =
∫ dA M3 =
∫ η dA
Area Area
η
O
2 4
I =
∫ η dA M4 =
∫ η dA
Area Area
M3 and M4 are needed for curved beams only and can be set to zero for straight
beams.
The option Library means that the section properties will be determined from a set
of predefined parametric sections available in StressCheck.

Using beam elements
1
The option Database means that the section properties will be obtained from
the AISC standard shape database as illustrated in this example. To assign a
section property from the AISC database select:
• Section Prop. tab > Select > All Elements > Selection > Database tab >
Section: W 10x15 > Accept.
As you click on any of the entries (W 10x15 in this example problem) in the
scrolling window of the dialog box (FIGURE 42), the corresponding area and
FIGURE 42 Section properties.
moment of inertia will be displayed in the Section Properties region. Note that
the 3rd and 4th moments also appear. These two moments are calculated by
StressCheck automatically.
Note that the Section Database box allows you to select the moment of inertia
so that it is consistent with the orientation of the beam (Ix or Iy). The default is
that the neutral axis coincides with the local x-axis of the section.

Using beam elements
1
Material properties To enter the material properties information select the Material tab in the Stress-
Check Input box and complete the following options:
• Define tab > Define > Linear > Selection > Click on the Browse button and
when the Material Library form appears, click on the steel ASTM-A36
record. When you select a material from the list, the name is automatically
entered in the input fields. > ID: mat > Material: Linear > Type: Isotropic >
• Assign tab > Select > All Elements > Selection > ID: mat > Accept.
The procedure for defining and assigning material properties for beam elements is
exactly the same as for any other element type. However, for beam elements only
linear isotropic materials can be assigned.
Load To enter the load data select the Load tab and provide the following information:
• Select > Edge > Traction > ID: LOAD > Direction: Norm./Tan. > Normal: -
q. With the Shift key depressed click on the two elements where the load is
to be applied and they will be highlighted, then click on the Accept button.
The load symbols will be displayed on those elements as shown in FIGURE 41.
Note that the object Edge was used to specify the traction load on a group of beam
elements. A traction applied to a beam element is understood to be in force per unit
length. The object Element can be loaded by body forces or temperature loading
only.
Constraint To enter constraint data select the Constraint tab and when the Constraint input box
appears provide the following information:
• Select > Node > Node > ID: CONST > Direction: XY > Data Type: Fixed >
System: Global > Turn on the switches in front of X and Y. Holding the
Shift key down click on two end nodes of the arch and then click on the
Accept button. A pinned support will be created on each node because the
nodal rotation was not fixed.
• Select > Node > Hinge. Click on the node located at the middle of the arch
and then click on any of the two elements attached to that node. The symbol
of a hinge (a circle with a dotted line pointing to the element selected with
the second mouse click) will be displayed as shown in FIGURE 41.
Note: When N-beam elements are attached to a node, the hinge has to be defined N-
1 times if all the elements are hinged. To do that, repeat the operation described
above selecting a different element each time.

Using beam elements
1
Execution The last step before executing the analysis is to create a solution record, that
identifies the load and constraint pair. Select the Solution ID tab, and when the
Solution ID dialog box appears provide the following information:
• Solution ID: SOL > Constraint ID: CONST > Load ID: LOAD. Click
Now you are ready to execute a linear analysis. Select the Compute Solution
icon from the Main Toolbar and complete as follows:
• Linear tab > Extension: Upward-p > p-limits:1 to 8
Iterative > Converge: None > Display: Sequence > Click on the Solve
button.
A sequence of eight finite element solutions will be obtained for polynomial

orders ranging from 1 to 8.
Post-processing After the execution is completed, select the View Results icon from the Main
Toolbar. Check the error in energy norm, and plot the deformed shape, the
bending moment, the shear force, and the axial force distribution over the arch.
Obtain the maximum value of the bending moment, shear force and axial force
and compute the support reactions.
Error Estimation
To compute the estimated relative error in energy norm, select the Error tab
from the Results window and complete the following information:
• Select > All Elements > Selection > Input tab > Solution: SOL > Run:
1 to 8 > Click on the Accept button.
The results will be displayed in tabular and in graphical form in the Graph win-
dow as indicated in FIGURE 43. Note that the error in energy norm is practi-
cally zero for p-levels greater than or equal to 5.
Deformed shape and Z-shape
To plot the deformed shape, select the Plot tab from the Results window and
complete the following information:
• Select > All Elements > Selection > Input tab > Solution: SOL > Run:
8 > Plot: Solution > Shape: Deform > Overlay toggle ON > Midsides:

Using beam elements
1
FIGURE 43 Relative error in energy norm for the beam problem.
6. Click on the Plot button to get the deformed and the undeformed shapes
together.
The result will be displayed on the main window as indicated in FIGURE 44.
It is possible to display a Z-shape plot over the mesh for the quantities which are
relevant for beams: Ux, Uy, Rz, Na, V1, Mz. To do that select Shape: Z- shape and
the desired function, and then click on Plot. FIGURE 45 shows the representation
of the bending moment over the arch.
• Select > All Elements > Selection > Input tab > Solution: SOL > Plot: Solu-
tion > Contour: Fringe > Shape: Z-shape > Overlay toggle OFF > Midsides:
10 > Functions: Mz > Click on the Plot button.
Because the z-shape is shown perpendicular to the xy-plane, the model has to be
rotated to see the representation. The Scale arrows can be used for controlling the
magnitude of the Z-shape.
Points Extraction
The points extraction option can be used for producing tabular and graph outputs of
the bending moment and axial and shear force distributions as described below.
Select the Points tab from the Results window.

Using beam elements
1
FIGURE 44 Deformed shape for the beam problem.
- Bending moment: When the Points Extraction dialog window appears, com-
plete the following information:
• Select > Any Boundary > Selection > Input tab > Solution: SOL >
Run: 8 - 8 > Func(s): Mz > # of pts: 18 > Click on the circle (Make
sure the Display Curves option in the Display Objects toolbar is on) >
The Bending Moment will be displayed in the Graph window as shown in FIG-
URE 46. Note that the independent variable in the graph was changed to the x-
coordinate which is measured from the center of the arch. To change the inde-
pendent variable, make sure that the graph windows has focus and then select
View > Indep. Variable from the main menu.
A closed-form solution for the bending moment can be obtained from the clas-
sical beam formulation. From Ref. [52], the bending moment is given by:
Mz= (qr2/2)*(cos θ + sin θ - 1), for θ = 0 to 90 deg.

Using beam elements
1
FIGURE 45 Bending moment, Mz. Z-shape representation.
FIGURE 46 Bending moment

distribution for the circular arch.

Using beam elements
1
Mz= - (qr2/2)*(cos θ - sin θ + 1), for θ = 0 to -90 deg.
where q, r and θ are as shown in FIGURE 40.
The maximum value of Mz can be obtained from the above expressions:

Mzmax= 0.207qr2 = 20700 lb.in at θ = π/4, which is very close to the maximum
value shown in FIGURE 46.
- Shear force: In the Points Extraction dialog window complete the following
information:
• Select > Any Boundary > Selection > Input tab > Solution: SOL >
Run: 8 - 8 > Func(s): V1 > # of pts: 19 > Click on the circle and then
The Shear Force will be displayed in the Graph window as shown in FIGURE
47. Again, the independent variable was changed to be the x-coordinate.
FIGURE 47 Shear force distribution

for the circular arch.
From Ref. [52], the shear force distribution obtained from the classical formu-
lation is:

Using beam elements
1
V1= - (qr/2)*(sin θ - cos θ), for θ = 0 to 90 deg.
V1= (qr/2)*(sin θ + cos θ), for θ = 0 to -90 deg.
and the maximum value is V1max= 500 lb at θ=0 deg.
- Axial force: In the Point Extraction dialog box complete the following informa-
tion:
• Select > Any Boundary > Selection > Input tab > Solution: SOL > Run: 8 -
8 > Func(s): Na > # of pts: 18 > Click on the circle and then click on the
Accept button.
The Axial Force will be displayed in the Graph window as shown in FIGURE 48.
Again, from Ref. [52], the axial force distribution obtained using the classical for-
mulation is:
Na= (qr/2)*(cos θ+ sin θ) - qr, for θ = 0 to 90 deg.
Na= - (qr/2)*(cos θ - sin θ), for θ = 0 to -90 deg.
and the maximum is Namax= -292.89 at θ= 45 deg, which coincides with the results
of StressCheck.
FIGURE 48 Axial force distribution

for the circular arch.

Using beam elements
1
Min/Max Extraction
This option can be used for determining the maximum value of the quantities
of interest. Select the Min/Max tab in the Results window. Complete the fol-
lowing information:
• Select > All Elements > Line > Input tab > Solution: SOL > Run: 1 to 8
> Function: Mz > Midsides: 15 > Maximum > Click on the Accept but-
ton.
The results will be displayed in the graph window as shown in FIGURE 49.
The estimated limit shown in the figure is very close to the maximum value
computed from the closed-form solution. Selecting the functions V1 and Na
FIGURE 49 Convergence of maximum bending moment.
from the Function box, the convergence plots shown in FIGURE 50 are
obtained for the shear and axial forces.

Using beam elements
1
FIGURE 50 Convergence of maximum shear and maximum axial force.
Reactions
To compute the support reactions, select the Resultant tab in the Results window.
Provide the following information:
• Check > Node > Selection > Solution: SOL > Run: 1 to 8 > Turn on the Fx
and Fy buttons in the Forces input area > Turn on the Mz button in the
Moments input area. Point to one of the end nodes of the arch and click the
left mouse button. The moment center is not relevant for beams.
The results will be displayed in the Table window for each run number as shown in
FIGURE 51 for the left support. The resultant values are provided in the global
directions (Fx, Fy, Mz).
This problem can be loaded from the Beam Handbook. Refer to the User’s Guide
on how to run a handbook problem in StressCheck.

Using beam elements
1
FIGURE 51 Reactions at left support of the circular arch.

2 Axisymmetric 2
Elasticity
There are many practical problems in engineering applications where the elastic body is axially sym-
metric, that is, it can be generated by rotating a plane figure about an axis, and furthermore, it is loaded
and constrained in such a way that symmetry of the solution is preserved. This chapter contains guide-
lines for the preparation of input data for axially symmetric problems in elasticity. For this type of prob-
lems there is no need to specify thickness. The computations are performed for a 1 radian sector.
Setting the analysis type

After opening a new database, from the Reference/Theory/Units selectors change
the default analysis to Axisymmetric Elasticity.
Analysis Guide Chapter 2: Axisymmetric Elasticity 71

Entering geometric and mesh data

2 Select Class: Geometry from the Main Menu Bar, or click on the Create Model
icon and select the Geometry tab. Note that instead of working with the X, Y,
and Z coordinates, you will be working with R and Z.
Consider the domain shown in FIGURE 52, which represents a bolt head with
a washer support, in tension.
a = 1.0
di = 1.0
Do = 2.5 di
Ew = 1.0e7 L
washer
hw = 0.5 Ew r
L = 2.0
f
hw
F = 2000
rf = 0.2
a
Do
FIGURE 52 Bolt head in tension (washer support).
The body can be generated by rotating a plane figure about the z-axis. There-
fore it can be modeled for analysis as shown in FIGURE 53.
Example Make sure you are using US units. To create the mesh shown in FIGURE 54,
consider first the geometric description:
• Geometry tab > Action: Create > Object: Rectangle > Method: Locate
> Data tab > Input toggle switch on > R: 0, Z: 0, Width: 0.5, Height: 3,
Rot-Z: 0 > Button: Accept.
72 Chapter 2: Axisymmetric Elasticity Analysis Guide

Z 2
S = 4F/π di2
FIGURE 53 Solution domain.
di/2 = 0.5
L = 2.0
rf = 0.2
Kn = Ew/hw
a = 1.0
R
Do/2 = 1.25
• Create > Rectangle > Locate > R: 0, Z: 0, Width: 1.25, Height: 1, Rot-Z: 0
> Accept
• Create > Line > Locate > R: 0.3, Z: 0.0, Length: 3.0, Angle: 90 > Accept
• Create > Fillet > Selection > Input switch on > Radius: 0.2. Using the
mouse, click on the upper line of the large rectangle and the right line of the
tall rectangle.
After these operations, the screen should look as shown in FIGURE 54a. Next, cre-
ate the nodes and elements as shown in FIGURE 54b. Most of the nodes can be cre-

ated by the intersection method. The nodes labeled A in FIGURE 54b were
created by projecting the nodes on the fillet to the corresponding lines, while
2 the node labeled B was created by projecting the node labelled X to the fillet.
Note: In axisymmetric analysis, the domain should be defined in such a way

that all nodes and elements of the mesh should be located at R ≥ 0 . If this con-
dition is violated, an error message will be issued.
(b) Mesh (a) Geometry
A A B
X A


2
The material properties are entered by the menu selection Class: Material, or by
selecting the Material tab in the StressCheck Input dialog window. The form differs
from its counterpart in the planar elasticity case only in that there is no need to
specify Plane Stress or Plane Strain.
For linear materials the options are isotropic and orthotropic. For isotropic materi-
als, the input coefficients are the same as for planar elasticity. For orthotropic mate-
rials, seven engineering constants must be provided: the modulii Err, Ett, Ezz, the
Poisson’s ratios vrt, vtz, vrz and the shear modulus Grz. Additionally, the coefficients
of thermal expansion αrr and αtt can be provided.
Example To define and assign the material properties select the following options:
• Material tab > Define tab > Action: Define > Object: Linear > Method:
Selection > ID: STEEL > Material: Linear > Type: Isotropic > Units: U.S. >
E: 2.9e7 > v: 0.295 > Accept
After the material properties are defined, you have to assign them to the elements.
For this purpose, provide the information indicated below:
• Assign tab > Action: Select > Object: All Elements > Method: Selection >
ID: STEEL > Accept.
Entering load data

To enter the load data select the Load tab in the StressCheck Input dialog window,
or Class: Load from the Main Menu Bar.
Types of loading
The types of loading available for axisymmetric analysis are the same as for planar
elasticity. The only difference is that in axisymmetric analysis, the counterpart of a
point load is a line load. Therefore, if a line load of magnitude w (force/length)
needs to be specified acting at a position ro, then the entry in the Point force field
should be the product of w times ro.

Entering load data
When defining a formula, in the R-Z plane the System Option in the Formula
form must be set to CARTESIAN. This is so because the generating section is
2 defined in the X-Y plane even though the analysis is performed in cylindrical
coordinates. For example, if a centrifugal force in the R-direction is to be spec-
ified, then the formula for the R-field should be:
n 2
ρ  2π ------ X
 60
where ρ is the mass density, n is the rotational speed in rpm, and X is the coor-
dinate that takes the role of R.
Example Proceeding with our example problem, enter the following data in the Load
input area:
• Load tab > Action: Select > Object: Any Curve > Method: Traction >
ID: LOAD > Direction: Norm./Tan. > Normal: =8000/pi > Select the
upper boundary of the model using the left mouse button > Accept.
Note that the equal (=) sign was entered before the expression 8000/pi. Press
the tab key to force the program to evaluate the expression. The load symbol
will be displayed on the mesh as shown in FIGURE 54b. The traction load is
simply the total load (F=2000) divided by the cross-sectional area of the bolt
(A= π di2/4).
Checking applied loads To check the magnitude of the applied loads follow the same procedure
described in the Planar Elasticity chapter.
In axisymmetric analysis, the resultant of the applied mechanical load is com-

puted as the integral of the applied traction over each edge of the selected ele-
ments and over a 1 radian sector.
Example For the case of the bolt head in tension, checking the applied mechanical loads
for the two elements where the load is applied, gives a force Fz=318.31, which
corresponds to the applied load F divided by 2π (F/2π=318.31).


2
To enter the constraint data select the Constraint tab in the StressCheck Input dialog
window or Class: Constraint from the Main Menu Bar. The types of constraints
available in axisymmetric analysis are the same as in planar elasticity. The only dif-
ference is that there is only one rigid body mode in axisymmetric analysis which is
a rigid body translation in the Z-direction.
Example To simulate the washer support shown in FIGURE 52 we use a spring constraint
with a spring constant given by the ratio between the modulus of elasticity and the
thickness of the washer:
Ew 10 7
K n = ------- = -------- = 2 × 10 7
hw 0.5
Complete the constraint input area as follows:

• Constraint tab > Action: Select > Object: Edge > Method: Spring Coeff. >
ID: CONST > Direction: Norm./Tan. > Normal: 2.0e7
With the left button of the mouse select the element edge where the spring is
applied and then click on Accept. The constraint symbols will be displayed on the
mesh as shown in FIGURE 54b.
Note: There is no need to constrain the Ur displacement component along the Z-

axis. StressCheck will impose Ur=0 for all element edges along R=0.

To assign a solution name to the desired constraint and load name pair, select
Class: Solution ID from the Main Menu Bar, or select the Solution ID tab from the
StressCheck Input dialog window.
Example To complete the solution record for this problem supply the following information:
• Solution ID tab > Solutions tab > Solution ID: SOL > Constraint ID:
CONST > Load ID: LOAD > Accept.


2 The discretization parameters and p-level can be specified independently over
each element, group of elements, or all elements in the mesh. They can be
fixed, bounded or variable. The polynomial degree can range from p=1 to 8.
The options under the p-Discretization tab are described in detail in the User’s
Guide. To define the parameters select Class: p-Discretization from the Main
Menu Bar or select the p-Discretization tab from the StressCheck dialog win-
dow.
Example Try product space for this problem:

Method: Selection > Space: Product > p-Discretization: Variable > p-
level: 1 > Accept

To execute a linear analysis select Edit > Solution from the Main Menu Bar, or
click on the Compute Solution icon from the Main Toolbar. The execution
options are the same as described for Planar Elasticity.
Example When the Solution dialog window appears, select the Linear tab and complete
the requested information:
Next, choose the SOLVE! tab to get the solution. Enter the following informa-
tion:
Iterative > Converge: None > Solve
This will produce a sequence of solutions for polynomial orders ranging from
1 to 8.


2
The procedures for quality assessment and extraction of engineering quantities are
basically the same as those described for Planar Elasticity, and are explained in the
following.

To obtain an error estimate, select the Error tab from the Results window, or select
Class: Error Estimation from the Main Menu Bar. Supply the following informa-
tion:
• Error tab > Input tab > Action: Select > Object: All Elements > Method:
Selection > Solution: SOL > Run: 1 to 8 > Estimate > Click on the Accept
button.
Example The estimated error in energy norm for the example problem is shown in FIGURE
55.
FIGURE 55 Relative error in energy norm.
Small variations in the results will occur if the finite element mesh is not created
exactly as described. This is because the finite element solution depends on the dis-
cretization.

2 The available standard functions in axisymmetric analysis are listed in Table 2.

In addition, any combination of the Standard StressCheck functions can be
computed through user-specified formulas or through the use of the calculator.
In Axisymmetric Elasticity, the displacement vector components ur(r, z) and

uz(r, z) are computed for each element. The strains are then computed as:
∂u r ∂u z ur ∂u r ∂u z
εr = , ε = , ε t = -----, γ rz = +
∂r z ∂z r ∂z ∂r
TABLE 2. Standard functions for Axisymmetric Elasticity

Er Normal strain εr in the radial direction.
Ez Normal strain εz in the Z-direction.

Et Normal strain εt in the circumferential direction.
Grz Shear strain γrz
E1 Principal strain ε1 in the rz-plane.
E2 Principal strain ε2 in the rz-plane.
Eeq Equivalent strain εeq
Ur Displacement component in the R-direction
Uz Displacement component in the Z-direction
Sr Normal stress σr
St Normal stress σt
Trz Shear stress τrz
S1 Principal stress σ1 in the rz-plane.

TABLE 2. Standard functions for Axisymmetric Elasticity
S2 Principal stress σ2 in the rz-plane. 2

Seq Equivalent stress σeq
Tmax Maximum shear stress τmax in the rz-plane.

Fmla Formula. Using this option, any mathematical expression containing
the standard functions can be computed for a given solution.
Calc Calculator. Using this option, any mathematical expression containing
standard functions can be computed for any arbitrary combination of
solutions.
The stress components are determined from the stress-strain relationships. For iso-
tropic materials for example, we have
σr 1–v v v 0 εr
σz v 1–v v 0 εz
E
= ------------------------------------- v v 1 – v 0
σt ( 1 + v ) ( 1 – 2v ) εt
1 – 2v
τ rz 0 0 0 --------------- γ
2 rz
The principal strains (ε1, ε2) and principal stresses (σ1, σ2) are computed in the r-z
plane as follows
εr + εz εr – εz 2 γ rz 2
ε 1, 2 = --------------- ±  --------------- +  ------
2  2   2
σr + σz σr – σz 2
σ 1, 2 = ----------------- ±  ----------------- + τ rz
2
2  2 

Note that the third principal strain (ε3) and stress (σ3) are:
2 ε3 = εt
σ3 = σt
The other quantities shown in Table 2 are as follows:
1 2 2 2
ε eq = ---------------------- × ( ( ε 1 – ε 2 ) + ( ε 2 – ε 3 ) + ( ε 3 – ε 1 ) )
2
2(1 + v)
1 2 2 2
σ eq = --- × ( ( σ 1 – σ 2 ) + ( σ 2 – σ 3 ) + ( σ 3 – σ 1 ) )
2
σ1 – σ2
τ max = ------------------
2

To display computed information in graphical form select the Plot tab from the
Results window.
Example To plot the deformed shape for the example problem, complete the following
information:
• Plot tab > Action: Select > Object: All Elements > Method: Selection >
Input tab > Solution: SOL > Run: 8 > Plot: Solution > Shape: Deform
> Overlay toggle ON > Midsides: 10 > Click on the Plot button.
The deformed shape for run number 8 of the example problem is shown in
FIGURE 56.

To compute minimal and maximal values of displacements, strains or stresses,
select Min/Max tab in the Results window.

FIGURE 56 Deformed shape.
Example FIGURE 57 shows the convergence of the maximum value of Sz as a function of

the number of degrees of freedom (DOF) for the example problem. To obtain the
results shown in the figure, select:
• Min/Max tab > Select > All Elements > Line > Input tab > Solution: SOL >
Run: 1 to 8 > Function: Sz > Midsides: 12 > Maximum > Accept.
Computing data in points

To extract results along element edges, boundary segments, points, etc., select the
Points tab from the Results window.
Example Complete the required information to compute the vertical displacement along the
edge with washer support:
• Points tab > Select > Edge > Selection > Input tab > Solution: SOL > Run:
8 to 8 > Func(s): Uz > # of pts.: 10. Select the element edge where spring
constraint was specified and click on the Accept button.

FIGURE 57 Convergence of the maximum

value of the stress component Sz.
FIGURE 58 shows the Uz displacement distribution for p=8 along the edge
with spring constraint. Note that the independent variable in the graph is the
coordinate R. The default independent variable is N (the point number). To
change the independent variable, make sure that the graph windows has focus
and then select View > Indep. Variable from the main menu.
FIGURE 58 Displacement along edge

with spring constraint.

Computing stress resultants

For Axisymmetric Elasticity, the stress resultants are computed for a 1 radian sec-
2
tor. Referring to FIGURE 59 ,
FIGURE 59 Stress resultants for axisymmetric analysis.
the directional resultants Fr and Fz, the hoop force (Ft), the projection of the hoop
force (Fh) and moment (Mh), and the moment of the directional resultants (Mo)
about a point (ro,zo) are given by:

2 1 N2 1 N2
--- ---
2 2
Fr =
∫ 1
– ---
cos θ dθ
∫
N1
T r r ds Fz =
∫ 1
– ---
dθ ∫ N1
T z r ds
2 2
θ
F h = – 2F t sin  ---
Ft =
∫∫ σ t dr dz  2
Ω
∫ ∫ σ zdrdz
Mh = Mt θ
Mt = t
Ω
N2
Mo =
∫ N1
( T z ( r – r o ) – T r ( z – z o ) )r ds
where Tr and Tz refer to the traction vectors computed from the stress compo-
nents and the direction cosines of the normal to the element edge (nr, nz).
T r = σ r n r + τ rz n z
T z = τ rz n r + σ z n z
Note: When computing the equilibrium of an axisymmetric element, the sum

of the radial force resultants should be equal in magnitude and opposite in sign
to the projection of the hoop force:
∑F r = –Fh
Similarly the sum of the moment resultant should be equal in magnitude and
opposite in sign to the projected hoop moment:
∑ Mo = –Mh

To compute stress resultants select the Resultant tab from the Results window and
the input area shown in FIGURE 60 will appear.
2
The extraction option depends on the object selected. Select Edge to compute the
integral of the tractions along an element edge. Select Element to compute the con-
tour integral of the traction over the element.
Enter the name of the solution for which you wish to compute stress resultants and
specify the desired run numbers. Enter the coordinates (ro,zo) about which the
moment is to be computed. Turn on the toggle switches in front of the desired
resultants then using the cursor click on the element edge for which the resultant is
to be computed.
If you are computing the resultants for a sequence of solutions then the graphical
display window will show the corresponding convergence path(s). The input area
will show the numerical values for the run with the highest number of degrees of
freedom only.
If you use the Action: Select, click on all the objects (edges or elements) you want
to compute the resultants and then click on the Accept button. If you use the
Action: Check, then the computation is performed the moment you click on the
edge or element. If you click on the Cancel button, the previously obtained results
in the input area will be cleared. If you select the next element edge and do not click
on Cancel then the values in the input area will be cumulative.
Note that the resultants are computed on the element-by-element basis. StressCheck
will use that element to which you point. Therefore you can check how well the
finite element solution approximates the action-reaction principle in an average
sense. Click Cancel, select the Action: Check, then point to an interelement edge
with the cursor first in one element then in a neighboring element. It may be easier
to simply shrink the elements to facilitate the edge selection. Ideally, the result
should be zero. Usually there is a remainder, an error, owing to the approximate
nature of the finite element solution.
Example For the example problem, compute the resultant force in the Z-direction, Fz, along
the boundary of the mesh where the load is applied.
• Resultant tab > Select > Edge Curve > Selection > Input tab > Solution:
SOL > Run:1 to 8 > Turn on the toggle switch in front of Fz. > Using the
cursor, click on the line along which the load is applied > Click on the
Accept button.

The small window at the right side of the Fz resultant force label will show the
force along the Z-axis. In this case F z = F ⁄ ( 2π ) = 318.3 , as shown in FIG-
2 URE 60.
FIGURE 60 Resultants input area for the example problem.

Model problem: Floyd’s vessel

2
Problem description
A flat-headed, thick-walled axisymmetric vessel is loaded by uniform internal pres-
sure of 2.61 psi. Due to symmetry only half of the generating section is shown in
FIGURE 61. It is assumed that the material is homogeneous and isotropic, the mod-
ulus of elasticity is E = 1435 psi and Poisson’s ratio is v = 0.49. Compute the loca-
tion, orientation and magnitude of the largest principal stress.
EA, CD: u n = T t = 0 Note: All dimensions are given in inches.

AB, BC: T n = T t = 0 (stress free)
DE: Tn = -2.61 psi, Tt = 0 0.6
z D C
0.15
E 3.15
1.5
A B r
6.0
FIGURE 61 Floyd’s vessel: Solution domain and boundary conditions.

The finite element mesh, consisting of 17 elements, is shown in FIGURE 62. In
designing this mesh certain knowledge about the exact solution was taken into con-

sideration. The exact solution is not analytic in the end points of the circular
fillet, and in point B. It is known that if the fillet radius were zero then there
2 would be a strong corner singularity in the point r = 5.4, z = 1.5 and a geomet-
ric mesh, that is a mesh graded in geometric progression with a common factor
of 0.15, would be used in conjunction with p-extension. If, on the other hand,
the solution were very smooth (for example the fillet radius would be large)
then a nearly uniform mesh would be best. The mesh design shown in FIGURE
62 is based on the expectation that, because the fillet radius is small, the solu-
tion is likely to be closer to the singular case than to the smooth case.

Geometry and mesh To create the finite element mesh, click on the Create Model icon from the Main
Toolbar and then select the Geometry tab.
2
1. Create two rectangles representing the outline of the vessel.
• Geometry tab > Action: Create > Object: Rectangle > Method: Locate >
Data tab > Input switch on > R: 0, Z: 0, Width: 6, Height: 1.5, Rot-Z: 0 >
• Create > Rectangle > Locate > R: 5.4, Z: 0, Width: 0.6, Height: 3.15, Rot-
Z: 0 > Accept
2. Create a fillet of radius 0.15 and two additional circular arcs of radius 0.25 and
0.40, respectively.
• Create > Fillet > Selection > Radius: 0.15. With the mouse click on the
upper line of the large rectangle and the left line of the small rectangle.
• Select > Fillet > Selection. Turn the Input toggle switch off and click on the
fillet just created. The location of its center and its radius is deposited in the
input area.
• Create > Circle > Locate > Turn the Input toggle switch back on. The cir-
cles to be created are located at the same r, z coordinates of the fillet. Turn
on the Repeat switch and enter 2, then Radius: 0.25+0.15, P1-Min: 270 + 0,
P1-Max: 360 + 0 > Accept.
3. Define auxiliary lines for mesh construction.

• Create > Point > Local > Turn Input toggle switch on > R=0, Z=0. Click on
the local system of the circles and choose Accept.
• Create > Line > Point > Length: 1, Angle: 0. Click on the last point created.
• Create > Line > Point > Length: 1, Angle: -25. Click on the same point.
• Create > Line > Point > Length: 2, Angle: -90. Click on the same point.
• Create > Line > 2-Point: Create a line by clicking on the point in the local
system and the point located at the lower right corner of the vessel (point B,
FIGURE 61).
• Create > Line > Locate > R: 5.4, Z: 2.05, Length: 1, Angle: 0 > Accept
• Create > Line > Locate > R: 4.85, Z: 1.50, Length: 2, Angle: -90 > Accept
4. Define all nodes.

• Mesh tab > Action: Create > Object: Node > Method: Locate > Data tab >
Input toggle on > R: 5.10, Z: 1.50. Click on the Accept button.
• Create > Node > Locate > R: 5.40, Z: 1.80 > Accept

• Create > Node > Intersection. Define the remaining nodes, using the
intersection of boundaries procedure.
2
5. Define all the elements of the mesh.
• Mesh tab > Create > Quadrilateral > Selection. Define fifteen quadri-
lateral elements by associating the elements with the appropriate nodes
as shown in FIGURE 62.
• Create > Triangle > Selection. Define two triangular elements by asso-
ciating the elements with the appropriate nodes as shown in FIGURE
62.
Material Properties Assign the material properties to all elements by selecting the Material tab in
the StressCheck Input dialog window and completing the following informa-
tion:
• Material tab > Define tab > Define > Action: Define > Object: Linear >
Method: Selection > ID: PROP > Material: Linear > Type: Isotropic >
Units: U.S. > E: 1435 (Araldite), v: 0.49 (nearly incompressible) >
Button: Accept. At this point you will receive a warning: “E matrix for
property PROP has a condition number greater than 20”. Ignore the
warning. For explanation, please refer to the User’s Guide.
To assign the material to the elements, select:

• Assign tab > Select > All Elements > Selection > ID: PROP > Accept.
Load Select the Load tab in the StressCheck Input window and assign the 2.61psi
internal pressure to the DE boundary by supplying the following information:
• Load tab > Action: Select > Object: Edge Curve > Method: Traction >
ID: LOAD > Direction: Norm./Tan. > Normal: -2.61. Using the mouse,
click on any of the element edges where the pressure is applied, and
then click on the Accept button. If you have difficulty selecting an ele-
ment edge, turn off curve display and try again.
Constraint Select the Constraint tab in the StressCheck Input window. Assign zero normal
displacement to the boundary segment DC (see FIGURE 63):
• Constraint tab > Action: Select > Object: Edge Curve > Method: Gen-
eral > ID: CONST > Direction: Norm./Tan. > Data Type: Fixed > Turn
on the toggle in front of Normal, and make sure that the toggle in front
of Tangent is turned off. With the mouse click on the boundary seg-
ment DC, and then select Accept.

There is no need to constrain the normal displacement along AE. The program will
impose Ur=0 for all element edges along R=0.
2
FIGURE 63 Schematic representation of the specified boundary conditions.
Execution Complete the Solution ID and the p-Discretization inputs. Execute the Linear Anal-
ysis upward from p=1 to p=8 in automatic mode.
CONST > Load ID: LOAD. Click on the Accept button.
• p-Discretization tab > Action: Select > Object: All Elements > Method:
Selection > Space: trunk > p-Discretization: Variable > p-Level: 1 > Accept
Select the Compute Solution icon from the Main Toolbar and when the dialog win-
dow appears complete as follows:
• Linear tab > Extension: Upward-p > p-limits: 1 to 8
tive > Converge: None. Click on the Solve button.
Post-processing After the execution is completed extract results from the finite element solution by
selecting the View Results icon from the Main Toolbar.

Error estimation
2 The converging sequence of finite element solutions for the linear analysis is
displayed in FIGURE 64. To obtain the error estimate select the Error tab and
supply the following information:
• Error tab > Input tab > Select > All Elements > Selection > Solution:
SOL > Run: 1 to 8 > Estimate > Accept.
FIGURE 64 Estimated error in energy norm. Trunk space.
Largest principal stress
To obtain the location (r, z), and magnitude (σ1) of the largest principal stress,
select the Min/Max tab from the Results window and provide the following
information:
• Min/Max tab > Input tab > Select > All Elements > Grid > Solution:
SOL > Run: 1 to 8 > Function: S1 > Midsides: 12 > Maximum toggle
ON > Accept.
The results displayed in tabular form in the Graph window show that the loca-
tion of the principal stress σ1 is on the boundary of the circular fillet, FIGURE
65.

FIGURE 65 Value and location of maximum σ1 for each run.
FIGURE 66 shows the convergence of σ1 as it appears in the Graph window when

the Chart option is selected. The first principal stress exhibits strong and obvious
convergence with respect to increasing the number of degrees of freedom (DOF).
Remark This problem has been studied by many investigators using an experimental (photo-
elastic) method, the boundary element method and the h- and p-versions of the
finite element method. The mesh shown in FIGURE 62, is the same as the one used
in Ref. [3], page 346. A challenging aspect of this problem is that Poisson’s ratio is
close to its limiting value (0.5). The material Araldite was used to make photoelas-
tic measurements possible.
Recommended exercises
1. Construct the mesh in such a way that the fillet radius can be changed by a single
command.
2. Plot the equivalent stress distribution for this problem.
3. Repeat the analysis using the product space and downward p-extension for p=1
to 8.

FIGURE 66 Maximum principal stress convergence graph.

3 Plate Bending
StressCheck is the first finite element analysis software product to introduce into professional practice
hierarchic modeling capabilities for structural plates made of isotropic or composite materials. The
important practical benefits are that control of the errors of idealization is made possible and the same
finite element mesh can be used for both the structural and strength analyses of plates. For a discussion
and summary of the theoretical background refer to the Advanced Topics Guide.
The implementation is limited to flat plates contained in the xy-plane and loaded normal to that plane
(in the z-direction) by distributed transverse loading, shear or moments.
The input data preparation and extraction procedures are very similar to those described for Planar Elas-
ticity and reviewed in the following by solving an example problem.

After opening a new database, from the Reference and Theory selectors change the
default analysis to Plate Elasticity.
Analysis Guide Chapter 3: Plate Bending 97


Select the Create Model icon from the Main Toolbar to create the geometry and
the finite element mesh. Consider, for example, the domain shown in FIGURE
67, which represents a uniformly loaded square plate with a central hole and all
edges simply supported. Due to symmetry, only one fourth of the domain is
considered for analysis.
3
r = 0.25
2.0
Y
t = 0.1
X
2.0
FIGURE 67 Square plate with a central hole.
Example To create the mesh shown in FIGURE 68, consider first the geometric descrip-
tion:
• Geometry tab > Create > Rectangle > Locate > Input switch on > X:
0.0, Y: 0.0, Z: 0.0, Width: 1.0, Height: 1.0, Rot-Z: 0.0 > Click on the
Accept button.
From the Views Toolbar select the Center Model icon.

• Create > Circle > Locate > Input switch on > X: 1.0, Y: 1.0, Z: 0.0,
Radius: 0.25, P1-Min: 180, P1-Max: 270, Rot-Z: 0.0 > Accept
98 Chapter 3: Plate Bending Analysis Guide

Assigning thickness
• Create > Circle > Locate > X: 1.0, Y: 1.0, Z: 0.0, Radius: 0.35, P1-Min:
180, P1-Max: 270, Rot-Z: 0.0 > Accept
• Create > Line > Locate > X: 0.0, Y: 0.1, Z: 0.0, Length: 1.0, Angle: 0.0 >
Accept
• Create > Line > Locate > X: 0.1, Y: 0.0, Z: 0.0, Length: 1.0, Angle: 90 >
Accept
Next construct the mesh shown in FIGURE 68: 3

• Mesh tab > Create > Node > Method: Intersection. Create a node at each
intersection. A total of 12 nodes should be created using this method. You
can create a node at every intersection by selecting Accept instead of select-
ing pair of curves on the screen. The program will request a confirmation
that you want to create a node at every curve intersection.
• Create > Node > Mid-Offset > Click on 2 circle nodes. A midside node
should result on the arc segment. Repeat for the other circle.
• Create > Quadrilateral > Selection, then point and click to four nodes which
define an element in any order. You can use the marquee pick to draw a box
around four nodes in a single operation. To do that, simply click on the
screen near one of the target nodes, then while holding down the left mouse
button, drag the mouse and draw a box around four nodes and then release
the left mouse button.
The resulting mesh should look like the one indicated in FIGURE 68.
Assigning thickness
For Plate Bending it is necessary to associate a thickness with the elements. To
assign a thickness select the Thickness tab in the StressCheck Input dialog window.
For plate bending, the elements must be assigned constant or parametric thickness.
Example Assign a uniform thickness of 0.1 to our example problem.

• Thickness tab > Select > All Elements > Selection > Thickness: 0.1 > Click
You can display the thickness by turning on the toggle switch in front of ID. To
check for consistency in the thickness input data for all elements in the mesh use

FIGURE 68 Finite element mesh for plate with a central hole.
Action: Check and click on the Accept button. The result of the check will be
displayed in an “Edit Window”.

The material properties are entered by selecting the Material tab in the Stress-
Check Input window. For plate analysis the material can be homogeneous or
laminated.
Homogeneous materials The types of homogeneous materials currently supported are isotropic and
orthotropic. If the material is homogeneous and isotropic then a shear correc-
tion factor is also required for plate models 1 and 2. The options for the shear
correction factor are:
Unity: The shear stiffness is not modified by a shear factor. The shear
factor is unity.

Energy: The shear factor is computed internally based on the Poisson’s ratio
of the material to minimize the error in the strain energy of the solution with
respect to the fully three-dimensional solution.
Deflect: The shear factor is computed internally based on the Poisson’s ratio
to minimize the error in the mid-plane deflection of the plate with respect to
the fully three-dimensional solution.
For further details about the use of shear correction factors refer to the Advanced 3
Topics Guide.
Laminated materials If the material is laminated (more than one lamina through the thickness), the
options are Laminate-Isotropic or Laminate-Orthotropic. If you select Type:
Lamin.Ortho, the Material Properties interface will look as shown in FIGURE 69.
FIGURE 69 Material input area for the Laminate Orthotropic type.

At this point you can define the material properties of an individual lamina,
specify the material name, the material coefficients and the thickness of the
lamina.
After the material properties have been defined, you must create a ply group
and a stacking sequence. For this purpose select the Assign tab, and choose
Type: Laminate-Stack. The Laminated Material Definition dialog window will
appear on the screen as shown in FIGURE 70.
3
FIGURE 70 Laminate plate properties.
Ply Group: Select the Ply tab in the Laminated Material dialog window and
define the ply groups that can be formed using the lamina properties. Specify

the ply group name, the lamina material name and the ply layout angles (the angle
in degrees for the orientation of the material 1-axis with respect to the global x-axis
for orthotropic properties) and click Accept.
Stack Sequence: Select the Stack tab in the Laminated Material dialog window and
define the combination of ply groups that defines the laminated plate. Specify the
stacking sequence name, the number of layers and the stacking sequence and click
on Accept. For plates only symmetric stacks are supported by StressCheck. The
stacking sequence name should be entered in the ID field of the input area to assign 3
laminated properties to the elements.
Any combination of isotropic and orthotropic layers is allowed. Therefore, compos-

ite and sandwich plates can be defined using these forms.
Note: Layer angles have to be specified beginning with the bottom layer. See exam-
ple.
Example Assignment of laminated material properties.
For the problem shown in FIGURE 68, let us consider the 8-ply orthotropic sym-
metric laminate (90/0/90/0)s shown in FIGURE 71, where each lamina is a 0.0125
inches thick graphite/epoxy composite material with the following properties:
EL = 25.0x106 ET = 1.0x106
GLT = 0.5x106 GTT = 0.2x106
vLT = 0.25 vTT = 0.49
where the subscript L indicates the direction parallel to the fibers, T is the trans-
verse direction, and vLT = -εTT/ εLL is the Poisson’s ratio. When the L-direction
coincides with the global x-axis, we refer to it as a 0o orientation. A laminate desig-
nation (90/0/90) for instance, means a 3-ply laminate with the central layer oriented
with the L direction parallel to the global x-axis, and the outer layers with the L
direction at 90o with the global x-axis.
The following information should be provided in Action: Define of the Material

input area (see FIGURE 69):
ID: GRAPHITE (any name can be used)

Type: Lamin.Ortho

E11: 25e6, E22: 1e6, E33: 1e6

v12: 0.25, v23: 0.49, v13: 0.25
G12: 0.5e6, G23: 0.2e6, G31: 0.5e6
Density: 1.5e-4 (needed for modal analysis only)
a11: 0.0, a22: 0.0, a33: 0.0, Thick: 0.0125
Button: Accept
3
Z
90 0.0125
0
90
0.1
0
X
90
0
90
FIGURE 71 8-ply orthotropic symmetric laminate (90/0/90/0)S.
The following information should be provided in Action: Assign of the Mate-

rial input area:
Type: Laminate-Stack
The dialog window for the Laminated Material Definition should contain the
information given below.
1. Ply tab (see FIGURE 70)
Ply Group Name: CROSSPLY (any name can be used)

Entering load data
Lamina Material Name: GRAPHITE (same name given in the material defi-
nition ID)
Ply layout angles: 90 ; 0 (creates a ply group of two layers at 90o and 0o ori-
entation with respect to the global x-axis. Note that the first layer is a 90
degree ply.)
Buttons: Accept
2. Stack tab (see FIGURE 70) 3

Stack Sequence Name: LAM (any name can be used)
Number of Layers: 8 (total number of layers in the laminate)
Symmetry: Symmetric (default)
Stacking sequence: 2*CROSSPLY (same name given in Form PlyGroup)
Buttons: Accept
Note: Because of symmetry, the stacking sequence is defined for half of the lami-
nate (4 layers) only.
After entering the ply group and stacking sequence information, enter the stack
sequence name (LAM) in the ID field of the Assign Material input area, select All
Elements, system: Global and click on Accept. This last operation will assign the
stacking sequence properties to all elements in the mesh (see FIGURE 70).
Note: When laminated properties are assigned, the program checks that the number
of layers times the thickness of each layer equals the value entered for the thickness
in the Thickness input form. If they don’t match, a warning message is issued at
execution time.
Entering load data

In the Input dialog window, select the Load tab and the input area will appear as
shown in FIGURE 72. For Plate Bending the loading methods and sign conventions
are illustrated in FIGURE 73 and described in the following:
Distributed load: (q) means that a distributed pressure (in force per unit area) is
imposed on a face or a group of faces. Distributed loads can be specified as con-
stant, parametric or formula. A positive distributed pressure is applied in the posi-
tive Z-direction.

Entering load data
FIGURE 72 Input area for load input.
Edge moment: (Mn, Mnt) means that a distributed moment (in force-length per
unit length) is applied along a boundary or element edge. The distributed
moment is a vector quantity with bending and twisting components. Therefore
two vector components must be specified in the normal-tangent reference
frame. A positive bending moment (Mn) is a vector in the direction of the neg-
ative tangent to the boundary (clockwise). A positive twisting moment (Mnt) is
a vector in the direction of the positive normal to the boundary (outward).
Moment components can be specified as constants, parameters or by formulae.
Edge shear: (Qn) means that a distributed shear force (in force per unit length)
is applied along a boundary or element edge. A positive shear force is a vector
in the direction of the negative Z-axis. The distributed shear can be specified as
a constant, parameter or by a formula.

Entering load data
Mid-surface
Fz
* t
3
z
Mn n
Qn
Mnt
y
x
FIGURE 73 Sign convention for plate loads. Positive quantities are indicated.
Point load: (Fz) means that a concentrated force is applied in the z-direction at any
location on the plate. To access this option select the object point or node. Refer to
the Advanced Topics Guide for additional information about the use of concen-
trated forces. Points forces can be applied anywhere within an element (including
the nodes).
Note: Edge moment and shear are intended to be used on external boundaries
only. However, they can be applied to internal boundaries or element edges, using
the following interpretation:
a) If the load is applied to a boundary segment (line, circle, ellipse, etc.) all the
element edges associated with the selected boundary will be loaded. This
means that the elements at both sides of the boundary will be loaded (the
effective load doubles). Use the check option to verify the applied load.
b) If the load is applied directly to element edges, only the selected elements
will be loaded, and the prescribed load will be applied.
Example Specifying a uniform distributed load.

Consider once more the problem shown in FIGURE 68 and apply a uniformly
distributed pressure in the negative Z-direction of magnitude q=100. Enter the
following information in the load input area:
• Load tab > Select > Face Surface > Distributed > ID: LOAD > System:
Global > Transverse: -100. Select any element face in the mesh and all
the faces which are slope continuous will be selected. Click on Accept.
See FIGURE 72 and FIGURE 75.
3 Note: The type of loading that can be applied is controlled by the Object selec-
tion. When Edge or Any Curve are active, the methods are Moment and Shear.
If Face is active, then the only method is Distributed Load.

In the Input dialog window, select the Constraint tab and the input area will
appear as shown in FIGURE 74. For Plate Bending, the constraint methods are
as follows:
Soft-simple support: means that the transverse displacement component will

be set to zero (Uz= 0) along the selected element edges or boundaries. The
adjective “soft” refers to the fact that no constraints are imposed on twisting
(i.e., the twisting moment is zero).
Built-in support: means that all three displacement components will be set to
zero (Ux= Uy= Uz= 0) along the selected element edges or boundaries. The
rotation component of each displacement is also set to zero.
Symmetry condition: means that the displacement component normal to the

edge or boundary will be set to zero (Un= 0). By definition, this condition is
applicable only to straight edges or boundaries. The symmetry condition can
be viewed as a guided support on straight boundary segments.
Antisymmetric condition: means that the displacement component tangent to

the edge or boundary and the transverse displacement will be set to zero (Ut=
Uz= 0). By definition, this condition is applicable only to straight edges or
boundaries. The antisymmetric condition is the same as a hard-simple support
on straight boundary segments.

FIGURE 74 Constraint input area for Plate Bending.
Rigid body constraint: means that the transverse displacement components of

three non-colinear nodes will be set to zero. When Node > Rigid Body is selected in
the input area, use the cursor to click on any three non-colinear nodes in the display
area, and the program will automatically set Uz= 0 for all of them.
Nodal constraint: means that the transverse displacement components of the

selected nodes will be set to zero.
Note 1: The type of constraint that can be applied is controlled by the Object selec-
tion. If the object edge or boundary is active, then the methods are soft-simple,
built-in, symmetry, and Antisymmetry. If the object is node, then the methods are
rigid body and single nodal constraints.
Remark: Point constraints are admissible only for preventing rigid body displace-
ment and rotation. The body must be in equilibrium under the action of the applied

loads. Point constraints should not be used in StressCheck under conditions

other than self equilibrating loading.
Note 2: Edge constraints are intended to be used on external boundaries only.

However, they can be applied to internal boundaries or element edges, using
the following interpretation:
a) If the constraint is applied to a boundary segment (line, circle, ellipse,

etc.) all the element edges associated with the selected boundary will be
3
constrained. If there are no edges attached to the selected boundaries,
then no constraints will be applied to the elements.
b) If the constraint is applied to element edges, both shared edges must be

selected (use the “shrink” option from the Display Control box to do that)
otherwise the execution will fail.
Example Specifying soft-simple support and symmetry conditions. Referring to FIG-

URE 68, let us apply symmetry conditions to the upper and right boundaries of
the plate, and soft-simple support at the other two boundaries. Enter the fol-
lowing information in the constraint input area:
• Constraint tab > Select > Any Curve > Symmetry > ID: CONST. With
the mouse click on the upper boundary and then holding the Shift key
click on the right boundary > Click on the Accept button. The symbols
of symmetric condition (circles) will be displayed along the bound-
aries.
• Constraint tab > Select > Any Curve > Soft Simple > ID: CONST.
With the mouse click on the left and lower boundaries and then click
on Accept. The symbols for soft-simple support (triangles) will be dis-
played along the boundaries.
FIGURE 75 shows the complete mesh with loading and constraint symbols.

To assign a solution name to the desired constraint and load name pair, select
the Solution ID tab from the Input dialog window and the interface will appear
as shown in FIGURE 76. Complete the required information as follows:

FIGURE 75 Boundary conditions for example problem.
• Solution ID tab > Solutions tab > Solution ID: SOL (Any name that is
meaningful to you) > Constraint ID: CONST > Load ID: LOAD > Accept.

To define the polynomial and model orders, select Class: p-Discretization tab from
the StressCheck dialog window. The input area will appear as shown in FIGURE
77.
The discretization parameters, p-level and model order, can be specified indepen-
dently over each element, group of elements, or all elements in the mesh. They can
be fixed, bounded or variable. The polynomial degree can range from p=1 to 8 and
the model order can range from 1 to 6 for plates of homogeneous materials, and
from 1 to 5 for plates of laminated materials. Refer to Theoretical Background for
details about plate model classification.

FIGURE 76 Input area for Solution ID.
If the p-discretization is specified as variable or bounded, the plate model can

be specified as fixed only. Model extension (plate model variable or bounded)
is supported for fixed p-level only.
Example Assignment of p-discretization parameters. Let us assign the product space and
plate model 1 to all the elements of the problem shown in FIGURE 68. Enter
the following information in the p-Discretization input area:
Method: Selection > Space: Product > p-Discretization: Variable >
Level:1 > Plate Model: Fixed > Level:1 > Click on the Accept button.
The p-discretization record will be included in the scrolling list as
shown in FIGURE 77.

FIGURE 77 Input area for p-Discretization.

To execute a linear analysis select the Compute Solution icon from the Main Tool-
bar. When the Solution dialog window appears (FIGURE 78) select the Linear tab
and complete the requested information.
The following extension options are available for Plate Bending:
Upward-p: The plate model is fixed and the polynomial degree is increased from
the lower to the upper limit.

FIGURE 78 Linear analysis interface for Plate Bending.
Downward-p: The plate model is fixed and the polynomial degree is

decreased from the upper to the lower limit.
Upward-model: The p-level is fixed and the model order is increased from the
lower to the upper limit.
Uniform-h: The p-level and plate model are fixed and the mesh is uniformly
refined up to the run limit.
Adaptive-p: The plate model is fixed and the polynomial order will be deter-
mined for each element and for each run based on an error indicator until con-
vergence is realized. The convergence options are described below.

or, in the case or upward model, from the initial plate model to the final model
order.
“Restart” is used when a sequence of solutions has been computed and you
wish to compute higher order solutions, retaining the computed data.

Run Mode: You may execute an “Automatic” sequence of solutions for the exten-
sion and limits selected. Or you may execute “Stepwise”, requiring you to click on
Solve following the solution for each p-level, m-level, or h-refinement.
Method: It may be either “Iterative” or “Direct”. Method: Iterative means that the
Iterative Solver (default) is used. Method: Direct means that the Direct Solver is
used.
Converge: Two options are available: “None” means that the number of runs is 3
controlled by the settings in the linear tab. “Error=2%” means that the number of
runs will be controlled by the global error of the solution. When the estimated rela-
tive error in energy norm is less than or equal to 2%, the execution will stop. The
highest polynomial order cannot go beyond the limits entered in the linear tab.
Additional convergence criteria can be defined in the Settings tab of the Results
interface, as described in the User’s Guide.
- Results: Engineering results may be monitored by using the layout switch to acti-
vate the Results layout. Select the desired Output controls in the Results class of
interest before activating the solution. Monitoring result functions is useful in the
case of large problems where sequential analysis executed stepwise is employed.
The monitored data provide information on the basis of which you can decide
whether to stop or continue the extension process. For small problems full p-exten-
sion should be performed.
become quite large, so it if often desirable to have them created in a scratch area
deleted by the program when they are no longer needed. Others files are left behind
these special temporary files (.wrk).

Browse: The Browse button is used to identify a directory path where you wish
to have intermediate work files created. These files are disposable files.
Buttons: The Solve button will activate the currently specified solution. The
Check button will produce a report summarizing parameters, elements distor-
tion, geometric and material properties, applied load, and the disk space
required to perform the currently specified solution.
3 Example Linear analysis for Plate Bending. Let us execute a linear analysis for the plate
problem considering an upward p-extension with fixed model. Enter the fol-
lowing information:
• Linear tab > Extension: Upward-p > p-limits: 1 to 8 > Model: 1.
Next, choose the SOLVE! tab to get the solution. Complete the following
information:
The top portion of the linear status box displays the status of the execution. The
message bar located on the main window displays which operation is being
performed for each run.

To perform post-processing operations select the View Results icon from the
Main Toolbar. The Results dialog window is shown in FIGURE 79.

To obtain an error estimate, select the Error tab from the Results window.
There are two interpretations of the relative error in energy norm for Plate
Bending:
a) If the plate model is fixed and a p-extension or h-extension is performed

with at least three runs in a sequence, then the estimated error in energy norm
represents a measure of the difference between the exact solution of the
selected plate model and the corresponding finite element solution. Typically, a

FIGURE 79 Results window.
sequence of solutions is computed as p ranges from 1 to 8 with fixed mesh and

model order.
b) If the p-level and mesh are fixed and the model order is increased in a sequence,
then the estimated error in energy norm is a measure of the difference between the
exact solution of the corresponding three-dimensional problem and the exact solu-
tion corresponding to each plate model. The finite element solution is only an
approximation to the exact solution for each plate model. Therefore the estimated
error in energy norm will represent a reasonable measure of the modeling error only
if the discretization error for each model is small. Typically, a sequence of solutions
is computed as the model order ranges from 1 to 6 (homogeneous plates) or from 1
to 5 (laminated plates) with a p-level fixed at a value for which the discretization
error is small (less than 1%). Before running a sequence of models it is recom-
mended to run a p-extension with the model order fixed at a low number (typically
1 or 2) to select the p-level for which the error of discretization is small. A descrip-
tion of the plate models implemented in StressCheck, and the strategy for model
selection is found in the Advanced Guide.

Note: Model 1 should be excluded when estimating the modeling error. This is
because Model 1, the Reissner-Mindlin model, is not a member of the hierar-
chic family of plate models but rather an extension of the hierarchic family to
plates of small thickness.
Example Estimated error in energy norm. Let us check the overall quality of the solution
of the example problem. After a sequence of linear solutions were obtained as
described in the previous section (Upward-p, p=1 to 8, model 1), select:
3 • Error tab > Select > All Elements > Selection > Solution: SOL > Run:
1 to 8 > Estimate > Click on Accept. The results are shown in FIGURE
80.
FIGURE 80 Relative error in energy norm for example problem.
Note: For the mesh and p-discretization parameters selected, the estimated
error in energy norm is less than 1% for run number 8 (1408 DOF). The mesh
for this problem was designed to minimize the effect of the boundary layer in
the energy of the solution. If the smallest error in energy norm is not accept-
able, it will be necessary to modify the finite element mesh in order to increase
the number of degrees of freedom (we are already using the product space).
Refer to the Advanced Guide, for recommended mesh design for Plate Bend-
ing.


The available standard functions in Plate Bending are listed in Table 3. The defini-
tion of strains and stresses are available in the Planar Elasticity chapter.
The “in-plane” stress components (σx, σy, τxy) are computed directly from the
finite element solution; that is, the strains are computed by differentiating the dis-
placement field, and the stresses by multiplying the strains with the material stiff-
ness matrix. The transverse stresses are computed by integration of the three- 3
dimensional equilibrium equations. This post-processing procedure for computing
the transverse shear (τxz, τyz) and transverse normal (σz) stresses has been tested
and the results published in Ref. [22].
TABLE 3. Standard functions for Plate Bending.

Ex Normal strain εx
Ey Normal strain εy
Ez Normal strain εz
Gxy Shear strain γxy
Gxz Shear strain γxz
Gyz Shear strain γyz
Ux Displacement component in the x-direction ux
Uy Displacement component in the y-direction uy
Uz Displacement component in the z-direction uz
Mx Bending moment Mx
My Bending moment My
Mxy Twisting moment Mxy
Sx Normal stress σx
Sy Normal stress σy
S1 First principal stress in the plane of the plate
S2 Second principal stress in the plane of the plate
S3 Normal stress σz
Seq von Mises (equivalent) stress computed using all six stress components

TABLE 3. Standard functions for Plate Bending.

Txy Shear stress τxy
Txz Shear stress τxz
Tyz Shear stress τyz
Qx Shear force Qx
3 Qy Shear force Qy
Tmax Maximum shear stress in the plane of the plate

Fmla Formula. Using this option, any mathematical expression containing the
standard functions can be computed for a given solution.
Calc Calculator. Using this option, any mathematical expression containing
standard functions can be computed for any arbitrary combination of
solutions.
The stress resultants are defined as follows:
The bending moments are:
h⁄ 2 h⁄ 2
Mx = 2 ∫ σx z dz , My = 2 ∫ σ y z dz (1)
0 0
where h is the thickness of the plate. The sign convention for bending moments
is shown in FIGURE 81(b). The double arrow indicates the sense of a right-
handed screw advancing in the direction of the arrows.
The twisting moments are defined as:
h⁄ 2
M xy = – 2 ∫ τx y z dz , M yx = – M xy (2)
0
The sign convention for twisting moments is shown in FIGURE 81(c).

The shear forces are defined as:
h⁄ 2 h⁄ 2
Qx = –2 ∫ τx z dz , Qy = –2 ∫ τ yz dz (3)
0 0
3
The sign convention for the shear forces is shown in FIGURE 81(a).
Qx
Qy
q(x,y)
Qy
Qx
(a)
Mxy
Mx
∩
∩
Myx
∩
My
∩
Myx
∩
My
∩
∩
Mx
Mxy
(b) (c)
FIGURE 81 Sign convention for stress resultants.

Plotting the deformed shape

To display computed information in graphical form select the Plot tab from the
Results window and the input area will appear as indicated in FIGURE 82.
Note that a Z-Plane coordinate is required for Plate Bending (the default value
is Z=0.0). The Z-Plane information is used to evaluate displacements, strain
and stress components, but ignored for stress resultants.
3 To display the deformed shape for run 8, complete the following information:
• Plot tab > Select > All Elements > Selection > Solution: SOL > Run: 8
> Plot: Solution > Shape: Deform > Overlay ON > Midsides: 10 > Z-
Plane: 0.0 > Click on the Plot button.
FIGURE 82 Deformed shape for example problem.
When plotting the deformed shape, the default Z-plane (Z=0) is recommended.
To visualize the deformed shape you may need to rotate the model.

Computing data in points

To extract results along element edges, boundary segments, points, or through-
thickness, select the Points tab from the Results window and the input area will
appear as indicated in FIGURE 83.
The possibility of selecting a through-thickness extraction is only applicable to

plates. The XY Thickness method can be used to display displacement, strain or
stress distributions in the thickness direction at any location specified by its global 3
coordinates (in the fields marked XYZ in the input area). The requested functions
are computed at several Z-locations starting Z=0.0 up to Z=h/2 (h = total plate
thickness). The number of points through the thickness depends on the entry in the
“# of pts” field of the input area and on the type of plate. For laminated plates, the #
of pts. is the number of midsides points (not including the interface between lami-
nae) for each layer. For homogeneous plates, the # of pts. is for the number of mid-
side points for the total plate thickness.
Example Through-thickness normal and shear stresses. To compute the normal stress σy at
the edge of the hole, complete as follows:
• Points tab > Action: Select > Object: Node > Method: XY Thickness >
Solution: SOL > Run: 8 to 8 > Function: Sy > # of pts.: 3 > Click on the
node at the edge of the hole (X: 0.75, Y: 1.0) > Accept
Selecting # of pts: 3 for the 8-ply laminate, the computation of the selected function
will be performed at 5 equally spaced points for each layer (3 midsides plus one at
each interface) and from Z=0.00 to Z=0.05 (that is, 4 layers). FIGURE 83 shows
the σy stress distribution in the Z-direction for run #8 at X=0.75, Y=1.00 (the edge
of the hole), while FIGURE 84 shows the shear stress distribution (τyz) for run #8 at
X=1.0, Y=0.0 (intersection of vertical symmetry line and support edge).
Note: When the method XY Thickness is selected, the z-coordinate in the input
field is suppressed. When computing displacement, strain or stress components at
points for a method other than XY Thickness, you must enter the z location in the z-
plane input field (the default value is z=0.0). If the z-coordinate is less than zero or
greater than half the plate thickness, an error message will be displayed by the pro-
gram. The proper range is 0 ≤ Z ≤ h ⁄ 2 . The Z-coordinate is irrelevant for the com-
putation of stress resultants.

FIGURE 83 Point extraction for example problem: σy(0.75, 1, z).
FIGURE 84 Shear stress distribution τyz(1,0, z)

for the example problem.


To compute minimal and maximal values of displacements, strains, stresses, or
stress resultants, select the Min/Max tab in the Results window and the input area
will appear as indicated in FIGURE 85. Follow the steps described below to com-
plete the input.
Example FIGURE 85 shows the entries in the input area and the results of the computation of 3
the minimum value of the transverse displacement component (Uz at Z=0.0) as a
function of the run number, for the example problem. Note that the load was
applied in the negative Z-direction, and therefore the largest deflection is negative.
The computation of minimal and/or maximal values is performed at the specified

Z-plane. The default value is Z=0.0. If the values of strain or stress components are
of interest, a Z-plane other than the default should be given because for Plate Bend-
ing all the strains and stresses are zero at the middle surface. Supply the following
input to obtain the graph shown in FIGURE 85:
FIGURE 85 Min/Max extraction for Plate Bending.

• Min/Max tab > Select > All Elements > Line > Solution: SOL > Runs:
1 to 8 > Function: Uz > Midsides: 10 > Minimum ON > Z-plane: 0.0 >
Click on Accept.
The results show the convergence of the transverse displacement (Uz) as a

function of the number of degrees of freedom (DOF).
Computing stress resultants

3
To compute stress resultants along element edges, select the Resultant tab in
the Results window and the input area will appear as shown in FIGURE 86.
The edge resultants available for plates (Vn, Mn, Mt) are defined as follows:
N2
Vn = ∫ ( Qx nx + Qy ny ) ds
N1
N2
Mn = ∫ ( M x n 2x + M y n 2y + 2M xy n x n y ) ds
N1
N2
Mt = ∫ [ M xy ( n 2x – n 2y ) – ( M x – M y ) n x n y ] ds
N1
where N1 and N2 refer to nodes (in counterclockwise order) which define an

element edge, Qx and Qy are the shear forces defined in equation (3), Mx and
My are the bending moments defined in equation (1), Mxy is the twisting
moment defined in equation (2), and nx, ny are the direction cosines of the nor-
mal to the element edge.

FIGURE 86 Input area for resultant extraction.
Example To compute the stress resultants for the plate problem of the example, complete as
follows:
• Resultant tab > Select > Edge Curve > Selection > Solution: SOL > Runs: 8
to 8 > Turn on the switches in front of Vn, Mn and Mt, select any edge
along the symmetry line at x=1.0 and all the edges which are slope continu-
ous will be highlighted > Click on Accept.
FIGURE 86 shows the computed resultants for the edges of the three elements
along the vertical symmetry line of the example problem. Note that the shear result-
ant (Vn) and the tangent moment resultant (twisting moment Mt) displayed in the
summation fields of the Resultant input area are very close to zero along the sym-
metry line (as expected). The normal moment resultant (Mn) is, of course, different
from zero.

Model problem: Built-in square plate
Problem description
A clamped square plate with uniform transverse unit load is shown in FIGURE
87. It is a 2a by 2a plate with a thickness of 0.1 inch and a=1.0 inches. The
material is isotropic, the modulus of elasticity is E = 30x106 psi and Poisson’s
3 ratio is v = 0.30. No shear correction factor is considered. Due to symmetry
conditions, only one quarter of the plate needs to be analyzed. Of interest is the
computation of the bending moment and shear distribution along the symmetry
lines, and the maximum deflection of the plate.
A C
AB, BD, DC, CA: u x = u y = u z = 0

2a
B D
2a
FIGURE 87 Solution domain and boundary conditions.

Boundary layer effects occur at the plate boundaries, and are characterized by
the fact that the solution “near” the boundary is substantially different from the
solution in the interior. All hierarchic plate models exhibit boundary layers,
and an important part of the energy of the solution is contained in them. There-
fore, the mesh design necessary to obtain accurate solutions for any given
member of the hierarchic sequence of models should properly account for the
boundary layers.

The first step is to design a finite element mesh that provides optimal rate of con-
vergence for the exact solution of the plate problem in the interior of the domain
without consideration for the boundaries. For smooth problems and p-convergence
this typically involves the use of uniform or quasi-uniform meshes. This will be
referred as the “coarse mesh”. For this case, the coarse mesh consists of one ele-
ment of size 1 unit. Once the coarse mesh is available, the boundary layers should
be accounted for by the use of the graded meshes. The relevant element size is the
dimension in the direction perpendicular to the boundary. For most practical prob-
lems one or two layers of graded elements toward the edges are sufficient to 3
account for the boundary layers effects. FIGURE 88 shows the finite element mesh
with one layer of graded elements used for this problem.
A typical value for the grading is r = 0.1 x a, where a is the size of the element of
the coarse mesh near the boundary. Refer to the Advanced Guide for additional
details about meshing in the presence of boundary layers.

Quarter model
ro
a
Geometry and mesh Because the boundary layer can significantly affect the solution, parametric grading
is recommended. In this way, the sensitivity of the solution to the choice of the
graded mesh size can be analyzed. To create the parameters for this problem, select

the Model Info icon from the Main Toolbar and when the dialog window
appears select the Parameters tab. Complete the following information:
• Name: a > Description: Half plate length > Value: 1.0 > Limit: > 0.0 >
Class: General.
• Name: ro > Description: Grading parameter > Value: 0.1 > Limit: > 0.0
> Class: General.
Click on the Accept button to save the parameters information given above.
3
Provide the geometric description of the plate, creating a square. From the
Main Toolbar select the Create Model icon and then select the Geometry tab
from the Input dialog window. Complete the following information:
• Geometry tab > Create > Rectangle > Locate > Input: X: 0.0, Y: 0.0, Z:
0.0, Width: a, Height: a, Rot-Z: 0.0 > Accept. From the Views Toolbar
select Center Model.
Define two lines for the boundary layer elements:

• Create > Line > Locate > X: 0.0, Y: ro, Z: 0.0, Length: a, Angle: 0.0 >
Accept.
• Create > Line > Locate > X: ro, Y: 0.0, Z: 0.0, Length: a, Angle: 90 >
Accept.
Define the position of the nodes at the intersection between lines (9 nodes).
• Mesh tab > Create > Node > Intersection. Click on the boundary seg-
ments near the intersection points where a node is to be located or click
on Accept.
Define the elements of the mesh.

• Create > Quadrilateral > Selection. Create four quadrilateral elements
by associating the elements with the appropriate nodes as shown in
FIGURE 88.
Thickness To assign thickness to the elements, click on the Thickness tab in the Stress-
Check Input dialog window. Select the following options:
• Thickness tab > Select > All Elements > Selection > Thickness: 0.1 >
Accept.
Material properties Assign the material properties to all elements by selecting the Material tab in
the StressCheck Input dialog window. Complete the following information:

• Material tab > Define tab > ID: prop > Units: U.S. > Material: Linear >
Type: Isotropic > Shear: Unity > E: 3.0e7, v: 0.3 > Accept.
• Material tab > Assign tab > Select > All Elements > Selection > ID: prop >
Type: Homogeneous > Accept.
Load Select the Load tab from the StressCheck Input dialog window, and complete the
following information:
• Load tab > Select > Face Surface > Distributed > ID: LOAD > Direction: 3
XYZ > Transverse: -1.0 > Click on any element in the mesh and all the
faces which are slope continuous will be highlighted. Click on the Accept
button.
Note: To have a better view of the load, rotate the model. The transverse load sym-
bols will be pointing in the negative Z-direction.
Constraint Select the Constraint tab from the StressCheck Input dialog window. Assign zero
normal displacement to the symmetry boundary segments and built-in condition to
the other two boundary segments (FIGURE 89):
• Constraint tab > Select > Any Curve > Symmetry > ID: CONST > With the
mouse click on the top line and then holding the Shift key click on the right
line > Click on Accept.
• Select > Any Curve > Built-In > ID: CONST > Select the bottom and left
lines and click on Accept.
Execution Complete the solution record and define the polynomial order, as indicated below:
CONST > Load ID: LOAD > Accept.
• p-Discretization tab > Select > All Elements > Space: Trunk > p-Discretiza-
tion: Variable > level: 1 > Plate Model: Fixed > level: 1 > Accept.
Now you are ready to start the computation. Select the Compute Solution icon from
the Main Toolbar and when the Solver dialog window appears select the Linear tab
and complete the following information:
• Linear tab > Extension: Upward-p > p-limits: 1 to 8 > Model: 1.
Next, choose the SOLVE! tab to get the solution. Complete the following informa-
tion:

FIGURE 89 Schematic representation of the constraints.
Results
You are ready now to extract results from the finite element solutions. Select
the View Results icon from the Main Toolbar.
Error Estimation Select the Error tab from the Results window and supply the following infor-
mation:
• Error tab > Select > All Elements > Selection > Solution: SOL > Run:
1 to 8 > Estimate > Click on Accept.

The converging sequence of finite element solutions for the linear analysis is dis-
played in FIGURE 90.
FIGURE 90 Estimated error in energy norm. Trunk space.
Point Extraction Bending and shear distributions: To get the bending moment (My) and the shear
(Qx) distribution along the top symmetry line, select the Points tab from the Results
window. Supply the following information:
• Points tab > Select > Edge Curve > Selection > Solution: SOL > Run: 8 to 8
> Function: My > # of pts.: 12. Using the mouse point to the top line and
then click on Accept.
The bending moment distribution is shown in FIGURE 91a. To obtain the shear
distribution along the same edge, select the function Qx and the shear distribution
will be as shown in FIGURE 91b. To display the moment and shear distribution as
a function of X, select View > Indep. Variable from the Main Menu Bar and when
the box appears check the toggle in front of X.
Min/Max Extraction To find the minimum deflection of the plate, select the Min/Max tab in the Results
window. Provide the following information:

(a) Bending moment distribution. (b) Shear distribution.
FIGURE 91 Bending moment and shear distributions.
• Min/Max tab > Select > All Elements >Line > Solution: SOL > Run: 1
to 8 > Function: Uz > Midsides: 12 > Minimum > Z-plane: 0.0 >
Accept.
The minimum deflection as a function of the number of degrees of freedom

(DOF) will be displayed in the Graph window. The magnitude of the minimum
Uz for run #8 is Uz = -7.6709e-06 at (x=1.0, y=1.0, z=0.0).
Remarks This example demonstrates that the good approximation properties of the p-
version, well established for two-dimensional and axisymmetric analyses are
also realized for Plate Bending problems.
The solution of this problem for the Kirchhoff plate model can be found in
Theory of Plates and Shells, by S. Timoshenko and S. Woinowsky-Krieger,
McGraw-Hill Book Company, Inc., 1959, page 197. At the center of the plate:
Uz = -7.338x10-6, Mx = My = -0.0924.
The solution for Model 1 (Reissner-Mindlin model) when no shear factor was
specified, was found to be: Uz = -7.671x10-6, Mx = My = -0.0919.

The difference in the central deflection is 4.5% while the difference in the bending
moment is only 0.5%. These differences are due to differences in the models. The
Kirchhoff model does not account for shear deformation while the Reissner-Mind-
lin model does. The aspect ratio of the plate is a/h = 2/0.1 = 20, and therefore the
shear deformation effect are small, but nevertheless measurable.
3
1. Referring to FIGURE 88, double the value of ro (ro=0.2) and run the problem
again. Observe that now the estimated error in energy norm at p=8 is only 0.01%,
but the values of Uz and the bending moment at the center of the plate are practi-
cally the same as before. (Uz = -7.67x10-6, Mx = -0.0919).
2. For ro=0.10, run a sequence of models for a fixed p-level. Select Linear Analysis
and then complete the following:
• Linear tab > Extension: Upward-Model > Model: 1 to 6 > p-level: 8.
tive > Convergence: None > Click on the Solve button.
Obtain the distribution of My along the top symmetry line as before. Note that the
results for models 2 to 6 give different value of the bending moment My at the
clamped edge. The boundary effect is not properly accounted for by plate model 1.
The value of My at the edge for models 2 to 6 is about 1.5 times that of model 1
(FIGURE 92).
FIGURE 92 Results for Upward-Model: My distribution along top symmetry line.


4 Three-Dimensional
Elasticity
The reader should be thoroughly familiar with the use of StressCheck for solving planar elastostatic
problems before attempting three-dimensional problems. Two options are available in StressCheck for
solving three-dimensional problems: 3D and Extrusion. The 3D option is the traditional approach in
which the model is created in 3D-space. In the Extrusion option the model is created in 2D-space and
then converted into a 3D problem by extrusion. These two options are discussed in the following.

After opening a new database, from the Reference/Theory/Units selectors make
sure the analysis is set to 3D Elasticity so that the proper input area will be available
for data entry. Many of the steps involved in the 3D model creation are similar to
those of Planar Elasticity.
Analysis Guide Chapter 4: Three-Dimensional Elasticity 137


Select Class: Geometry from the Main Menu Bar, or click on the Create Model
icon and select the Geometry tab. To illustrate the main features of geometric
data entry in 3D an example is worked out below.
Problem description
Consider the notched cantilever plate in bending shown in FIGURE 93. The
problem is characterized by seven parameters: The plate length (L), width (W),
and thickness (th), the support width (b) and height (h), the fillet radius (ra)
and the applied load per unit length (Ps). The material is plain carbon steel
Ps
th
ra W
L b
FIGURE 93 Notched plate configuration.
with a modulus of elasticity E=200 GPa and Poisson’s ratio is v=0.3. The
objective of the analysis is to compute the maximum von Mises stress for the
default value of the parameters. The steps to create a parametric solid model,
the finite element mesh, define and assign material properties, loading, con-
straints, and execution parameters for this problem are described in the follow-
ing.
138 Chapter 4: Three-Dimensional Elasticity Analysis Guide

Geometry construction First, using the unit selector set the units to mm/N/sec/C. To create the parameters
for this problem, select the Model Info icon from the Main Toolbar and when the
dialog window appears select the Parameters tab. Complete the following informa-
tion:
• Name: b > Description: Support width (b>th/2) > Value: 40.0 > Limit: >0.0
> Class: General.
• Name: h > Description: Support height (h>th+2ro) > Value: 60.0 > Limit:
>0.0 > Class: General.
• Name: L > Description: Plate length (L>ro) > Value: 100.0 > Limit: >0.0 >
Class: General.
• Name: W > Description: Plate width > Value: 100.0 > Limit: >0.0 > Class:
General.
• Name: th > Description: Plate thickness > Value: 20.0 > Limit: >0.0 > 4
Class: General.
• Name: ra > Description: Fillet radius > Value: 6.0 > Limit: >0.0 > Class:
General.
• Name: Ps > Description: Line load in N/mm > Value: 6.0 > Class: General
After entering all the parameters click on the Accept button.
The Description field is optional when defining parameters. The Limit field when
given, indicates the limiting value the parameter can take. If there is no entry for the
Limit field it is assumed that the parameter can take any value.
To protect mesh integrity it is necessary to enforce relationships among certain

parameters. This step is not required, but is recommended. For this purpose select
the Rules tab and include the relationships as shown in FIGURE 94.
Once the parameters have been defined we can proceed to construct the model.
Select the Create Model icon from the Main Toolbar and then select the Geometry
tab. First, create two solid boxes in parametric form:
• Create > Box > Locate > Solid button: On > Input switch On > X: 0, Y: 0,
Z: 0, Width: L, Height: W, Depth: th, Rot-X: 0.0, Rot-Y: 0.0, Rot-Z:0.0 >
Accept.
• Create > Box > Locate > Solid button: On > Input switch On > X: (L+b)/2,
Y: 0, Z: (-h+th)/2, Width: b, Height: W, Depth: h, Rot-X: 0.0, Rot-Y: 0.0,
Rot-Z:0.0 > Accept.
• Create > Body > Bool-Union. With a left mouse button select the two boxes
just created and then click on Accept.

4 FIGURE 94 Rules tab settings for example problem.
Select the Center Model icon from the Views Toolbar and the solid body just
created should look as shown in FIGURE 95a
The next step is to create the two fillets along the intersecting lines between the
to boxes and two auxiliary lines:
• Create > Blend Edge > Rolling Ball > Radius: ra. Select the two lines
at the intersection of the two boxes (see FIGURE 95a) and then click
on Accept. If you have difficulties selecting the lines, simply switch to
wireframe mode (see FIGURE 95b).
• Create > Point > Locate > Input toggle switch ON > X: (L+th)/2, Y: W/
2, Z: (-h+th)/2 > Accept.
• Create > Point > Projection > Turn off the Display Surface icon .
Using the last point created, project this point to the lines indicated in
FIGURE 95c by holding Control-Shift, clicking on the appropriate
lines, and choosing Accept.
• Create > Line > 2-Point. Make sure the Curve/Surface selector is in
curve mode. Create a vertical line by clicking on the first point and the
point directly above it. Repeat by selecting the third and fourth points
to create a second vertical line (see FIGURE 95c).
The description of the solution domain is now complete.

box 2
(a) (b)
box 1
Intersection lines between boxes

4
(c)
point by location
FIGURE 95 Geometry for example

problem.
points by projection
Creating the mesh To create the nodes and elements for the solution domain defined above, proceed as
follows (see FIGURE 96):
• Create > Node > Point > Accept. A node will be created at each point loca-
tion. A total of 24 nodes (labeled A in FIGURE 96) should be created by
attachment to the same number of points.
• Create > Node > Mid-Offset. Move the cursor to the display area and click
on two nodes along the fillet. Repeat at the other end. A total of 4 nodes
(labeled B in FIGURE 96) are created using the mid-offset method.
• Create > Node > Midpoint. Move the cursor to the display area, click on
two nodes, and a node will be created at the midpoint between them. Repeat
as many times as needed. A total of 8 nodes (labeled C in FIGURE 96) are
created using the midpoint method.

• Create > Node > Projection. Move the cursor to the display area, click
on a node, hold Control-Shift, click on a line, and a node will be cre-
ated on the line at the shortest distance from the selected node. Repeat
as many times as needed. A total of 4 nodes (labeled D in FIGURE 96)
are created using the midpoint method.
A
A: Nodes to Points A
B: Mid-offset Nodes
D
C: Midpoint Nodes
B C
D: Projection Nodes C
C
4 B D D
B C
C
C
C
B
D
A
C
A A
FIGURE 96 Node location for example problem.
Node creation is now complete, 40 nodes in total. Your screen should look as
shown in FIGURE 96. The next step is to create the elements according to Fig-
ure 97. Proceed as follows:
• Mesh tab > Create > Hexahedron > Selection. Move the cursor to the
display area and click on 8 nodes that define an element in any order.
You may also click and draw a box around 8 nodes, or any combina-
tion of marquee pick and individual node selection until 8 nodes are
selected. A total of 10 hexahedral elements should be created.
• Create > Pentahedron > Selection. Move the cursor to the display area
and click on 6 nodes that define a pentahedral element. The first 3

nodes will define a triangular face of the pentahedron. Two elements should
be created.
The finite element mesh should look as indicated in FIGURE 97.
FIGURE 97 Finite element mesh for example problem.
Elements types
Four types of elements are available in 3D-Elasticity: hexahedral (brick), pentahe-
dral (wedge), tetrahedral and beam elements.
Hexahedral elements: They are constructed by selecting 8 nodes in any order.
Pentahedral elements: They are constructed by selecting first 3 nodes, in any order, that
represent one triangular face and then the other 3 nodes.
Tetrahedral elements: They are constructed by selecting 4 nodes in any order.
Beam elements: They are constructed by selecting 2 nodes in any order.
If two neighboring nodal points lie on a boundary, for example a circle, then that
boundary is automatically associated with the element edge. If three or four nodes
lie on a surface, for example a cylinder, then the surface is automatically associated
with the element face. In the case of 3D beam elements, only straight beams are
supported.

When the elements are displayed, an inscribed 3x3 mesh represented by dotted
lines, appears on the screen. These lines are there to help you see the element
clearly. If you do not wish these lines to be displayed then select View/ Display
Options Toolbar/ Element Handles and the inscribed mesh will disappear.
Vanishing angles Elements with vanishing angles (zero vertex angle) are acceptable for linear
analysis. They require a special post processing procedure (which is performed
automatically by the program) and their main advantage is that facilitate mesh-
ing in filleted regions or when different materials are used. All three solid ele-
ment types can have one or more vanishing angle. For example, in the mesh
shown in FIGURE 98, the elements around the fillet region have one zero ver-
tex angle.
FIGURE 98 Mesh with zero vertex

hexahedral elements.

To enter the material properties select the Material tab in the StressCheck Input
dialog window, or select Class: Material from the Main Menu Bar. The mate-
rial properties module is handled in two parts: the definition and the assign-
ment of the material properties.

Defining material properties: In 3D elasticity the material can be specified as lin-

ear, elastic-plastic or nonlinear elastic. For linear materials the choices are isotro-
pic, transversely isotropic, orthotropic, anisotropic, laminate isotropic and laminate
orthotropic. For isotropic, temperature-dependent material properties can be
defined. For transversely isotropic materials, the plane of isotropy is the yz-plane of
the material axes. After selecting the material type, the input area automatically
adjusts to provide the appropriate number of fields into which the material proper-
ties are to be entered. For elastic-plastic and nonlinear elastic materials refer to the
Advanced Guide.
Assigning material properties: Next you need to assign the material name to the
element or group of elements which are to have the specific properties. You may
select all elements, elements one-by-one, or a group of elements. If the assignment
is element by element then simply click on the elements to which the property is to
be assigned. If the assignment is to a group of elements then drag the cursor to 4
enclose the desired group of elements. The selected elements will be highlighted. If
you wish to remove one or more elements from the group then hold the Ctrl key
while clicking on the elements to be removed from the selected group. If the mate-
rial properties are orthotropic or transversely isotropic, and the orientation of the
material axes are not coincident with the global coordinate system, you can enter
the name of the local coordinate system in the System field. To complete the
assignment process, click on the Accept button. To check whether a material was
assigned to each element, select Action: Check and Object: All Elements. Once the
elements of interest are highlighted click on Accept. A confirmation report will
appear on the screen.
For orthotropic and transversely isotropic materials, a coordinate system other than
the global can be specified. If the selected coordinate system is Cartesian (x, y, z),
the material 1-axis is aligned with the x-axis of the selected coordinate system, the
material 2-axis is aligned with the y-axis and the material 3-axis is aligned with the
z-axis. If the selected coordinate system is cylindrical (r, θ, z), then the material 1-
axis is aligned with the θ-direction, the material 2-axis is aligned with the r-direc-
tion and the material 3-axis is aligned with the z-axis.
For the assignment of laminated composite properties see “Laminate Composite

Analysis” on page 186 The assignment of isotropic, orthotropic and temperature-
dependent material properties are described in the examples below.
Example 1 To define the isotropic material properties for the notched plate problem complete

• Material tab > Define tab > ID: steel > Material: Linear > Type: Isotro-
pic > Units: S.I. > E: 2e5, v: 0.30 > Accept. Note that the options Plane
Stress/Plane Strain are not available in 3D (they are only relevant in
Planar Elasticity).
To assign the material ID “steel” to all elements in the mesh, select the follow-
ing options:
• Assign tab > Select > All Elements > ID: steel > Accept.
Example 2 Consider the definition and assignment of orthotropic material properties in

cylindrical coordinates. The mesh shown in FIGURE 99, consisting of eight
hexahedral elements, represents a cylinder made out of an orthotropic material
in which the material 1-axis is in the circumferential direction, the material 2-
axis is in the radial direction and the material 3-axis in the longitudinal direc-
4 tion. The material coefficients are entered in the direction of the material axes
as shown in the figure.
FIGURE 99 Definition of orthotropic

material coefficient in the direction of the
material axes.
The material is then assigned to all the elements in the mesh and the direction
of the material 1-axis is shown on the model when the material ID toggle is on.

As shown in FIGURE 100, the assignment setting includes the selection of the
coordinate system (SYS1 in this example) which has to be defined as a cylindrical
system.
FIGURE 100 Assignment and display of

orthotropic properties in cylindrical system
Temperature-dependent material properties

To specify temperature-dependent material properties, check the Temp Dep switch
in the Define Material tab of the Input dialog window. Only linear isotropic mate-
rial can be defined as temperature-dependent.
The modulus of elasticity (E), the Poisson’s ratio (v), the density (Dens.) and the
coefficient of thermal expansion (a(th)), may be defined as a constant, a parameter,
a formula or a table. Entering a constant or a parameter is straightforward, simply
click on the appropriate box and key in the constant value or the parameter name. If
the coefficients are specified by formula (as a function of temperature) then the for-
mula name is entered into the input area. The results of the evaluation of the for-
mula will be displayed in the Graph window. If the coefficients are specified by
tabular data, enter the table name in the corresponding fields. The program will fit

the tabular data using least square fitting and the results will be displayed in the
Graph window. Any combination of constant, parameter, formula or table can
be used for the four material coefficients.
Note: When temperature-dependent material properties are defined and

assigned to elements, only Linear, Bending and Modal analyses can be per-
formed. Margin Check and Nonlinear analyses are not supported.
Formula option FIGURE 101 illustrates the case when the modulus of elasticity is defined by
the formula ‘E_T’ as a function of temperature. In this example:
• If Temp < 200 then: E = 200,000
• If 200 <= Temp < 400 then: E = 200,000 - 100,000 (Temp/200 - 1)
• If Temp > =400 then: E = 100,000
4
In the Formula dialog box, X (the independent variable) is the temperature.
The formula is entered as follows (see FIGURE 101b):
if(X<200;200000;if(X>=400; 100000;200000 - 100000*(X/200 -1)))
The evaluation of the formula is shown in a graph window as shown in FIG-

URE 101c.
Note: To set the range of temperatures shown in the graph window, create the
parameters (Note that nl is lowercase NL):
• _nl_min: Initial temperature value to be included in the graph
• _nl_max: Final temperature value to be included in the graph
• _nl_count: number of points between min and max.
Table option FIGURE 102 illustrates the case when the Poisson’s ratio is defined by a table
as a function of the temperature. The name of the table (V_T) is entered in the
corresponding fields of the Material input area. The program performs a least
square fitting of the tabular data and displays the fitted curve in the Graph win-
dow. For temperatures outside the tabular data range a constant value is
assumed. To create the tabular data, select the Tables tab from the Input dialog
box. A pair of values is entered by suppling the following information (FIG-
URE 102a):
• Tables tab > Select > Location > Constant > ID: V_T > Type: Material
> X: key in the temperature > Y: enter the Poisson’s ratio > Accept

(b) Formula form
(a) Material input area
(c) Graphical representation of formula
FIGURE 101 Temperature-dependent material properties: Formula option.
Repeat for as many entries as needed. A table should contain at least two entries
with different values of temperature. The results of the least square fitting is shown
in FIGURE 102b.

Entering load data
(b) Graph with fitted tabular data
(a) Table input area
FIGURE 102 Temperature-dependent material properties: Table option.
An example of the use of temperature-dependent material properties is pre-

sented in the Advanced Guide.
Entering load data

To enter the load data select the Load tab in the StressCheck Input dialog win-
dow, or Class: Load from the Main Menu Bar.

Entering load data
Types of loading
Several types of loading are available in 3D elasticity: Traction, spring, body
forces, thermal, bearing and point loads and moments. The definitions are as fol-
lows:
Traction loading: means that a distributed load (in force per unit area) is imposed
on a surface or element face. If the traction is attached to a surface, all the element
faces associated with that surface will be loaded. Traction is a vector quantity, and
therefore three components must be given. These may be in the normal-tangent ref-
erence frame, in the global system, or an arbitrary local system. If the reference
frame is the normal-tangent then only a normal traction can be applied.
Traction loads can also be applied to beam elements if the object Edge or Edge
curve is selected in the user interface. In the case of beams, tractions are in force per 4
unit length.
Spring loading: means that load is applied through a displacement imposed on a

distributed spring which has been specified on an element face or surface. Spring
loading can be applied in normal-tangent, global or local coordinate systems. The
only restriction is that the spring coefficients and the imposed spring displacements
must be in the same reference frame. To specify the values of the spring coeffi-
cients, refer to the next section.
Body force loading: The body is loaded by some volumetric load, such as by grav-
ity or inertia due to acceleration. The body force is in terms of force per unit vol-
ume. If the acceleration vector a={ax, ay, az} and the mass density (ρ) are known
then enter the products:
X constant = ax ρ
Y constant = ay ρ
Z constant = az ρ
Body forces can be applied to elements in the global or any local Cartesian coordi-
nate system.
Thermal loading: means that a temperature distribution (in degrees) will be speci-
fied on elements. Temperature distribution obtained from a heat transfer analysis
are also available in 3D. Thermal loading can be applied to elements only.
Fit-Auto and Fit-Manual: Capability to specify imported point load from a global
model to a group of faces of a 3D model. The procedure assigns point loads to ele-

Entering load data
ment faces, calculate interpolated traction functions on those faces, and incor-
porate the tractions into a load case for use in the finite element solution
process. For specific technical aspects of the implementation of the capabilities
see “Global-Local Analysis” on page 206
Bearing loading: Normal sinusoidal traction applied to a circular hole by

specifying the magnitude and direction, or the two vector components (Fx, Fy).
Bearing loads can be applied to a cylinder or to a group of faces on a cylindri-
cal pattern. A coordinate system must be selected with the z-axis directed
along the center of the cylinder. The angle under the Mag/Dir option is mea-
sured from the positive x-axis of the system (positive counter clockwise).
Finally, the position of the bearing resultant along the z-axis must be provided
under the Zo-input. Zo=0 means that the resultant is applied at the mid-height
of the hole. If Zo is different from zero, then the applied traction will include
4 the effect of bending. The traction applied to the element faces around the hole
is determined from the following equation:
F 2F 12
T n = – ---------- cos ( θ – α ) 1 + -----2- Z o ( Z – Z G ) if ( T n < 0 )
πRh h
Zo
T n = 0 if ( T n > 0 )
h
ZG
Z
X
Note: The values of h and ZG are computed
Y automatically by the program from the location
of the nodes on the selected faces.
Tn
F
R
α
θ
X

Entering load data
Point Force: The elastic body is loaded by a concentrated force F={Fx, Fy, Fz}.
Point loads can be applied to points or nodes in the global directions only. The
points can be located inside an element for 3D solid elements. For beam elements,
only nodal forces in the global directions are possible.
Point Moment: Applicable to nodes of beam elements, a point moment can have
three components in global coordinates (Mx, My, Mz).
Note: Shell Traction, Shell Moment and Shell Shear are reserved for shell elements
which are currently under development.
You may load an object (such as a face, surface or element) or a group of objects.
An object is selected by pointing to it and clicking on the left mouse button. A
group can be selected by dragging the cursor until the desired group is completely
enclosed in a square. On releasing the left mouse button, the selected group will be 4
highlighted. If the group includes objects which were not to be selected then indi-
vidual objects can be removed from the group by selecting them while holding
down the Ctrl key of the keyboard. If you selected traction loading then the object
to be loaded may be a surface, a face surface or an element face, or a group of sur-
faces or faces.
The loading data may be constant, parametric or a formula.
Example To specify the loading for the notched plate problem shown in FIGURE 93, enter
• Load tab > Action: Select > Object: Face Surface > Method: Traction > ID:
LOAD > Direction: XYZ > Z: -Ps/th. Move the cursor to the display area
and click on any of the two faces at the free end of the plate, then click on
the Accept button. A distributed traction pointing downward will be dis-
played on the element faces. If you selected the wrong element face, simply
select Cancel and try again.
The load line Ps was converted to a traction by dividing by the plate thickness th.
The total force applied to the plate will be the product of the line load multiplied by
the width of the plate (F=Ps W=600 N). To check the total applied load to the plate,
enter the following information:
• Check > All Elements > ID: LOAD > Moment-X: 0.0, Moment-Y: 0.0,
Moment-Z: 0.0 > Accept.
The program will integrate the applied tractions and report the sum of the forces in
the global coordinate directions and the moment about the three global axes. The

Entering load data
moment center was selected at the center of the global coordinate system
which is located at the center of the base of the first box created during the
solid model construction. The result will be displayed in the Report window as
FIGURE 103 Load check report and input area.
Section properties in formula loads

It is possible to incorporate section properties (cross-sectional area, moments
of inertia, etc.) of a group of faces from 3D-solids into a formula used for the
application of traction loadings. Once the group of faces is selected, the section
properties are computed by the program before the tractions are applied. The
steps involved are as follows:

Entering load data
1. Create a local Cartesian coordinate system in such a way that the xy-plane of the
local system is in the same plane as the group of faces to be loaded. To create a
local coordinate system, select Class: Geometry and follow the procedure described
in the User’s Guide, Chapter 3: Model Input.
2. Create a load record by selecting Class: Load, enter the Load ID, select the group
of faces for which the section properties should be computed. The faces should all
be flat and contained in the same plane. Select the local system defined in step 1,
and enter the formula name to be created in step 3.
3. Create a formula with the name given in step 2 that uses the Section Property
Function name(s) shown below:
2
Yc
1
θ
Xc
c
(X’c,Y’c)
Y’
Local system
X’

Entering load data
FUNCTION
NAME EXPLANATION
SPArea A (cross sectional area)
SPCtrx X’c (x-coord. of centroid w.r.t. local system X’Y’Z’)
SPCtry Y’c (y-coord. of centroid w.r.t. local system X’Y’Z’)
SPMomx Ixc (moment of inertia about the centroidal axis Xc parallel to X’)
SPMomy Iyc (moment of inertia about the centroidal axis Yc parallel to Y’)
SPCross Ixy (cross moment of inertia about Xc,Yc)
SPMom1 I1 (Moment of inertia about the principal axis-1)
4
SPMom2 I2 (Moment of inertia about the principal axis-2)
SPAng θ (Angle in degrees between the Xc-axis and the principal axis-1)
At the time of the formula evaluation to compute the load vector associated
with the prescribed traction, the program will compute all the section proper-
ties of the selected faces, and use the ones included in the formula definition.
Suppose we are interested in applying a bending moment M about the weak
axis of the cross-section of the beam-like structure shown in FIGURE 104.
Having a local coordinate system located at the base of the section as shown,
the normal traction that should be applied to the group of faces is given in
terms of the applied moment and the section properties as follows:
M
T n = – ------ ( y' – y' c )
I xc
The corresponding Formula dialog window is shown in FIGURE 105.
Checking applied loads To check the magnitude of the applied mechanical loads, select the action
Check, the object All Elements or Any Element and the Load ID to be
checked. Enter the coordinates of the moment center in the fields Moment-X,
Moment-Y, Moment-Z and click on Accept (if you selected the option Any
Element, you need to pick the elements of interest before clicking on Accept).

Y’ 4
Y’
X’ X’
Z’ SYS1
FIGURE 104 Use of section properties in formula loading.
The results include the three resultant vector components Fx, Fy Fz and the three
moment components Mx, My, Mz.

To enter the constraint data select the Constraint tab in the StressCheck Input dialog
window, or Class: Constraint from the Main Menu Bar.
You may constrain an object (such as a node, face or surface) or a group of objects
(nodes, faces or surfaces). An object is selected by pointing to it and clicking on the
left mouse button. A group can be selected by dragging the cursor until the desired
group of objects is completely enclosed in a square frame. If the group includes
objects which were not to be selected then individual objects can be removed from

FIGURE 105 Traction loading using section properties.
the group by selecting them while holding down the Ctrl key of the keyboard.
The object selected controls the type of constraint that can be applied.
Types of constraints
The type of constraints available in 3D elasticity, such as General, Rigid Body,
or Spring Coefficient are explained in the following:
General: One, two or three displacement components can be prescribed to one

or more surfaces or faces in the global or in a local coordinate system. Several
commonly occurring situations have been implemented as special cases as
shown below:
Built-In: All three displacement components are set to zero. The selected
faces can be flat or curved.
Symmetry: The normal displacement component is set to zero. Symme-
try constraints are applicable only to flat faces. A check is automatically

performed every time symmetry is specified to make sure that the selected
face or faces are flat. A warning message is issued if symmetry is applied to
faces which are not flat.
Anti symmetry: Both tangential displacement components are set to zero.
Anti symmetry constraints are applicable only to flat faces. A check is auto-
matically performed every time anti symmetry is specified to make sure that
the selected face or faces are flat.
Fit-Auto and Fit-Manual: Capability to specify imported nodal constraints

from a global model to a group of faces of a 3D model. The procedure assigns
constraints to element faces, calculate interpolated displacement functions on
those faces, and incorporate the displacements into a constraint case for use in
the finite element solution process. For specific technical aspects of the imple-
mentation of the capabilities see “Global-Local Analysis” on page 206
4
Soft-Simple, Pinned and Sliding: Applicable to 3D-shells only (under devel-
opment).
Spring Coefficients: When an elastic body is constrained by means of distributed

springs, or loaded by displacements imposed by distributed springs, then the spring
coefficients modify the stiffness matrix. For this reason the spring coefficients are
entered as constraints. Spring coefficients can be specified in the direction of the
global or a local coordinate system or in the direction Normal/Tangent to a face or
group of faces. For Normal/Tangent, the tangent direction means both tangents.
In the case of General constraints the data type may be Fixed, Constant, Parametric
or Formula. Fixed means that the selected displacement component is set to zero. If
the displacement is constant along a surface or face then the constant value must be
entered into the appropriate field. A parameter or parametric expression can be
used also to specify a prescribed displacement. The value of the displacement is
computed from a parametric expression (up to 20 characters long) entered in the
corresponding input field. Any combination of previously defined parameters and
constants can be used. For the formula type, the formula name is entered into the
appropriate box. When a formula is defined in a coordinate system, other than the
global system, then the name of the coordinate system must be entered also.
In the case of Springs the data type may be Constant, Parametric, or Formula.
Rigid Body: StressCheck provides convenient means for specifying rigid body
constraints for solid models. The user selects three non-colinear nodes, and Stress-
Check creates a right-handed local coordinate system, the x-axis of which is
directed from the first node to the second. In this local coordinate system, all dis-

placement components are constrained for the first node; two displacement
components are constrained for the second node, and only one displacement
component for the third node.
Remark: For solid elements, point constraints are admissible only for prevent-
ing rigid body displacement and rotation. The body must be in equilibrium
under the action of external forces. Unlike the errors introduced by concen-
trated forces, errors introduced by point constraints are not localized, and the
reactions at point constraints are dependent on the mesh and the polynomial
degree of elements. For this reason multipoint constraints should not be used in
StressCheck.
Node: Individual nodes can be constrained or their displacement prescribed in

one or more directions. In the case of beams, three displacements and three
rotations in global coordinate directions can be prescribed.
4
Hinge: Releases rotational degree of freedom associated with a beam element
at that node. A hinge is specified by selecting a node and, while holding the
shift key, one element attached to that node. If more than two elements are
hinged at a node, multiple records can be created selecting the same node and a
different element each time. The rotation(s) to be released is defined in the
beam coordinate system, such that Rx is the twist rotation, and Ry, Rz are the
bending rotations.
Contact Zone: To specify the spring constant for a contact problem. For addi-
tional details refer to the Multi-body Contact chapter in the Advanced Guide.
Example To specify the built-in constraint to the notched plate of the example problem,
• Constraint tab > Select > Face Surface > Built-In > ID: CONST. Move
the cursor to the display area and click on any of the two element faces
at the fixed end of the plate and then click on Accept. The symbols of
the built-in (clamped) constraint will be displayed on the element faces
attached to that plane as shown in FIGURE 106.

FIGURE 106 Constraint input area and symbols.

Because StressCheck allows more than one load and constraint cases to be defined,
it is necessary to associate a unique solution name with each desired constraint and
load name pair. To do this, select Class: Solution ID from the Main Menu Bar, or
select the Solution ID tab from the StressCheck Input dialog window.
Example To complete the solution record for the notched plate of the example problem sup-
ply the following information:
CONST > Load ID: LOAD > Click on the Accept button.
By this action a unique solution name (SOL) has been assigned to the constraint
(CONST) and load (LOAD) name pair created previously.


To specify the discretization parameters select Class: p-Discretization from the
Main Menu Bar, or select the p-Discretization tab from the StressCheck dialog
window. In StressCheck, typically uniform p-extensions are used. However, it
is possible to select any group of elements and fix the polynomial degree for
those elements. All the options under the p-Discretization tab are described in
the User’s Guide.
If uniform p-extension, using the trunk space, is to be employed then nothing

needs to be done because this is the default option for StressCheck. If uniform
p-extension using the product space is to be employed, the model must have
hexahedral elements only. In this case the product space must be assigned to all
elements.
4
Example Let us assign the trunk space to all the elements of the example problem and
• p-Discretization tab > Select > All Elements > Space: Trunk > p-Dis-
cretization: Variable > p-Level: 1 > Click on the Accept button.

To execute a linear analysis select Edit: Solution from the Main Menu Bar, or
click on the Compute Solution icon from the Main Toolbar. When the Solution
dialog window appears then select the Linear tab and complete the requested
information. The following options are available for 3D Elasticity (FIGURE
107):
Extension: Choosing “Upward-p” means that the solution will be computed

from the minimum to the maximum p-levels specified under p-limits. This
option requires more CPU time than the “Downward-p” but requires less disk
space. The highest possible p-level in StressCheck is 8.
Choosing “Downward-p” means that the solution will be computed from the
maximum to the minimum p-level specified under p-limits. Generally, the
downward sequence is faster but requires more disk space than the upward. For
the highest p-level, the upward and downward options yield identical results.
For the lower p-levels there will be minor differences owing to the fact that the

4
maximal number of integration points are used when the downward option is
selected. You may give the p-limits in either order.
Choosing “Uniform-h” means that h-extensions (uniform mesh refinement) can be

performed when all boundary conditions were specified on “boundaries” rather
than element edges. When boundary conditions are prescribed on boundaries then
the loads and constraints are inherited by those elements which have an edge on the
boundary. StressCheck will generate finite elements by uniformly subdividing each
standard finite element.
If you select “Uniform-h” then you need to specify Run Limit, which is the number
of h-refinements to be performed. For example, at each refinement level each hexa-
hedral element is subdivided into eight elements. You also have to specify the p-
level which will remain the same for all the runs.
Choosing “Adaptive-p” means that the polynomial order of the elements with a
variable or uniform p-discretization assignment will be determined based on the
values of the error indicator for each run until convergence is realized. The Run
limit input field refers to the maximum number of adaptive runs the program will
perform if convergence is not realized. The p input field is to provide the initial p-
level for the adaptive p-extension.The convergence options are explained below.


or, in the case of h-extensions, from the current mesh to the number of refine-
ments specified by the Run Limit entry.
“Restart” is used when a sequence of p-levels has been computed and you wish
to compute higher p-levels, retaining the computed data. In this case simply
increase the upper p-limit to its desired value and select “Restart”. In the case
of large problems it is good practice to run initially from p=1 to 4, check the
results, then if necessary, increase the p-limits until the desired level of preci-
sion is reached.
4 Run Mode: You may execute an “Automatic” sequence of solutions for the
extension and limits selected. Or you may execute “Stepwise”, requiring you to
click on Solve following the solution for each p-level or h-refinement.
Method: It may be either “Iterative” or “Direct”. Method: Iterative means that

the Iterative Solver (default) is used. Method: Direct means that the Direct
Solver is used. The direct solver typically requires more memory than the iter-
ative solver for the same size problem. the direct solver is more efficient when
very thin domains are represented with solid elements.
Converge: Two options are available: “None” means that the number of runs
is controlled by the settings in the linear tab. “Error=5%” means that the num-
ber of runs will be controlled by the global error of the solution. When the esti-
mated relative error in energy norm is less than or equal to 5%, the execution
will stop. The highest polynomial order cannot go beyond the limits entered in
the linear tab. Additional convergence criteria can be defined in the Settings
tab of the Results interface, as described in the User’s Guide.
- Results: Engineering results may be monitored by using the layout switch to

activate the Results layout. Select the desired Output controls in the Results
class of interest before activating the solution. Monitoring result functions is

useful in the case of large problems where sequential analysis executed stepwise is
employed. The monitored data provide information on the basis of which you can
decide whether to stop or continue the extension process. For small problems full p-
extension should be performed.
become quite large, so it if often desirable to have them created in a scratch area
deleted by the program when they are no longer needed. Others files are left behind
these special temporary files (.wrk). 4
Browse: The Browse button is used to identify a directory path where you wish to
have intermediate work files created. These files are disposable files.
Buttons: The Solve button will activate the currently specified solution. The Check
button will produce a report summarizing parameters, elements distortion, geomet-
ric and material properties, applied load, and the disk space required to perform the
currently specified solution.
Example Execute a linear analysis for the notched plate considering an upward p-extension.
In the Solver dialog window enter the following information:
tive > Convergence: None > Display: Sequence > Click on the Solve but-
ton.
The execution time depends on your computer processor and memory.

To access post-processing features of StressCheck you must select Edit: Results
from the Main Menu Bar, or select the View Results icon from the Main Toolbar.

Example To obtain a global error estimate, select the View Results icon from the Main
Toolbar, the Error dialog window will come out together with the Graph win-
dow. For the example problem, the interface settings and the tabular results of
the estimated relative error in energy norm are shown in FIGURE 108.
FIGURE 108 Estimated relative error in energy norm.
The maximum value of the equivalent (von Mises) stress can be computed
selecting the Min/Max tab and completing the input area as follows:
• Min/Max tab > Action: Select > Object: Face > Method: Grid > Input
tab > Solution: SOL > Run: 1-8 > Function: Seq > Midsides: 15. With
the mouse click on the top face of the element in the fillet region where
the maximum stress is expected. > Click on Accept.
The equivalent stress is computed at (15+2)*(15+2) grid points over the ele-
ment face and the maximum value is Seq=11.92 MPa at p=8. FIGURE 109
shows the convergence of the maximum equivalent stress.

FIGURE 109 Convergence of the maximum von Mises stress.

Model problem: Plate in tension
Problem description
Consider a plate in tension with a reinforced circular hole which includes a fil-
let between the plate and the stiffening ring, as shown in FIGURE 110. The
material properties are that of a standard carbon steel ASTM A-36, SI units.
4 W
th
rf
L
h
a
B
FIGURE 110 Plate with reinforced hole in tension.
We are interested in creating a 3D description for this problem with all dimen-
sions given in parametric form. An outline of the steps for creating the geome-
try and finite element mesh, applying the load and enforcing the constraints is

Geometry and mesh Before starting, set the units selector to mm/N/sec/C. There are seven geometric
parameters necessary to describe the problem (a, B, h, L, rf, th, W) and one loading
parameter (S). To create the parameters for this problem, select the Model Info icon
from the Main Toolbar and when the dialog window appears select the Parameters
tab, complete the following information:
• Name: a, Description: Hole diameter, Value: 1.0, Limit: >0.0, Class: Gen-
eral.
• Name: B, Description: Stiffening ring diameter, Value: 2.0, Limit: >0.0,
Class: General.
• Name: h, Description: Stiffening ring thickness, Value: 2.0, Limit: >0.0,
Class: General.
• Name: L, Description: Plate length, Value: 8.0, Limit: >0.0, Class: General.
• Name: rf, Description: Fillet radius, Value: 0.3, Limit: >0.0, Class: General. 4
• Name: th, Description: Plate thickness, Value: 0.5, Limit: >0.0, Class: Gen-
eral.
• Name: W, Description: Plate width, Value: 4.0, Limit: >0.0, Class: General.
• Name: S, Description: Remote stress, Value: 1.0, Limit: >0.0, Class: B.
Cond.
To protect mesh integrity it is necessary to enforce relationships among certain

parameters. To incorporate these requirements select the Rules tab and complete the
information as shown in FIGURE 111.
FIGURE 111 Relational Expressions for example problem.

Making use of symmetry (geometry and loading), it is possible to work with

only one-eighth of the problem (there are three planes of symmetry). This sym-
metry consideration will simplify model creation and reduce running time. To
produce the geometric description shown in FIGURE 112, create a body con-
sisting of a box and a hollow cylinder, and then blend the intersection edge to
create the fillet:
• Geometry tab > Create > Box > Locate > Solid switch ON, Input
switch ON > X: 0.0, Y: 0.0, Z: -th/2, Width: L, Height: W, Depth: th,
Rot-X: 0.0, Rot-Y: 0.0, Rot-Z: 0.0 > Accept.
• Create > Cylinder > Locate > Solid switch ON, Input toggle switch ON
> X: 0.0, Y: 0.0, Z: -h/2, Radius: B/2, Height: h, Rot-X: 0.0, Rot-Y: 0.0,
Rot-Z: 0.0 > Accept.
• Create > Cylinder > Locate > Solid switch ON, Input toggle switch ON
4 > X: 0.0, Y: 0.0, Z: -h/2, Radius: a/2, Height: h, Rot-X: 0.0, Rot-Y: 0.0,
Rot-Z: 0.0 > Accept.
• Create > Body > Bool-Union. With the left mouse button, click on the
first cylinder and then on the box. Then click on Accept. The box and
the larger cylinder became a single body.
• Create > Body > Bool-Subtract. With the left mouse button, click on
the last body created in the step above and then on the second cylinder,
in that order. Then click on Accept. A hole is inserted into the body.
• Create > Blend Edge > Rolling Ball > Input toggle switch ON >
Radius: rf. select the circular edge between the cylinder and the box
(FIGURE 112a) and then click on Accept. This only needs to be done
on one side of the body because we are applying symmetry conditions.
The next step is to create three planes to clip the solid such that only one eighth
of the model remains:
• Create > Plane > Locate > Input toggle switch ON > X: 0.0, Y: 0.0, Z:
0.0, Width: 1.5*L, Height: 1.5*W, P1-min: -0.5, P1-max: 0.5, P2-min:
-0.5, P2-max: 0.5, Rot-X: 0.0, Rot-Y: 0.0, Rot-Z: 0.0 > Accept. The
plane must be larger than the body for the clipping operation.
• Create > Body > Clip-Back. With the left mouse button, click on the
body and then on the plane created in the step above, in that order. Half
the solid is removed (see FIGURE 113a and b).
We have to repeat two more times with two different planes:

Edge to blend Edge blend
(a) (b)
FIGURE 112 Geometry for model problem.
• Create > Plane > Locate > Input toggle switch ON > X: 0.0, Y: 0.0, Z: 0.0,
Width: 1.5*L, Height: 2*h, P1-min: -0.5, P1-max: 0.5, P2-min: -0.5, P2-
max: 0.5, Rot-X: 90.0, Rot-Y: 0.0, Rot-Z: 0.0 > Accept.
• Create > Body > Clip-Front. With the left mouse button, click on the body
and then on the plane created in the step above, in that order (see FIGURE
113c and d).
• Create > Plane > Locate > Input toggle switch ON > X: 0.0, Y: 0.0, Z: 0.0,
Width: 1.5*h, Height: 1.5*W, P1-min: -0.5, P1-max: 0.5, P2-min: -0.5, P2-
max: 0.5, Rot-X: 0.0, Rot-Y: -90.0, Rot-Z: 0.0 > Accept.
• Create > Body > Clip-Front. With the left mouse button, click on the body
and then on the plane created in the step above, in that order (see FIGURE
113e and f).
The geometric description is now complete. One possible discretization for the
solution domain defined above is shown in FIGURE 115. Before you start with

(b)
4 (a)
Clipping plane
(d)
(c)
(e)
FIGURE 113 Solid clipping operations.
(f)

node creation, it is convenient to turn the display of surfaces off and turn them back
on when needed. The nodes and elements were created as follows:
• Mesh tab > Create > Node > Point > Accept. A node will be created at each
point location. A total of 16 nodes (labeled A in FIGURE 114a) should be
created by attachment to the same number of points.
• Create > Node > Mid-Offset. Click on two nodes on the fillet region and a
node will be created on the fillet. A total of 8 nodes (labeled B in FIGURE
114b) should be created by this method.
A
A
A: Nodes to Points
B 4
A B: Mid-offset Nodes
A C: Mid-offset Nodes
A
A D: Projection Nodes
(b)
A B
A A C C
(a)
D
A
FIGURE 114 Node location for model problem. D

(c)
D
• Create > Node > Mid-offset. Click on two nodes to create a node as a mid-
offset between the selected nodes. A total of 2 nodes (labeled C in FIGURE
114c) should be created by this method

click on a node, hold Control-shift, click a curve/surface, and a node
will be created on the curve/surface at the shortest distance from the
selected node. Eight nodes were created by the method projection
(labeled D in FIGURE 114c). When projecting nodes to surfaces, the
Display Surfaces icon must be turned on, and when projecting nodes to
curves, it must be turned off.
Node creation is now complete. The next step is to create the elements as
shown in FIGURE 115. A total of 9 hexahedral elements have been defined by
selecting the appropriate nodes in arbitrary order.
FIGURE 115 Finite element mesh

for model problem.
Material To define the material properties complete the following information:

• Material tab > Define tab > ID: steel > Material: Linear > Type: Isotro-
pic > Units: S.I. > E: 2e5, v: 0.295 > Accept. Note that the options
Plane Stress/Plane Strain are not available in 3D (they are only relevant
in Planar Elasticity).
To assign the material ID “steel” to all elements in the mesh, select the follow-
ing options:
• Assign tab > Select > All Elements > ID: steel > Accept.

Load and Constraints Apply a parametric uniform traction to the right plane of the plate (make sure that
the Display Surfaces icon is on):
• Load tab > Action: Select > Object: Any Surface > Method: Traction > ID:
LOAD > Direction: Norm./Tan. > Normal: S. Move the cursor to the dis-
play area and click on the rightmost plane of the box, then click on the
Accept button. A distributed traction pointing in the direction of the out-
ward normal will be displayed on the element faces attached to the plane. If
you selected the wrong plane, simply click again until the proper surface is
highlighted.
Symmetry boundary conditions must be applied to the three planes of symmetry.

After selecting the Constraint tab in the input area, enter the following information:
• Constraint tab > Action: Select > Object: Any Surface > Method: Symme-
try > ID: CONST. Move the cursor to the display area and click on the three 4
surfaces of the plate which lie on the planes of symmetry while holding
down the Shift key on your keyboard. Click Accept.
The load and constraint symbols are shown in FIGURE 116. Complete the execu-
tion parameters as discussed before, and obtain a sequence of solutions from p=1 to
8.
FIGURE 116 Load and constraint symbols for model problem.
Note: This problem is available from the Tutorial Handbook: Doubler3D.sci. Refer
to the User’s Guide to execute a problem from the Handbook library.

Extrusion
Extrusion
The Extrusion option in StressCheck provides a simple way to investigate the
effects of out-of-plane loads and constraints on bodies which are essentially
two-dimensional. Extrusion is applicable only for components that are defined
in the xy-plane (Planar reference) and have piecewise constant thickness. The
loads and constraints (symmetry, antisymmetry, built-in) are automatically
converted to their 3D equivalent when Extrusion is performed. Once a model
has been extruded, it is possible to add to or modify existing load and con-
straint records before executing the analysis.
Model Problem: Extruded Pipe Model

4 A thick-walled pipe of internal radius Ri = 0.5 inches and external radius Re =
2.0 inches, is subjected to a uniform internal pressure of Pi = 1000 psi. The
material properties are those of a nearly incompressible material: the Modulus
of elasticity is E = 30 x 106 psi and the Poisson’s ratio is v = 0.45. The objec-
tive of the analysis is to determine the maximum principal stress to within 5%
relative error.
AB: u n = T t = 0 (symmetry) Y C
BC: T n = T t = 0 (stress free)
CD: u n = T t = 0 (symmetry)
DA: T t = 0, T n = -1000 psi D

(internal pressure)
A B X
FIGURE 117 The solution domain and Ri
boundary conditions. Re
The model will be constructed as a two-dimensional elasticity problem, con-

sidering only a portion of the cross-section of the pipe and assuming plane-
strain conditions. One-eighth of the cross-section will be considered, and sym-
metry boundary conditions will be imposed on the two radial boundaries of the

Extrusion
sector (FIGURE 117). The boundary conditions are given with reference to the nor-
mal (n) and tangent (t) directions for each side. The positive normal is outward and
the positive tangent direction is counterclockwise. The symbols un, ut represent the
displacement vector components, Tn, Tt represent the traction vector components.
Geometry and mesh From the Reference and Theory selectors make sure the analysis is set to Planar
Elasticity and the units are set to in/lbf/sec/F, then select Edit: Input and Class:
Geometry from the Main Menu Bar, or click on the Create Model Icon to begin
model construction (refer to FIGURE 118).
FIGURE 118 Pipe sector. Single

quadrilateral element.
• Geometry tab > Create > Circle > Locate > Repeat # = 2 (turn Repeat
switch ON and then type 2 in the box) > Input toggle switch ON > X: 0.0 +
0.0, Y: 0.0 + 0.0, Z: 0.0 + 0.0, Radius: 0.5 + 1.5, P1_min: 0.0 + 0.0, P1-
Max: 360.0 + 0.0, Rot-Z: 0.0 + 0.0 > Accept. From the Views Toolbar click
on the Center Model icon. As a result of this operation two circles of radii
0.5 and 2.0 are now displayed at the center of the “Main Window”.
Create two lines of length 3.0, which are 45 deg. appart from each other and with
the origin at the same location as the center of the circles:

Extrusion
• Geometry tab > Create > Line > Locate > Input toggle switch ON >
Repeat #= 2, X: 0.0 + 0.0, Y: 0.0 + 0.0, Z: 0.0 + 0.0, Length: 3.0 + 0.0,
Angle: 0.0 + 45.0 > Accept
Locate four nodes at the intersection of the lines and circles:

• Mesh tab > Create > Node > Intersection. Select a circle and a line by
positioning the mouse in the “Model Window” area and clicking on the
boundaries close to the location of the node to be created. Repeat the
operation four times. After each selection, the symbol of a node (a
small square) will be displayed at each location. You can create all
intersection nodes in one operation by clicking on the Accept button
instead of creating each node manually. If you use the Accept button
method, you can delete the node at the origin if you wish by Select >
Node > Selection, click on the node at the origin, and click Delete.
4
Create a quadrilateral element by selecting the four nodes created in the previ-
ous step:
• Create > Quadrilateral > Selection. Click on each one of the four nodes
in any order, or use the marquee pick to group all four nodes by draw-
ing a box around them. A quadrilateral element will be displayed, as
shown in FIGURE 118. Note that the element has two curved sides.
The program recognizes that the nodes are related to the circles and
automatically makes the corresponding element edges curved. A refer-
ence grid of dashed lines (element handles) is displayed over the ele-
ment for purposes of visualization only. If you wish to suppress the
grid, go to the Display Options Toolbar and turn off the Element Han-
dles icon.
Thickness To assign a thickness to the element, select Class: Thickness or click on the
Thickness tab in the StressCheck Input dialog window. Supply the following
input:
Accept. A summary of the thickness assignment record is inserted in
the pulldown menu.
Materials The material properties module is handled in two parts: One is the definition of
the material properties and the other is the assignment of material properties to
the elements. Select the Material tab in the StressCheck Input dialog window
and choose the following options:

Extrusion
• Material tab > Define tab > ID: prop (Give a name that will identify the
material properties) > Material: Linear > Type: Isotropic > Units: U.S. >
Case: Pl. Strain (Plane-strain) > E: 3.0e7 (Modulus of elasticity), v: 0.45
(Poisson’s ratio), Dens.: 0.0 (Density. Not required for this problem), a(th):
0.0 (Coefficient of thermal expansion. Not required for this problem) >
Click on the Accept button to complete the material definition operation.
Note: A warning message will appear in the execution window saying that the con-
dition number of the material stiffness matrix is larger than 20. When you define
material properties, StressCheck automatically computes the smallest and largest
eigenvalues of the material stiffness matrix ( λ min and λ max , respectively). When
the condition number of the material stiffness matrix, that is, the ratio λ max / λ min , is
greater than 20 then a warning message appears in the execution window. The rea-
son for this is that the energy error and the root-mean-square error in stresses are
close only when λ max / λ min is not large. You may proceed with the analysis but 4
exercise care when computing stresses from the finite element solution. For further
information and an example see Section 11.4.2 of Ref. [3]
The assignment of the material properties to elements is performed by selecting the

following options:
• Assign tab > Select > All Elements > ID: prop (the same name given in the
DEFINE action) > Click on the Accept button to complete the material
assignment operation.
The material definition and assignment are now complete. A summary of the mate-
rial record is inserted in the scrolling listbox in the Input Area.
Loads The next activity is to apply the uniform normal traction (pressure) to the inside
diameter of the tube. Select the Load tab from the StressCheck Input dialog win-
dow and supply the following information:
• Load tab > Select > Any Curve > Traction > ID: LOAD (Any name can be
used) > Direction: Norm./Tan. > Normal: -1000.0 (turn the toggle switch
ON and enter the value). The negative sign indicates that the direction of
the load is opposite to the positive -outward- normal to the element. Move
the cursor toward the Model Window and select the inner circle (this is the
boundary that you want the load applied to). The program will find all ele-
ments that are bounded to the selected curve, and the load will be associ-
ated with those elements. > Tangent: 0.0 (or disabled) > Click on the
Accept button to complete the operation. A summary of the load record is

Extrusion
inserted in the scrolling list box and the load symbols are displayed as
Constraints Symmetry boundary conditions must be applied to the edges of the sector. This
means that the normal displacement component should be set to zero. Select
the Constraint Tab from the StressCheck Input dialog window and complete as
follows:
• Constraint tab > Select > Any Curve > Method: Symmetry > ID:
CONST. Using the mouse, select the two straight lines on the model
while holding down the Shift key and then click on Accept. A sum-
mary of the constraint record is inserted in the scrolling list box and the
constraint symbols are displayed along the element sides as shown in
FIGURE 119.
4
FIGURE 119 Model Window and Constraint Interface.
The next step is to select Extrude Elasticity from the Reference and Theory
Selectors. The program will convert the 2D problem into a 3D-solid problem,
as shown in FIGURE 120. Note that the original nodes that defined the 2D
quadrilateral element are the only ones visible. This is a reminder that we are
dealing with a solid created by extrusion.

Extrusion
FIGURE 120 Extruded pipe model.
Updating the model When extruding a 2D problem it is a good practice to carefully consider whether
the boundary conditions defined in 2D are complete in 3D or not. In some cases
they will be complete, but in general they will not. In this example, we assumed
plane-strain conditions in the 2D description. Therefore we need to impose normal
constraints on the front and back faces of the hexahedral element to enforce the
zero displacement in the z-direction. To do this, select the Constraint tab from the
Input dialog window and complete the input area as follows (FIGURE 121):
• Constraint tab > Select > Face > Symmetry > ID: CONST (Same name as
used before in 2D). Select the two flat faces (back and front) of the hexahe-
dral element by clicking on the corresponding element handles while hold-
ing down the Shift key, and then click on Accept. A summary of the new
constraint record is added to the scrolling list and the constraint symbols are
displayed on the element faces.
Note that the original constraint information provided in 2D was automatically con-
verted to its 3D equivalent when the model was extruded. Also note that the new
constraints were attached to faces rather than surfaces because there are no surfaces
defined for this problem.

Extrusion
FIGURE 121 Constraints for the Extruded pipe problem.
Extrusion constraints
As explained above, when extruding a 2D model it is necessary to check if the
constraints are sufficient or not. The following cases illustrate additional model
constraints that are required when certain 2D models are extruded. The four
cases below illustrate the rules to convert 2D nodal constraints, and to specify
constraints on the extrusion side.

Extrusion
Double symmetry FIGURE 122 shows how double symmetry constraint applied in 2D should be com-
plemented with a single nodal constraint in the Z-direction in 3D applications.
(b)
(a) 1 addition
symmetry Uz=0
(one node)
FIGURE 122 Double symmetry: (a) planar - (b) extrude.

4
Single symmetry FIGURE 123 shows how single symmetry plus a nodal constraint applied in 2D
should be complemented in 3D applications by the addition of two more nodal con-
straints in the Y and Z-directions.
(b)
(a) 2 additions
symmetry
Uy=0
(two nodes)
node constraint
Uy=0
FIGURE 123 Single symmetry: (a) planar - (b) extrude.

Uz=0
(one node)
Symmetry-antisymmetry FIGURE 124 shows how symmetry, anti symmetry and nodal constraints applied
in 2D should be complemented in 3D applications by adding one nodal constraint
in the Y-direction.

Extrusion
(b)
(a) 1 addition
symmetry
Uy=0
nodal constraint (two nodes)
Uy=0
antisymmetry
FIGURE 124 Symmetry and antisymmetry: (a) planar -

(b) extrude.
4
Double antisymmetry FIGURE 125 shows how double anti symmetry plus a nodal constraint applied
in 2D should not be changed when extruding.
(a) (b)
antisymmetry no change
nodal constraint
Uy=0
FIGURE 125 Double antisymmetry: (a) planar -

(b) extrude. Uy=0
(one node)
Solution ID Select the Solution ID tab from the StressCheck Input dialog window and sup-
ply the following information:
CONST > Load ID: LOAD > Click on the Accept button.
By this action a unique solution name (SOL) has been assigned to the con-
straint (CONST) and load (LOAD) name pair created previously.

Extrusion
Execution You are now ready to start the computation. Click on the Compute Solution icon
from the Main Toolbar and when the Solution dialog window appears select the
Linear tab and complete the requested information as done before:
• Linear tab > Extension: Upward-p > p-limits: 1 to: 8.
tive > Convergence: None. Next, click on the Solve button. A sequence of
solutions of increasing polynomial order (from p=1 to p=8) will be
obtained.
Extraction of results The procedures for estimating the error in energy norm, plotting the data of interest,
etc., are the same as those described for the 3D analysis. Following the same steps,
the results shown in FIGURE 126 will be obtained.
FIGURE 126 Results for the Extrusion.
Note about thickness It is possible to have piecewise thickness assignments in 2D and then extrude. The
program takes care of the thickness transition automatically. The only consideration
is that the stresses in the transition region should be interpreted in an average sense.

Laminate Composite Analysis
In reality, if a sharp transition exists in 3D, then the stresses in that region
would be unbounded. In the extrusion these stresses are finite, however. The
smoothening is performed in an integral sense, so that the sharp transition does
not exist when extruding. Numerical testing on the effect of thickness transi-
tions in extrusion indicated that the stresses away from the transition are very
close to the stresses obtained solving a 3D-solid (non-extruded) model.
Summary As demonstrated by this very simple example problem, the Extrusion option
can be used for any problem defined in the 2D reference system. Once the
model is extruded, the loads and constraints can be edited before executing the
analysis. Care must be exercised in ensuring that the three-dimensional body is
properly constrained. For example, if in the pipe problem we did not enforce
the plane-strain condition, then a rigid body translation along the z-axis would
have been possible. To prevent rigid body displacement, it would have been
4 necessary to specify a nodal constraint in the z-direction (any node can be con-
strained).

Individual plies and sub-laminate properties for 3D analysis can be easily and
conviniently input in StressCheck. The material coefficients are entered in the
material directions and then assigned to elements, together with the ply orien-
tation, in either a local coordinate system or as mapped laminates. When a set
of plies needs to be combined in a single element (sub-laminate), the properties
of the sub-laminate are obtained by homogenization (See Laminate composites
for planar and 3D problems).
Provide the material coefficients of a single orthotropic ply in the direction of

the material axes (FIGURE 127a) by selecting the material type: Lamin.Ortho
under the Define tab.
Nine engineering coefficients (E11, E22, E33, G12, G23, G31, ν12, ν23, ν13) in
the direction of the material axes and the ply thickness must be provided. The
three coefficients of thermal expansion (α11, α22, α33) and the mass density
can also be entered, but they are not required unless there is thermal loading or
a modal analysis is needed.
After the material properties for a single ply are entered, select the Assign tab
to provide the ply group and stacking sequence information.

The ply group information includes the angular orientation of each ply in the group,
in accordance with the following convention: A positive ply angle (θ) is measured
as a counterclockwise rotation about the z-axis of a local coordinate system (xyz).
The z-axis of the local system is aligned with the material 3-axis as shown in FIG-
URE 127b.
z≡3 1
θ x
y
3 Y
1 X 2
2
‘plane-strain’ 4
plane
Z
The ply groups are assembled into a stack by providing the total number of layers in
the laminate, defining whether the stack is symmetric or not, and entering the stack-
ing sequence based on the ply group names. This procedure provides great flexibil-
ity, since various ply groups of different material properties can be assembled
together to represent a single sub-lamina.
Finally, the stack is assigned to the elements in the mesh by selecting the type :
Laminate-Stack, and identifying the local coordinate system, the z-axis of which is
assumed to coincide with the material 3-axis and must be perpendicular to the plane
of the ply (FIGURE 127b). The case of a single ply assigned to an element is
treated as a particular case of a sublaminate with one orthotropic layer.
It is also possible to assign laminated plies directly by selecting the type: Laminate-
Ply instead of having to assign a stack, which in turn references a ply definition.
This feature is particularly useful when the same material properties are used for all
the plies in the stack.
For a single ply, the type: Lamina can be used for assigning laminate material with-
out the need for defining the ply group or stacking sequence.

Assigning laminate properties
The assignment of laminate properties (single ply and sub laminate) in a cylin-
drical system will be illustrated with an example. Consider a 12-ply laminated
cylinder under internal pressure with the dimensions shown in FIGURE 128.
The stacking sequence is [0/90/0/90/…] starting from the inside of the cylin-
der, and the properties in the material directions are: E11=25x106,
E22=E33=1.0x106, G12=G13=0.5x106, G23=0.4x106, ν12=ν23=ν13=0.25,
t=0.01. Because of symmetry, only one-eighth of the cylinder is considered for
the analysis (a 45o sector). For a 0o-ply, the material 1-axis is in the θ-direc-
tion, the material 2-axis is in the z-direction, and the material 3-axis is in the
radial direction. For a 90o-ply, the material 1-axis is in the z-direction, the
material 2-axis is in the θ-direction, and the material 3-axis is in the radial
direction.
4
Two cases are considered: in Case 1, one element is included for each ply; in
Case 2, the central six plies are considered as a single element with homoge-
nized properties. Both finite element meshes are shown in FIGURE 128. These
models can be loaded from the Tutorial Handbook: Laminate.sci for Case 1
and Sublaminate.sci for Case 2.
FIGURE 128 Laminated

cylinder under internal

• To define and assign the material properties for Case 2, the following oper-
ations are required (refer to FIGURE 129):
FIGURE 129 Definition and

assignment of laminated
properties in a cylindrical system.
• Enter the material coefficients in the material directions by selecting the

Define Tab in the material input form using the type: ‘Lamin.Ortho’. The
name MAT was used in this example as shown in FIGURE 129.
• Create the ply groups by providing a Ply Group Name, the laminated mate-
rial ID and the ply layout. The definition of the ply group CENTRAL is

shown in FIGURE 129. Two more ply groups were created for this
problem: PLY0 and PLY90.
• Create the stacks by providing the Stack Sequence Name, the number
of layers (plies) and the stacking sequence (the names of the ply group
or groups created in the previous step). The definition of the stack
HOMO, used for the six homogenized central plies, is shown in FIG-
URE 129. Two more stacks were created for this problem: LAM0 and
LAM90.
• Assign a stack to a group of elements. In the Assign tab of the material
input, select the type: Laminate-Stack, select the stack ID, the coordi-
nate system, the corresponding color, and the elements of the mesh. In
the example shown in FIGURE 129, the stack HOMO is assigned to
the central element of the mesh in the cylindrical system SYS2 which
has an arbitrary orientation relative to the global system XYZ. Two
4 more assignments are required in the same local system: LAM0 is
assigned to the 0o-plies and PLY90 to the 90o-plies.
FIGURE 130 shows the estimated relative error in energy norm for cases 1 and
2 corresponding to a sequence of finite element solutions for polynomial
orders ranging from 1 to 8 (regular solids and trunk space). In both cases, the
estimated error is under 1% for p ≥ 3. The value of the estimated potential
energy when the central six plies are replaced by a homogenized layer (Case 2:
0.00339994) is very close to that of the case when each ply is explicitly
included (Case 1: 0.00341857). FIGURE 131 shows the deformed configura-
tion for each case. Note that the maximum radial displacement for Case 1
(7.31x10-5) is also very close to that of Case 2 (7.27x10-5).
Finally, FIGURE 132 shows the hoop-stress contours for cases 1 and 2. Note
that the maximum (1801) and minimum (48.9) values of the hoop stress for
Case 1 are very close to the corresponding maximum (1791) and minimum
(48.4) for Case 2. Both the maximum and the minimum are located away from
the central homogenized region of the cylinder.
Mapped laminates
A capability is available to align laminated orthotropic material properties with

the curvature of an element for both single plies and sub-laminates, given an
angle orientation or stacking sequence, respectively. Two approaches are avail-
able for the definition of laminate properties with general curvature: a general
mapped laminate (Lam-Map Volume), and an edge-based mapped laminate

FIGURE 130 Estimated relative error in energy norm for Case 1 (top) and Case 2 (bottom).
FIGURE 131 Deformed configurations for Case 1 (left) and Case 2 (right).

4 FIGURE 132 Hoop stress distribution for Case 1 (left) and Case 2 (right).
approach (Lam-Map Edge). In both options, the principal material directions

are computed based on the selection of element faces and edges from the finite
element mesh as shown in FIGURE 133.
Select element faces

Select element edges
FIGURE 133 Assignment of Lam-Map Edge and Lam-Map Volume options.

Lam-Map Edge In the Lam-Map Edge option, the material 1-axis follows the natural coordinate
defined by the direction of the selected edge and remains parallel to the edge at all
points in the domain of the element. At every point within an element a unit vector
is created parallel to the tangential direction of a selected element edge. The mate-
rial 1-axis will follow the natural coordinate defined by the direction of the selected
edge throughout the element. A second unit vector is created in the direction nor-
mal to a selected element face, in such a way that an angle ply will have its angular
orientation defined as a rotation about the second vector relative to the direction of
the first vector.
This option is applicable to situations in which laminated components are created

by stacking pre-preg plies. Material direction arrows throughout each element can
be visualized to verify that the ply angle orientations provided conform to the prob-
lem to be analyzed. This option is illustrated in FIGURE 134 for a single 0o-ply
when the selected edges are as shown in FIGURE 133. Note that the arrows repre- 4
FIGURE 134 Example of Lam-Map Edge option assigned to elements.
senting the material 1-axis direction remain parallel to the selected element edges

even though the opposite element edge of each element is not parallel with the
selected edge.
Lam-Map Volume In the Lam-Map Volume option, the material 1-axis follows the natural coor-
dinate defined by the direction of the selected edge and transitions uniformly
between element faces. At any given point within an element, a unit vector is
created tangent to the direction of a standard coordinate passing through the
point. There are three standard coordinate directions at any given point in the
element, and the one selected corresponds to the direction associated with a
given edge specified by the user. For a general curvature element, this direction
represents a general curve in 3D space. A second unit vector is created in the
direction normal to a selected element face. An angle ply will have its angular
orientation defined as a rotation about the normal direction (positive counter-
clockwise) and relative to the direction of the tangent vector. In other words,
4 the material 1-axis of the orthotropic property follows the natural coordinate
defined by the direction of the selected edge everywhere in the element.
This option is useful when modeling injection molded parts that include cut
fibers which orient themselves with the mold. This option is illustrated in FIG-
URE 135 for a single 0o-ply when the selected edges are as shown in FIGURE
FIGURE 135 Example of Lam-Map Volume option assigned to elements.

133. Note that the arrows representing the material 1-axis direction does not remain
parallel to the selected element edges and blend smoothly from the selected element
edge to the opposite element edge.
Assignment of mapped laminates

Mapped laminates can be assigned to hexahedral or pentahedral elements only.
Select the Material tab from the input interface, select then the Define sub tab, and
enter the material properties as LaminOrtho. Then select the Assign sub tab and
proceed as follows (FIGURE 136):
1. Select face 2. Control-Shift and select edge
FIGURE 136Input interface for the assignment of

mapped laminates.
• Select > Face (or Face Surface) > ID: select material from combo > Type:
select Lam-Map Vol or Lam-Map Edge > Angle: value of ply angle. Click
on the faces of the target elements using the left mouse button (the normal
to the face defines the axis of rotation for the ply definition), then hold Con-
trol-Shift and pick the edges on the same or opposite face of the faces
selected > Accept.

Thin Solids
The input field Angle can be the value of the ply angle in degrees, a user
defined parameter with the value of the ply angle in degrees, or a ply sequence
for sub-laminates. For example, FIGURE 136 shows an alternative input for a
14-ply sub-laminate instead of a single ply input. The angle input field is lim-
ited to 31characters.
When the object Face Surface is used, all the faces which are slope-continuous
to the one selected will be highlighted. When an edge is selected afterwards
while holding down the Control and Shift keys, all the edges that are slope con-
tinuous to it will be highlighted.
Visualization There are three options to facilitate visualization of laminated composite

attributes. These options are accessible in the pull-down menu found next to
the Display Materials icon in the main tool bar as shown in FIGURE 137. The
4 options are to display the material attributes by name, with arrows, or using
colors.
• If “Name” is selected, the material ID that has been assigned to each
element will be display at the centroid of the element.
• If “Arrows” is chosen, the principal direction of the material fiber will
be represented by arrows displayed at uniformly spaced points within
each element. The density of the arrows may be controlled in the View
Controls interface.
• If “Color” is selected, each element will be displayed using a color
from a user-defined color table, according to the angle of the principal
material direction relative to the global coordinate system or assigned
local coordinate system. The laminate color table may be defined in the
Assign tab of the Material Input Class interface.
In the example shown in FIGURE 137, the material 1-direction arrows are dis-
played over the elements.
Thin Solids
Thin solid elements are formulated into StressCheck to allow independent con-
trol of the p-discretization in the transverse (‘thickness’) and in-plane direc-
tions. This allows substantial savings in computational time, without
compromising the quality and accuracy of solution, when laminated compos-
ites are modeled.

Thin Solids
FIGURE 137 Material 1-direction arrow display for mapped laminates
Two types of thin solids elements are available: Hexahedral and pentahedral ele-
ments. In the case of hexahedral elements, any of the three possible directions can
be selected to represent the ‘thin direction’. In the case of pentahedral elements,
only the direction connecting the triangular faces can be declared as the ‘thin direc-
tion’.
Degrees of freedom The number of degrees of freedom associated with a thin solid element depends on
the polynomial order in the ‘in-plane’ uv-direction (p-level) and on the polynomial
order in the ‘thin’ w-direction (q-level). For q = p the element is a regular solid, and
for q fixed and p variable the element is a thin solid. The number of degrees of free-
dom (DOF) equals the number of shape functions per field (n) multiplied by the
number of fields (3 in three-dimensions): DOF = 3n. The number of degrees of
freedom for different p-levels and values of q is shown in Table 4 for the trunk and
product spaces of thin hexahedra and for the trunk space of pentahedral elements.

Thin Solids
Table 4: Number of Degrees of Freedom (DOF) for thin solid elements

DOF for thin solids
p-level Hexahedra Pentahedra
Trunk Space Product Space Trunk Space
q=p q=1 q=2 q=p q=1 q=2 q=p q=1 q=2
1 24 24 36 24 24 36 18 18 27
2 60 48 60 81 54 81 45 36 45
3 96 72 84 192 96 144 78 60 69
4 150 102 126 375 150 225 126 90 108
4 5 222 138 174 648 216 324 192 126 156
6 315 180 231 1029 294 441 279 168 213
7 432 228 297 1536 384 576 390 216 279
8 576 282 372 2187 486 729 528 270 354
q-discretization Any fixed value of q between 1 and 8 can be assigned to a thin-solid element.
The ‘thin’ direction is determined based on the input provided by the user in
the p-discretization interface. By selecting the face of an element, the thin
direction is understood to be in the direction normal to the selected face. This is
illustrated FIGURE 138 which shows the setting of the p-discretization inter-
face and the element faces selected to indicate the thin-direction. In this exam-
ple, the q-level was fixed at 1 in the direction perpendicular to the selected
faces.
Multiple p-discretization assignments are possible for a given problem. Some

elements may be assigned a value of q=1, others a different value of q, and still
others be treated as regular solids. The rule is based on the object selection in
the p-discretization interface. The default record is an ALL record, in other
words all the elements in the mesh are assigned p=1 variable (regular solid). If
a SET record is added, the new record will apply to the selected elements, and
anything not included in the SET record will be controlled by the ALL record.
For example, consider the problem shown in FIGURE 139 that represents a
sector of a pipe composed of six thin plies, three at each side of a thicker core.
The p-discretization interface contains 3 records: One ALL record and two
SET records. Note that one set record applies to the three plies above the core

Thin Solids
FIGURE 138 p-Discretization interface to

assign q-level to thin solids.
and the other record to the three plies below the core. No SET record is available
for the element of the core, therefore the ALL record applies to it.
When using the Select/Face or Select/Face Surface, there is a checkbox called

“Multi-layer”. If checked, then StressCheck automatically selects the correspond-
ing face of every element that is connected in attached layers. If a blank element is
encountered, the propagation will be interrupted. NOTE: There can be unintended
consequences when multi-layer face propagation is performed in a curved area,
because the propagation may not be appropriate.
Example Laminated strip under sinusoidal load: Consider a 16-ply [4(0/90)]S simply sup-
ported (a/h=20) laminated composite under plane strain conditions subjected to a
sinusoidal traction as shown in FIGURE 140. Each ply is of the same material and
thickness (t=0.0625 mm) with the following properties:
EL = 1.38x105 MPa, ET = 9.3x105 MPa, GLT = 4.6x105 MPa, GTT = 3.1x105 MPa

Thin Solids
FIGURE 139 Multiple p and q-discretization records for the same problem.
νLT = 0.3, νTT = 0.5
where L indicates the direction parallel to the fibers and T is the transverse
direction.
FIGURE 140 Laminated strip under sinusoidal traction. Notation.

Thin Solids
Because of symmetry, only half of the beam was considered for the analysis. The
mesh shown in FIGURE 141 consists of 16 extruded quadrilateral elements, one for
each ply.
4
FIGURE 141 Laminated strip: Mesh and boundary conditions.
Plane strain conditions were enforced in 3D by prescribing zero normal displace-

ments to the front and back faces of the 5-units deep strip (FIGURE 142). Half of
the load was applied to the top surface and the other half to the bottom surface to
preserve the anti-symmetry conditions.
FIGURE 142 Laminated strip mesh. 3D view.

Thin Solids
The 2D problem description can be loaded from the Tutorial Handbook: Beam-
Laminate.sci. After loading the problem, extrude it, add the constraints in the
Z-direction to simulate plane-strain conditions and then specify the q-level to
the elements as described before.
This problem represents an infinite strip under sinusoidal normal traction and
therefore it is free from any boundary layer effects. The solution is first
obtained when each element is considered as a regular solid and uniform p-
extension is used. The results are then compared with those corresponding to a
uniform p-extension using thin solids with q=1, q=2 and q=3. In all cases the
direct solver option in the StressCheck solver interface was used and the results
include:
• The estimated relative error in energy norm.
• The convergence of the vertical displacement Uy at x=0 as a function
4 of the number of degrees of freedom (DOF).
• The contour plot corresponding to the normal stress σx displayed over
the deformed shape.
• The through-thickness normal stress distribution σx along one edge on
the symmetry face (x=0 in FIGURE 141).
• The through-thickness shear stress distribution τxy along one edge on
the anti-symmetry face (x=a/2 in FIGURE 141).
The results are shown in FIGURE 143 for the solution corresponding to regu-
lar solids. FIGURE 144 shows the results for the thin solid with q=1, q=2 and
q=3.
The results clearly show the ability of the thin solids to approximate the three-
dimensional problem as the q-level is increased. The global behavior (energy,
deflection) of the laminated strip and the local behavior (in-plane normal stress
σx) is well represented by the thin solids with q=1. Note, however, that the
transverse shear stress is not represented well by the q=1 model, except in the
average sense. For the case q=1, the shear stress distribution is piecewise con-
stant. This is because the approximation of the displacement in the transverse
direction is linear, and therefore the transverse shear strain is constant within
each layer. Because the material properties change at the interface between
plies, the shear stress distribution is piecewise constant. For the cases q=2 and
q=3, the shear stress distribution is practically identical to the one obtained for
the regular solids.

Thin Solids
FIGURE 143 Laminated strip. Results for regular solids (q=p).

Thin Solids
FIGURE 144 Laminated strip.

Results for thin solids.

Thin Solids
Table 5 presents a summary of the results for p=8. The first column shows the
model order as determined by the q-level. The last entry in the column corresponds
to the regular solid. The second column shows the corresponding degrees of free-
dom (DOF) while the third column is the total potential energy of the solution. The
fourth column shows the vertical deflection at the center of the strip Uy(0,-h/2,0)
and the fifth column is the normal stress σx(0,-h/2,0) at the bottom surface in the
symmetry plane.
Table 5: Results for laminated strip at p=8
Model DOF Potential Uy(0,-h/2,0) σx(0,-h/2,0)

Energy
q=1 1666 -3.010570 -0.24085 394.6
q=2 2578 -3.011574 -0.24093 394.2
4
q=3 3234 -3.011574 -0.24093 394.2
p=q 4274 -3.011574 -0.24093 394.2
Selecting the q-level The minimum q-level that should be assigned to a thin solid is problem-dependent.
It depends on the topology, type of loading and the goals of the computation. It is
recommended that for every new class of problem being considered, a q-extension
be performed to ascertain that proper convergence is achieved. In other words, run
a p-extension for a fixed q-level and repeat for a different q-level. Check the sensi-
tivity of the results to increasing values of the q-level.
The following guidelines are intended to provide some a-priori information in

selecting the proper q-level:
• For membrane dominated problems q=1 is typically sufficient.
• For bending dominated problems q=2 or 3 may be required depending on
the type of problem. For laminated composites where several plies are
explicitly included in the mesh, q=1 is sufficient if details of the transverse
stress distribution are not of interest. Otherwise, q greater or equal to 2
should be used.

Global-Local Analysis
Global-local analysis refers to the possibility of importing point loads or point
constraints from a subdomain of a global shell model and their application to a
StressCheck solid model for a detail analysis. These two importation
approaches will be addressed in the following.
Importation of point loads

A capability is available to import point load information (forces and moments
per unit length) from a global model into StressCheck, assign these loads to
element faces, calculate interpolated traction functions on those faces, and
incorporate the tractions into a load case for use in the finite element solution
4 process. A description of the implementation and examples are included in the
following.
Basic functionality The functionality allows for the visualization of the imported load data
obtained from the solution of a global shell model, and their assignment to a set
of selected element faces. Once the global forces and moments are assigned to
element faces, interpolating functions are calculated for each force (Fx, Fy, Fz)
and moment (Mx, My, Mz) component. Both the point forces and moments are
understood to be per unit length and in the global coordinate directions, and
are positive when in the direction of the positive coordinate axes (FIGURE
145). Once the data points are assigned to the local mesh, the quality of the
continuous fitted functions over the imported data can be assessed by perform-
ing a quality check.
The object type point load is specified by: Object number, Case ID, global
coordinates (X, Y, Z), three force components (Fx, Fy, Fz), three moment com-
ponents (Mx, My, Mz), and the shell thickness. The resultant force and
moment associated with a point load object can be displayed using two scaled
resultant vectors, one for the resultant force and one for the resultant moment.
Assignment of multiple load cases is also possible.
The procedure to compute continuous traction functions is described in the fol-

lowing. The quality of fitting is evaluated by a least square error measure asso-
ciated with each component.
Point load fitting There are two basic steps in converting point load data into continuous traction
functions needed by StressCheck: (a) Fitting a curve over the spatial distribu-

Y Fy My
X Fx Mx
Z Fz Mz
Global Coordinate System Positive forces Positive moments
FIGURE 145 Sign convention for point forces and moments.

4
tion of the selected data points, and (b) fitting the force/moment components as a
function of a single parameter along the fitted curve.
(a) Curve fitting: First, the spatial distribution of the data points is approximated
with a space curve using a spline representation as a function of a single parameter
s. Let XP , YP , ZP be the global coordinates of an arbitrary point P, from the N
points imported from the global model to be associated with a group of faces of the
local model. Then the parametric representation of the curve fitted through the
points can be written in vector form as follows (FIGURE 146):
r ( s ) = x ( s )i + y ( s )j + z ( s )k EQ 1
where i , j , k are the unit vectors in the direction of the global X, Y and Z axes
1 2
Z P
s N
r(s)
X Y
FIGURE 146 Curve fitting through data points as a function of a single parameter s.
respectively. Mapping functions are computed (and the corresponding inverse map-

ping functions) relating a point on the spline curve with the global coordinates
of that point:
x = x ( s ), y = y ( s ), z = z(s) EQ 2
The algorithm to fit the spline curve is based on a Parasolid internal function
that requires the points to be ordered from one end to the other end of the target
curve. The ordering may be determined in two ways: the ordering can be spec-
ified by the user, or it can be determined automatically by StressCheck based
on the distance between neighboring points. To order automatically, the pro-
gram begins with an arbitrary first point and finds the closest point to it. This
point becomes the next point in the sequence and the new end point of the
order. This procedure continues, adding new points to either end of the ordered
list of points until all points have been added to the list. The ordered list of
4 points is used for constructing a spline curve using a standard Parasolid func-
tion. The curve is displayed on the screen for visual feedback to the user. The
result of the fitting is the set of three parametric equations of the spline given in
EQ 2. For a given value of the parameter – 1 ≤ s ≤ 1 , it returns the correspond-
ing global coordinates (x, y, z).
(b) Force/Moment fitting: Next, the force and moment components per unit
length and the shell thickness are approximated in the parameter space of the
curve using least squares fitting. Let h(s) represent a force, moment or shell
thickness along the curve, then h(s) can be written as a linear combination of
known polynomial functions φi(s) multiplied by unknown coefficients ai:
M
h(s) =
∑ a φ (s)
i i EQ 3
i=1
Let hˆk ( x k ,y k ,z k ) be the known value of the function at point k. Because of the
relations given by EQ 2 and their inverse, the known value in global coordi-
nates can be mapped into the parameter space of the curve:
hˆk ( x k ,y k ,z k ) → hˆk ( s k ) . The unknown coefficients ai in EQ 3 are determined by
minimizing the square of the difference between the given data points and the
approximating function:
 N 
∂  ˆ ( s ) – h( s ) ]2  = 0
∂ a j  ∑ [ h k k k 

EQ 4
k=1
 

Substituting EQ 3 into EQ 4 and performing the indicated operations, a linear sys-

tem of equations in the unknown coefficients ai is obtained:
N M N
ˆ
∑∑ a i φ i ( s k )φ j ( s k ) = ∑ h ( s )φ ( s ),
k k j k j = 1 ,… , N EQ 5
k = 1i = 1 k=1
where N is the number of data points to be fitted along the curve, M is the polyno-
mial order of the fitting and φi(s) are one-dimensional polynomial functions in the
range – 1 ≤ s ≤ 1 given in Ref. [3]. The quality of fit is determined by computing the
relative error in the least squares sense, defined as follows:
N
ˆ
∑ [h (s ) – h(s )]
2
k k k
k=1 EQ 6 4
e(%) = -------------------------------------------------- × 100
N
hˆ ( s )
∑
2
k k
k=1
Once the polynomial fitting for the space curve x ( s ) connecting the data points and
the three forces F x ( s ) , F y ( s ) , F z ( s ) , three moments M x ( s ) , M y ( s ) , M z ( s ) and
thickness h(s) are available, the traction components in the global system at any
point on the face of an element are computed as follows (FIGURE 147):
• Project the point of interest Q(x, y, z) onto the curve x ( s ) .
• Find the global coordinates (x’, y’, z’) of the projected point Q’(s).
• Compute the traction components Tx, Ty, Tz at point Q as:
Fx( s ) 12 12
T x ( x, y, z ) = ------------
- – ------------ M ( s ) ( y – y′ ) + ------------ M y ( s ) ( z – z′ )
h ( s ) h3 ( s ) z 3
h (s)
Fy ( s ) 12 12
T y ( x, y, z ) = ------------
- – ------------ M ( s ) ( z – z′ ) + ------------ M z ( s ) ( x – x′ )
h ( s ) h3 ( s ) x 3
h (s)
Fz ( s ) 12 12
T z ( x, y, z ) = ------------
- – ------------ M ( s ) ( x – x′ ) + ------------ M x ( s ) ( y – y′ )
h ( s ) h3 ( s ) y 3
h (s)
EQ 7

Tz
Q(x,y,z)
x( s) Tx
Ty
Z Fz
Q’(s)
Mz
Q’
Fx Fy
Q’
X Y Mx My
FIGURE 147 Traction components at a point on an element
4
The global tractions Tx, Ty, Tz are used during the solution of the local model at
the required integration (Gauss) points on the element faces to determine the
corresponding load vector terms.
Data format The format of the point load record (*4410) for the StressCheck input file is as
follows.
Sample record:
*4410
1 ,0 ,case1 ,0 ,0 ,0 ,1.0000000e+000 ,0.0000000e+000 ,0.0000000e+000
0.0000000e+000 ,1.0000000e-001 ,0.0000000e+000
1.0000000e+000 ,1.2000000e+000 ,1.399999e+000
1.0000000e-001 ,2.0000000e-001 ,3.000000e-001
Description:
number ,0 ,caseID ,0 ,0 ,0 ,thickness ,0.0 ,0.0
x ,y ,z
Fx ,Fy ,Fz
Mx ,My ,Mz
number - integer number assigned to a particular point load record. The num-
ber must be unique, independent of Case ID.
Case ID - character string up to 15 characters which identifies the global load
case.
thickness - shell thickness at location of data point.

x, y, z - location of point load in global coordinates (single precision).

Fx, Fy, Fz - Force components per unit length (single precision).
Mx, My, Mz - Moment components per unit length (single precision).
All other values shown as zeros are for possible future use, and should be set to
zero. All entries should be followed by a blank, and then followed by a comma as
shown. The real values may be given in scientific or in fixed decimal format.
User Interface Once the point load information is imported into StressCheck using the File >
Append Input option and the local solid model is available, select the Load tab of
the Model Input interface to assign a set of point load objects to a group of faces on
the local model.
When the object Face or Face Surface is selected, two methods are available for the
4
assignment of the point load data: Fit-Auto and Fit-Manual, as shown in FIGURE
148.
• Fit-Auto: The selected point load data will be automatically ordered by the
program from one end to the other end of the target curve. This allows the
use of the marquee-pick option to select a group of point load objects.
• Fit-Manual: The curve fitted through the spatial location of the point load
data will be performed following the pick order. This allows the user to
bypass the automatic ordering process in the event that it does not perform
as expected.
Once the name of the load has been entered, simply select from the list of Case ID’s
to enable the display of associated point load data. As shown in FIGURE 148, point
load objects are shown as small triangles on the screen. The display options for the
selected Case ID are:
• Locations: Shows input point load data on screen (triangles).
• Symbols: Shows 2 vectors per point with the resultant input force (single
arrow head) and moment (double arrow head).
• Labels: point load object number.
Object Creation: The Edit Definitions button shown in FIGURE 148 enables the
table shown in FIGURE 149 for the modification of existing input data or the cre-
ation of new point load objects. The first column in the table is the case ID, the sec-
ond is the object number, and the third column (Status) is used for indicating
whether the point load object has been assigned to a load ID and the number of
times it has been used. Columns 4 to 6 contain the global coordinates of the point
load record (X, Y, Z), while columns 7 to 12 contain the three forces (Fx, Fy, Fz)

spline curve fitted through data points
point load objects
point load symbols
FIGURE 148 Load interface to assign point load data to element
and three moments (Mx, My, Mz). The last column shows the shell thickness
used in the global model.
To add a new point load object, simply enter values into the text fields in the
edit region of the window, and click Add. The new record will appear in the
table shown in the bottom portion of the window. The window may be
stretched to view the entire table. To edit an object definition, simply click on
the object’s entry in the table and the corresponding data will be transferred to
the input fields in the edit region of the window. Update the values and click
the Replace button to record the changes. Note that the display symbol corre-
sponding to the point load object record selected in the table will be high-
lighted in the model window.

Edit Region
FIGURE 149 Interface to edit point load data.
To select objects in the table, just click anywhere on the corresponding row of the
table. The row will be highlighted in red, and the corresponding object will be high-
lighted in the model window. Use the arrow key to move up or down in the table,
and simultaneously transfer the data to the input fields and highlight the object in
the model window. The object will remain highlighted until the Cancel action
is selected.
To delete an object definition, simply click on the object’s entry in the table then
click the Delete button. To delete multiple objects at one time, hold the SHIFT key
while selecting records to highlight multiple object in the table, then click Delete.
To purge all object definitions, click the Purge button. If you purged all the data
points by mistake and want to restore them back, click the Undo button .
When load assignment records are selected from the list box at the top of the load
input window, the corresponding faces and point load objects will be highlighted
automatically in the model window. The corresponding point load records will also

be highlighted in red in the table in the lower portion of the Case Definition
window. Use the <Previous and Next> buttons to automatically locate the next
(or previous) highlighted record in the table.
Object Assignment: To create a load record based on point load data, select
the faces of the elements by pointing and clicking with the left button of the
mouse. Next, while holding down the Ctrl and Shift keys of the keyboard
simultaneously, select the point load objects from the screen using the left but-
ton the mouse. Objects may be selected one at a time, or using the marquee
pick. Finally, click the Accept button. The spline curve fitted through the data
points created automatically by the program will be displayed as shown in
FIGURE 148. Note that there are two different spline curves, because two
records were created. When the boundary of the local solid model is piecewise
continuos, one load record should be created for each boundary segment.
4
To change an assignment record, click the record shown in the list box at the
top of the Load input window. The corresponding faces and point load objects
will be highlighted in the model window. To change faces or point objects, use
the standard object selection techniques. When finished, click the REPLACE
button.
Checking the Quality of Fitting: To check the quality of fit of the forces,
moments and thickness, simply select the action Check from the combo box in
the Load interface and the object Face and click on the Accept button. The %
relative error (see EQ 6) for each force and moment component and the thick-
ness for each set# and for each Load ID are tabulated in a report window as
FIGURE 150 Typical quality of fitting report.

Importation of point constraints

The functionality allows for the visualization of the imported input displacements
and rotations obtained from the solution of a global shell model, and their assign-
ment to a set of selected element faces. Once the global displacements and rotations
are assigned to element faces, interpolating functions are calculated for each dis-
placement (Ux, Uy, Uz) and rotation (Rx, Ry, Rz) component. Both the displace-
ments and rotations are understood to be in the global coordinate directions, and are
positive when in the direction of the positive coordinate axes. Once the data points
are assigned to the local mesh, the quality of the continuous fitted functions over
the imported data can be assessed by performing a quality check. The calculated
displacement functions can be assigned to any constraint case.
Basic functionality The object type point constraint is specified by: object number, case ID, global 4
coordinates (X, Y, Z), three displacements components (Ux, Uy, Uz), and three
rotation components (Rx, Ry, Rz) in the global coordinate directions. The displace-
ment and rotations associated with a point constraint object can be displayed using
two scaled resultant vectors, one for the displacements and one for the rotations.
The procedure to compute continuous displacement functions is described in the

following. The quality of fitting is evaluated by a least square error measure associ-
ated with each displacement component.
Displacement fitting There are two basic steps in converting point constraint data into continuous dis-
placement functions needed by StressCheck: Fitting a curve over the spatial distri-
bution of the selected data points, and fitting the displacement/rotation components
as a function of a single parameter along the curve.
Curve fitting: First, the spatial distribution of the data points is approximated with
a space curve using a spline representation as a function of a single parameter s. Let
XP, YP, ZP be the global coordinates of point P, of the N points imported from the
global model to be associated with a group of faces of the local model. Then the
parametric representation of the curve fitted through the points is (FIGURE 151):
r ( s ) = x ( s )i + y ( s )j + z ( s )k EQ 8
where i ,j ,k are the unit vectors in the direction of the global X, Y and Z axes
respectively. Therefore mapping functions are computed (and the corresponding
inverse mapping functions) that relate a point on the spline curve to the global coor-
dinates of that point:

1 2
Z P
s N
r( s)
FIGURE 151 Curve fitting through data points
as a function of a single parameter s.
X Y
x = x ( s ), y = y ( s ), z = z(s ) EQ 9
4
The algorithm to fit the spline curve is based on a Parasolid internal function
that requires the points to be ordered from one end to the other end of the target
curve. The ordering may be determined in two ways: The ordering can be spec-
ified by the user, or can be determined automatically based on the distance
between neighboring points. To order automatically, the program begins with
an arbitrary first point and finds the closest point to it. This point becomes the
next point in the sequence and the new end point of the order. This procedure
continues, adding new points to either end of the ordered list of points until all
points have been added to the list. The ordered list of points is used for con-
structing a spline curve using a standard Parasolid function. The curve is dis-
played on the screen for visual feedback to the user.
Displacement/Rotation fitting: Next, each displacement and rotation compo-

nent is approximated in the parameter space of the curve using least squares
fitting. Let u(s) represent a displacement or rotation component along the
curve, then it is possible to write u(s) as a linear combination of known polyno-
mial functions φi(s) multiplied by unknown coefficients ai:
M
u(s) = ∑ a φ (s)
i i EQ 10
i=1
Let uˆk ( x k ,y k ,z k ) be the known value of the function at point k. Because of

the relations given by EQ 9 and their inverse, uˆk ( x k ,y k ,z k ) → uˆk ( s k ) . The
unknown coefficients in EQ 10 are determined by minimizing the square of the
difference between the data points and the approximating function:

 N 
∂  2
∂ a j 
ˆ
∑
[ uk ( sk ) – u ( sk ) ]  = 0

EQ 11
k = 1 
Substituting EQ 10 into EQ 11 and performing the indicated operations, a linear
system of equations in the unknown coefficients ai is obtained:
N M N
ˆ
∑∑ a i φ i ( s k )φ j ( s k ) = ∑ u ( s )φ ( s ),
k k j k j = 1 ,… , N EQ 12
k = 1i = 1 k=1
where N is the number of data points to be fitted along the curve, M is the polyno-
mial order of the fitting and φi(s) are one-dimensional polynomial functions in the
range – 1 ≤ s ≤ 1 given in Ref. [3]. The quality of fit can be determined by com-
puting the relative error in the least squares sense, defined as follows: 4
N
ˆ
∑ [u (s ) – u(s )]
2
k k k
k=1
e(%) = -------------------------------------------------- × 100 EQ 13
N
uˆ ( s )
∑
2
k k
k=1
Once the polynomial fitting for the space curve x ( s ) connecting the data points, the
three displacement u x ( s ) , u y ( s ) , u z ( s ) , and three rotations R x ( s ) , R y ( s ) , R z ( s )
are available, the displacement components in the global system at any point on the
face of an element are computed as follows (FIGURE 152):
• Project the point of interest Q(x, y, z) onto the curve x ( s ) .
• Find the global coordinates (x’, y’, z’) of the projected point Q’(s).
• Compute the displacement components Ux, Uy, Uz at point Q as:
U x ( x, y, z ) = u x ( s ) – R z ( s ) ( y – y′ ) + R y ( z – z′ )
U y ( x, y, z ) = u y ( s ) – R x ( s ) ( z – z′ ) + R z ( x – x′ )
U z ( x, y, z ) = u z ( s ) – R y ( s ) ( x – x′ ) + R x ( y – y′ )
EQ 14

Uz
Q(x,y,z)
x( s) Ux
Uy
Z uz
Q’(s)
Rz
Q’
ux uy Q’
X Y Rx Ry
FIGURE 152 Displacement components at a point on an element face.
4
The global displacements Ux, Uy, Uz are used during the solution of the local
model at the required Gauss points on the element faces to determine the con-
straint equations to be applied to the global stiffness matrix of the problem.
Format The format of the point constraint record (*3810) to be incorporated into a
StressCheck input file (*.sci) is as follows.
Sample record:
*3810
1 ,0 ,case1 ,0 ,0 ,0 ,0.0000000e+000 ,0.0000000e+000 ,0.0000000e+000
0.0000000e+000 ,1.0000000e-001 ,0.0000000e+000
1.0000000e+000 ,1.2000000e+000 ,1.399999e+000
1.0000000e-001 ,2.0000000e-001 ,3.000000e-001
Description:
number ,0 ,caseID ,0 ,0 ,0 ,0.0 ,0.0 ,0.0

x ,y ,z
Ux ,Uy ,Uz
Rx ,Ry ,Rz
number - integer number assigned to a particular point constraint record. The
number must be unique, independent of Case ID.
Case ID - character string up to 15 characters which identifies the load case.
x, y, z - location of point constraint in global coordinates (single precision)

Ux, Uy, Uz - displacement components of the point constraint (single precision)

Rx, Ry, Rz - rotation components of the point constraint (single precision)
All other values shown as zeros are for possible future use, and should be set to
zero. All entries should be followed by a blank, and then followed by a comma as
shown. The real values may be given in scientific or in fixed decimal format.
User Interface Once the point constraint information is imported into StressCheck and the local
model is available, the user selects the Constraint tab of the Model Input interface
to assign a set of point constraint objects to a group of faces of the local model.
When the object Face or Face Surface is selected, two methods are available for the
assignment of the point constraint data: Fit-Auto and Fit-Manual, as shown in FIG-
URE 153.
4
spline curve fitted through data points
point constraint objects
point constraint symbols
FIGURE 153 Constraint interface to assign point constraint data to element faces.

Fit-Auto: The selected point constraint data will be automatically ordered by

the program from one end to the other end of the target curve. This allows the
use of the marquee-pick option to select a group of point constraint objects.
Fit-Manual: The curve fitted through the spatial location of the point con-
straint data will be performed following the pick order. This allows the user to
bypass the automatic ordering process in the event that it does not perform as
expected.
Once the name of the constraint has been entered, the user can select from the
list of Case ID’s to enable the display of associated point constraint data. As
shown in FIGURE 153, point constraint objects are shown as small circles on
the screen. The display options for the selected Case ID are:
Location: Shows input points-displacement data on screen (circles).

4
Symbols: Shows 2 vectors per point with the resultant input displacement (sin-
gle arrow head) and rotation (double arrow head).
Labels: point constraint object number.
Object Creation: The Edit Definitions button enables the interface for modi-
fying existing input data or creating new point constraint objects (FIGURE
154). The first column in the table is the case ID, the second is the object num-
ber, and the third column (Status) is used for indicating whether the point con-
straint object has been assigned to a constraint case and the number of times it
has been used. Columns 4 to 6 contain the global coordinates of the point con-
straint record (X, Y, Z), while columns 7 to 12 contain the three displacements
(Ux, Uy, Uz) and three rotations (Rx, Ry, Rz).
To add a new point constraint object, simply enter values into the text fields in
the top portion of the window, and click Add. The new record will appear in
the table shown in the bottom portion of the window. The window may be
stretched to view the entire table. To edit an object definition, simply click on
the object’s entry in the table and the corresponding data will be transferred to
the input fields in the top portion of the window. Update the values and click
the Replace button to record the changes. Note that the display symbol corre-
sponding to the point constraint object record selected in the table, will be
highlighted in the model window.
To select objects in the table, just click anywhere on the corresponding row of
the table. The row will be highlighted in red, and the corresponding object will

FIGURE 154 Interface to edit point constraint data.
be highlighted in the model window. Use the arrow key to move up or down in the
table, and simultaneously transfer the data to the input fields and highlight the
object in the model window.
To delete an object definition, simply click on the object’s entry in the table then
click the Delete button. To delete multiple objects at one time, hold the Shift key
while selecting records to highlight multiple object in the table, then click Delete.
To purge all object definitions, click the Purge button.
When constraint assignment records are selected from the list box at the top of the
Constraint input window, the corresponding faces and point constraint objects will
be highlighted automatically in the model window. The corresponding point con-
straint records will also be highlighted in red in the table in the lower portion of the

Case Definition window. Use the <Previous and Next> buttons to automati-
cally locate the next (or previous) highlighted record in the table.
Object Assignment: To create a constraint record based on point constraint

data, the user should select the faces of the elements by pointing and clicking
with the left button of the mouse. Next, while holding down the Ctrl and Shift
keys simultaneously, select the point constraint objects from the screen using
the left button the mouse. Objects may be selected one at a time, or using the
marquee pick. Finally, click the Accept button. The spline curve fitted through
the data points created automatically by the program will be displayed as
To change an assignment record, click the record shown in the list box at the
top of the Constraint input window. The corresponding faces and point con-
straint objects will be highlighted in the model window. To change faces or
4
point objects, use the standard object selection techniques. When finished,
click the Replace button.
Checking Quality of Fitting: To check the quality of fit of the displacements

and rotations, simply select the action Check from the combo box in the con-
straint interface and click on the Accept button. The % relative error (see EQ 6)
for each displacement and rotation component and for each Constraint ID are
tabulated in a report window as shown in FIGURE 155.
FIGURE 155 Typical error of fitting report.
Model Problem: Plate with a T-stiffener

FIGURE 156 shows the dimensions of the plate with a T-stiffener terminating
at a distance from the loaded edge. The skin is 0.08” thick and the stiffener web
and flange are 0.11” thick. The material properties and boundary conditions are

also shown in the figure. This problem will be analyzed using both global-local
Material: E=10x106 psi, ν=0.3
FIGURE 156 Plate with T-stiffener.

Dimensions and boundary
conditions.
techniques.
Global shell model The global model was prepared in PATRAN using 4-noded shell elements. After
the solution was obtained using ABAQUS, the grid point forces and moments in
the local region shown in FIGURE 157 were extracted and converted to forces/

moments per unit length before importing them into StressCheck. The resultant
FIGURE 157 Local region where the nodal forces and moments were extracted.
nodal forces and moments were converted into forces and moments per unit
length as follows (FIGURE 158):
• Skin nodes: The nodal force/moment is divided by the sum of one half
of the length of the element edge at each side of the node and assigned
to the node as a force/moment per unit length.
• Flange or web nodes: The same rule as for skin nodes applies to the
interior nodes but for the corner nodes (one at each side of the stiffener
web and one at the top of the stiffener flange), only half of one element
edge is used to calculate the force/moment per unit lenght.
Global solid model To have a reference solution, the complete model was also analyzed in Stress-
Check using thin solids (q=3) and performing a uniform p-extension for p=3 to
8. The details of the model creation are omitted here, but the file can be loaded
from the Tutorial Handbook: T_StiffenerGlobalModel.sci.
The results of the analysis include the global estimated error in energy norm,
the displacement contour over the deformed shape, and the contour plot of the
first principal stress as shown in FIGURE 159. Because there is a strong singu-

All interior nodes and corner nodes in skin:
Pk stiffener
ai+1 F k = ----------------------------
- ai+1
ak – 1 + ak + 1
Pi Pi
ai-1 i ai-1 i
skin
Corner nodes in web or flange:
Pk
aj+1 F k = ----------
-
aj+1
j ak + 1 j
Pj aj-1 Pj
4
FIGURE 158Rule for converting nodal forces/moments (P) into force/
moments per unit length (F).
larity at the stiffener termination, the maximum stress is unbounded and the plot-
ting range was limited to 0-100,000.
Additionally, the average stress Seq over the faces of the three elements around the
stiffener termination on the skin was computed for all p-levels and the results are
shown in FIGURE 160. As can be seen in the figure, the average stress converges
to a limiting value of Seq=38,700 over an area of 0.33, even though the stress along
the edge at the stiffener termination is unbounded.
Local solid model with point loads
A local model was built using thin solid elements and the mesh is shown in FIG-
URE 161. Note that a layer of elements was created around the boundary of the
local model where the tractions will be applied. This problem can be loaded from
the Tutorial Handbook: T_StiffenerLocalTraction.sci.
The converted point load information from the global shell model was imported
into StressCheck and assigned to the faces of the local solid model. A total of six
load records were created, one for each side of the skin, one for the stiffener web
and one for the stiffener flange as shown in FIGURE 162. Note that the point
located at the intersection of the stiffener web and flange is assigned twice, one to
the web faces and the other to the flange face.

(a)
(b)
4
(c)
FIGURE 159 Results from the global solid

model. (a) Error estimate, (b)
Displacement contour and deformed
shape for p=8, (c) First principal stress
contour (range 0-100,000) for p=8.
The quality of fit and the overall equilibrium of the fitted tractions is shown in
FIGURE 163. As can be seen, the error in the least squares fitting is significant
for some of the components, and the overall balance of forces and moments is
reasonable with some unbalance in the forces Fx and Fz and the moment My.

FIGURE 160 Global solid model: Average stress over the faces of three elements in the
stiffener termination region.
FIGURE 161 Mesh for the local

solid model.
For that reason, distributed spring constraints were specified in a few faces instead
of using the rigid body constraints as would be appropriate for a perfectly equili-

Point load assigned to the

skin element faces

stiffener flange element faces

stiffener web element faces
FIGURE 162 Load records created with the imported point load information.
brated system. Very soft directional spring were used to minimize the reaction
at these locations as shown in FIGURE 164.
The solution of the local model was obtained by performing p-extension from
p=3 to 8 and the results are shown in FIGURE 165. The overall deformation
and the stress distribution are similar to those obtained for the global solid
model and shown in FIGURE 159. Finally, FIGURE 166 shows the average
equivalent stress over the faces of the three elements around the stiffener termi-
nation. The results again are very close to those computed for the full solid glo-
bal model and shown in FIGURE 160.

(a)
4
(b)
FIGURE 163 (a) Error in the fitting of the point load data for the local solid model, and
(b) Equilibrium check of the fitted tractions.
Kz=100
Kx=1000
Ky=Kz=100
FIGURE 164 Spring constraints for the local solid model.

(a)
4 (b)
(c)
FIGURE 165 Results from local solid model. (a) Error estimate, (b) Deformed shape for p=8,
(c) First principal stress contour (range 0-100,000) for p=8.
Local solid model with point constraints
The same local model described in the previous section was analyzed by
importing displacements and rotations instead of forces and moments. The
point constraint information (nodal displacement and rotations) extracted from
the global shell model was imported into StressCheck and assigned to the faces
of the local solid model. A total of six constraint records were created, one for

FIGURE 166 Average stress Seq over the faces of three elements in the stiffener termination region (skin side)
4
each side of the skin, one for the stiffener web and one for the stiffener flange as
SET 23
SET 32
SET 22
SET 21
(a)
SET 18 SET 31
FIGURE 167Local solid model with

point constraint data. (a) Location of
point constraint data, (b) fitted face
(b) constraints.

Post-processing
The quality of fit of the displacements and rotations for each group of faces is
shown in FIGURE 168. Note that the largest fitting error is 6.93% for the rota-
tion Rx of set 21 which is associated with the shell faces on the stiffener end.
4
FIGURE 168 Quality of fit for the point displacement and rotation data.
The material properties were modified for the layers of elements around the
model (shown in different color in FIGURE 167) by setting Poisson’s ratio to
zero. All the elements in the mesh were declared as thin solids with q=3, and
the solution was obtained by performing a uniform p-extension from p=3 to 8.
This problem can be loaded from the Tutorial handbook:
T_StiffenerLocalDisplacement.sci
The results shown in FIGURE 169 are very close to those obtained for the
local model loaded by tractions (FIGURE 165 and FIGURE 166) and the refer-
ence global solid model (FIGURE 159 and FIGURE 160).
Post-processing
In StressCheck the post-processing windows Plot, Min/Max and Points let you
select either total or mechanical strains for extraction. This feature is imple-
mented for 2D plane-strain, extrusion and 3D-elasticity solutions, both linear
and nonlinear. Additionally, all six components of stress/strain can be output in
the lamina (material) coordinate system for both flat and cylindrical elements
when laminate composites are used. Also the generalization of the stress/strain
extraction with user-defined coordinate systems, includes general user-defined

Post-processing
(a)
(b)
4
(c)
FIGURE 169 Results from local solid model with imposed displacements. (a) Deformed
configuration for p=8, (b) First principal stress contour (range 0-100000) for p=8, (c)
Average stress Seq over the faces of three elements in the stiffener termination region.
Cartesian or cylindrical coordinate systems that can have any arbitrary rotation in
3D space.

Post-processing
Total/mechanical strains
By definition, the mechanical strain tensor in the global coordinate system is

computed as the difference between the total strain tensor and the thermal
strain tensor:
( total ) ( therm )
{ ε } ( mech ) = { ε } – {ε}
In StressCheck, the total strains are computed in the global system as the deriv-
atives of the global components of the displacement vector (ux, uy, uz):
∂
0 0
∂x
( total )
 ε  ∂
4  x  0 0
∂y
 εy 
  ∂  ux 
 εz  0 0  
{ε}
( total )
=  = ∂ z
  uy 
 γ xy  ∂ ∂
0  
  ∂ y ∂x  uz 
 γ yz 
  ∂ ∂
 γ zx  0
∂z ∂y
∂ ∂
0
∂z ∂x
The thermal strains are computed based on the coefficients of thermal expan-
sion (CTE) given in the material directions and the specified thermal loading.
In the global coordinate system, the thermal strain tensor is computed as:
( therm )
 εx   α xx 
   
 εy   α yy 
   
( therm )  εz   α zz 
{ε} =   = ∆T  
 γ xy   α xy 
   
 γ yz   α yz 
   
 γ zx   α zx 

Post-processing
where ∆T is the temperature change, and the αij are the transformed CTEs after
rotation from the material directions into the global coordinate system. For isotro-
pic materials, the CTEs are constant, so no transformation is needed. For orthotro-
pic, transversely isotropic or laminate materials, the CTEs are rotated applying the
same transformation rules used for strains (see for example, Experimental Stress
Analysis by J.W.Dally and W.F.Riley, 2nd Edition, Section 2.3).
A toggle switch in the Plot, Min/Max and Points interface of StressCheck allows
the user to select if the requested strain for output is total (Et) or mechanical (Em).
FIGURE 170 shows the Plot interfaces in the default condition (total strains) and
after the toggle switch is selected (mechanical strain).
FIGURE 170 Update of Plot interface to extract total/mechanical strains.
The mechanical strains can be computed in the global system, in any local (Carte-
sian or Cylindrical) system and in the material principal directions. If a principal
strain is requested and the mechanical strain switch is on, then the program first

Post-processing
computes the mechanical strain tensor in the global system and then solves for
the corresponding principal strain as described in the Planar Elasticity chapter.
As an example of the implementation, consider a rectangular strip (2 units

long, 0.25 unites high and 1 unit deep) under plane strain conditions subjected
to a uniform temperature change (∆T = -415o). The following properties are
considered in the material directions:
E11 = 23.5×106, E22 = E33=1.5×106, G12 = G13 = 0.88×106, G23 = 0.58×106
ν12 =ν23 = ν13 = 0.3, α11 = 4×10-7, α22 = α33 = 2×10-5.
As shown in FIGURE 171, the material 1-axis is aligned with the global X-
4 axis, while the material axes 2 and 3 are aligned with the global axes Z and Y,
respectively. Turn the Display Material icon on to see double-headed arrows in
the direction of the material 1-axis displayed over the mesh. Also shown in the
figure are the relations of the CTEs and Poisson’s ratio between the two sys-
tems. Plane-strain conditions are enforced by setting to zero the normal dis-
placement on the faces perpendicular to the Z-axis.
FIGURE 171 Strip with uniform ∆T.

Post-processing
The exact values of the total and mechanical strains for this problem are:
total –4
εx = ( α x + α z ν zx ) ∆T = – 3.249 × 10
mech total –4
εx = εx – α x ∆T = – 1.589 × 10
total –2
εy = ( α y + α z ν zy ) ∆T = – 1.079 × 10
mech total –3
εy = εy – α y ∆T = – 2.490 × 10
total
εz = 0
4
mech total –3
εz = εz – α z ∆T = 8.3 × 10
This problem was meshed using two pentahedra and one hexahedral element, and it
can be loaded from the Tutorial Handbook: StrainTM.sci. A single solution for p=1
is sufficient in this case because the strains are constant. The deformed configura-
tion is shown in FIGURE 172. As expected, the contraction in the X-direction is
FIGURE 172 Deformed and undeformed configuration for the strip.
much smaller than in the Y-direction and the contraction in the Z-direction is pre-
vented by the normal constraints.

Post-processing
Finally, FIGURE 173 shows the total and mechanical strains at a point as com-
puted by StressCheck. The results are identical to the exact values shown
above. Note that the mechanical strains are labeled with (M) next to the corre-
sponding component.
TOTAL STRAINS
4
MECHANICAL STRAINS
FIGURE 173 Total and mechanical strains at a point.
Stress/strains in material directions
All six components of stress/strain can be output in the lamina (material) coor-
dinate system for both flat and cylindrical elements when laminate or orthotro-
pic material properties are specified.
Laminate in Cartesian For the case of laminate properties in Cartesian systems (flat laminates), con-
sider the three coordinate systems indicated in FIGURE 174: The XYZ-global
system, the xyz-local Cartesian system where the laminated properties are
defined, and the 123-material system. The orientation of the local system rela-
tive to the global system is arbitrary, and the material 3-axis is always in the
direction of the local z-axis. The following steps are involved in computing the
stress/strain in the material directions:
1 - Compute all six stress/strain components in the global directions XYZ. This
is required because StressCheck solves the problem in global coordinates, and

Post-processing
FIGURE 174 Global, local and material coordinates for flat laminates.
therefore the displacement vector and the material properties are available in the
global system.
2 - Compute all six stress/strain components in the xyz-system by rotating the glo-
bal components using the direction cosine matrix of the local system.
{ ε } xyz = [ T ε ] { ε } XYZ
{ σ } xyz = [ T σ ] { σ } XYZ
T
{ ε } = { ε x ε y ε z γ xy γ yz γ zx }
T
{ σ } = { σ x σ y σ z τ xy τ yz τ zx }
EQ 15
The transformation matrices are given by:

Post-processing
2 2 2
l1 l2 l3 l1 l2 l2 l3 l3 l1
2 2 2
m1 m2 m3 m1 m2 m2 m3 m3 m1
2 2 2
[ Tε ] = n1 n2 n3 n1 n2 n2 n3 n3 n1
2l 1 m 1 2l 2 m 2 2l 3 m 3 l 1 m 2 + m 1 l 2 l 2 m 3 + m 2 l 3 l 3 m 1 + m 3 l 1
2m 1 n 1 2m 2 n 2 2m 3 n 3 m 1 n 2 + n 1 m 2 m 2 n 3 + n 2 m 3 m 3 n 1 + n 3 m 1
2n 1 l 1 2n 2 l 2 2n 3 l 3 n1 l2 + l1 n2 n2 l3 + l2 n3 n3 l1 + l3 n1
2 2 2
l1 l2 l3 2l 1 l 2 2l 2 l 3 2l 3 l 1
2 2 2
m1 m2 m3 2m 1 m 2 2m 2 m 3 2m 3 m 1
4 2
n1
2
n2
2
n3 2n 1 n 2 2n 2 n 3 2n 3 n 1
[ Tσ ] =
l1 m1 l2 m2 l3 m3 l1 m2 + m1 l2 l2 m3 + m2 l3 l3 m1 + m3 l1
m1 n1 m2 n2 m3 n3 m1 n2 + n1 m2 m2 n3 + n2 m3 m3 n1 + n3 m1
n1 l1 n2 l2 n3 l3 n1 l2 + l1 n2 n2 l3 + l2 n3 n3 l1 + l3 n1
EQ 16
 x  l1 l2 l3  X 
   
 y  = [ R ] { X } = m1 m2 m3  Y  EQ 17
   
z
  n1 n2 n3  Z 
In Eq.17, [R] is the direction cosine matrix of the local system relative to the
global coordinate system.
3 - Compute the stress/strain components in the material system by rotating the

xyz-components about the z-axis. Referring to FIGURE 174 and the transfor-
mation equations of step 2, we have:
In this case, the rotation matrix [R] is given by:
cos θ sin θ 0
[ R ] = – sin θ cos θ 0 EQ 18
0 0 1

Post-processing
 ε 11   εx 
   
 ε 22   εy 
   
 ε 33   εz 
{ ε } 123 =  = [ T ε ]  
 γ 12   γ xy 
   
 γ 23   γ yz 
   
 γ 31   γ zx 
xyz
 σ 11   σx 
    4
 σ 22   σy 
   
 σ 33   σz 
{ σ } 123 =  = [ T σ ]  
 τ 12   τ xy 
   
 τ 23   τ yz 
   
 τ 31   τ zx 
xyz
These transformations are applicable for 2D and 3D. For the case of 2D-plane
strain, the global stress components τyz and τzx are set to zero before the transfor-
mation into the local system is performed. Similarly, the global strain components
εz=γyz=γzx=0 for plane-strain.
Laminate cylindrical For the case of laminate properties in cylindrical systems, similar transformations
are performed. Referring to FIGURE 175, the following operations are performed
to determine the stress/strain in the material directions:
1 - Compute all six stress/strain components in the global directions XYZ.
2 - Compute all six stress/strain Cartesian components in the xyz-system by rotating

the global components using the direction cosine matrix of the local system accord-
ing to Eq.15 to Eq.17.

Post-processing
FIGURE 175 Notation for coordinate system tranformation.
3 - Compute the angular position θ of the extraction point in the cylindrical

system, and rotate the Cartesian stress/strain tensor according to the following
expressions:
εr σr
εθ σθ
εz σz
{ ε } rθz = = [ T ε ] { ε } xyz { σ } rθz = = [ T σ ] { σ } xyz
γ rθ τ rθ
γ θz τ θz
γ zr τ zr
[Tε] and [Tσ] are given by Eq.16 and the rotation matrix is the same as Eq.18.
4 - Compute the stress/strain components in the material system by rotating the

cylindrical components through an angle α:

Post-processing
ε 11 σ 11
ε 22 σ 22
ε 33 σ 33
{ ε } 123 = = [ T ε ] { ε } rθz { σ } 123 = = [ T σ ] { σ } rθz
γ 12 τ 12
γ 23 τ 23
γ 31 τ 31
[Tε] and [Tσ] are again given by Eq.16, and the rotation matrix is:
4
0 cos α sin α
[ R ] = 0 – sin α cos α
1 0 0
To have access to the computation of the stress/strain in the material directions, a

system type material is available in the post-processing windows of the user inter-
face. As shown in FIGURE 176 for the case of the Plot interface, the option Mate-
rial is available in the system combo box.
The same system option is available in the Points and Min/Max interfaces. Note
that the labels for the strain/stress components are unchanged when the Material
option is selected. The interpretation is based on the transformation equations pre-
sented above. Table 6 summarizes the interpretation of the StressCheck functions
based on the system selection.
Table 6: StressCheck functions for different reference systems.

StressCheck Function Global/Local Cartesian Local Cylindrical Material Direction
Sx / Ex σx / εx σ r / εr σ11 / ε11
Sy / Ey σy / εy σθ / εθ σ22 / ε22
Sz / Ez σz / ε z σ z / εz σ33 / ε33
Txy / Gxy τxy / γxy τrθ / γrθ τ12 / γ12
Tyz / Gyz τyz / γyz τθz / γθz τ23 / γ23
Txz / Gxz τxz / γxz τrz / γrz τ31 / γ31

Using 3D-beam elements
FIGURE 176 Plot interface with

the material system option.
Note: For formula extraction, the System Option in the Formula form controls
the interpretation of the system type in the Plot interface.

Straight 3D-beam elements are available under 3D Elasticity based on the
Timoshenko beam formulation. The displacement components at any point on
the beam is given by:
u x ( x, y, z ) = u 0 ( x ) – θ ( x )y + φ ( x )z
u y ( x, y, z ) = v 0 ( x ) – ψ ( x )z
u z ( x, y, z ) = w 0 ( x ) + ψ ( x )y
where u0(x), v0(x), w0(x) represent the displacements of the beam centerline
and ψ(x), φ(x), θ(x) represent rotations about the x-, y- and z-axis of the beam
local system (BLS), respectively (FIGURE 177).

x
y v0
φ
ψ
u0 2
Beam local system (BLS) x
θ
w0 Z
1
z
X Y
Global coordinate system (GCS)
FIGURE 177 Beam element. Notation

4
Beam element A beam element is defined by selecting two nodes in any order. The order in which
the nodes are picked when creating the elements define one axis of the BLS. The
beam x-axis is defined in the direction from the first to the second node. The direc-
tion of the other two axes depend on the coordinate system selected during the
assignment of the section properties. The z-axis of the BLS will be parallel the z-
axis of the selected local system. The y-axis of the BLS is computed as the cross
product between the positive z-axis and the positive x-axis.
If nodes are attached to curves, the beam elements will be defined as straight since
curved 3D-beams are not currently supported.
Section Properties To define the section properties select the Section Prop. tab from the StressCheck
Input dialog window. Two options are available: User and Library. The option User
means that the section properties will be provided by the user. The required inputs
are the cross sectional area (A), the moment of inertia about the centroidal y-axis
(Iy), the moment of inertia about the centroidal z-axis (Iz) and the cross moment of
inertia (Izy). Also the coordinate system to be used to determine the direction of the
z-axis of the BLS must be selected.

y
2
A =
∫ dA Iz =
∫ y dA
Area Area
O
z
2
Iy =
∫ z dA I zy =
∫ zy dA
Area Area
4
The option Library means that the section properties will be determined from
a set of predefined parametric sections available in StressCheck as described in
the User’s Guide. Only constant section properties along the span can be spec-
ified to a beam element.
Beam loads The following loads are currently supported for 3D-beam elements: point loads
and moments, body forces and distributed tractions. A description follows:
Point Force: Concentrated force with components Fx, Fy, Fz applied to nodes
in the global coordinate directions.
Point Moment: Concentrated moment with components Mx, My, Mz applied

to nodes. The moment components must be provided in global coordinates,
and a positive moment is a vector in the positive axis direction.
Beam Traction: Distributed load per unit length applied to element edges.
Traction loads can be applied in global/local Cartesian coordinate directions
(Tx, Ty, Tz) or in the Normal/Tangent reference (Tny, Tnz, Tax). Tny means a
distributed load normal to the beam in the xy-plane of the BLS, Tnz is normal
to the beam in the xz-plane of the BLS, and Tax is and distributed axial load.
Body force loading: Volumetric load, such as by gravity or inertia due to

acceleration. The body force is in terms of force per unit volume and can be
applied to elements in the global or any local Cartesian coordinate system (X,
Y, Z).

Beam constraints Two options are available for beam constraint when the object node is selected: sin-
gle node and hinge.
Single node: The constraint can be fixed or constant and each displacement and
rotation in the global directions can be controlled independently. For example, to
define a node as simply-supported, set the displacement components (X, Y, Z) to
fixed and leave the rotation components (Rx, Ry, Rz) unchecked.
Hinge: Releases rotation degree of freedom associated with a node. A hinge is

specified by selecting a node and, while holding the shift key, one element attached
to that node. If more than two elements are hinged at a node, multiple records can
be created selecting the same node and a different element each time. The rota-
tion(s) to be released is understood in the BLS, such that Rx is the twist rotation,
and Ry, Rz are the bending rotations.
4
Beam results The quantities available for post-processing are shown in TABLE 7.
TABLE 7. Standard functions for 3D-beams.
Symbol Explanation
Ux, Uy, Uz Displacement components in global coordinate directions
Rx, Ry, Rz Rotation components in global coordinate directions
Mx Twisting moment. Vector in the direction of the x-axis of the BLS.
My Bending moment. Vector in the direction of the y-axis of the BLS.
Mz Bending moment. Vector in the direction of the z-axis of the BLS.
V1 Shear 1. Force in the direction of the y-axis of the BLS.
V2 Shear 2. Force in the direction of the z-axis of the BLS.
Na Axial force in the direction of the x-axis of the BLS.
The sign convention for the displacements is that they are positive when in the
direction of the positive axes. A rotation is positive if the rotation vector of a right-
handed system advance in the direction of a positive axis. A positive axial load is in
tension and positive moments and shear forces are shown in FIGURE 178.
Example The utilization of the 3D-beam elements will be illustrated with an example. Con-
sider the beam/column structure shown in FIGURE 179. The material is steel
(E=29x106 psi, ν=0.295) and the section is a hollow circular pipe of 4.0 inch exter-

Positive Axial and Shear

Positive Moments
FIGURE 178 Sign convention for 3D-
4 beams.
nal diameter and 0.8 inch wall thickness. A 100 lb concentrated load is applied
at one end and is clamped at the other end as shown in the figure.
70”
Fy=100 lb
50”
100”
Z X
FIGURE 179 3D-beam, example problem.

To create the finite element description of this problem, start by selecting 3D >
Elasticity > in/lbf/sec/F from the Reference/Theory/Units toolbar. Select the Create
Model icon and then Geometry tab to create a local coordinate system at (0, 0, 0)
with a 90 degree rotation about the Y-axis:
• Create > System > Locate > X: 0.0, Y: 0.0, Z: 0.0, Rot-X: 0.0, Rot-Y: 90,
Rot-Z: 0.0 > Accept
Next, switch to the Mesh tab and create 4 nodes by the method location:
• Create > Node > Locate: X: 0.0, Y:0.0, Z: 0.0 > Accept
Repeat for nodes at (0, 100, 0), (70, 100, 0), (70, 100, 50). To create the 3 beam ele-
ments for this problem, select the object Beam and with the left mouse button select
nodes 1 and 2, then nodes 2 and 3, and finally nodes 3 and 4 (FIGURE 180).
4
2
FIGURE 180 Beam elements and nodes for the

example problem.
Next we assign section properties to the beam elements. Select the Section Prop.
tab and the option Library. Select the circular tube section and enter the values of
a=2.0 and b=1.2, then select beams 1 and 2, System: Global and click on Accept.
Because beams 1 and 2 are contained in the global XY-plane, the z-axis of the BLS
will be parallel to the global Z-axis. Next change the system to SYS1, select beam

3, and click on Accept. For beam 3, the z-axis of the BLS will be parallel to the
z-axis of SYS1 created above.
Next, create and assign the material properties by selecting the Material tab
and entering the following:
• Define tab > Define > Linear > Selection > ID: MAT > Material: Lin-
ear > Type: Isotropic > Shear: Unity > E: 29e6 > v: 0.295. Click
Accept.
• Assign tab > Select > All Elements > Selection > ID: MAT. Click
Accept.
Next, select the Load tab and create the point load record:
• Select > Node > Point Force > ID: LOAD > Fx: 0.0, Fy: -100, Fz: 0.0.
Select node number 4 and then click on Accept.
4
Select the Constraint tab to fix the displacement and rotations of node 1 to sim-
ulate the built-in condition:
• Select > Node > Node > ID: CONST > Data type: Fixed > Turn on the
switches in front of X, Y, Z, Rx, Ry, Rz, select node number 1 and click
on Accept.
Select the Solution ID tab and create a solution record:

• Solution ID: SOL, Constraint ID: CONST, Load ID: LOAD > Accept
The description of the problem is now complete. You can load this problem
from the Beam Handbook: SignPost.sci.
Solution: Obtain a sequence of solutions by running a p-extension from p=1 to

6. Select the Compute Solution icon and complete the Linear tab:
• Upward-p > p-limits: 1 to 6, then switch to the Solve! tab
• Initialize > Automatic > Iterative > None > Solve
Results: After the solution is complete, select the View Results icon and then
the Error tab and complete as follows:
• SOL > Run: 1 to 6 > Estimate > Accept
The estimated relative error in energy norm is shown in FIGURE 181. Note
that the potential energy remains constant for any run greater than or equal to
3. That is because the exact solution for this problem corresponds to a polyno-
mial of degree 3.

FIGURE 181 Estimated relative error in energy norm for the example problem.
4
To obtain the deformed shape and the bending and torsion moment plots, select the
plot tab and complete as follows:
• Select > All Elements > Selection > Input tab > Solution: SOL > Run: 6 >
Plot: Solution > Shape: Deform > Overlay toggle ON > Midsides: 10. Click
on the Plot button to get the deformed and the undeformed shapes together
(FIGURE 182a).
• Select > All Elements > Selection > Input tab > Solution: SOL > Run: 6 >
Plot: Solution > Contour: Fringe > Shape: Z-shape > Overlay toggle OFF >
Function: Mz > Midsides: 10. Click on the Plot button to get the plot of the
moment Mz (FIGURE 182b).
Similar plots can be obtained for all other quantities of interest

4 (a) Deformed shape (b) Bending moment Mz
FIGURE 182 Plot results for the 3D-beam example problem.

5 Modal and Buckling
Analyses
This chapter contains a description of the procedures used in executing a Modal Analysis and bifurca-
tion (eigenvalue) Buckling Analysis. The goal of a Modal Analysis is to find the natural frequencies and 5
mode shapes. StressCheck provides a capability for taking into account the effects of initial stresses on
the computed natural frequencies. The goal of a Buckling Analysis is to find the load factor correspond-
ing to a given loading condition, and the corresponding buckling shapes. These two types of analyses
are discussed together because the problem setup and post-solution procedures are almost identical.
Modal Analysis is available for planar, plate bending and three-dimensional problems; while Buckling
Analysis is available only for three-dimensional problems.
It is recommended that you become familiar with the Planar Elasticity, Plate Bending and 3D-Elasticity
chapters before performing Modal or Buckling Analysis.
Input data preparation

Preparation of the input data consists of the following steps:
• Opening a database
• Description of the solution domain
Analysis Guide Chapter 5: Modal and Buckling Analyses 253

• Specification of material properties. Make sure that you provide the

specific density (mass per unit volume) for the material when perform-
ing a modal analysis
• Specification of loading conditions (Modal Analysis without pre-stress
does not require load specification)
• Specification of constraints
• Association of a solution name with a constraint name and load name
(buckling and pre-stress modal only)
• Specification of the discretization parameters
Recommended p-levels
Modal and bifurcation Buckling Analyses are applicable to beams, plates,
shells and similar three-dimensional bodies. In general, when one dimension of
a body is much smaller than the others, locking will occur at low p-levels. It is
recommended therefore that p-levels should range from not less than p=3 to at
least p=6.
5 Vibration under pre-stress

Modal Analysis is used for the computation of the frequencies and associated
mode shapes at which a linear elastic body will tend to vibrate once it was set
into motion in the absence of external loading. Therefore specification of load-
ing is not required for a standard Modal Analysis. However, in many practical
problems the natural frequency is influenced by the existence of pre-stress. For
example, a rotating helicopter blade vibrates at a higher frequency than a sta-
tionary one. StressCheck has the capability to account for the effect of pre-
stress. If this option is used then StressCheck first solves the linear problem,
then, utilizing the linear solution, computes the geometric stiffness matrix,
which modifies the elastic stiffness matrix for the eigenvalue computation. The
pre-stress option is available only for three dimensional problems, including
extruded problems.
Units
The units for the material properties should be given in a consistent set. For
example, if using US units, the force is given in pound force (lbf), the time in
seconds (sec), the length in inches (in), the pressure or stress in lbf/in2 (psi), the
254 Chapter 5: Modal and Buckling Analyses Analysis Guide

Modal analysis
mass in lbf/g (lbf-sec2/in), and the specific density in lbf/g-in3 (lbf-sec2/in4), where
g is the gravitational acceleration (g=386.1 in/sec2). Using SI units, the force is
given in Newtons (N), the mass in kilograms (kg), the time in seconds (sec), the
length in meters (m), the pressure or stress in N/m2 (Pa), and the density in kilo-
gram per cubic meter (kg/m3). For both the US and SI units, the natural frequencies
are reported in cycles per second (Hz) and in radians per second. If you chose a
time unit other than seconds, the results should be interpreted as cycles per unit
time or radians per unit time.
For Modal Analysis, when the modulus of elasticity is given in MPa (N/mm2), then
the density should be given in Ns2/mm4. This change is necessary because:
2 2
kg Ns – 12 Ns
- = 10 ----------4- .
1 ------3 = 1 --------
4
m m mm
Consider, for example, an aluminum alloy. If the modulus of elasticity is E=75200

MPa, then the density should be ρ=2800 10-12 Ns2/mm4. In the list of standard
materials provided with StressCheck, these are the units used for E and ρ.
5
Modal analysis
Consider the square plate with the square opening shown in FIGURE 183. Of inter-
est is the computation of the natural frequencies and mode shapes.
FIGURE 183 shows the solution domain consisting of a square plate of length
a=10.0 inches with a centered square opening of size b=5.0 inches and a thickness
h=0.10 inches. All four sides of the plate are soft simply supported (Uz=0) and the
material properties are those of an ASTM-A36 carbon steel (E=29.5 x106 psi,
v=0.295, ρ=7.36 x 10-4 lbf-sec2/in4).
After opening a new database, from the Reference and Theory Selectors change the
default analysis to Plate Elasticity so the proper input area will be available for data
entry.

Modal analysis
FIGURE 183 Square plate with square opening.
Create the finite element mesh for the full plate. The use of symmetry condi-
tions is not recommended for Modal Analysis because symmetric structures
5 have non-symmetric modes of vibration. Assign constant thickness to all ele-
ments. Select ASTM A-36 from the Material Library (set the shear factor to
Energy) and assign that material to all elements. Specify soft-simple support
conditions to all external boundaries of the plate. The eight-element mesh for
this problem is shown in FIGURE 184. In general, it is not necessary to grade
the mesh in the vicinity of sharp corners when only the lowest natural frequen-
cies and mode shapes are of interest.
You can load this problem from the Tutorial Handbook. Activate the Hand-
book Analysis interface by choosing the Handbook Library icon from the Main
Toolbar. When the Handbook interface appears, click on the Browser icon of
the Model Info tab to access the Model Browser. Choose the Tutorial Hand-
book, and double click on the file blevins11-8-2.sci. The problem will be
loaded into StressCheck. Refer to the User’s Guide for more information about
the Handbook interface.
Executing a Modal Analysis

To execute a Modal Analysis, select the Compute Solution icon from the Main
Toolbar and when the Solver dialog window appears select the Modal tab. The

Modal analysis
FIGURE 184 Mesh and constraints for example problem.
Modal Analysis input area shown in FIGURE 185 will appear. The available 5
options are as follows:
FIGURE 185 Modal Analysis input area for Plate Bending.

Modal analysis
Extension: the options in Modal Analysis are Upward-p, Downward-p, and

Uniform-h. These options are described in the Planar Elasticity chapter.
Type: Two output options are available: Frequencies only or Frequencies and
modes. “Frequencies only” means that the natural frequencies will be com-
puted and stored but not the mode shapes. “Frequencies and modes” means
that the natural frequencies and the corresponding mode shapes will be com-
puted and stored. The second option requires more CPU time than the first.
p-limits: Select a range, such as 3 to 6. If you have just created a problem then
it is good practice to select (say) p=1 to 3. This will run quickly and if there is
any problem with the input data then appropriate error messages will appear.
Remember that you need a range of at least three solutions to be able to obtain
an error estimate based on p-extension.
Frequency: Two options are available: Range and Interval. “Range” means
that the requested output will be computed from the frequency number i to the
frequency number j, where i and j are selected in the Frequency numbers
fields. “Interval” means that the requested output will be computed for all fre-
quency values greater than or equal to the lower limit (in Hz) specified in the
first Frequency values field and less than or equal to the upper limit specified
5 in the second Frequency values field.
Pre-stress: StressCheck has the capability to take into account the effect of
pre-stress on the natural frequency of vibration. To exercise this option you
must be performing a three-dimensional analysis and the body must be prop-
erly loaded and constrained. StressCheck will compute the stresses and make
the appropriate corrections to the stiffness matrix.
Next select the SOLVE! tab and the following options will be available:
Execute: The options are Initialize and Restart. “Initialize” means that the
sequence will be computed from the first p-limit or, in the case of h-extensions,
from the current mesh to the number of refinements specified by the Run Limit
entry. “Restart” is used when a sequence of p-levels has been computed and
you wish to compute higher p-levels (if running Upward), or lower p-levels (if
running Downward), retaining the computed data. In this case simply increase
the upper (or lower) p-limit to its desired value and select Restart.
Run Mode: The options are Automatic and Stepwise. You may execute an
Automatic sequence of solutions for the extension and limits selected. Alterna-

Modal analysis
tively, you may execute Stepwise, requiring you to click on the Solve button fol-
lowing the solution for each p-level or h-refinement.
Method: The options are Iterative or Direct. “Iterative” means that the Iterative
Solver (default) is used. “Direct” means that the Direct Solver is used.
Converge: This option is not available for modal analysis.
Display: StressCheck will allow you to monitor one of the following while the
solution proceeds:
- The Sequence switch is used to monitor the progress of the solution. The elements
are highlighted when they are processed into the solver.
- Engineering results may be monitored by using the layout switch to activate the
Results layout. Select the desired Output controls in the Results class of interest
before activating the solution.
Monitoring result functions is useful in the case of large problems where sequential
analysis executed stepwise is employed. The monitored data provide information
on the basis of which you can decide whether to stop or continue the extension pro-
cess. For small problems full p-extension should be performed. 5
Example Let us execute a Modal Analysis for the plate problem described above. Consider-
ing downward p-extension, enter the following information in the Solver dialog
window:
• Modal tab > Extension: Downward-p > Type: Frequencies and modes > p-
limits: 8 to 1 > Frequency: Range > Frequency numbers: 1 to 5.
tion:
tive > Click on the Solve button.
The top portion of the Modal dialog window displays the status of the execution.
The message bar located on the main window displays which operation is being
performed for each run.
Note: If Modal Analysis is performed after a linear analysis, both sets of solution
records will be available in the database. The Initialize option in the Modal Solver
deletes only the records of existing modal solutions, while the same option in the
Linear Solver deletes ALL existing records. Therefore, if a linear analysis is per-

Modal analysis
formed after a Modal Analysis has been executed, all solution records for the
Modal Analysis will be lost.
Post-solution operations
There are two basic post-solution operations which are relevant for modal
analysis: Frequency convergence and display of the mode shapes. The fre-
quency convergence is performed under the Error tab in the Results dialog
window, while the display of the deformed shapes is done under the Plot option
as explained in the following.
Frequency convergence From the Main Menu Bar select the View Results icon and when the Results
dialog window appears select the Error tab (FIGURE 186). To obtain an error
estimate, there must be at least three runs in a sequence. The available solution
names and run numbers are displayed in the scrolling list. The solution corre-
sponding to each eigenpair (frequency and mode shape) is identified by the
name provided in the Solution ID class followed by an underscore (_) and a
number. For example, referring to FIGURE 186, the solution corresponding to
the first eigenpair is labeled SOL_00001.
5
To check the convergence of the first two natural frequencies for the plate
problem, complete the Error estimation input area for each solution and click
on the Accept button. The estimated error for the first natural frequency is plot-
ted in FIGURE 187 as a function of the number of degrees of freedom (DOF).
FIGURE 188 shows the tabular results for the first two natural frequencies.
Plotting of mode shapes To obtain a deformed configuration, select the Plot tab from the Results dialog
window (FIGURE 189). To display the mode shape associated with a given
frequency, select the corresponding solution name and run number, select
Shape: Deformed and then click on the Plot button.
For the first natural frequency complete the Plot option input area as follows
(options other than defaults are indicated):
• Plot tab > Select > All Elements > Solution: SOL_00001 > Run: 1 >
Plot: Solution > Shape: Deform > Midsides: 3 > Click on the Plot but-
ton.
To visualize the mode shape of the first natural frequency of the problem, you
need to rotate the model. Use the Views Toolbar to rotate the model until you
get the desired view (see FIGURE 189).

Modal analysis
FIGURE 186 Input area for error estimation.
The main window has a legend that includes the solution ID, the run number, the
degrees of freedom, the natural frequency, the mode shape number, and the maxi-
mum and minimum values of the displayed function. The maximum and minimum
values have only relative significance, since mode shapes are known only up to an
arbitrary multiplier.
Note: StressCheck computes repeated eigenvalues. If you plot the second and third
mode shapes for this problem, both have the same, or very nearly the same, natural
frequency but different mode shapes.

Modal analysis
FIGURE 187 Convergence of first natural frequency on log-log scale.
FIGURE 188 Convergence of first two natural frequencies.

Modal analysis for beams
FIGURE 189 Mode shapes corresponding to the first natural frequency.

5

Consider the pinned-pinned multispan continuous beam with pinned intermediate
supports shown in FIGURE 190. The objective of the analysis is to find the first
natural frequency and the corresponding mode shape. The typical span, L= 200 in,
is taken to be the distance between adjacent supports. The material properties are
those of the steel ASTM-A36 (E= 29x106 psi, ν= 0.295, ρ= 7.36x10-4 lbf-sec2/in4).
The beam is an AISC W 10x15 (A= 4.41 in2, I= 68.9 in4). US units are used.
Model creation After opening a data base, set the analysis mode to Planar Elasticity.

L L L L
FIGURE 190 Multispan beam with intermediate supports.
Select the Model Info icon from the Main Toolbar and when the Model Infor-
mation dialog window appears select the Parameters tab. Provide the following
information:
• Name: L > Description: Characteristic Span > Value: 200 > Limit: >0,
Class: General > Click on the Accept button.
Provide the geometric description of the beam by selecting the Create Model
icon and supplying the following information:
• Geometry tab > Create > Line > Locate > Input toggle switch ON > X:
0.0, Y: 0.0, Z: 0.0, Length: 4*L, Angle: 0.0 > Accept.
5 • Mesh tab > Create > Node > Offset > Repeat # = 5, Offset: 0.0 + 0.25 >
Click on the line created in the previous step > Turn the Repeat switch
off.
• Create > Beam > Selection. Create four beam elements by connecting
pairs of consecutive nodes.
To define the section properties select the Section Prop. tab from the Input dia-
log window, and supply the following information:
• Select > All Elements > Selection > Click on Database tab.
After you click on the Database tab, the AISC Section Database box will
appear (FIGURE 191). As you click on any of the entries in the scrolling win-
dow of the box, the Area and Moment of Inertia will be displayed in the Sec-
tion properties region. Once you selected a section (W 10x15 in this example
problem), you can store the properties by clicking on the Accept button. Note
that in the Section Properties Input dialog window the 3rd and 4th Moments
also appear. These two moments are calculated and assigned to the elements by
StressCheck automatically when you click on Accept.

FIGURE 191 AISC Section Database box.
To enter the material properties information select the Material tab from the Input
dialog window and supply the following information:
• Material tab > Define > Linear > Selection > Define tab > Units: U.S. >
Material: Linear > Type: Isotropic. Click on the Browse button and from
the standard list of materials appearing in the Material Library, click on the
ASTM-A36 and its properties will be entered automatically. > ID: MAT >
• Assign tab > Select > All Elements > ID: MAT > Accept.
To enter constraint data select the Constraint tab from the Input dialog window and
provide the following information:
• Constraint tab > Select > Node > Node > ID: CONST > Direction: XY >
Data type: Fixed > System: Global > Turn on the switches in front of X and
Y > Click on the left most node and then click on Accept. A pinned support
will be created on that node because the rotation was not fixed.

• Select > Node > Node > ID: CONST > Direction: XY > Data Type:
Fixed > System: Global > Turn on the switch in front of Y only > Click
on the rest of the nodes while holding the Shift key > Click on Accept.
Finally, select the Solution ID tab and provide the following information:
CONST > Click on Accept.
Solution Now you are ready to start the computation. Click on the Compute Solution
icon and when the Solver dialog window appears supply the following infor-
mation.
• Modal tab > Extension: Upward-p > Type: Frequencies and modes > p-
limits: 3 to 8 > Frequency: Range > Frequency numbers: 1 to 1.
Iterative > Display: Sequence > Click on the Solve button.
5
After the execution is complete, click on the View Results icon and when the
dialog window appears select the Error tab. Complete the following informa-
tion:
• Error tab > Select > All Elements > Selection > Solution: SOL_00001
> Run: 1 to 6 > Click on the Accept button.
The results will be displayed in tabular form in the Report window as shown in
FIGURE 192. Note that the estimated relative error is practically zero for run
#4 which corresponds to a p-level of 6.
The first natural frequency obtained from Ref. [53], Table 8-3d using the clas-
sical formulation of beams is f1= 30.81 Hz, which compares well with the
value of 30.6 shown in FIGURE 192.
To plot the mode shape associated with the first natural frequency, select the
Plot tab and complete the following information:
• Plot tab > Select > All Elements > Solution: SOL_00001 > Run: 6 >
Plot: Solution > Shape: Deform > Midsides: 15 > Click on the Overlay
button > Click on the Plot button. The mode shape is shown in FIG-
URE 193.

Buckling analysis
FIGURE 192 Multispan beam: Convergence

of the first natural frequency.
FIGURE 193 Mode shape corresponding to the first natural frequency.
Buckling analysis
Consider the circular disc of outside radius a=100 inches, inside radius b=50 inches
and thickness th=2.0 inches, fixed in the direction perpendicular to the plane of the
disc along the external boundary and subjected to a uniform radial compression σ
as shown in FIGURE 194. The material properties are those of a plain carbon steel
(E=30 x 106 psi, v=0.3).

Buckling analysis
σ
σ
th
5
FIGURE 194 Circular disc under radial compression.
Set the analysis mode to Planar Elasticity and create a 2D-finite element mesh
of the full disc. Assign constant thickness and isotropic material properties to
all the elements in the mesh. Apply a constant unit normal traction on the out-
side boundary (the normal traction should be specified as negative) and apply
rigid body constraints. The finite element mesh and boundary conditions are
shown in FIGURE 195a.
After the problem is created in 2D, change the analysis mode from Planar to
Extrude and add the constraint in the Z-direction to the element faces on the
outer surface of the disc:
• Constraint Tab > Select > Any Curve > General > ID: CONST > Data
Type: Fixed > Check the Z toggle box > Click on the circle that defines
the outer edge of the disc > Click Accept.

Buckling analysis
(a) 2D-mesh (b) Extruded mesh

5
FIGURE 195 Mesh and boundary conditions for buckling problem.
The mesh and boundary conditions created in 2D will be converted to their 3D

equivalent (see FIGURE 195b). Then, define the Solution ID. Refer to the Extru-
sion section in the Three-Dimensional Elasticity chapter for instruction on how to
perform an extrusion and specify the constraints on the extruded elements. This
problem can be loaded from the StressCheck Library, Tutorial Handbook: Buck-
ling.sci.
Executing a Buckling Analysis

To perform a Buckling Analysis, StressCheck first solves the linear problem corre-
sponding to the specified loads and constraints, then, utilizing the stress field com-
puted from the linear solution, computes the geometric stiffness matrix, which is
used for the eigenvalue computation. Therefore, after the Buckling Analysis is
completed, two sets of solutions are available: the linear solution, which establishes
the pre-buckling stress state and the eigenvalue buckling solution.

Buckling analysis
To execute a bifurcation Buckling Analysis select the Compute Solution icon

from the Main Toolbar and when the Solution dialog window appears select
the Buckling tab (FIGURE 196). The available options are as follows:
FIGURE 196 Input area for Buckling Analysis.
5
Extension: the options in Buckling Analysis are Upward-p, Downward-p, and
Uniform-h. These options are described in the Planar Elasticity chapter.
Type: Two options are available: Buckling load only or Buckling load and
shapes. “Buckling load only” means that the buckling load factors will be com-
puted and stored but not the mode shapes. “Buckling and shapes” means that
the buckling load factors and the corresponding buckling mode shapes will be
computed and stored. The second option requires more CPU time than the first.
p-limits: Select the range of p-values for which you wish the analysis per-
formed. Remember that in the case of “thin” bodies locking will occur at low
p-levels; so, let p range between from not less than 3 to at least 6.
Load numbers: Enter the range of eigenvalues you wish to compute. Very
likely, you will be interested in the lowest eigenvalue which is the smallest
multiplier (load factor) for the load identified by the load name in the solution
ID form. In that case enter “1”.
After completing the information in the Buckling tab, select the SOLVE! tab.
The SOLVE! window requires the following information:

Buckling analysis
Execute: The options are Initialize and Restart. “Initialize” means that the
sequence will be computed from the lowest p-level to the highest or, in the case of
h-extensions, from the current mesh to the number of refinements specified by the
Run Limit entry.
“Restart” is used when a sequence of p-levels has been computed and you wish to
compute higher p-levels (if running Upward), or lower p-levels (if running down-
ward), retaining the computed data. In this case simply increase the upper (or
lower) p-limits to its desired value and select Restart.
Run Mode: The options are Automatic and Stepwise. You may execute an Auto-
matic sequence of solutions for the extension and limits selected. Alternatively, you
may execute Stepwise, requiring you to click on the Continue button following the
solution for each p-level or h-refinement.
Method: The options are Iterative or Direct. “Iterative” means that the Iterative
Solver (default) is used. “Direct” means that the Direct Solver is used.
Converge: This option is not available for buckling analysis.
Display: StressCheck will allow you to monitor one of the following while the
5
solution proceeds:
- Engineering results may be monitored by using the layout switch to activate the
Results layout. Select the desired output controls in the Results class of interest
before activating the solution.
Monitoring result functions is useful in the case of large problems where sequential
analysis executed stepwise is employed. The monitored data provide information
on the basis of which you can decide whether to stop or continue the extension pro-
cess. For small problems full p-extension should be performed.
- The Sequence switch is used to monitor the progress of the solution. The elements
are highlighted as they are included into the solver.
Example To compute the first buckling load and shape for the circular disc under radial com-
pression (FIGURE 194), considering an upward p-extension from p=5 to 8, enter
• Buckling tab > Extension: Upward-p > Type: Buckling and shapes > p-lim-
its: 5 to 8 > Load numbers: 1 to 1.

Buckling analysis
Next, choose the SOLVE! tab to get the solution. Complete the requested infor-
mation as shown below:
Iterative > Display: Sequence > Click on the Solve button.
The top portion of the buckling status box displays the status of the execution.
The message bar located on the main window displays which operation is
being performed for each run.
Post-solution operations
There are two basic post-solution operations which are relevant to buckling
analysis: buckling load convergence and display of the buckling shapes. The
buckling load convergence is performed under the Error estimation option in
the Results dialog window, while the display of the deformed shapes is done
under the Plot option as explained in the following section.
Load convergence Select the Error tab from the Results dialog window. To obtain an error esti-
mate, there have to be at least three runs in a sequence. The available solution
5 names and run numbers are displayed in the scrolling list of the input area. The
solution corresponding to each eigenpair (buckling load and mode shape) is
identified by the name provided in the Solution ID input form followed by an
underscore (_) and a number followed by the letter B. For example, the solu-
tion corresponding to the first buckling load is labeled SOL_0001B. The linear
solution is also available in the scrolling list of the input area with the name
SOL (assuming that SOL is the name given in the Solution ID input form).
To check the convergence of the buckling load for the circular disc under radial
compression described above, click on the last record of the buckling solution
in the scrolling list of the input area and click on the Accept button. The esti-
mated error for the buckling load factor is shown in FIGURE 197. The buck-
ling load obtained from Roark’s Formulas for Stress and Strains (Ref. [51],
Table 35, Case 12, page 688) is σ = 2769 psi, which is very close to the value
computed with StressCheck: σ = 2702 psi at p=8. Note that we specified unit
radial compression, and therefore the load factor is the same as the buckling
stress.
Plotting buckling shape Select the Plot tab from the Results dialog window. To display the buckling
mode shape associated with a given buckling load, complete as follows:

Using the eigensolver
FIGURE 197 Convergence of buckling load.
• Plot tab > Select > All Elements > Solution: SOL_0001B > Run: 4 > Plot: 5
Solution > Shape: Deform > Midsides: 3 > Click on the Plot button.
The buckling mode shape will be displayed in the Graphic window as shown in
FIGURE 198.

There are two eigensolvers in StressCheck. One solver is based on the Lanczos
method, the other is based on the QR method. The Lanczos solver is used by
default. To select the QR solver, specify the parameter “_qr”.
When running a downward p-extension, the element stiffness and mass (resp. geo-
metric) matrices corresponding to the highest p-level are saved and reused for all
lower p-levels. Thus, the downward p-extension is more efficient than the upward.
Some problems may develop in using the eigensolvers. The possible remedies are
discussed in the following:

FIGURE 198 First buckling mode shape.

5
Problem: Frequency (buckling load) number i is non-monotonic

When the frequency or buckling load values do not decrease monotonically as
the number of degrees of freedom are increased (see under Frequency Conver-
gence and Load Convergence on how to check for convergence), determine for
which run number (or p-level) the value increases, and then try one of the fol-
lowing:
Remedy 1: Specify a higher accuracy by setting parameter _nfig>6 (_nfig is

the number of decimal digits of accuracy desired in the eigensolver. By default
_nfig=6).
Remedy 2: Switch to the QR eigensolver by defining the parameter “_qr”

(arbitrary value) and recompute at those p-levels for which the eigenvalues are
not monotonic.

Problem: Program appears to be running too long

You are executing a modal or Buckling Analysis in automatic mode, and at a given
run number (p-level) the program seems to become very slow. In that case, press
the escape (Esc) key while the focus is in the Graphic Window. When the program
stops, try one of the following:
Remedy 1: If the tolerance is too small, the solver may take too long to compute
the eigenpairs. Request a lower accuracy by setting the parameter _nfig<6.
Remedy 2: Switch to the QR solver by defining the parameter “_qr” for the p-level
for which the slow performance is detected. Run in Stepwise mode and switch back
to the Lanczos solver by deleting the parameter “_qr” after the run is completed.
Note Rigid body constraints alone should not be used in Modal and Buckling Analysis.
The reason is that rigid body modes are admissible solutions and the corresponding
eigenvalues are zero. If you run a problem which is not fully constrained, the eigen-
solver will find the zero eigenvalues. To speed-up the computation however, it is
recommended that you specify the parameter “_k_indef” (arbitrary value) before
executing the Modal/Buckling Analysis.
5


6 Conduction Heat
Transfer
This chapter contains guidelines for the preparation of input data for steady state heat conduction prob-
lems. In addition, post-solution procedures are illustrated by examples. StressCheck can be used for
solving linear, steady state heat conduction problems with prescribed temperature, flux and convective
boundary conditions. The effects of temperature-dependent material properties and radiation can also be
accounted for using the nonlinear procedure described in the Advanced Guide. The computed tempera-
ture distribution can be used in a very convenient way for computing the load vector for thermo-elastic
problems.
The implementation covers planar, axisymmetric, and three-dimensional problems. 6

Preparation of the input data consists of the following steps:
• Opening a database and selecting the analysis type
• Entering geometric data: solution domain and finite element mesh
• Specification of thickness (for planar problems)
• Specification of material properties
• Specification of flux boundary conditions, the temperature of the convec-
tive medium and heat generated per unit volume and unit time
Analysis Guide Chapter 6: Conduction Heat Transfer 277

Model problem: Pipe with cooling fins
• Specification of temperature boundary conditions and the convective

heat transfer coefficients
• Association of a solution name with the flux and temperature boundary
conditions
• Assignment of an extension process.
An explanation of the available options is presented through a model problem


A circular pipe has an internal radius of 25 mm, an external radius of 30 mm.
Ten evenly spaced, tapered cooling fins are attached to the pipe. Taking advan-
tage of twenty-fold symmetry, the solution domain is as shown in FIGURE
199. The material is an aluminum alloy, the coefficient of thermal conduction
(k) of which is 0.206 W/mmoC. The internal wall temperature is 500oC. Cool-
ing is by convection. The coefficient of convective heat transfer (hc) is
1.2x10-3 W/mm2 oC and the temperature of the convective medium is 20oC.
Therefore the mathematical model is formulated as follows:
E Circular arc,
F
6 Subtended angle: 2.5o
18o C
7.5o
A B
25 5 15
FIGURE 199 Circular pipe with cooling fins. Solution domain.
278 Chapter 6: Conduction Heat Transfer Analysis Guide

Solve the heat conduction problem:
2 2
∂ u ∂ u
+ = 0
2 2
∂x ∂y
subject to the following boundary conditions:
(a) Along AB and EF symmetry condition is prescribed, that is, the flux is zero:
∂u
qn = –k = 0
∂n
where n represents the normal to the boundary, k is the coefficient of heat conduc-
tion and u(x,y) is the temperature.
(b) Along BC, CD, and DE convective boundary condition is prescribed:
qn = hc ( u – uc )
where hc is the coefficient of convective heat transfer, also called film coefficient,
and uc is the temperature of the convective medium.
(c) Along FA the temperature of 500oC is prescribed. 6

The objective is to determine the temperature distribution and the heat dissipated by
the finned pipe per unit length.
Selecting analysis type After opening a new database, from the Reference and Theory selectors change the
default analysis to Planar Heat Transfer, and select the units mm/N/sec/C..

Geometry and mesh Click on the Create Model icon from the main tool bar and select the Geometry
tab. To create the mesh shown in FIGURE 199, consider first the geometric
description:
• Geometry tab > Create > Circle > Locate > Turn on the Repeat switch
and enter 2 > Input toggle switch ON > X: 0.0, Y: 0.0, Z: 0.0, Radius:
25 + 5, P1-Min: 0.0, P1-Max: 18.0, Rot-Z: 0.0 > Click on the Accept
button.
• Create > Circle > Locate > Input toggle switch ON > X: 0.0, Y: 0.0, Z:
0.0, Radius: 45, P1-Min: 0.0, P1-Max: 2.50, Rot-Z: 0.0 > Click on the
Accept button. > Click on the Center Model icon in the Views toolbar.
Create the nodes and elements by selecting the Mesh tab in the StressCheck
Input window, and proceeding as follows:
• Mesh tab > Create > Node > Offset > Input toggle switch ON > offset:
0.0. With the mouse cursor select the three circles.
• Create > Node > Offset > Offset: 2.5 > Select circle of radius = 45.
• Create > Node > Offset > Offset: 7.5 > Select the first and second cir-
cles.
• Create > Node > Offset > Offset: 18.0 > Select the first and second cir-
cles.
• Create > Quadrilateral > Selection. Define three quadrilateral elements
Thickness To assign thickness, click on the Thickness tab in the StressCheck Input dialog
6 window. Assign unit thickness to all the elements of the model problem by
selecting:
Material Properties To enter the material properties select the Material tab in the StressCheck Input
dialog window and the input area shown in FIGURE 200a will appear. The
specification of material properties is in two parts: the definition of material
properties and the assignment of material properties to the elements.
The coefficients of heat conduction kxx, kxy, kyy and, optionally, the constant
Q>0 may be specified. The coefficients of heat conduction are defined by the
following relationship:

(a) (b)
FIGURE 200 Input area for material properties.

6
∂u
qx k xx k xy ∂ x
= –
qy k xy k yy ∂u
∂y

where qx, qy are the heat flux vector components and u(x,y) is the temperature.
The coefficients of heat conduction are subject to the restrictions:
kxx>0, kyy >0 and kxx kyy - kxy2>0.
The constant Q is zero in the case of the heat conduction problem. If you select
Q>0 then StressCheck will solve the elliptic Helmholtz equation, namely:
∂  ∂u ∂u ∂ ∂u ∂u
– k + k xy  –  k xy + k yy  + Qu = f ( x, y )
∂ x  xx ∂ x ∂ y ∂ y  ∂ x ∂ y
subject to the specified boundary conditions.
As in the case of the elasticity problems, StressCheck provides a standard list

of material properties. These must be understood as generally accepted average
values. Actual material properties may vary substantially from these average
values. It is your responsibility to ensure that the material data are prop-
erly defined for your application.
To define material properties select the Define tab and select the Browse but-
ton, depending if you wish to select kxx, kxy, kyy from the standard material list
provided by StressCheck. If you choose the standard list then specify whether
US or SI units are to be used. The US units for kxx, kxy, kyy are understood to be
Btu/sec in oF, the SI units are understood to be in W/mmoC. This means, of
course, that the length dimensions must be in inch (resp. mm) units when US
(resp. SI) units are chosen. Similarly, the temperature must be specified in
6 Fahrenheit (resp. Celsius) units when US (resp. SI) units are chosen. Of course,
you may use any consistent system of units when you enter the material prop-
erties yourself.
For the model problem, kxx=kyy=0.206 should be entered. For isotropic materi-
als kxy is always zero.
• Material tab > Define tab > ID: MAT > Material: Linear > Units: SI >
kxx: 0.206, kxy: 0.0, kyy: 0.206 > Click on the Accept button.
• Assign tab > Action: Select > Object: All Elements > Method: Selec-
tion > ID: MAT > Click on the Accept button.
Flux The flux boundary condition data include the following:

• flux specified along boundaries, element edges or faces
• temperature of the convective medium

Select the Flux tab in the StressCheck Input dialog window and the input area will
appear as shown in FIGURE 201. Enter a name into the field next to ID. Next select
the boundaries or edges (resp. elements) for which flux boundary conditions (resp.
heat source data) are to be prescribed by pointing to the appropriate edges and ele-
ments in the Model window.
FIGURE 201 Flux boundary condition.
For the model problem, enter the following information about the temperature of
the convective medium:
• Flux tab > Select > Edge > Conv. Temp. > ID: FLUX > Value: 20 > Select
the element edges labeled BC, CD, DE in FIGURE 199 while holding the
Shift key > Click on the Accept button.
The convective boundary condition data for the model problem is shown in FIG-
URE 201. The zero convective temperature is a natural boundary condition and
does not have to be entered.

As in the case of the model problem, in many cases zero flux is specified in
connection with symmetric boundary conditions. The case of zero flux is a
natural boundary condition and hence does not have to be explicitly pre-
scribed. Nonzero flux values are specified either as constants, parameters, or
by a formula. The details are described in the User’s Guide. Having entered the
formula into the formula dialog window, dismiss it and enter the formula name
in the appropriate field of the flux input area.
Heat source Heat source data are also entered through the same input area. The object
selected may be an element, a group of elements, or all elements. By heat
sources we understand heat generated (or absorbed) per unit volume. If US
units are used then the heat source data are in Btu/sec in3 units. If SI units are
used then the heat source data are in J/sec mm3.
When you click on the Accept button, the specified boundary condition data
will be entered into the database. Confirmation will appear in the scrolling list
in the input area.
Remark: Equilibrium requirement.
When flux values are prescribed along the entire boundary in 2D then the fol-
lowing condition must be satisfied:
∫ q tds + ∫ ∫ f( x, y )tdxdy = 0
n
Γ Ω
6
where qn is the normal component of the flux vector at the boundary, t is the
thickness, f(x,y) is the heat generated per unit volume, Γ is the boundary of the
solution domain Ω. This is analogous to the equilibrium requirement in
mechanics which states that the tractions and body forces must be in equilib-
rium. Analogously to the “rigid body constraint” in mechanics an arbitrary
temperature value must be prescribed at a node.
Temperature To access the input area shown in FIGURE 202, select the Temperature tab in
the StressCheck Input window and enter a name next to ID. Select the bound-
aries, edges or faces for which temperature boundary conditions or convective
boundary conditions are to be specified. Select Temperature or Conv. Coeff.,
depending on whether you wish to prescribe temperature or convective coeffi-
cients on the selected boundaries or edges. Next enter the temperature or con-
vective coefficient as a constant, a formula, or a parameter.

FIGURE 202 Temperature and coefficients of convective heat transfer.
6
The details for entering formulas are described in the User’s Guide. The entry form
for parameters appears when selecting the Model Info icon from the Main Toolbar.
Finally, click on the Accept button. This will cause the boundary condition to be
entered into the database. Confirmation will appear in the scrolling list.
For the model problem, enter the following information about the inside tempera-
ture and the convective coefficient:
• Temperature tab > Select > Any Curve > Temperature > ID: TEMP >
Value: 500 > Select the leftmost circular boundary > Click on the Accept
button.
• Select > Edge > Conv. Coeff. > ID: TEMP > Value: 0.0012 > Select the
edges labeled as BC,CD,DE in FIGURE 199 while holding the Shift key >
Click on the Accept button. The finite element mesh with the boundary
condition symbols is shown in FIGURE 203.

FIGURE 203 Pipe with cooling fins: Mesh and boundary conditions.
Remark: Continuity requirement
The prescribed temperature along the boundary must be a continuous function.

No abrupt changes from one element to the next are allowed. StressCheck will
issue an error message when it encounters a discontinuity in the prescribed
temperature boundary condition.
Remark: Convective boundary conditions
The coefficient of convective heat transfer hc is related to the conduction pro-

cess in the thin stationary layer of fluid at the wall surface and modifies the
6 stiffness matrix of the system. For this reason the film coefficient has to be
treated as a principal boundary condition and hence entered in the Temperature
input class.
The temperature of the convective medium uc, on the other hand, affects the
right-hand side of the system of equations, therefore has to be treated as a natu-
ral boundary condition and hence entered in the Flux input class.
Defining Solution ID Select the Solution ID tab in the StressCheck Input window, and the input area
shown in FIGURE 204 will be displayed on the screen. Enter the following
information:
• Solution ID tab > Solutions tab > Solution ID: SOL > Temperature ID:
TEMP > Flux ID: FLUX > Accept.

FIGURE 204 Input area for Solution ID.
Executing analysis Next select the Compute Solution icon from the Main Toolbar. When the Solver 6
window appears complete the following information:
• Linear tab > Extension: Upward-p > p-limits: 1 to 8 (FIGURE 205 a).
tive > Convergence: None > Display: Sequence > Click on the Solve button
(FIGURE 205 b).
Since we did not specify the discretization parameters, the trunk space was the
default setting for the analysis.
Extracting results Select the View Results icon from the Main Toolbar and then the Error tab. FIG-
URE 206 shows the estimated relative error in energy norm for runs 1 to 8. Note
that the rate of convergence is small (less than 1). The reason for this is that, due to
the presence of the re-entrant corner (point D in FIGURE 199), the exact solution
has a singularity. Therefore convergence is slower. If we are interested in the total

FIGURE 205 Solver input window: (a) Linear tab - (b) SOLVE tab.
FIGURE 206 Convergence of the potential energy.

heat loss, or the temperature only, then the solution is sufficiently accurate for engi-
neering purposes.
The temperature contours, shown in FIGURE 207 for p=8 do not change apprecia-
bly when p is changed from 6 to 8. It is recommended that you verify this. Select
the Plot tab from the Results dialog window and complete as follows:
• Plot tab > Select > All Elements > Selection > Solution: SOL, Run: 8 >
Plot: Solution > Contour: Fringe > Functions: U > Midsides: 10 > Plot.
6
FIGURE 207 Temperature contours, p=8.
Similarly, check the heat loss by selecting the Heat Flow tab in the Results window.
• Heat Flow tab > Select > Edge > Selection > Solution: SOL > Runs: 8 to 8
> Click on the two edges associated with the inner wall of the pipe.
You will find that at p=8 the heat loss is 10.856 W/mm for the sector analyzed when
measured along the inner wall. For the entire pipe this figure must be multiplied by
20, since we took advantage of the 20-fold symmetry in defining the solution
domain.

The “Standard Functions” available in StressCheck are listed in Table 8.
TABLE 8. Standard functions. Heat Transfer.
Symbol Explanation
U Temperature
Ux Temperature gradient in the x-direction (Ur)
Uy Temperature gradient in the y-direction (Uz)
Uz Temperature gradient in the z-direction
Qx Heat flux per unit area in the x-direction (Qr)
Qy Heat flux per unit area in the y-direction (Qz)
Qz Heat flux per unit area in the z-direction
Fmla Formula. Using this option, any mathematical expression containing the standard
functions can be computed for a given solution.
Calc Calculator. Using this option, any mathematical expression containing standard
functions can be computed for any combination of solutions.
Note: Symbols in brackets ( ) are appropriate for axisymmetric analysis.
Exercise 1 Repeat the solution of the model problem using the three element mesh and
product space. You will find that the accuracy is slightly higher, but the data of
interest, the temperature distribution and the total flux, are not changed appre-
6 ciably. This shows that these data are virtually independent of the discretiza-
tion and hence must be close to their exact values.
Exercise 2 Change the reference from Planar Heat Transfer to Extrude Heat Transfer. Run
the problem again after assigning new three-dimensional material properties to
all elements (kxx: 0.206, kyy: 0.206, kzz: 0.206, kxy: 0.0, kxz: 0.0, kyz: 0.0) .
This will perform a fully three dimensional solution. Since the flux is under-
stood to be zero when no boundary conditions are specified, the extruded solu-
tion will be the same as the planar one. The error estimate will be slightly
different, however, due to the fact that the number of degrees of freedom is dif-
ferent.
Exercise 3 Change the thickness of the model problem to 100 mm and define a tempera-
ture distribution by a formula,

u = 500 + 2z
on the inner surface. This can be done by editing the appropriate record in the
StressCheck Temperature input window. On displaying the temperature distribution
for p = 8, you will find the results shown in FIGURE 208.
FIGURE 208 Temperature distribution for the model problem of Exercise 3. 6
Note: When a 2D mesh is extruded, the elements will extend from -t/2 to t/2 in the
z-direction, where t is the assigned thickness to the elements. That means that the
temperature given by the above formula will vary from 400οC at z=-50 to 600oC at
z=50.
After extrusion, make sure all the material coefficients are defined (kzz must be >
0).

Verification example
The purpose of this example is to demonstrate that StressCheck yields the cor-
rect results for temperature, heat flux, total heat flow, and potential energy.
Problem description
Consider a thick-walled pipe made of a material with a coefficient of heat con-
duction k = kxx = kyy = 1.0, kxy = 0.0. The internal radius is 1.0 and the external
radius is 2.0. The internal wall temperature is 10, the external wall is cooled by
convection. The coefficient of convective heat transfer (hc) is 1.0, the tempera-
ture of the convective medium (uc) is also 1.0 and unit thickness is assumed.
These data are chosen to make hand calculations simple.
This is, essentially, a one-dimensional problem because the temperature is a

function only of the radial distance from the center. A 45-degree sector of the
pipe is shown in FIGURE 209.
6 45o
A B
FIGURE 209 Pipe sector.
1.0 1.0
The boundary conditions are as follows:

1 ∂u
• Along AB and CD: q n = – k --- = 0
r ∂θ
• Along DA: u = 10
∂u
• Along BC: q n = – k = hc ( u – uc ) .
∂r

Geometry and mesh First, set the Reference/Theory/Units toolbar to Planar > Heat Transfer > in/lbf/sec/
F. To construct the geometric description and mesh, select the Create Model icon
from the main tool bar and perform the following steps:
• Geometry tab > Create > Circle > Locate > Repeat #= 2 > Input toggle
switch on > X: 0.0 + 0.0, Y: 0.0 + 0.0, Z: 0.0 + 0.0, Radius: 1.0 + 1.0, P1-
Min: 0 + 0, P1-Max: 45 + 0, Rot-Z: 0 + 0 > Click on the Accept button.
• Mesh tab > Create > Node > Offset > Repeat#= 2 > Input toggle switch on
> Offset: 0 + 45. Move the mouse cursor to the Model window and click
near each circular arc to create a total of 4 nodes.
• Mesh tab > Create > Quadrilateral > Selection. Create the element by
selecting the four nodes.
Material • Material tab > Define tab > ID: PROP > Material: Linear > Units: US >
kxx: 1.0, kxy: 0.0, kyy: 1.0 > Click on the Accept button.
• Material tab > Assign tab > ID: PROP > Select > All Elements > Selection
> Click on the Accept button.
Boundary conditions • Flux tab > Select > Any Curve > Convective Temperature > ID: FLUX >
Value: 1.0. > Select the rightmost circular arc > Click on the Accept button.
• Temperature tab > Select > Any Curve > Temperature > ID: TEMP >
Value: 10.0 > Use the mouse cursor to select the leftmost circular arc > 6
• Temperature tab > Select > Any Curve > Convection Coefficient > ID:
TEMP > Value: 1.0 > Select the rightmost circular arc > Click on the
Accept button.
Solution ID • Solution ID tab > Solutions tab > Solution ID: SOL > Temperature ID:
TEMP > Flux ID: FLUX > Click on the Accept button.

The exact solution

We are interested in solving:
2 2
∂ u
1 ∂u 1 ∂ u
∆u = + --- + ----- = 0
∂r
2 r ∂ r r2 2
∂θ
subject to the boundary conditions indicated above. It is easily verified that the
exact solution is:
u ( r ) = c 1 ln ( r ) + c 2
where:
18
c 1 = – ---------------------- = – 7.54308
ln ( 4 ) + 1
c 2 = 10.0
The temperature at the outer wall is therefore:
6 18
u ( 2 ) = – ----------------------ln ( 2 ) + 10 = 4.77154
ln ( 4 ) + 1
The exact value of the normal flux at the outer boundary is:
c1
q n = – k ∂u = – ----- = 3.77154
∂r r = 2 2
The exact value of the total heat flux through the 45 degree sector of the pipe is
π
---
4
∫ –k∂r rdθ = –c  --4- = 5.92432

∂u π
q = 1

By definition, the potential energy is that functional which is minimized by the

exact solution over the set of all admissible functions. For details refer to Ref. [3].
In this case the potential energy is defined by:
π π
--- ---
42 4
2 2
∫∫  ∂u +  ∂u
∫ [ u ( 2 ) – u ] r dθ
1
π EX = --- k
 ∂ x  ∂ y
dx dy + 1--- h c c
2
2 2
01 0
By substituting the exact solution and the given values for k, hc and uc, we have:
π EX = 15.487536 + 11.1718956 = 26.6594319
The finite element solution

Using one finite element, trunk space, at p=8 StressCheck yields for the tempera-
ture of the outer wall:
u FE ( 2 ) = 4.77154
showing agreement with the exact solution to at least 6 digits. This high accuracy is
due to the fact that the exact solution can be easily approximated by the mapped
polynomials. 6
StressCheck reports the flux vector components in the x-y system, denoted by Qx
and Qy. Select the Points tab from the Results window and the input area shown in
FIGURE 210a will appear. Enter the solution name, the desired run number, and
select the Function: Fmla. In the Formula dialog window (FIGURE 210b), enter an
arbitrary formula name, for example Qn, into the Name field, select the system
option in the formula to be Cylindrical, and define Qn as:
Qn = Q cos ( t ) + Q sin ( t )
x y
This is the radial component of the flux vector. Choose Accept in the Formula(e)
window, close it, and select Qn from the Fmla pulldown menu under the Points tab.

(b) Formula dialog window.
(a) Results dialog window. FIGURE 210 Points Extraction.
Selecting any point on the outer edge will yield for p=8:
Qn = 3.77152
6
which is very close to the exact value, q n = 3.77154 .
To find the value of the total heat flux, select the Heat Flow tab in the Results
window, and the input area shown in FIGURE 211 will appear. Enter the fol-
lowing information:
• Heat Flow tab > Action: Check > Object: Edge > Method: Selection >
Solution: SOL > Run: 8 to 8 > Click on the outer boundary. For p=8
StressCheck will report the value:
q = 5.924296
FE
which is very close to the exact value.

FIGURE 211 Input area for heat flow.
To estimate the relative error in energy norm, select the Error tab from the Results
dialog window. Enter the solution name and run numbers 1 to 8. StressCheck will
report the number of degrees of freedom, potential energy, rate of convergence and 6
estimated error in energy norm, as shown in FIGURE 212.
As seen in FIGURE 212, at p=8, trunk space, one element, StressCheck reports:
π FE = 26.6594315
which is, once again, very close to the exact value.
Exercise Repeat this example using (a) two triangular elements, (b) axial symmetry and one
quadrilateral element, (c) axial symmetry and two triangular elements.
In the axially symmetric case you will find that the total heat flux reported by
StressCheck is:
( q FE ) AX = 7.54305

FIGURE 212 Convergence of the potential energy. Trunk space.
and the potential energy at p=8, trunk space, one quadrilateral element, is:
( π FE ) AX = 33.9438425
This is because in the axisymmetric case StressCheck reports data for a one-
radian sector. In the planar case we used a 45 degrees sector (π/4 radians). To
6 compare the total heat flux and potential energies, it is necessary to multiply
the axisymmetric data by π/4. Thus, for
π π
( π FE ) PLANAR = --- ( π FE ) AX = --- × 33.9438425 = 26.6594316
4 4
which is exactly the same result as reported for the planar quadrilateral ele-
ment. Similarly we have:
π
( q FE ) PLANAR = --- ( q FE ) AX
4

Stresses caused by thermal loading
Of course, the temperature values and the heat flux require no such adjustment. The
proper geometric description for axisymmetric analysis is shown in FIGURE 213.
Z
Boundary conditions:
AD: u= 10 D C
BC: qr = hc (u-uc) 1.0
hc = 1.0 A B
uc = 1.0 1.0 1.0 R
FIGURE 213 Pipe sector for axisymmetric analysis.

In many cases we are interested not only in the temperature distribution but also in
the stress maxima caused by thermal loading. StressCheck makes it very conve- 6
nient to “import” the solution from the thermal problem into the elasticity problem
and to compute the load vector for the thermal loading. The procedure is illustrated
in the following through the model problem shown in FIGURE 199. For the pur-
poses of computing the thermally induced stresses, we use E=7.0x104 MPa,
ν=0.365, a(th)=2.35x10-5/ oC, Plane Stress.
Referring to FIGURE 199, the boundary conditions for the elasticity problem are:
Along AB and EF symmetry boundary conditions are prescribed: un=Tt=0. Along
the other boundary segments stress-free boundary conditions Tn=Tt=0 are pre-
scribed.
To solve the thermally loaded elasticity problem after the thermal problem has been
solved, switch from Planar Heat Transfer to Planar Elasticity. Enter and assign the
material properties for the elasticity problem using a name which is different from

the name used for the material properties in the case of the Planar Thermal
problem as shown below:
• Material tab > Define > Linear > Selection > Define tab > ID:
PROP_EL > Material: Linear > Type: Isotropic > Case: Pl. Strain > E:
7.0e4, v: 0.365, a(th): 2.35e-05 > Click on the Accept button.
• Assign tab > Select > All Elements > Selection > ID: PROP_EL >
Next define the loading as indicated below.

• Load tab > Select > All Elements > Thermal > ID: LOAD_TH >
Import Thermal Solution toggle on > Solution: SOL_TH > Run: 8 >
Thermal ID: SOL_TH, Run #: 8 > Click on the Accept button.
FIGURE 214 Thermal loading.

Note that the load ID must be different from the ID given to the thermal case.
Next define the constraint conditions for the elasticity problem as follows:
• Constraint tab > Select > Edge > Symmetry > ID: SYM > Select the edges
corresponding to AB and EF in FIGURE 199 > Click on the Accept button.
Specify the Solution ID, giving an unique name to the solution of the elasticity
problem, which is different from the solution name of the thermal problem. In this
example problem, we used SOL_TH as the name of the solution for the thermal
problem, and SOL_EL for the elasticity problem.
Next we solve the elasticity problem from p=1 to 8 and we find that in this case the
maximal stress is steadily increasing. No convergence to a limiting value is evident.
The reason for this is the presence of the reentrant corner. The stress, corresponding
to the exact solution of the elasticity problem, is not finite. In order to ensure that
the elastic stress is finite, it is necessary to introduce a fillet to round the reentrant
corner.
Introducing a fillet of radius 1.5 mm at the reentrant corner, the mesh is modified as
shown in FIGURE 215. Upon solving the thermal problem again, we find that the
FIGURE 215 Mesh with fillet.
heat flux is: q = 10.52 W/mm, which is very close to the value found for the cross
section without fillets, indicating that small fillets do not alter the heat flow signifi-
cantly.

Next we solve the elasticity problem and we find that the maximum stress
occurs near the fillet and its value is 75 MPa. The contour plot for the first prin-
cipal stress is shown in FIGURE 216.
6 FIGURE 216 Contour plot for the first principal stress.
Note: After modifying the mesh, check the boundary conditions to see that
they are properly applied.
This problem can be loaded from the Tutorial Handbook: FinFillet.sci.

ADVANCED GUIDE

StressCheck
Volume 4
ADVANCED GUIDE
Release 7
August, 2005

Copyright 2005
COPYRIGHT NOTICE
U.S.A.

rights reserved.

Portions of StressCheck are owned by UGS, Inc. Copyright  1998. All rights reserved.
DISCLAIMER

or change.
1
1 Nonlinear Analysis
StressCheck is the first finite element software product to implement material and geometric nonlinear-
ities within the framework of the p-version of the finite element method. The implementation covers:
• Nonlinear materials: Small-strain, small-displacement, with elastic-plas-
tic and nonlinear-elastic material characterization for plane-stress, plane-
strain, axisymmetric and three-dimensional problems.
• General nonlinear: Nonlinear-strain, large-displacement, with elastic,
elastic-plastic and nonlinear elastic material characterization for plane-
stress, plane-strain, axisymmetric and three-dimensional problems.
• Cold working: Automated analysis of cold-worked fastener holes, with
elastic-plastic material characterization for plane-stress and plane-strain
problems.
• Fastened connections: A nonlinear capability to analyze frictionless shear
fastened structural connections under plane-stress or plane-strain condi-
tions.
• Margin check: Nonlinear analysis capability (material or general nonlin-
ear) with failure criteria control for plane-stress, plane-strain, axisymmetric
and three-dimensional problems.
• Nonlinear heat transfer: Radiation and temperature-dependent material
properties for steady state heat transfer problems in two- and three-dimen-
sion.
Advanced Topics Chapter 1: Nonlinear Analysis 1

Nonlinear materials: Elastic-plastic problems
1
To solve a nonlinear problem, a linear problem must be solved first. A nonlinear material stress-
strain law must be specified to perform a material nonlinear or cold working analysis. For a general
nonlinear analysis and fastened connections, the material may be specified as either linear or non-
linear. During any nonlinear execution, normal springs are checked to see if they remain under
compression (attached to the boundaries).
It is recommended that you become thoroughly familiar with linear analysis before attempting to
perform a nonlinear analysis.

The implementation is based on the von Mises yield criterion and the deforma-
tion theory of plasticity under the assumption of small strain, small displace-
ment. The purpose and scope of this implementation are as follows Ref. [42]:
1. Realistic mathematical models of real physical systems must have a capabil-

ity to provide initial estimates for the effects of nonlinearities at a low compu-
tational cost. The deformation theory of plasticity serves this purpose well for a
large class of practical problems.
2. The effects of a single overload event on structures made of ductile materials

are of substantial practical importance. Such effects can be well represented by
mathematical models based on the deformation theory of plasticity provided
that the plastic flow is contained, that is, the plastic zone is surrounded by elas-
tic material [29], [30].
3. The propagation of cracks in strain-hardening materials is generally corre-

lated with the J-integral. The J-integral is based on the deformation theory of
plasticity [31], [32].
4. An important feature of StressCheck is that engineering data can be conve-

niently extracted from the finite element solutions in the post-solution phase.
The deformation theory of plasticity makes it feasible to extend this into the
elastic-plastic regime because the data storage requirements are small.
Remark: The p-version is not susceptible to Poisson’s ratio locking and hence
correct limit loads are obtained. In conventional FEA (h-version) locking
occurs when the displacement formulation is used. For this reason alternative
formulations, generally known as mixed methods, must be employed. See, for
example, [33].
2 Chapter 1: Nonlinear Analysis Advanced Topics

1
To solve an elastic-plastic problem, a linear problem must be first solved and an
elastic-plastic stress-strain relationship must be specified. With the exception of the
material properties, the input data preparation is identical to that for linear prob-
lems.
Elastic-plastic material properties

To specify elastic-plastic material properties, select Class: Material from the Main
Menu Bar and the material input form will appear. Select Material: Elastic-Plastic.
Decide whether you wish to enter your own material property, or select material
properties from a standard list provided by StressCheck. To use material properties
from the standard list, click on the Browser button located at the side of the material
ID and when a list of material designations appears in the window of the Material
Library form, click on the material name to automatically enter the property data in
the Material input window. Finally, click on Accept.
StressCheck supports four types of stress-strain relationships for elastic-plastic

materials: Elastoplastic, Bilinear, 5-Parameter, and Ramberg-Osgood. These are
illustrated in FIGURE 1 and described in the following:
Elastoplastic means elastic-ideally-plastic materials. Such materials are character-

ized by three parameters: the modulus of elasticity (E), Poisson’s ratio (v) and the
yield stress (Sy) (FIGURE 1a). Upon entering these parameters into the input area,
the stress-strain law will be displayed in a graph window.
Bilinear means linear-elastic, linear strain-hardening materials, characterized by

four parameters: the modulus of elasticity (E), Poisson’s ratio (v), the yield stress
(Sy) and the tangent modulus (Et) which characterizes strain hardening (FIGURE
1b).
5-Parameter means that the material is characterized by five parameters: the mod-
ulus of elasticity (E), Poisson’s ratio (v), the yield stress (Sy), the strain-hardening
modulus (Et), the starting point on the linear strain-hardening part of the stress-
strain relationship given by the strain (Eps-2) which must be greater than Sy/E, and
the corresponding stress (S2) which must be greater than Sy. The stress-strain law is
linear in the elastic range and for strains greater than Eps-2. The two linear seg-
ments are joined by a cubic spline, as shown in FIGURE 1c.
Ramberg-Osgood materials are characterized by four parameters: the modulus of

elasticity (E), Poisson’s ratio (v), the stress (S70E), which is the stress corresponding

1
to the intersection of the stress-strain curve with a line which passes through
the origin and has the slope of 0.70E, and an exponent n in the expression [54]:
σ 3 S 70E  σ  n
ε = E-- + - --------
- --------- .
7 E  S 70E
Typical values of n range between 4 and 90. Fractional values are permitted for
n (FIGURE 1d). An alternative expression for the Ramberg-Osgood stress-
strain curve is given in the MIL-HDBK-5J handbook: Metallic Materials and
Elements for Aerospace Vehicle Structures. The expression results in the same
stress-strain curve and it is given as [71]:
n1
= E-- + 0.002  -------
σ σ
ε
 σ02
where σ02 is the 0.2% offset yield strength of the material. The relationship
between the two set of parameters is the following:
n = n1
n -
-----------
n–1
σ 02
S 70E = -----------------------------
1⁄n
-
 0.014
------------- 
 3 E
Obtaining a linear solution

Having entered the input data, execute a linear analysis. It is recommended that
you obtain a sequence of solutions and estimate the relative error in the energy
norm. At least one of the solutions in the sequence should have a small relative
error (typically under 1% for 2D-problems and under 5% for 3D-problems). If
this were not the case, then either refine the mesh or switch to the product
space (or both) and repeat the linear analysis. In general, you should expect the
accuracy of the nonlinear solution to be somewhat lower than the accuracy of
the linear solution. In the case of plane strain, axisymmetric, and 3D problems,
for example, the plastic strain is subject to the incompressibility constraint,
which is not present in the case of the linear solution.

σ σ
Sy Et
Sy 1
E E
1 1
ε ε
(a) Elastoplastic material (b) Bilinear material
σ σ
Cubic spline
S2 Et S70E
1
Sy 0.70E
1
E
E
1
1
Eps-2 ε ε
(c) 5-Parameter material (d) Ramberg-Osgood material
FIGURE 1 Elastic-plastic stress-strain relationships supported by StressCheck.

1
Executing material nonlinear analyses
Prior to executing a Nonlinear analysis, you should develop a plan in accor-
dance with the following guidelines:
• Decide what are the data of interest: For example, the limit load, the
extent of the plastic zone, the J-integral, strain maxima, etc.
• Decide how accurately you wish to compute the data of interest. If you
wish to compute them to less than X percent relative error then the cor-
responding data in the linear solution should be accurate to within 0.1X
and 0.5X. The tolerance selected for the nonlinear solution should be
also between 0.1X and 0.5X relative error.
• Make sure that the applied load or imposed displacement will not
cause excessive plastic strains, except in the immediate vicinity of sin-
gular points. Remember that the small strain, small displacement the-
ory is used. The plastic zone should be completely confined by an
elastic zone.
Select the Nonlinear tab in the Solver dialog window and the user interface for
executing nonlinear solutions will appear (FIGURE 2). In the scrolling list you
will see the list of linear solutions. Proceed as follows:
FIGURE 2 Nonlinear execution area.
1. Click on the linear solution which is to be the starting solution for the non-
linear iterations. The solution name, run number, type, and associated degrees
of freedom (DOF) appear in the scrolling list of the dialog window. The corre-

1
sponding discretization will remain fixed in the course of the nonlinear solution
process. The relative error in energy norm of the linear solution you select should
be small (typically under one percent), and the estimated relative errors in the data
of interest should be less than the error tolerance for the nonlinear solution, as
explained above.
2. Enter a name for the nonlinear solution (Nonlin ID:) which is different from the
name of the linear solution (Linear Sol:). For example, you can modify the name of
the linear solution by adding the letters NL (for nonlinear).
3. Select the analysis Type: Material (NLMat).
4. Select the Convergence, the Technique and the Tolerance(%). The only tech-
nique available for material nonlinear analysis is the direct integration. The Con-
vergence choices are: Stress or Energy. If you choose Stress, StressCheck will
continue the iterations until in each Gauss point the difference between the equiva-
lent stress from the current iteration and the reference equivalent stress from the
stress-strain curve is less than the specified tolerance.
(k)
Denoting the equivalent stress in the kth Gauss point at the ith iteration by σ̃ i , the
(k)
reference stress from the uniaxial stress strain curve by σ i , and the tolerance by
τ, this criterion is satisfied when
(k) (k)
σ̃ i – σ (ik ) ≤ τ σ̃ i . (1)
If you choose Energy then the error criterion is based on the total apparent potential
energy πi . By definition, the total apparent potential energy of the ith iteration is:
1
πi = -2 ( X Ti K i – 1 X i ) – ( rX i )
where xi is the solution vector at the kth iteration; Ki-1 is the stiffness matrix corre-
sponding to xi-1 and r is the load vector. The energy criterion is satisfied when
πi – πi – 1 ≤ τ πi . (2)
It is recommended that you choose the energy criterion when the solution has sin-
gular points and when the objective is to compute limit loads.
The iteration number and the error, that is, the actual value of τ (percent) will be
displayed for each iteration cycle in the status area of the solver interface.

1
5. Select the number of iteration cycles you wish StressCheck to perform auto-
matically and enter this number into the field next to Iteration Limit. Stress-
Check will stop when the iteration limit is reached, unless the specified error
tolerance is reached first. When StressCheck stops at the iteration limit you can
examine the results and decide whether to continue or not. If you wish to con-
tinue, then simply reset the Iteration Limit and click on the Solve button.
6. Select the Load steps. This option is used for applying the load in increments
from the minimum to the maximum specified load in equal increments. When
the load steps toggle is turned on, you will be able to select the load parameter
and enter the maximum value for the load (FIGURE 2). The minimum load
value is displayed but cannot be changed. You also will have the option of
selecting the Save Load Steps toggle. This option will save all converged inter-
mediate load steps for further post-processing. Each saved step will be identi-
fied with the name given in step 2, an underscore ‘_’ followed by the run
number of the starting linear solution. For example, NSOL_8 means that the
solution NSOL was initiated from the linear solution #8. The run number asso-
ciated with each nonlinear solution will correspond to each load step
Next, select the SOLVE! tab and complete as follows:
FIGURE 3 Solver Execution Interface.
7. Select the mode of execution. The options are: Initialize and Restart. Select
Initialize when you begin an elastic-plastic analysis, select Restart when you
wish to continue from an existing elastic-plastic solution.

1
8. Select Run Mode. The options are Automatic and Stepwise. If you select Auto-
matic then the iterations will continue until you reach the specified iteration limit or
satisfy the specified error tolerance. If you select Stepwise then the iteration process
stops after each cycle, and you need to click on the Solve button to go to the next
iteration cycle.
9. Select Method. The choices are Iterative and Direct. Iterative means that the Iter-
ative Solver (default) is used. Direct means that the Direct Solver is used.
10. Select Display. The choices are Sequence and Results. You may monitor the
progress of the solution by displaying the Sequence; this will shade out each ele-
ment as the elemental stiffness matrix is computed. You can monitor the value of
functions at points, maximum values of functions or contour plot of functions dur-
ing the nonlinear iteration by turning on the Results toggle switch.
11. Click on the Solve button to start the Nonlinear analysis.
After the analysis is completed, the nonlinear solution just obtained will be avail-
able in the scrolling list of the post-processing dialog box with the name you
entered in step 2, the same run number and DOF as the linear solution selected, and
with a type NLMat. Selecting a different run number of the linear solution to start a
nonlinear iteration will result in a new nonlinear solution record with different
degrees of freedom and run number than the previous one. Both nonlinear solutions
will be available for extraction. You can obtain as many nonlinear solutions as there
are linear solutions. In this way you can check the convergence characteristics of
the nonlinear solutions.
Model problem: Plate constrained by fasteners

An aluminum plate is constrained by two long rows of fasteners, arranged as shown
in FIGURE 4. The plate is 1/8 inches thick, the material is 6061-T6 aluminum. The
plate is loaded in its own plane only by an imposed displacement ∆, as shown.
Assuming that the fasteners can be represented by a distributed spring characterized
by the spring rate kn=5.0 x 107 lbf/in3, and the elastic-plastic stress-strain relation-
ship for the plate can be represented by the Ramberg-Osgood law with the follow-
ing material constants: E=1.01x107 lbf/in2, v=0.365, S70E=3.5x104 lbf/in2, n=31,
Plane-stress; determine the force corresponding to the imposed displacement
∆=8.5x10-3 inches and plot the plastic zone, that is the region in which the equiva-
lent strain exceeds the value of 3.465x10-3.

0.3”(diam.)
0.5”
(typ)
0.5” 0.5” 2.5” ∆
FIGURE 4 Plate constrained by two rows of fasteners.
Geometric description First, set the Reference/Theory/Units to Planar > Elasticity > in/lbf/sec/F. Tak-
ing advantage of symmetry, the solution domain and boundary conditions are
as shown in FIGURE 5. An auxiliary line to facilitate mesh construction is also
included in the figure.
The symbols T and u refer to the traction and displacement vectors and the sub-
scripts n, t refer to the normal and tangential directions, respectively. Because
the exact solution of this problem is smooth, the finite element mesh is laid out
as shown in FIGURE 6. The minimum number of four-sided elements,
required for representing the geometric description, is sufficient (9 -elements
in this case).
Solution On performing a planar (2D) linear analysis, p-extension, trunk space, we find
that the estimated relative error in energy norm is under 1 percent for p = 7.
Furthermore, computing the stress resultants for the edge EF (see FIGURE 5),
it is seen that the force is substantially independent of p for p > 6. Therefore we

J I G F
H 0.5
C
A B D E
0.5 0.5 0.5 2.0
Boundary conditions
AB Symmetry un = 0, Tt = 0
BC Spring Tn = -kn un, Tt = 0
CD Free Tn = 0, Tt = 0
DE Symmetry un = 0, Tt = 0
EF Imposed displacement un = ∆, Tt = 0
FG Symmetry un = 0, Tt = 0
GH Free Tn = 0, Tt = 0
HI Spring Tn = -kn un, Tt = 0
IJ Symmetry un = 0, Tt = 0
JA Free Tn = 0, Tt = 0
FIGURE 6 Mesh layout and schematic representation of boundary conditions.

1
perform the material nonlinear solution for p = 7. Specifying Type: Material,
Convergence: Stress, Tolerance: 1.0%, the solution is obtained in 14 iterations.
The plastic zone is shown in FIGURE 7. To obtain FIGURE 7, select the Plot
tab from the Results dialog window, and select the following options:
• Plot tab > Action: Select > Object: All Elements > Method: Selection >
Click on the nonlinear solution in the scrolling list > Plot: Solution >
Contour: Fringe on > Functions: Eeq (the equivalent strain) > Mid-
sides: 7 (this defines the data mesh) > Range min: 0.0, Range max:
6.93e-3 > Intervals: 2. This will ensure that there will be a contour line
at 3.465x10-3, which is one half of the specified range. Finally, select
the Gray button and then click on the Plot button.
The plastic zone will be shown in dark. The results indicate that the solution is
of good quality: The contour line does not “jump” as it passes over element
boundaries.
FIGURE 7 The plastic zone.
To compute the stress resultant, select the Resultant tab from the Results dialog
window, click on the nonlinear solution in the scrolling list; make sure that the
Fx button is on and click on the edge corresponding to the edge EF in FIGURE
5. You will obtain Fx=1325, that is, the force corresponding to the imposed
displacement ∆ is 1325 lbf (per 1/2 inch length).

1
Remark: Unlike in the case of linear problems, the amount of CPU time to execute
a Nonlinear analysis depends on the magnitude of loading. In this example loading
was by imposed displacement and the resulting plastic zone was confined by elastic
material. The number of iterations depends on the value of ∆: For smaller values of
∆ the solution is closer to the linear solution and hence smaller number of iterations
are needed. For larger values of ∆ larger number of iterations are required. As ∆ is
increased, the values of the strain increase also and the applicability of the small
strain and small displacement assumptions becomes questionable. Thus, the prob-
lem of modeling errors must be considered. In general, it is good practice to use the
linear solution to estimate the level of loading which will not cause extensive plas-
tic flow, the treatment of which would lie outside of the scope of small strain mod-
els.
In general, the largest strain of the linear solution will become even larger for the
nonlinear solution. The spread of the plastic zone can be anticipated by examining
the equivalent stress (Seq) contour plot of the linear solution. The size of the plastic
zone will be larger than the region for which Seq > Syield.
Exercise Investigate the sensitivity of the results obtained with respect to the spring stiffness
kn .
Discussion concerning the Exercise: It is best to determine the spring rate of fasten-
ers experimentally. Unfortunately, experimental data are rarely available. A simple
model is described in the following which will provide an order of magnitude esti-
mate of spring stiffness for fasteners. In problems such as this, it is important to
take into consideration uncertainties in the input data. For further discussion on fas-
tener models see Ref. [4] and [5].
Visualize the fastener as a cylindrical body subjected to sinusoidally distributed

normal stress over half of its perimeter, as shown in FIGURE 8a. From consider-
ation of equilibrium, the shear stress distribution can be approximated by the for-
mula:
VQ
τ = ------
It
where V is the shear force, equal and opposite to the resultant of the sinusoidally
distributed normal stress; Q=Q(y) is the static moment with respect to the centroi-
dal x-axis of part of the circular cross section bounded by the line y=constant. This
is the shaded area in FIGURE 8b. I is the moment of inertia and t is the chord length

y
Sinusoidal y
normal stress
σn
(a) t y
x x
(b)
Body force
to represent
shear
FIGURE 8 Fastener model and notation.
shown in FIGURE 8b. Since the effects of bending can be generally neglected,
the shear stress can be replaced by body forces of the same distribution. Solv-
ing this two-dimensional problem makes it possible to compute the ratio
between the normal stress and the normal displacement along the loaded part
of the fastener. While this ratio is not constant, it is fairly close to being con-
stant in the vicinity of the maximum normal stress. Therefore we can approxi-
mate kn by the ratio of the maximum normal stress and the corresponding
normal displacement.
The Handbook utility contains a pre-defined problem for estimating kn. The
fastener pin is characterized by its diameter, Young’s modulus and Poisson’s
ratio. Plane strain is assumed. To perform the computation, select the “Spring
stiffness for a pin” problem from the 2D-Basic handbook (file: spring_rate.sci),
and follow the instructions to obtain an approximate value for the spring rate.
Computation of residual stresses

StressCheck makes it possible to estimate residual stresses caused by a loading
into the plastic range, followed by unloading. The method is applicable only
when the unloading does not produce plastic deformation. This is illustrated by
two examples. One in which the loading is applied by tractions, and another in
which the loading is the result of an imposed displacement.

1
Example 1 Loading by tractions. Consider a simply-supported beam loaded by uniform nor-
mal pressure Pr (lb/in2), as shown in FIGURE 9. The beam has a square cross sec-
tion and the material is elastic-ideally plastic with the following properties (ASTM
A-36): E = 29x106 psi, v = 0.295, σy = 36000 psi, plane-stress. The objective of the
analysis is to find the normal stress distribution along the symmetry section (C in
FIGURE 9) and estimate the residual stress in that section once the load is
removed.
y
Pr
y
z
h
x
VA VB
A C B b
L/ 2 L/ 2
Pr = 170 lb/in2
b = h = 1 in
FIGURE 9 Simply-supported beam (not to scale).
L = 20 in
Solution: Taking advantage of symmetry, the solution domain and boundary condi-
tions are as shown in FIGURE 10. The symbols T and u refer to the traction and
displacement vectors and the subscripts n and t refer to the normal and tangential
directions, respectively.
The tangential traction along edge IH corresponds to the parabolic shear stress dis-
tribution from the support reaction at B. From elementary beam theory:
P bL
r
V
B
= ---------
2
-
VB b h2
T = τ = ----- -  ---- – y 2
t Ib 2  4 

y
FG Symmetry: un = 0, Tt = 0
F I
x GH Free: Tn = 0, Tt = 0
h/2
G H IH Imposed traction: Tn = 0, Tt = 60 Pr (0.25-y2)
L/2
FI Imposed traction: Tn = -Pr, Tt = 0
where I = b h3/12, is the moment of inertia of the cross section. Therefore,
P L 2
T
t
r
- [ 1- – ( -y ) ] =
= 3 ------ 60P r ( 0.25 – y2 )
h 4 h
Because the exact solution for this problem is smooth, using only one finite
element is sufficient. The mesh is shown in FIGURE 11. Since the sum of the
forces in the y-direction resulting from the specified traction is equal to zero, a
nodal constraint in the y-direction has been specified at H to prevent rigid body
translation.
FIGURE 11 Finite element mesh and boundary conditions.
On performing a planar linear analysis, uniform p-extension, trunk space, from

p=1 to 5, we find that the estimated relative error in energy norm is zero for
p ≥ 4 (note that the exact solution given by the beam theory is a polynomial of

1
degree 4). Therefore we perform the material Nonlinear analysis for p=4. Specify-
ing the Convergence: Stress, Tolerance: 0.5%, we obtain the solution in less than 15
iterations. To compute the stress distribution along the symmetry section, select the
Points tab in the Results dialog window, choose Select > Edge > Selection, click on
the nonlinear solution of the scrolling list and enter Run 4 to 4, select # of pts: 25,
and click on the constrained edge FG.. Select the functions Sx and Calc. from the
StressCheck Function box. Use the calculator box as explained in the User’s Guide,
entering the names of the linear and nonlinear solutions; the corresponding run
numbers (four in our case); and the following formula:
Formula: Sx_2 - Sx_1
where the subscript 1 is associated with the linear solution and the subscript 2 with
the nonlinear solution. Click on the Accept button. The results are shown in FIG-
URE 12. The formula provides an estimation of the residual stresses along the sym-
metry section of the beam which is the difference between the normal stress of the
linear solution from the normal stress of the elastic-plastic solution.
According to elementary beam theory, the bending moment at section C is given by
2
P r bL
M
c
= -----------
8
- = 8500 in-lb

FIGURE 12 Nonlinear and residual stress distribution along center line.
and the elastic, elastic-plastic and residual stress distributions are as shown in
FIGURE 13, where yY represents half the thickness of the elastic core and is
obtained from the following expression [41]:
2
1 2y Y
– 3-  --------
3
M = -2 M Y 1
 h 
where M is the bending moment at the symmetry section, MY is the value of M

at the onset of yielding:
2
bh
MY = ------- σ Y
6
and c = h/2.
On comparing FIGURE 12 and FIGURE 13 the close agreement between the

solutions of the two-dimensional problem and the beam theory is apparent.

y y y
yY=0.2
h/ 2
σx + σx = σx
h/ 2
36 ksi 51 ksi 15 ksi
elastic-plastic linear residual
FIGURE 13 Elastic, elastic-plastic and residual stress distribution.
There are two relevant differences between the elasticity and the beam theory mod-
els of this problem. First, in elasticity the applied load produces a different state of
stress at the top surface of the beam than at the bottom surface of the beam, which
is ignored by the beam theory. Second, in the two-dimensional elasticity model,
Poisson’s ratio effects are present which are not accounted for by the beam theory.
Example 2 Loading by imposed displacement. Consider a square plate with a central hole as
shown in FIGURE 14. An oversized mandrel is driven through the hole to induce a
plastic zone around it. The objective of the analysis is to estimate the residual
stresses after the mandrel is removed. The interference δ (the difference between
the diameter of the mandrel and the diameter of the hole) was simulated by a dis-
tributed spring along the edge of the hole with an imposed radial displacement
equal to half the magnitude of the interference (δ/2). The spring coefficient was
selected to simulate the radial stiffness of the steel mandrel. FIGURE 14 shows the
configuration of the plate and the finite element mesh with the boundary condi-
tions.
To define the model, select Planar > Elasticity > in/lbf/sec/F under Reference/The-
ory./Units. Because of symmetry, only a quarter of the plate was considered for the
analysis. Note that a layer of elements was placed around the hole on a circle of
radius d.

Plate Material 2024-T3:

Y E = 10.6 x 106 psi
ν = 0.33
W σY = 48000 psi
X Plane Stress
d
Mandrel Spring Coefficient
Kn = 15.35 x 107 lb/in3
W = 0.875 in
d = 0.1945 in
th = 0.071 in (thickness)
δ = 0.002 in (interference)
FIGURE 14 Plate configuration and mesh.
Nonlinear analysis: The elastic-plastic analysis was performed using a stress-

strain law corresponding to an elastic-perfectly-plastic material. A sequence of
linear solutions were obtained for polynomial degrees ranging from 1 to 8. The
quality of the solution was assessed by computing the estimated relative error
in energy norm. For p=8 the error was less than 0.1%. The run corresponding
to p=8 was used as the initial solution for the nonlinear iteration. Specifying a
tolerance of 0.1% in stress, the material nonlinear analysis takes less than 10

1
iterations. FIGURE 15 shows the size of the plastic zone and the von Mises stress
distribution along the horizontal symmetry line.
FIGURE 15 Nonlinear analysis: Plastic zone and von Mises stress distribution.
Residual Stresses: The effect of unloading of the plastic zone cannot be repre-
sented by the deformation theory of plasticity. Therefore, an alternative procedure
can be used to estimate the residual stresses after the mandrel is removed. Unlike in
the case of loading by tractions, it is not possible to simply combine the linear and
nonlinear solutions as described in the previous example. However, a superposition
procedure is still possible. The basic idea is that when the mandrel is removed, the
free-traction condition has to be satisfied in the radial (normal) direction around the
hole.
The traction distribution around the hole when the mandrel is still in place can be
computed from the linear and nonlinear solutions. The average radial stress for the
linear solution is TL=-51100 psi, while for the nonlinear case is TNL=-42450 psi. To
compute the radial stress, select the Points tab in the Results window, select the
function Sx and the name of a local cylindrical coordinate system in the corre-
sponding fields, and select the circle with the mouse cursor. If your model does not

1
have a local cylindrical coordinate system, create one at the origin before pro-
ceeding (ignore any warnings about the model having been changed since the
last solution was computed). Note that when the local system type is cylindri-
cal, the function Sx corresponds to the radial stress Sr.
(a)
(b)
FIGURE 16 Radial stress distribution around the hole:

(a) linear solution, (b) nonlinear solution.
To compute the average radial stress, simply select the option Average, the
function Sx, the name of the local cylindrical system and the object Edge.
Select the two edges around the hole and the click on Accept. The program will
compute the integral average of the radial stress. Refer to Post-Processing
Operations in the User’s Guide for explanation about average stress computa-
tions.
Since the radial stress around the hole is fairly uniform (note the range of Seq
values on the y-axis in FIGURE 16), the residual stress (σR) is estimated by
subtracting the linear solution from the nonlinear solutions after multiplying
the linear solution by the ratio between the average radial stress of the nonlin-
ear solution and the average radial stress of the linear solution:
T NL 42450
σ R = σ NL – --------
- σ = σ NL – --------------- σ L
TL L 51100
This transformation is required to satisfy the stress-free condition along the

hole when the mandrel is removed. The residual stress along the symmetry line

1
is shown in FIGURE 17 as determined using the above expression. In this case the
calculator used the following formula:
Sy_1 - 42450/51100*Sy_2
where Sy_1 is the σy stress component corresponding to the nonlinear solution and
Sy_2 is the σy stress component corresponding to the linear solution. This
approach for estimating the residual stresses is valid if (a) the stress distribution
around the hole is nearly constant, and (b) if the value of the residual stress is below
the yield strength of the material (i.e., reverse yielding does not occur).
FIGURE 17 Residual stress distribution along symmetry line.
An automatic procedure for the computation of residual stresses due to cold-work-

ing operations in holes is one of the options of the Nonlinear capabilities of Stress-
Check. See Coldwork analysis for details.

Isotropic Exponential Material Properties
1
Isotropic Exponential Material Properties
StressCheck supports a nonlinear elastic material law based on an exponential
strain energy density function of the form:
2CW L 1 T
W = WL e , W L = --- { ε } [ E ] { ε }
2
where W is the strain energy density, WL is the linear strain energy, C is a con-
stant, e is the base of the natural logarithm, [E] is the linear material stiffness
matrix and {ε} is the strain tensor.
This strain energy density function has the following characteristics:

• W depends on three parameters: E is the slope of the one-dimensional
stress-strain curve at ε=0, ν is the Poisson’s ratio and C is a constant
with units of length2/force.
• When {ε} goes to zero, W also goes to zero.
• The constant C controls the type of stress-strain relationship:
If C = 0, W = WL
If C > 0, W > WL
If C < 0, W < WL
The stress components are obtained by differentiation of the strain energy

function:
∂W 2CW L
{σ} = = ( 1 + 2CW L )e [E]{ε}
∂{ε}
This relation is used during the nonlinear iteration procedure when solving a
material nonlinear problem.
To specify an isotropic exponential material in StressCheck, select the Create

Model icon from the Main Toolbar and then the Material tab. In the Material
combo box select Nonlinear Elastic and in the Type combo box select Iso.-
Exp., and provide the required input coefficients: E, ν, C as indicated above. A
one-dimensional representation of the nonlinear stress-strain curve will be
available after clicking on the Accept or Plot button of the material input box.
FIGURE 18 shows the stress-strain curve for the case of E= 790 and C = 0.00,
0.01, -0.01 respectively.

General nonlinear analysis
C = 0.0 C = 0.01 C = -0.01
FIGURE 18 Stress-strain curve for an isotropic exponential material.

The implementation includes nonlinear strains and large deformations with the
material description being linear-elastic or elastic-plastic. The General Nonlinear
Analysis capability makes it possible to satisfy equilibrium in the deformed config-
uration. This is very important when the deformed configuration differs substan-
tially from the undeformed one. For elastic-plastic materials, the von Mises yield
criterion is used together with the deformation theory of plasticity.
To solve a general nonlinear problem, a linear problem must be solved first.

• If the material is specified as linear-elastic then the general nonlinear analy-
sis will account only for changes in geometry (geometrically nonlinear
problems).
• If the material is specified as elastic-plastic, then the analysis will account
for changes in geometry and material during the iteration process.
The input data preparation to run a general nonlinear problem is identical to that for
linear problems. However, there are some important considerations which must be
kept in mind when preparing the data for a general nonlinear analysis:
a) Loads: If a traction is specified in the normal/tangent reference frame, it

becomes a “follower load”, that is, will remain normal/tangent throughout the
deformation. If a traction is specified in a global or local system, the direction of the
load will remain fixed. Bearing or formula loads can also be specified for general
nonlinear analysis.

1
b) Constraints: Rigid body constraints are not meaningful for general nonlin-
ear problems. The reason is that a body originally in equilibrium under the
applied tractions is likely to become non-equilibrated due to the deformation.
This is because the loaded boundaries may shrink, stretch, or simply move,
causing the resultants of the applied tractions to change. Also symmetry and
antisymmetry should be used carefully, because under large deformations sym-
metric or antisymmetric conditions may not be preserved.
b) Springs: When distributed

Un Dn
springs are specified over element
edges (Planar and Axisymmetric)
or faces (3D), and the spring coef-
Y Kn
ficients are defined in the global or
in a local coordinate system, they s
are treated in the same way as in
the linear analysis, namely they
remain “attached” to the bound-
aries whether under tension or (e)
compression. If the spring coeffi-
cients are specified in the normal
direction however, they will
X
remain attached to the boundaries
only if they are in compression and will be set to zero if in tension. If Un repre-
sents the normal component of the displacement along a boundary (edge or
face) of an element (e), Dn is the imposed normal displacement on the spring
and Kn is the spring constant, then in the course of the nonlinear analysis the
spring constant at each integration point along the boundary (s) is evaluated as
follows:
Kn(s) = Kn, if (Dn - Un) < 0
Kn(s) = 0, if (Dn - Un) ≥ 0
By definition, a normal displacement is positive in the direction of the positive

(outward) normal to the boundary.
Executing general nonlinear analysis

Having entered the input data, execute a linear analysis. It is recommended that
you obtain a sequence of solutions with p ranging from 1 to 8. Next, estimate
the relative error in energy norm. At least one of the solutions in the sequence

1
should have a small relative error. In general, you should expect the accuracy of the
nonlinear solution to be somewhat lower than the accuracy of the linear solution.
Select the Compute Solution icon from the Main Toolbar, then select the Nonlinear
tab and proceed as follows (FIGURE 19):
1. Click on the linear solution which is to be the starting solution for the nonlinear
iterations. The solution name, run number, type, and associated degrees of freedom
(DOF) appear in the scrolling list of the dialog box. The corresponding discretiza-
tion will remain fixed in the course of the nonlinear solution process.
2. Enter a name for the nonlinear solution which is different from the name of the
linear solution. For example, you can modify the name of the linear solution by
adding the letters NL (for nonlinear).
3. Select the Type: General (NLGen).
4. Select Convergence, Technique and allowable Tolerance. For general nonlinear

analysis, the convergence option and the effect of the specified tolerance depends
on the type of material.
• For linear-elastic materials, the iterations will continue until the difference
between two consecutive solutions is less than the tolerance, regardless of
the convergence option selected. Denoting by Xi the solution vector at the
ith iteration, and the tolerance by τ, this criterion is satisfied when:
1 1
--- ---
 i – 1 2 
2
i 2
2
∑ ∑
i
 ( Xk – Xk )  ≤ τ ( Xk )  . (3)
 k
  k

The significance of equation (3) is that the iteration will proceed until the
difference in the deformed configuration between the current and previous
cycle is less than the prescribed tolerance.
• For elastic-plastic materials, the convergence option “Stress” means that
the iteration will proceed until the convergence criteria given by equations
(1) and (3) are both satisfied. For the convergence option “Energy”, the
convergence criterion of equation (3) is the only one considered.
Two techniques are available for general nonlinear problems: The direct integra-
tion, which is the default should be used for most problems because is faster and
can handle the vast majority of applications. The Newton-Raphson algorithm is

1
applicable for those cases in which there is a strong coupling between mem-
brane and bending effects when analyzing thin domains.
5. Select the number of iteration cycles you wish StressCheck to perform auto-
matically and enter this number into the field next to Iteration Limit. Stress-
Check will stop when the Iteration Limit is reached unless the specified error
tolerance is reached first. When StressCheck stops at the Iteration Limit you
can examine the results and decide whether to continue or not. If you wish to
continue then simply enter a higher value for Iteration Limit, then select the
SOLVE! tab, and click on the Solve button.
6. Select the Load Steps button. This option is used for applying the load in
increments from the minimum to the maximum specified load in equal incre-
ments. The load step factor is applied to the selected load parameter. The Load
Steps are the number of steps from the minimum to the maximum load that are
applied to any selected load parameter defined in the model. If the load is to
remain constant, turn Load Steps off.
In most problems involving large deformations, the load has to be applied in

steps to achieve convergence. Therefore the selection of the initial value of the
load and the number of load steps are essential to perform a general nonlinear
analysis. The current implementation does not support an automatic load step
selection, so some guidelines are provided in the following:
• Select the initial load in such a way that the overall elastic deformation
of the object under analysis is of the order of 10% of the characteristic
dimension of the object. For example, considering the case of a cantile-
ver beam-like structure under transverse loading, the magnitude of the
initial load should be selected such that the maximum deflection com-
puted from the linear solution is not larger than 10% of the length of
the beam.
• Select the number of load steps based on the ratio between the maxi-
mum load and the initial load. The larger the ratio, the larger the num-
ber of load steps. If the error displayed in the status region of the
nonlinear execution box increases two or more times in a row with
respect to the previous solution then the number of load steps is not
large enough. Increase the number of load steps and repeat the proce-
dure.
When the Load Steps option is turned on, you will be able to select the load
parameter and enter the maximum value for the load (FIGURE 19). The mini-
mum load value is displayed but cannot be changed. You also will have the

1
option of selecting the Save Load Steps toggle. This option will save all converged
intermediate load steps for further post-processing. Each saved step will be identi-
fied with the name given in (2) above, an underscore ‘_’ followed by the run num-
ber of the starting linear solution. For example, NSOL_8 means that the solution
NSOL was initiated from the linear solution #8. The run number associated with
each nonlinear solution will correspond to each load step.
FIGURE 19 Typical settings for General Nonlinear analysis.
7. Select the SOLVE! tab to set the mode of execution. The options are: Initialize
and Restart. Select Initialize when you begin a general Nonlinear analysis, and
select Restart when you wish to continue from an existing nonlinear solution.
matic then the iterations will continue until you reach the specified iteration limit or
satisfy the specified error tolerance. If you select Stepwise then the iteration process
stops after each cycle, and you need to click on the Solve button to go to the next
iteration cycle.
10. Select Display. You can monitor the value of functions at points, maximum val-
ues of functions or contour plot of functions during the nonlinear iteration. Display-
ing the deformed configuration while executing general nonlinear problem in

1
Stepwise mode is a good strategy for visually checking whether the selected
load step factor is adequate.
11. Click on the Solve button to start the nonlinear analysis.
After the analysis is completed, the nonlinear solution just obtained will be
inserted in the scrolling list of the Results dialog box with the name you
entered in step 2, the same run number and DOF as the linear solution selected,
and with a type NLGen. Selecting a different run number of the same linear
solution to start a nonlinear iteration will result in a new nonlinear solution
record with different degrees of freedom and run number than the previous
one. Both nonlinear solutions will be available for extraction. You can obtain
as many nonlinear solutions as there are linear solutions. In this way you can
check the convergence characteristics of the nonlinear solutions with respect to
increasing number of degrees of freedom.
Example 1: Cantilever beam

Consider the cantilever beam shown in FIGURE 20. The beam has a square
cross section of unit size, linear-elastic material properties characterized by
E=12000 lb/in2 and ν=0.20 and it is loaded by a normal uniform pressure Pr
(lb/in2).
y
Pr
CROSS SECTION
MA
NA x
h
A B
VA L b
L = 50 in E = 12000 lb/in2 Pr = 0.08 lb/in2

b = h = 1.0 in v = 0.2
FIGURE 20 Cantilever beam under uniform pressure (not to scale).

1
The objective of the analysis is to compute the end deflection of the beam and the
reactions at the built-in edge. From elementary beam analysis, the end deflection
given by the linear theory is:
– Pr bL 4
y max = --------------- = – 62.50 in
8EI
3
bh
where I = --------- is the moment of inertia of the beam cross sectional area.
12
The values of the bending moment, shear and axial load at the built-in edge are:
2
P r bL
MA = ----------- = 100 in-lb
2
VA = P r bL = 4 lb
NA = 0
Note that the maximum deflection predicted by the linear theory is larger than the
length of the beam, hence the solution is well beyond the range of validity of linear
beam theory.
Solution: The finite element mesh and boundary conditions are shown in FIGURE
21. Because the exact solution of this problem corresponding to theory of elasticity
has corner singularities at the built-in end, a small quadrilateral element was
located at the fixed edge with a size equal to the beam depth to isolate the singular
points. This problem can be loaded from the Tutorial Handbook (file:
BeamNLG.sci).
The load has been defined acting normal to the upper surface of the beam, in terms
of a parameter Pr, with a value of 0.008, which corresponds to one tenth of the tar-
get load. Note that when Pr=0.008 lb/in2, the maximum deflection of the beam will
be of the order of 10% of its length.
On performing a planar linear analysis, uniform p-extension, trunk space we find

that the estimated relative error in energy norm is under 1% for p=6. Furthermore,
when computing the stress resultants for the built-in edge, it is seen that the shear
force and bending moment are practically independent of p for p>2, therefore we

FIGURE 21 Finite element mesh for example problem.
perform the general nonlinear solution for p = 6, 7 and 8 so that the dependence
of the nonlinear solution on the number of degrees of freedom can be exam-
ined.
To run the nonlinear analysis select the following execution parameters from
the nonlinear dialog window:
• Nonlinear tab > Linear Sol: SOL > Run: 6 > Nonlinear ID: NLSOL >
Type: General (NL Gen) > Convergence: Stress > Technique: Direct
Int. > Tolerance (%): 0.01 > Iteration Limit: 15 > Load Steps: 5 > Load
Param.: Pr > Load Max: 0.08 (do not turn on the Save Load Steps tog-
gle).
Iterative > Display: Results > Solve button.
After setting the plot interface in the Results dialog window, as indicated in
FIGURE 22, click on the Solve button. Please note that the Scale has been set
to 1.0, so that the deformed shape is not automatically scaled by the program
after each iteration. FIGURE 22 shows the settings and the deformed configu-
ration of the beam at the end of the nonlinear analysis for p=6.

FIGURE 22 Plot window and deformed configuration for nonlinear run #6.
Repeat the analysis for p=7 and 8. To do this, simply select the linear solution run
#7 from the scrolling list of the Nonlinear window, switch back to the SOLVE! win-
dow, and click on Solve. After the solution is completed, repeat for run #8.
To compute the beam end deflection, select the Points tab in the Results dialog win-
dow. Click on the last nonlinear solution available in the scrolling list, select
Object: Node and click on the lower node at the free end of the beam. Select the
function Uy and then click on the Accept button. The results are shown in FIGURE
23. Note that the deflections computed from the three nonlinear solutions are very
close to each other and that they converge to a limit value (-40.86) which is sub-
stantially different from the one computed from elementary beam analysis.
To compute the bending moment and the shear and axial forces at the built-in edge,
select the Resultant tab from the Results dialog window, click on the last nonlinear
solution available in the scrolling list, select Action: Check, Object: Edge, and turn
on the Fx, Fy, Mz buttons. Click on the built-in edge to perform the computation.
The results will be shown in the Results window and in the Graph window as
shown in FIGURE 24. Note that because the load follows the beam, the axial force

FIGURE 23 Beam end deflection as a function of DOF.
is now different from zero, and in fact larger than the shear force. Also the
bending moment is less than that computed from the linear beam theory.
Example 2: Lug with bearing load

Consider the model problem shown in FIGURE 25. The model represents a lug
subjected to a bearing load of magnitude F=15000 lb at 30 degrees angle with
respect to the global X-axis, and constrained by two fasteners. The fasteners
are represented by distributed normal springs (kn=1x108) acting in the perime-
ter of each fastener hole. Elasto-plastic material properties are considered for
the lug (E=29x106, v=0.3, σy=36000). The objective of the analysis is to
determine the normal (radial) stress distribution along each fastener hole.
This problem can be loaded from the Tutorial Handbook. Select the Handbook
Library icon from the Main Toolbar, and when the Library interface appears,
click on the Model Browser icon and navigate to the Tutorial folder. You can
scroll through the list of entries using the scroll bar. If you double-click on the
“Lug.sci” problem, the corresponding model will be loaded into StressCheck
for analysis. After the problem is loaded, select the Compute Solution icon
from the Main tool bar and when the Solver interface appears select the Linear
tab. Set to run from p=1 to 8 in trunk space. Select the SOLVE! tab and click
on the Solve button. Once the linear analysis is completed, switch to the Non-

FIGURE 24 Resultants at the built-in edge

of the beam.
linear tab. Execute a General Nonlinear analysis selecting the following execution
parameters from the nonlinear solver user interface:
• Nonlinear tab > Linear Sol.: SOL > Run: 8 > Nonlinear ID: NSOL > Type:
General (NLGen) > Convergence: Stress > Technique: Direct Int. > Toler-
ance: 0.5% > Iteration Limit: 5
• SOLVE! tab > Initialize > Automatic > Iterative > Solve button.
The nonlinear execution should be completed after 3 iterations. To extract the radial
stresses along each hole, select the Points tab from the Results interface and com-
plete the following information for the top hole:
• Points tab > Select > Any Boundary > Selection > Solution: NSOL > Run:
8-8 > Func(s): Sx > # of pts.: 359 > P1-Min:0, P1-Max: 1 > System: SYS4.
Click on the circle of the top fastener hole and then click on the Accept but-
ton.
You may have to turn the boundaries on by selecting the Display Curves icon from
the Display Objects Toolbar. The results are shown in FIGURE 26a.

F=15000
FIGURE 25 Lug under bearing load. All dimensions are in inches.
By selecting 359 intermediate points along the circle, they will be located at
one degree intervals. Note that when the local system type is cylindrical, the
function Sx corresponds to the radial stress Sr.
To extract the radial stress distribution for the lower fastener hole, complete the
point extraction box as follows:
• Points tab > Select > Any Boundary > Selection > Solution: NSOL >
Run: 8 to 8 > Func(s): Sx > # of pts: 359 > System: SYS3. Click on the
circle of the lower fastener hole and then click on the Accept button.
The results are shown in FIGURE 26b.
Note that the radial stresses are nearly zero along about half of the hole perim-
eter (where the spring “detached” from the boundary) and compressive along
the other half (where it remained “attached” to the boundary). Along the lower
hole a minor oscillation in the stress distribution is apparent near the point of
detachment. This behavior is typical whenever the detachment point is not
coincident with a node.
To compare these results with the linear solution, change the solution name in
the point extraction box to SOL and repeat the extraction procedure as before.
The results are shown in FIGURE 27. Note that in the linear solution both ten-
sile and compressive radial stresses developed around each hole.

SYS4
(a)
(b)
SYS3
FIGURE 26 Radial stress distribution for nonlinear solution.
Example 3: Single lap joint
Consider a single lap-joint specimen consisting of two aluminum plates bonded by

a layer of adhesive as shown in FIGURE 28. This case represents the type of prob-
lems for which the Newton-Raphson integration algorithm overcomes the limita-
tions of the direct integration procedure when performing a geometric nonlinear
analysis.
Description The specimen is constrained in the normal direction along two edges on the left
side, and along one edge on the right side, as shown in FIGURE 28. A load is
applied at one end as a distributed traction with an initial value PL=1000 lb. The
overlap length is Lo=0.5 in., the plate thickness is tp=0.03 in., and the adhesive

SYS4
(a)
(b) SYS3
FIGURE 27 Radial stress distribution for the linear solution.
thickness is ta=0.01 in. The material properties for the aluminum plates are
those of a linear elastic material (E=10.0x106 psi, ν=0.33), and for the adhe-
sive, those of an elastic perfectly plastic material (E=251,900 psi, ν=0.34,
σY=11,600 psi). This problem is analyzed using 2D plane-strain conditions and
can be loaded from the Tutorial Handbook (file: SingleLapJoint1.sci).
Linear solution The 2D finite element mesh was constructed using 54 quadrilateral and 22 tri-
angular elements and the mesh was graded in geometric progression towards
the singular points as shown in the enlarged view in FIGURE 28. A sequence
of linear solutions was obtained for polynomial orders ranging form p=1 to
p=6. The estimated relative error in energy norm, as a function of the number
of degrees of freedom (DOF), and the deformed configuration in scale 1:1 for
the solution corresponding to p=6, are shown in FIGURE 29.

FIGURE 28 Single lap-joint specimen. Notation and mesh.
Nonlinear solution The solution corresponding to p=6 (2957 DOF) was used as the starting solution for
the general nonlinear analysis, with the execution settings are shown in FIGURE
30:
General (NLGen) > Convergence: Energy > Technique: Newton-Raph. >
Tolerance: 0.5% > Iteration Limit: 20 > Load Steps: 7 > Load Param.: PL >
Load Max.: 4000 > Save Load Steps on.
• SOLVE! tab > Initialize > Automatic > Iterative > Solve button.
Seven load steps were specified in 500 lb increments and the convergence criterion
was set to energy with an allowable value of 0.5%. When performing a general
nonlinear analysis, both the material and geometric nonlinearities are accounted for.
Because of the singularities present at both ends of the adhesive layer, an energy
convergence criterion was selected for this problem.

FIGURE 29 Single lap-joint. Linear solution: Estimated relative

error in energy norm and deformed shape.
As the solution proceeds, the status windows under the SOLVE! tab interface
shows the progress of the execution. For example, the status shown in FIG-
URE 30 reports that a general nonlinear analysis is running iteration 3 of 20 of
the third load step out of 7, and the error of the previous iteration was 1.87% in
energy which is larger than the allowable tolerance of 0.5%.
Results The deformed shape in scale 1:1 is shown for three values of the load in FIG-
URE 31, while FIGURE 32 shows the von Mises stress distribution in the
adhesive for PL=1000, 2000 and 3000 lb.
The largest load for which convergence was realized was PL=3000 lb. For
PL=3500 lb, the elements around the singularities became so distorted that
convergence within the specified tolerance was not realized in 20 iterations
(the solver will fail while running load step #6). The adhesive layer became
fully plastic and that was a contributing factor for the large distortion of the
elements in the adhesive. This is shown in FIGURE 33, where the deformed

FIGURE 30 Single lap-joint. Execution settings for general nonlinear analysis.
FIGURE 31 Single lap-joint. Nonlinear solution: Deformed shape

for various loads.
configuration of the lap-joint is shown for the nonlinear solution corresponding to

the last iteration. The extreme distortion of the elements near the two singular
points is clearly visible.
Alternative mesh An alternative meshing was considered around the singular points to prevent the
high distortion of the elements for large plastic deformation of the adhesive. The
new mesh consisting of 46 elements (44 quads and 2 triangles) is shown in FIG-
URE 34. Note that in the new mesh two semi-circular regions were created to

FIGURE 32 Single lap-joint. Nonlinear solution: von Mises stress

distribution in the adhesive.
FIGURE 33 Single lap-joint. Nonlinear solution: Deformed shape for PL=3500 lb.

1
remove the sharp termination of the adhesive. This mesh can be loaded from the
FIGURE 34 Single lap-joint. Alternative mesh.
Tutorial Handbook (file: SingleLapJoint2.sci).
As before, a sequence of linear solutions was obtained for polynomial orders rang-
ing form p=1 to p=6, and the solution corresponding to p=6 (1849 DOF) was used
as the starting solution for the general nonlinear analysis with the same execution
settings as shown in FIGURE 30. The estimated relative error in energy norm for
the sequence of linear solutions and the deformed configuration in scale 1:1 for the
solution corresponding to p=6, are shown in FIGURE 35. As before, the error in
energy norm is under 1% for p=6.
The largest load for which convergence was realized in this case was PL=3500 lb.
After that, convergence cannot be realized because the plasticity in the adhesive is
uncontained. This can be seen in FIGURE 36 which shows the von Mises stress
distribution in the adhesive for several values of the load. There are two advantages
associated with the use of the alternative mesh around the singularities. One is that
the distortion of the elements due to the combined effect of large deformations and
plasticity is minimized; and two, the number of iterations to achieve convergence
for each load increment is reduced.

Coldwork analysis
FIGURE 35 Single lap-joint. Linear solution: Estimated relative

error in energy norm and deformed shape for the alternative mesh.
Coldwork analysis
StressCheck provides a unique capability for simulating the effects of cold
working. This capability is based on the deformation theory of plasticity and
utilizes superposition. The cold-worked body is represented by a plane stress
or plane strain model. A detailed description follows.
When a mandrel is introduced into a hole in a structural component, a plastic

region develops around the hole, the size of which depends on the interference
level, the material properties of the mandrel, and the material properties and
topological configuration of the component. The extent of the plastic zone can
be accurately determined using the deformation theory of plasticity. When the
mandrel is removed, the traction-free condition in the radial direction around
the hole is restored, and residual stresses are locked into the component.

Coldwork analysis
FIGURE 36 Single lap-joint. Nonlinear solution: von Mises stress

distribution in the adhesive for the alternative mesh.
One important limitation of the deformation theory of plasticity is that unloading

events are not represented, and therefore the residual stresses must be determined
indirectly by superposition. When an external load is subsequently applied to the
part, the stress pattern will be affected by the residual stresses created during the
cold-working operation, and again a superposition procedure must be used. The
superposition procedure implemented within the Coldwork option can be initiated
after the linear analysis of the component for the user’s specified loads and con-
straints was completed. Once the hole to be cold-worked is identified, the following
analyses are performed:
1. An elastic-plastic analysis of the component when the mandrel is introduced into

the hole.
2. A linear analysis of the component with the specified constraints and loaded with
a self-equilibrated radial traction around the cold-worked hole, the distribution of
which is determined from the elastic-plastic analysis indicated above.
When the mandrel is removed, the traction-free condition must be restored in the
radial (normal) direction around the cold-worked hole. The traction distribution
around the hole when the mandrel is still in place is computed from the elastic-plas-
tic solution:

Coldwork analysis
1
T NL = σ x cos 2θ + σ y sin 2θ + 2τ xy sin θ cos θ (4)
where σx, σy and τxy are the nonlinear stress components around the hole in
the global coordinate system.
The effect of removing the mandrel is

accounted for by solving a linear prob- y
lem with the cold-worked hole loaded σy
with the radial traction distribution
induced by the mandrel. The traction
τxy
distribution given by (4) is approxi- σx
TNL
mated by a Fourier series expansion in
the form: θ
x
N
a
T R = ----0- +
2 ∑ a cos ( nθ ) + b sin ( nθ ) (5)
n n
n=1
where TR is an approximation of TNL, θ

is the angle measured from a polar coordinate system located at the center of
the hole, and the coefficients ai and bi are computed from:
2π 2π 2π
1 1 1
a 0 = ---
π ∫ 0
T NL dθ , a n = ---
π ∫ 0
T NL cos ( nθ ) dθ , b n = ---
π ∫ 0
T NL sin ( nθ ) dθ
The integrated traction is self-equilibrated, provided a1 = b1 = 0, therefore this

constraint is enforced. The expansion includes 10 terms, and TNL is computed
at 360 equally spaced points around the hole. The quality of fit is determined
by computing the least squares error between the actual and approximating
functions as follows:
360
∑ [T (θ ) – T
2
R k NL ( θ k ) ]
2πk-
e L2 = k=1
---------------------------------------------------------- , θ k = -------- (6)
360 360
∑
2
TR ( θk )
k=1

Coldwork analysis
1
Once the traction distribution is computed, the program performs a linear analysis
of the component with the originally specified constraints and when the only
applied loading is the traction around the hole given by Eq. (5).
After completion of the Coldwork analysis, the following information is available:

• The elastic-plastic solution due to the cold-working operation.
• The solution corresponding to the residuals after the mandrel was removed.
• The solution of the problem with the specified loads and the residual
stresses combined.
The applicable stress-strain laws include elastic-perfectly plastic, elastic-strain

hardening and Ramberg-Osgood laws. The von Mises yield criterion in the context
of plane stress or plane strain conditions is used. Because the resulting stress field is
determined by superposition, reverse plasticity cannot be accounted for by the
model. StressCheck issues a warning message if in a particular application reverse
plasticity is detected. Reverse plasticity is understood to mean that upon removal of
the mandrel a compressive zone develops in which the maximum equivalent plastic
strain exceeds a tolerance, the value of which is 0.2%.
Executing a coldwork analysis

The component to be analyzed is defined by providing its topology, thickness,
material properties, constraints and loading. The material properties of the compo-
nent must be defined as elastic-plastic. Also, the linear material properties of the
mandrel used in the cold working operation have to be provided. Execute a linear
analysis and obtain a sequence of solutions using p-extension so that the quality of
the solution can be estimated. The estimated relative error in energy norm should be
small, typically below 5% (preferably below 1%) before proceeding with the rest of
the analysis.
Next, select the Coldworking tab in the Solver interface and proceed as follows:
1.Click on the linear solution which is to be the starting solution for the coldwork
analysis. The solution name, run number, type, and associated degrees of freedom
(DOF) appear in the scroll window of the dialog box. The corresponding discretiza-
tion will remain fixed in the course of the coldwork solution process.
2. Enter a name for the nonlinear solution which is different from the name of the
linear solution. For example, you can modify the name of the linear solution by
adding the letters NL (for nonlinear).

Coldwork analysis
1
3. Select the Convergence option (Stress is recommended for Coldwork) and
the allowable tolerance (1% or less).
4. Enter the number of iteration cycles into the Iteration Limit field. Stress-
Check will stop when the iteration limit is reached, except when the specified
error tolerance is reached first. The Coldwork analysis is completed even if
convergence was not achieved after the specified number of iterations. If the
error is not small enough, provide a higher value for the iteration limit and exe-
cute the coldwork analysis again.
5. Provide information about the hole to be cold-worked:

• Hole ID: The name of the circle corresponding to the hole to be cold-
worked. You can click on the combo box next to Hole ID, and select
from the list of circles available. The one selected will be highlighted
on the main display area.
• Interference: Enter the diametric interference level, which is the differ-
ence between the diameter of the mandrel and the diameter of the hole.
You can enter a numerical value or the name of a previously defined
parameter into the field.
FIGURE 37 Solver interface for coldwork analysis.
• Mandrel Mat. ID: Enter the material ID given to the mandrel in the
material input box. If you did not define a material property for the

Coldwork analysis
1
mandrel, select Class: Material from the Main Menu Bar and provide the
material ID, modulus of elasticity and Poisson’s ratio. The scrolling list
next to Mandrel Material ID contains a list of all materials defined in the
current session.
Only linear isotropic material properties can be specified for the mandrel.
The input material coefficients are used to compute the radial spring stiff-
ness of the mandrel (Kr) from the following expression:
2E
K r = ------------------------------------------
D ( 1 + ν ) ( 1 – 2ν )
where E and ν are the modulus of elasticity and Poisson’s ratio of the man-
drel material, and D is the diameter of the hole.
6. Select the mode of execution Initialize, Run Mode: Automatic and Method: Iter-
ative.
7. Select Display: Results. You can monitor the value of functions at points, maxi-
mum values of functions or contour plot of functions during the nonlinear iteration.
8. Click on the Solve button to start the coldwork analysis.
Note: The use of symmetry conditions for the cold-worked hole are not possible in
the current implementation.
Example: Attachment lug

Consider the attachment lug shown in FIGURE 38. The lug is 0.96” thick and the
material type is Ramberg-Osgood with the stress-strain law characterized by the
following parameters: E=16x106 psi, ν=0.31, S70E=137000 psi, n=33, σY=131000
psi (plane-strain). The lug is loaded through the hole by a bearing load F=163000
lb, and a diametric interference level of 0.03” is specified to cold work the hole.
The objective of the analysis is to find the residual stresses after the hole has been
cold-worked, and the combined effect of the residual stresses plus those due to the
bearing load along the ligament A-B.
The material properties for the mandrel are: E=30x106 psi, ν=0.30. A parametric
description of this problem is available in the Tutorial Handbook (file: mda2.sci).

Coldwork analysis
FIGURE 38 Attachment lug configuration.
The 9-element mesh used for this problem is shown in FIGURE 39. The bear-
FIGURE 39 Mesh and boundary conditions.
ing load is simulated as a sinusoidal traction around the hole, and normal con-
straints were imposed at the left end of the lug. A nodal constraint was imposed
on the lower left node to prevent rigid body translation in the y-direction.

Coldwork analysis
1
Solution: On performing a planar (2D) linear analysis for p-levels ranging from 1
to 8, product space, we find that the estimated relative error in energy norm at p=8
(1262 DOF) is 0.12%. After the linear solution is completed, perform a Coldwork
analysis for p=8. Specify the Convergence option: Stress, Tolerance: 0.5%, Nonlin-
ear ID: NSOL, Hole: CIR1, Interference: 0.03, Iteration Limit: 15, Mandrel Mat.
ID: mandrel (name given to the material ID of the mandrel). The elastic-plastic
solution, representing the effect of inserting the mandrel into the hole, is obtained in
10 iterations.
The elastic-plastic solution is followed by the linear solution corresponding to the

traction given in equation (5) applied to the cold-worked hole. At the end of the
analysis, the message shown below is displayed.
During the coldwork analysis, the maximum residual von Mises stress (σVMmax) is
computed around the cold worked hole by superposition, and compared against the
yield strength (σY) of the material. If σVMmax > σY, reverse plasticity occurred and
the percent deviation is indicated. If σVMmax < σY however, the message will indi-
cate that no reverse plasticity was detected.
The formula used for representing the traction around the hole in equation (5) is
available in the database after the Coldwork analysis is completed. To examine the
formula, select the Formulae icon from the Main Toolbar, and select the formula
_MANDREL from the formula form.
After the analysis is completed, a set of solutions is available in the scroll window
of all post-processing dialog boxes. The details (assuming that SOL is the name of
the original linear solution, and NSOL is the ID given in the nonlinear status box)
are given in the following table.
Solution ID Solution TYPE Description

SOL LIN Linear solution of original problem
NSOL NLMAT Elastic-plastic solution with mandrel inserted

Coldwork analysis
1
Solution ID Solution TYPE Description
NSOL_R CWRes Residuals after mandrel is removed
NSOL_L CWork Combination of original problem plus residuals
Results: FIGURE 40 shows the elastic-plastic von Mises stress distribution

obtained from the solution NSOL. The plastic zone around the hole represents
the region for which the total von Mises strain is greater than or equal to the
yield strain of the material: ε Y = ( σ Y ⁄ E ) = 0.0081875 .
Plastic Zone
FIGURE 40 von Mises stress distribution after cold-working.
FIGURE 41 shows the normal (σx) residual stress distribution along the liga-
ment A-B as a function of the distance from the edge of the hole, and the sec-
ond principal stress distribution (S2) over the entire lug after the mandrel is
removed. The solution NSOL_R was used to obtain the fringe plot and the
stress graph.
Finally, FIGURE 42 shows the σx stress distribution along ligament A-B and
the distribution of the first principal stress over the entire lug due to the applied

Fastened structural connections
FIGURE 41 Residual stress distribution.
bearing load without the effect of the residual stresses (as obtained from the solu-
tion SOL - FIGURE 42a) and with the added effects of the residuals (as obtained
from the solution NSOL_L - FIGURE 42b). The graph shows the σx stress distri-
bution along A-B while the fringe plot shows the distribution of the first principal
stress. Note that the maximum normal stress at the edge of the hole is about 56% of
the value when no residuals are considered.

StressCheck provides an advanced capability which allows for the computation of
both the structural and the strength responses of fastened structural connections.
The implementation accounts for the following:
• Partial contact between fastener and plate,
• Fastener shear stiffness,
• Material and geometric nonlinearities,
• Interference fitting,
• Initial clearance between fastener and plates,

(a) without residuals (b) with residuals
FIGURE 42 First principal stress distribution.
under the assumption that the effects of bending are negligible and there is no
friction between the contact surfaces. In other words, all dominant effects in
frictionless shear connections are accounted for in a two-dimensional setting.
The main component of the implementation is a special fastener element which

consists of a rigid core with two degrees of freedom (Ux, Uy) connected to the

1
planar body by a distributed radial (normal) spring as illustrated in FIGURE 43.
Uy distributed spring (Kr)

hole
Ux
rigid core
planar body
FIGURE 43 The fastener element.
The spring coefficient (Kr) is determined from the elastic solution of a disc under
radial compression (plane-strain condition), as the ratio between the applied pres-
sure and the radial displacement. The expression for the spring stiffness, evaluated
by StressCheck automatically, is:
2E
K r = ------------------------------------------
D ( 1 + ν ) ( 1 – 2ν )
where E and ν are the modulus of elasticity and Poisson’s ratio of the fastener mate-
rial, and D is the diameter of the hole.
The rigid core can be connected directly to the rigid core of other fasteners or using
a link element, it can be fixed in one or both directions, it can be loaded by a force
or an imposed displacement. Interference fitting or initial clearance can be speci-
fied also for each fastener element. This makes it possible to assign clearance or
interference fitting at each connecting plate location independently. If two plates
are connected by a fastener, for example, interference fitting can be specified for
the part of the fastener going through one of the plates and neat fit can be assigned
to the part of the fastener going through the other plate, or any other combination of
neat fit, interference fitting or initial clearance.

1
When a link element is used for connecting two fastener elements, the force-
displacement relation between the connected fasteners becomes
 F x1   u x1 
  1 –1 0 0  
 F x2  –1 1 0 0  u x2 
  = Ks  
 F y1  0 0 1 –1  u y1 
  0 0 –1 1  
 F y2   u y2 
where Ks is the spring stiffness of the link, in force per unit length, given under
the material definition tab of the model input interface. The link element
allows to incorporate the shear/bending stiffness of the fastener into the model
independent of the radial (propping) stiffness of the shank. The radial stiffness
is taken into consideration by the spring stiffness Kr of the fastener.
The fastener implementation is illustrated with an example in the following.
Executing a fastener analysis

The analysis of a fastened structural connection requires two steps: First, a lin-
ear analysis is performed using p-extension to assess the quality of the numeri-
cal solution. Second, a General Nonlinear analysis is performed to account for
all nonlinear effects present in the connection (contact, clearance, geometric
and material nonlinearities).
Therefore, to perform a fastener analysis follow the same procedure described

for the execution of a general nonlinear analysis.
Example: Six-fastener splice plate

Consider the 6-fastener splice plate subjected to a remote stress σ=10,000 psi
as shown in FIGURE 44. The plates are of aluminum alloy 2014-T6
(E=10.9x106psi, ν=0.397, S70E=53000 psi, n=18.5), 1.0 inch wide, 6.0 inches
long and 1/8 inch thick. The two splice plates are of the same material and

1
dimensions but are only 1/16 in thick. Six 5/16 in diameter steel fasteners
(E=29x106 psi, ν=0.295) are used to connect the plates. The fastener installation is
such that interference fitting develops in all six holes as the fastener goes through
the 1/8” plates. The diametral interference is 0.0015 inches or about 0.5% of the
fastener diameter. Neat fit is assumed for holes number 1 and 2 in both splices and
a 0.002 inches clearance is considered for hole number 3 on both splice plates.
5/16 dia.
3 2 1 1 2 3
σ
1.0
σ
6.0
1/16 splice
1/8 plate
0.5 1.0 1.0

12.0
FIGURE 44 Six-fastener splice plate. Units: inches (not to scale).
The objective of the analysis is to determine the load distribution among fasteners
and the local stress distribution around the most highly stressed holes in the plate
and the splice. Because of symmetry, only one half of the splice will be discretized.
Even though further reduction is possible (1/4 model) the use of 1/2 model is pre-
ferred in this case to facilitate visualization. The finite element mesh and boundary
conditions are shown in FIGURE 45. Note that there is a ‘layer’ of elements around
each fastener hole to improve the quality of the solution in the region where the
maximum stresses are expected. This problem can be loaded from the Tutorial
Handbook (file: Splice6.sci) or can be created from the beginning according to the
description that follows.
After opening a new database, make sure that the analysis type is Planar Elasticity,
and the units selector is set to in/lbf/sec/F. From the Main Toolbar select the Create
Model icon, and when the Input dialog window appears select the Geometry tab
and proceed as follows:

Splice 1
2 Main Plate
Splice 2
Detail
FIGURE 45 Mesh and boundary conditions for the splice plate.
Main plate Create a rectangle at (0.0, 0.0, 0.0) of width=6, height=1 that represents the
outline of the main plate. Create 3 circles at (0.5, 0.5, 0.0); (1.5, 0.5, 0.0) and
(2.5, 0.5, 0.0) of radius 5/32. Then create three more circles of radius=1.5*5/32
at the same locations using the method Local. To facilitate post-processing,
Edit the coordinate system of each circle and change the data type to Cylindri-
cal. Create three auxiliary lines at 45 degree angle emanating from each hole.
One way of doing that is to create a point at the origin of each local system, and
then create a line from each point with a length of 0.75 and angle 45. The geo-
metric description should look as indicated in FIGURE 46a. After that, create
the nodes (using the methods of intersection, projection, midpoint, etc.) and the
quadrilateral elements so that the mesh of the plate looks like the one shown in
FIGURE 45. Then create a fastener element at each hole location. To do that
simply select:

(a) Plate
(b) Splice
FIGURE 46 Geometry for the plate and one splice.
• Mesh tab > Create > Fastener, and click on each one of the fastener holes.
The symbol of a fastener will be displayed after each click (see detail in
FIGURE 45).
Splice 1 It is convenient to create a parameter that will control the separation between plate
and splice to facilitate the visualization of the model. Also, since the analysis type
is planar, a separation is always required to avoid conflicts when elements of differ-
ent plates are at the same location. The magnitude of the separation has no effect in
the results, however. Select the Model Info icon from the Main Toolbar and create a
parameter Zdist=1.0, for example.
Next, create a rectangle at (0.0, 0.0, Zdist) of width=3, height=1 that represents the
outline of splice 1. Create 3 circles at (0.5, 0.5, Zdist); (1.5, 0.5, Zdist) and (2.5, 0.5,
Zdist) of radius 5/32. Then create three more circles of radius=1.5*5/32 at the same
locations. Two auxiliary lines can be created at 45 deg. emanating from two of the
holes as shown in FIGURE 46b. Finally, proceed as before to create the nodes, the
quadrilateral elements and the fastener elements for this splice plate.
Splice 2 To create the next splice plate, proceed as follows: Group all objects that represent
splice 1 (geometry entities and mesh). You can use the Action: Select > Object:
Any Object to accomplish that. Next, in the input region of the geometry box enter
X:0.0; Y:0.0; Z:-2*Zdist; Rot-X: 0.0, Rot-Y: 0.0, Rot-Z: 0.0, and then click on
Copy. A new splice will be created at a distance equal to (-2*Zdist) from splice 1 so

1
that the main plate will be located between the two splices as shown in FIG-
URE 45. Furthermore, since the Z-coordinate was given parametrically, if you
change the value of the parameter Zdist, both splice plates will move simulta-
neously. The parameter Zdist can be set to any value greater than zero (Zdist >
0).
Thickness Select the Thickness tab from the Input dialog window and assign a constant
thickness of 0.125 to all the elements of the main plate and a constant thickness
of 0.0625 to all the elements of splices 1 and 2.
Materials Select the Material tab and retrieve two materials from the Standard material
database:
• Define tab > Material: Elastic-Plastic > Units: U.S. > Type: Ramberg-
Osgood > Case: Plane Stress > Click Browse next to ID and select
2041-T6 > ID: PLATE > Accept.
• Define tab > Material: Linear > Units: U.S. > Type: Isotropic > Case:
Plane Stress > Click Browse next to ID and select ASTM-A36 > ID:
BOLT > Accept.
Next, assign the material ‘PLATE’ to all the elements in the main plate and
splices and the material ‘BOLT’ to all fastener elements.
• Assign tab > Select > All Elements > Material ID: PLATE > Accept.
• Assign tab > Select > Fastener > Material ID: BOLT. Select all 9 fas-
tener elements and click on the Accept button.
The material assignment is now complete.
Loads To specify the traction loading acting on the plate, select the Load tab and pro-
vide the following information:
• Load tab > Select > Any Curve > Traction > ID: load > Direction:
Norm./Tan. > Normal: 10000. Click on the line of the main plate where
the load has to be applied and then click on the Accept button.
Next assign the interference to fasteners 1, 2, and 3 on the main plate:

• Load tab > Select > Fastener > Interference > ID: load > Diam. Interf.:
0.0015. With the mouse cursor select fasteners 1, 2 and 3 on the main
plate, and then click on the Accept button.
To assign a clearance to fastener 3 on both splices, complete as follows:

1
• Load tab > Select > Fastener > Interference > ID: load > Diam. Interf.: -
0.002. With the mouse cursor select fastener 3 in both splice plates and then
click on the Accept button. A negative number is used to indicate a gap.
The load symbols will be displayed in the main window, as shown in FIGURE 45.
Constraints To specify the constraints, select the Constraint tab and proceed as follows. First
create a constraint record for the symmetry condition:
• Constraint tab > Select > Any Curve > Symmetry > ID: const. With the left
mouse button click on the two lines that represent the symmetry boundary
of both splice plates and then click on the Accept button (see FIGURE 45).
Remember to hold down the shift key when selecting more than one object.
Next create the constraint records connecting the fastener elements (rigid connec-
tion will be assumed. The case of a link connection is discussed later):
• Constraint tab > Select > Fastener > Connection > ID: const. While holding
down Shift, click on three fastener elements which are at the same xy-loca-
tion in splice 1, splice 2 and main plate (location 1 in FIGURE 45) and then
click on the Accept button.
Repeat two more times for the other 6 fasteners (2 groups of 3 fasteners each at
locations 2 and 3) to create the additional constraint records. The Connection con-
straint symbol is a filled circle at the center of the fastener as shown in FIGURE 45.
Note: If the fasteners are not connected, they will still fill the hole but will perform
as if they were broken fasteners, and will not transfer load from one plate to the
other.
Finally, it is necessary to prevent a rigid body displacement in the Y-direction. That

can be accomplished by fixing any node or fastener in the required direction. Con-
sider fixing a fastener as follows:
• Constraint tab > Select > Fastener > Displacement > ID: const > Data Type:
Fixed. Turn on the toggle switch in front of Y only and then click on one
fastener on the main plate and then click on the Accept button.
This completes the constraint assignment for this problem.
Execution Fill in the Solution ID record and the p-Discretization by assigning product space
and p=8 to all elements in the mesh. Then perform a linear analysis, using down-
ward p-extension, p-limits: 8 to 5. After the linear solution is completed, check the
relative error in energy norm. The results should be similar to those shown in FIG-

1
URE 47. Note that the estimated relative error in energy norm is less than 1%
for run #1 which corresponds to p=8. The results you obtain may be slightly
different from the ones shown here if your mesh is not identical to the one
shown in FIGURE 45.
FIGURE 47 Estimated relative error in energy norm for the linear solution.
Next select the Nonlinear tab from the solver interface and select linear run # 1
as the starting point for the nonlinear iteration corresponding to a general Non-
linear analysis. Set Iteration Limit to 10, the Convergence criterion to Stress,
the Technique to Direct integration, and the allowable Tolerance to 0.5%. It
should take 5 iterations to complete the General Nonlinear analysis.
Results The von Mises (Seq) stress distribution obtained from the nonlinear solution is
shown in FIGURE 48. Note that the maximum stress regions are located
around the holes in the plate due to the effect of the interference fitting of the
fasteners with a maximum of 53.1 ksi. Also note that the region around holes 1
and 2 in the splice plates are more highly stressed than the region around hole 3
which has a clearance fit fastener.
To compute the load transferred by each fastener to the splice plates, select the
Resultant tab from the Results dialog window, and complete as follows:
• Resultant tab > Select > Fastener > Selection > Solution: NSOL (or
any name you gave to the nonlinear run) > Run: 1 to 1 > Forces input

1 2 3
FIGURE 48 von Mises Stress distribution for nonlinear solution. Fasteners

with rigid connections.
area: Turn on the toggle switch in front of Fx, click on fastener 3 of splice
1, and then click on fastener 3 of splice 2 while holding Shift. Finally, click
The results will be displayed in the graph window (Table tab) as shown in FIGURE
49 and in the summation area of the Resultant dialog window. The total load trans-
ferred through fastener 3 in both splice plates is 244 lb. Repeating for the other fas-
teners, the total load transferred through fasteners 2 and 1 in both splice pates are
559 lb and 447 lb, respectively.
FIGURE 49 Force resultant for fastener 3 in both splice plates.

1
Finally, consider the radial (bearing) pressure distribution around fasteners 2
and 3 in the splice and main plate. Select the Points tab from the Results dialog
window and complete as follows:
• Points Tab > Select > Any Boundary > Solution: NSOL (or any name
you gave to the nonlinear run) > Run: 1 to 1 > Function: Sx > # of pts.:
359 > Boundary P1-Min: 0.0, P1-Max: 1.0 > System: SYS4 (the name
of the cylindrical coordinate system corresponding to fastener hole 3 in
the main plate). Click on the circle corresponding to fastener 3 in splice
1 and then click on Accept.
The results will be displayed in the graph window as shown in FIGURE 50a.
Note that by selecting the function Sx and a cylindrical system, the radial stress
is reported. By selecting 359 points for the extraction, each point is 1 degree
apart from each other. Repeating the operation for fastener 3 in the main plate
and fastener 2 in splice 1 and plate, the results shown in FIGURE 50b-d are
obtained. For fastener 2, the cylindrical coordinate system SYS3 was used for
the extraction of Sx.
Note that the contact area for fastener 3 in the splice plate (FIGURE 50a) is
smaller than for fastener 2 in the splice (FIGURE 50c) since a clearance was
specified for fastener 3 in the splice. Also note that due to the interference
level, full contact is maintained in both fasteners holes of the main plate (FIG-
URE 50b and d).
Adding links We next consider replacing the rigid fastener-to-fastener connections with link
elements. First, delete the connection constraint records created before. In the
Constraint tab of the model input interface, select a record identified as con-
nection and then click on the Delete button. Repeat for the other two similar
records in the interface.
Next, select the Mesh tab and create the link elements as follows:
• Create > Link > Connection. With the left button of the mouse click on
one fastener element in a splice an then on the corresponding fastener
element in the plate. Continue until all six links are created as shown in
FIGURE 51.
Next, we need to define and assign the stiffness properties to the links. For
example, consider Ks=100,000 lb/in for all six links. Select the Material tab
and complete as follows:

(a) Fastener 3 - Splice (b) Fastener 3 - Main plate
(c) Fastener 2 - Splice (d) Fastener 2 - Main plate
FIGURE 50 Bearing stress distribution around holes 2 and 3. Fasteners

with rigid connections.
• Define tab > Material: Linear > Type: Stiffness > ID: Link > Stiffness:
100000 > Accept.

FIGURE 51 Splice plate with fastener conneccted by link elements.
• Assign tab > Select > Link > Material ID: Link. Select all six link ele-
ments and click on the Accept button.
Run the linear and nonlinear analysis again following the same steps indicated
before for the case of the rigid connections. This time, given the added flexibil-
ity of the connection, the number of iterations to achieve the desired tolerance
increases by one.
The results of the nonlinear analysis of the splice plate with link connections
are shown in FIGURE 52. Note that the von Mises stress distribution is very
similar to that shown in FIGURE 48 for the case of rigid connections, and the
maximum value is practically the same as before.
The load transferred by the fasteners are quite different, however. The loads
computed from the fasteners in the splice plates are F1=437, F2=450 and
F3=363. The comparison between the two cases is shown in Table 1. Note that
fastener #3, with the 0.002 inches clearance, carries more load with the link
connection and therefore fastener #2 is loaded less than when the connection is
rigid.

1 2 3
FIGURE 52 von Mises Stress distribution for nonlinear solution. Fasteners with link
connections.
TABLE 1. Comparison between rigid and link fastener connections

for the splice plate problem.
Fastener # Fastener Load [lb]
Rigid connection Link connection

1 447 437
2 559 450
3 244 363
Finally, FIGURE 53 shows the bearing stress distribution around holes #2 and #3 of
the splice and main plate.

Margin Check analysis
(a) Fastener 3 - Splice (b) Fastener 3 - Main plate
(c) Fastener 2 - Splice (d) Fastener 2 - Main plate

FIGURE 53 Bearing stress distribution around holes 2 and 3. Fasteners with link connections.

The Margin Check is a nonlinear analysis capability (material or general non-
linear) with failure criteria control. To solve a Margin Check problem a linear
problem must be solved first. A nonlinear material stress-strain law must be
specified if a material nonlinear analysis is involved. The load must be defined
as a function of a parameter to perform a Margin Check analysis.
After the linear problem is solved, the nonlinear analysis is performed for the
original load until convergence is achieved. The load is then incremented

1
according to the specified step and the nonlinear solution is obtained for the new
load. At the end of each load step, the program checks if any of the predefined mar-
gin criteria was exceeded. A report is issued summarizing the load at which each
criterion was exceeded, and a solution is kept for further post-processing. The
implementation is illustrated with an example in the following.
Problem description
A rectangular plate with a circular hole in the center (FIGURE 54) is loaded by a
constant traction S=20000 psi. The plate has a thickness h=1 in, length L=15 in,
width W=5 in and the hole diameter is a=1.6 inches. The material is elastic-per-
fectly-plastic with a modulus of elasticity E=29x106 psi, Poisson’s ratio v=0.295
and yield strength σY=36000 psi (plane-stress).
The objective of the Margin Check analysis is to find the value of the applied trac-
tion S for which the following margin criteria are exceeded:
FS × Seq
MS1 = 1 – ----------------------
Ux
MS2 = 1 – --------- ε eq
MS3 = 1 – ---------- (7)
Syield U all
ε eqo
where FS is the factor of safety (FS=1.2), Syield is the yield strength of the material
(Syield=36000 psi), Seq is the average von Mises stress along the ligament EF, Uall
is the allowable displacement along AB (Uall=0.0065 in), Ux is the average dis-
placement along AB, εeq is the maximum equivalent strain along EF and εeqo is
the allowable value (εeqo=0.006 in/in).
Exploiting the symmetry of the model, only one fourth of the plate needs to be ana-
lyzed. FIGURE 55 shows the mesh used to perform the computation.
This problem can be loaded from the Tutorial Handbook (file: PlateHole.sci). After
loading the problem, adjust the parameters and execute the linear analysis, upward
p-extension from p=1 to 8.
It is convenient to define this problem in parametric form and include a few extra
parameters to be used in the definition of the margin of safety equations shown
above. Click on the Model Info icon and when the Model Editor window opens
select the Parameters tab and create the necessary parameters as shown in FIGURE
56. These parameters are already available if you loaded the problem from the

1
C E B
2a a W S
D A
L
FIGURE 54 Rectangular plate with a central hole.
FIGURE 55 Margin Check example problem.

1
Tutorial Handbook Also, enter the formulas for MS1, MS2, and MS3 in the formu-
lae window according to FIGURE 57.
FIGURE 56 Model Information: Parameters tab.
FIGURE 57 Formula box with margins of safety.

1
Entering Margin criteria
The margin criteria are predefined extraction procedures based on extraction

functions (typically formulas). These extraction procedures can be defined
from the Points extraction or the Min/Max extraction options of the Results
interface.
To define the margin criteria specified by the second of Eq. (7), for example,
go to the Results dialog window and select the Points tab. Choose the Settings
tab and supply the following information:
• Points tab > Settings tab > Select > Any Boundary > Selection > Solu-
tion: NSOL (for Margin Check any name in this field will be ignored)
> Run: 8 to 8 (same for this field) > Func: Fmla > # of pts: 30 > Aver-
age switch on > Formula: MS2 > Name: Displ (this is the name that
will appears in the scroll window of the Margin Check analysis form).
With the mouse cursor select the line where the load is applied and
then click on the Save button.
If you choose Displ as a margin criterion during the execution of the Margin
Check, the analysis proceeds until the evaluation of the formula MS2 gives a
value equal or less than zero. The formula is evaluated after each load incre-
ment takes place. FIGURE 58 shows the settings of the Points extraction dia-

1
log window for the Displ margin criterion. Next consider the margin criterion given
FIGURE 58 Points extraction - Margin criterion: Disp.
by the formula MS1 in Eq. (7). The criterion is applied to the two edges along
boundary EF as follows:
• Points tab > Settings tab > Select > Edge > Selection > Solution: NSOL >
Run: 8 to 8 > Func: Fmla > # of pts: 3 (when the object edge is selected, the
number of points is ignored in determining the average) > Average switch
on > Formula: MS1 > Name: Ligament. With the mouse cursor select the
two edges along the ligament where symmetry constraints are specified and
then click on the Save button. The settings are shown in FIGURE 59.
Finally, the margin criterion given by the formula MS3 in Eq. (7) is defined by
selecting the Min/Max tab from the Results window, choose the Settings tab and
supplying the following information:
• Min/Max tab > Settings tab > Select > Any Boundary > Line > Solution:
NSOL > Run: 8 to 8 > Func: Fmla > Midsides: 30 > Minimum > Formula:
MS3 > Name: Max_Eeq. With the mouse cursor select the EF boundary
and then click on the Save button.
Note that the minimum of MS3 corresponds to the maximum of Eeq. The settings
are shown in FIGURE 60.

FIGURE 59 Points extraction - Margin criterion:

Ligament.
Executing a Margin Check analysis

After the linear analysis is complete, check the quality of the linear solution.
The estimated relative error in energy norm should be small, typically below
5%, before proceeding with the rest of the analysis. Next, select the Margin
Check tab from the Solver interface and the dialog window shown in FIGURE
61 will appear. Proceed as follows:
1. Click on the linear solution which is to be the starting solution for the Mar-
gin Check analysis. The linear solution name, run number, type of analysis,
and associated degrees of freedom (DOF) appear in the scroll window of the
dialog box.
2. Select the Type of nonlinear analysis, the choices are General and Material
Nonlinear. For this problem select Material. For a small-strain, small-displace-
ment, elastic-plastic analysis, the option Material should be used.

FIGURE 60 Min/Max extraction - Margin criterion:

Max_Eeq.
3. Select Convergence. The choices are Stress or Energy. Refer to Executing mate-
rial nonlinear analyses for additional details about convergence options. Select
Stress for this problem.
4. Define the Tolerance. For this problem set the tolerance to one percent.
5. Enter the Iteration Limit, that is the number of iteration cycles you wish Stress-
Check to perform automatically for each load step. StressCheck will continue until
the Iteration Limit is reached, unless the specified error tolerance is reached first.
For this problem set the Limit to 50.
6. Set the Load. From the Param. pull down menu select the parameter S that
defines the load; its Initial value will be provided automatically. Specify the maxi-
mum load (Limit) and the load Step. For our problem we chose Sult (24000 psi) to
be the maximum and we want the load to be increased in equal increments of DelS
(100 psi) from the minimum to the maximum. Note that the name of a parameter
was entered in the input field for the Step and Limit.

1
7. Select Margin Criteria. The lower scroll window shows the different margin
criteria extraction options available. The selection is made by left clicking on
the records while holding down the Shift key. When several margin criteria are
selected, the Stop at 1st M.S. failure toggle switch can be used. FIGURE 61
shows the settings for this problem. In this case, three margin criteria are
FIGURE 61 Margin Check Solver.
selected. This implies that after each load step, the values of MS1, MS2 and
MS3 should remain positive in order to proceed with the next load increment.
Once the first margin criterion is exceeded, the program will continue because
the toggle stop at 1st M.S. failure is off. FIGURE 62 shows the setting of the
SOLVE! tab after the margin analysis is complete.
FIGURE 62 SOLVE! interface for example problem.

1
FIGURE 63 shows the values of the applied traction for which the margin criteria is
exceeded. The margin check summary is displayed in a report window once the
analysis is completed. A record is also kept in a log file, that can be retrieved at any
time by selecting File from the main menu and then View Session Log. It follows
that the traction should not exceed 22300 psi if no margin criterion should be
exceeded.
FIGURE 63 Margin Check summary for example problem.
After the Margin Check analysis is completed, several solution records are avail-
able for post-processing depending on the results of the analysis. A solution record
with the name _LAST_NL (last nonlinear run) is always available, even if no mar-
gin criteria were exceeded after the load increments. In addition, for each exceeded
margin there will be a solution record composed of the name of the margin criterion
followed by a number representing the load step at which the margin was exceeded.
For example, for the margin criterion Disp, the solution record is _DISP_25, since
it took 25 load steps to exceed this margin. FIGURE 64 shows the von Mises stress
distribution corresponding to each one of the exceeded margins.
The ‘Ligament’ criterion was exceeded for load step #24 while the ‘Disp’ criterion
was exceeded at load step #25. For that reason the difference in the stresses distri-
bution between the two cases is almost negligible. The ‘Max_Eeq’ criterion was
exceeded for load step #31 and the stress distribution corresponding to this load
(23000 psi) can be differentiated from the other two cases.

Nonlinear heat transfer
FIGURE 64 Stress distribution (Seq) for the load corresponding to each margin failure.

The implementation accounts for radiation boundary conditions and tempera-
ture-dependent thermal conductivity for steady state heat transfer problems in
two- and three-dimension. Temperature-dependent material properties can be
specified as a formula or in tabular data. Radiation boundary conditions can be
specified on any boundary (curve in 2D or surface in 3D) even if convective

1
boundary conditions are specified on the same boundary. For thermo-elastic analy-
sis, the material properties required for the elasticity analysis can also be defined as
temperature-dependent as described in Temperature-dependent material properties.
The details are described in the following.
Temperature-dependent material properties

To specify temperature dependent material properties, from the Input dialog win-
dow select the Material tab and then the Define tab. Select Action: Define, Object:
Temp. Depend., Method: Selection, and the Units to be used (US, SI or Other).
The coefficients of thermal conductivity (Kxx, Kxy, Kyy, etc.) may be defined as a
constant, a parameter, a formula or a table. Entering a constant or a parameter is
straightforward, simply click on the appropriate box and key in the constant value
or the parameter name. If the coefficients are specified by formula (as a function of
temperature) then the formula name is entered into the input area. The results of the
computation of the formula will be displayed in a graph window. If the coefficients
are specified by tabular data (as pair of values of temperature and thermal conduc-
tivity), enter the table name in the corresponding fields. The program will fit the
tabular data using least square fitting and the results will be displayed in a graph
window.
Formula option FIGURE 65 illustrates the case when the material coefficients Kxx=Kyy are
defined by the formula ‘k_coef’ as a function of temperature, and Kxy=0. The ref-
erence temperature (Rf. Tmp) is the temperature used to compute the thermal con-
ductivity during the linear analysis. This value should be estimated based on the
characteristics of the problem. In general, a value corresponding to room tempera-
ture is adequate.
In this example, the thermal conductivity is defined as:
k_coeff = coef1 + coef2*X + coef3*X^npow
where coef1, coef2, coef3 and npow are parameters defined in the Model Informa-
tion dialog window (Parameters tab) shown in FIGURE 65b, and X (the indepen-
dent variable) is the temperature. The evaluation of the formula is shown in a graph
window as shown in FIGURE 65c. The rule used by the program to select the range
of temperatures to be shown in the graph window is as follows:
• If Rf. Tmp > 100 then: Tmin = Rf. Tmp/2, Tmax = 2*Rf. Tmp
• If Rf. Tmp <= 100 then: Tmin = -Rf. Tmp, Tmax = 150

1
To change the default temperatures range used in the graph, create the parame-
ters:
• _nl_min: Initial temperature value to be included in the graph
• _nl_max: Final temperature value to be included in the graph
• _nl_count: number of points to between min and max.
and click on the Plot button of the input area shown in FIGURE 65a.
Table option FIGURE 66 illustrates the case when the material coefficients Kxx=Kyy are
defined by a table as a function of the temperature and Kxy=0. The name of the
table (AGPD) is entered in the corresponding fields of the Material input area.
The program performs a least square fitting of the tabular data and displays the
fitted curve in a graph window. For temperatures outside the tabular data range
a constant value is assumed.
(b) Graph with fitted tabular data
FIGURE 66 Temperature-dependent material

properties: Table option.
(a) Table input area

(b) Formula and parameter forms

(a) Material Input area
FIGURE 65 Temperature dependent material properties: Formula option.

1
To create the tabular data, select the Tables tab from the Input dialog window.
A pair of values is entered by suppling the following information (FIGURE
66a):
• Tables tab > Select > Location > Constant > ID: AGPD > Type: Mate-
rial > X: enter the temperature value > Y: enter the thermal conductiv-
ity > Click on the Accept button
Repeat for as many entries as needed. A table should contain at least two
entries with different values of temperature. The results of the least square fit-
ting is shown in FIGURE 66b.
Radiation boundary conditions

When two bodies exchange heat by radiation then the heat exchange is propor-
tional to the difference of the fourth power of their temperature:
4 4
q n = κf s f e ( u – u R ) (8)
where qn is the heat flux per unit area (in W/m2 units or equivalent), κ is the
Stefan-Boltzmann constant (κ=5.669 x 10-8 W/m2oK4), fs is the shape factor
function, fe is the emissivity function (0 < fe ≤ 1), u is the surface temperature
of the body being modeled and uR is the surface temperature of near radiating
body.
The boundary conditions corresponding to radiating heat transfer can be speci-

fied by selecting the Temperature tab from the Input dialog window. The fol-
lowing information will be required (FIGURE 67):
• K: Stefan-Boltzmann constant (typical value: 5.669x10-8 W/m2 oK4)
• FS: Shape factor (depends on the geometry or radiating bodies)
• FE: Emissivity function (0 < FE ≤1)
• Rad. Temp.: Surface temperature at near radiating body - uR in Eq. (8).
Units should be consistent with those of Stefan-Boltzmann constant.
K, FS, FE and Rad. Temp. can be specified as constant, parametric or formula

depending on the spatial variables.
Notes:

FIGURE 67 Temperature interface for Radiation boundary conditions.
• When radiation boundary conditions are specified, use temperature in oK

for all input fields that depend on temperature.
• If radiation and convective heat transfer are specified make sure the film
coefficient has units consistent with the Stefan-Boltzmann constant.
Executing nonlinear heat transfer analyses

To solve a nonlinear analysis, a linear problem must be solved first. The thermal
conductivity must be temperature-dependent and/or radiation must be specified to
perform a nonlinear heat transfer analysis. Radiation boundary conditions are
ignored during the linear solution.

1
Select the Nonlinear tab in the Solver window and the user interface for exe-
cuting nonlinear solutions will appear (FIGURE 68). In the scrolling list you
will see the list of linear solutions. Proceed as follows:
FIGURE 68 Nonlinear execution interface.
1. Click on the linear solution which is to be the starting solution for the non-
linear iterations. The solution name, run number, type, and associated degrees
of freedom (DOF) appear in the scrolling list of the dialog window. The corre-
sponding discretization will remain fixed in the course of the nonlinear solu-
tion process. It is recommended that you select a linear solution with a small
relative error in energy norm (typically under 5%, preferably under 1%).
2. Enter a name for the nonlinear solution which is different from the name of
the linear solution.
3. The analysis Type: Material is the only option. This includes both the tem-
perature-dependent thermal conductivity and radiation boundary conditions.
4. Select the Convergence and the Tolerance. The Convergence choices are:
Temperature or Energy. If the temperature option is selected, the iteration will
proceed until the difference in temperature between two consecutive cycles at
each integration point in every element is less than the specified tolerance. If
the option energy is selected, the convergence criterion given by equation (3)

1
on page 27 will be used. In general, it is recommended that you choose the energy
criterion when the solution has singular points.
5. Select the number of iteration cycles you wish StressCheck to perform automati-
cally and enter this number into the field next to Iteration Limit. StressCheck will
stop when the iteration limit is reached, unless the specified error tolerance is
reached first. When StressCheck stops at the iteration limit you can examine the
results and decide whether to continue or not. If you wish to continue, then simply
reset the Iteration Limit and click on the Solve button.
6. Select the mode of execution. The options are: Initialize and Restart. Select Ini-
tialize when you begin an analysis, select Restart when you wish to continue from
an existing solution.
matic, then the iterations will continue until you reach the specified iteration limit
or satisfy the specified error tolerance. If you select Stepwise, then the iteration pro-
cess stops after each cycle, and you need to click on the Solve button to go to the
next iteration cycle.
9. Select Display. The choices are Sequence and Results. You may monitor the
progress of the solution by turning on the Sequence toggle switch, or you can mon-
itor the value of functions at points, maximum values of functions or contour plot
of functions during the nonlinear iteration by turning on the Results toggle switch.
10. Click on the Solve button to start the Nonlinear analysis.
After the analysis is completed, the nonlinear solution just obtained will be avail-
able in the scrolling list of the post-processing dialog window with the name you
entered in step 2, the same run number and DOF as the linear solution selected, and
with a type NLMat. Selecting a different run number of the linear solution to start a
nonlinear iteration will result in a new nonlinear solution record with different
degrees of freedom and run number than the previous one. Both nonlinear solutions
will be available for extraction. You can obtain as many nonlinear solutions as there
are linear solutions. In this way you can check the convergence characteristics of
the nonlinear solutions.

1
Model problem 1: Temperature-dependent properties
Consider the case of a circular pipe with temperature-dependent thermal con-
ductivity. The pipe has an internal radius of 25 mm, an external radius of 30
mm. Ten evenly spaced, tapered cooling fins are attached to the pipe. Taking
advantage of twenty-fold symmetry, the solution domain is as shown in FIG-
URE 69. The material is an aluminum alloy, the coefficient of thermal conduc-
E Circular arc,
F
Subtended angle: 2.5o
18o C
7.5o
A B
25 5 15
FIGURE 69 Circular pipe with cooling fins. Solution domain.
tivity (k) of which is 0.206 W/mmoC when the temperature is less than or
equal to 20oC and varies with increasing temperature according to the follow-
ing equation:
2
k = 0.1180556 + 0.00455069T + ( – 0.0000055 )T
The internal wall temperature is 500oC. Cooling is by convection. The coeffi-

cient of convective heat transfer (hc) is 1.2x10-3 W/mm2 oC and the tempera-
ture of the convective medium is 20oC. You can construct this model problem
by following the steps described in the Conduction Heat Transfer chapter, in
the Analysis Guide.
After opening a database, activate the Handbook interface by choosing the

Handbook Library icon from the Main Tool Bar. When the handbook interface

1
appears, click on the Browser icon of the Model Info tab to access the Model
Browser. Choose the Tutorial Handbook, and double click on FinNL.sci (Pipe with
cooling fins). It will be loaded into StressCheck.
To specify the coefficients of thermal conductivity (Kxx, Kxy, Kyy, etc.) by for-
mula click on the Formulae icon, and when the Formula dialog window appears
enter the following information (see FIGURE 70a):
• Name: CoefForm > System Option: Cartesian > Use Constants button on >
C1: 0.1180556 > C2: 0.00455069 > C3: -0.0000055 > Formula:
if(X<=20;0.206;C1+C2*X+C3*X^2) > Click on the Accept button.
Next select the Material tab from the Input dialog window. Enter and assign the
material properties for the nonlinear problem using a name which is different from
the name used for the material properties in the linear case. Supply the following
information:
• Material tab > Define > Temp. Depend. > Selection > Define tab > ID:
PROP_TH > Material: Temp. Dep. > Units: SI > kxx: CoefForm > kxy: 0 >
kyy: CoefForm > Q: 0 > Rf. Tmp: 20 > Click on the Accept button (see
FIGURE 70b).
If you click on the Plot button a Graph window will display the thermal conductiv-
ity coefficients as a function of the temperature (see FIGURE 70c). Next, assign
PROP_TH to all the elements in the mesh:
• Material tab > Assign tab > Select > All Elements > Selection > ID:
PROP_TH > Accept.
Having entered the input data, execute a linear analysis (Upward-p extension, p-
limits:1 to 8). To execute the nonlinear analysis, select the Nonlinear tab in the
Solver window and supply the following information:
Material (NL Mat) > Convergence: Energy > Tolerance(%): 0.01 > Itera-
tion Limit: 5 (see FIGURE 68).
tive > Solve button.
The nonlinear solution should be obtained in 3 iterations. FIGURE 71 shows the

temperature distribution obtained from the linear (a) and the nonlinear (b) solu-
tions. Note that the maximum temperature in both cases is the same (500oC). How-
ever, the minimum temperature for the linear run is 363 oC while for the nonlinear
run is 466 oC.

(a) Formula definition form
(b) Material input area
FIGURE 70 Pipe with cooling fins.

Nonlinear material properties.

(a) Linear
(b) Nonlinear
FIGURE 71 Temperature distribution from linear and nonlinear solutions.
Model problem 2: Radiation boundary conditions

Consider the component shown in FIGURE 72 which represents an electronic chip
with four cooling fins. The base of the chip is at constant temperature (To=773K)
and the cooling is by convection and radiation. The chip is 4.7 mm by 3.5 mm by
0.6 mm, and there are four fins 3.0 mm long and 0.2 mm thick separated 1.1 mm.
The fillet size is equal to the fin thickness.
The coefficient of thermal conductivity is k=0.206 W/mm K, the coefficient of con-

vective heat transfer is hc=0.0012 W/mm2 K and the temperature of the convective

1.1 1.1 1.1
0.2
3.0
0.2
3.5
0.6
4.7
FIGURE 72 Model problem for radiation. All dimensions in mm.
media is Tc=300K. For the radiation boundary conditions the following values
are considered: The Stefan-Boltzmann constant κ=5.669x10-12 W/m2K4, the
shape factor fs=1.0, the emissivity fe=0.5 and the surface temperature of near
radiating body uR = 300K.
This model problem can be loaded from the Tutorial handbook (file: Cooling-
FinRadiation.sci). The finite element mesh is shown in FIGURE 72. Convec-
tive boundary conditions are applied to all the faces except at the bottom of the
chip where the constant temperature To is specified. Radiation boundary con-
ditions are specified in the faces indicated in FIGURE 73.
Having entered the input data, execute a linear analysis:

• Linear tab > Extension: Upward-p > p-limits:1 to 7.
Iterative > Convergence: None > Solve button.
Check the overall quality of the solution by computing the estimated relative
error in energy norm. The solution corresponding to p=7 will be used as the
starting point for the nonlinear analysis. Remember that during the linear solu-

radiation surface
radiation surface
FIGURE 73 Radiation boundary conditions for

model problem.
tion, the radiation boundary conditions are ignored, and therefore convective cool-
ing is the only one accounted for. The estimated error in energy norm and the
temperature distribution from the linear solution are shown in FIGURE 74.
To execute the nonlinear analysis, select the Nonlinear tab in the Solver dialog win-
dow and supply the following information:
Material (NLMat) > Convergence: Temperature > Tolerance(%): 0.5 > Iter-
ation Limit: 5

FIGURE 74 Results from the linear solution. Convective cooling only.
The nonlinear solution should be obtained in 3 iterations. The temperature dis-

tribution corresponding to the combined convective and radiation cooling is
shown in FIGURE 75. Note that the maximum temperature in both cases is the
same (773K). However, the minimum temperature for the linear run is 686K
while for the nonlinear run is 644K. The figure also shows the values of the
flow (the integral of the heat flux) for the four top faces of the fins obtained

1
from the linear and nonlinear solutions. As expected, the flow loss is greater when
radiation and convection cooling takes place.
Linear
Nonlinear
FIGURE 75 Results from nonlinear

solution. Convective and radiation
cooling.
Thermo-elasticity
In this section we address the solution of a temperature-dependent material prop-

erty problem in heat transfer and elasticity. Given the thermal conductivity as a
function of the temperature k=k(T), a nonlinear heat transfer solution is obtained
and the temperature distribution is used to determine the stresses caused by the

1
thermal loading. The modulus of elasticity E(T), the Poisson’s ratio ν(T), and
the coefficient of thermal expansion α(T), are also functions of the tempera-
ture, T.
Problem description Consider a short section of two Inconel 718 pipes connected by a thin fin.
There is a fluid inside each pipe at different temperatures. The dimensions are
as shown in FIGURE 76. Only half section is shown in the figure, since the
actual length is 1.5 inches.
FIGURE 76 Connected pipes.

Description.
The thermal conductivity as a function of temperature is given in tabular form

from room temperature to 2000 oF. The inside temperature of the left pipe is
1800 oF and that of the right pipe is 500 oF. Cooling is by convection, and the
film coefficient is hc=2.0 Btu/hr/in2/oF, and the temperature of the convective
medium is Tc=100 oF (see FIGURE 77).
This problem can be loaded from the Tutorial Handbook (file: Connected-
Pipes.sci). The finite element mesh, consisting of 15 extruded hexahedral ele-
ments is also shown in FIGURE 77. Because of symmetry, only half the pipe
sector was discretized. The material properties of the Inconel 718 alloy as a

FIGURE 77 Connected pipes: Boundary conditions for the heat

transfer problem.
function of temperature are shown in Table 2 for both the heat transfer and the elas-
ticity problem.
The material coefficients are entered into StressCheck using the table option
described in Section "Temperature-dependent material properties" on page 79 in
this guide and also in the Analysis Guide. A reference temperature of 200 oF was
used for the linear analysis. For this temperature, the thermal conductivity is k=0.60
Btu/in/h/oF.
TABLE 2. Material properties for Inconel 718
T k E x 106 ν α x 10-6
[oF] [Btu/h/in/oF] [psi] [-] [1/oF]
RT 0.55 29.0 0.29 6.84
200 0.60 28.4 0.29 7.12
400 0.69 27.5 0.28 7.45
600 0.78 26.7 0.27 7.68
800 0.86 25.8 0.27 7.86
1000 0.95 24.9 0.28 8.06
1200 1.03 23.7 0.29 8.35
1400 1.12 22.1 0.30 8.78

1
TABLE 2. Material properties for Inconel 718
T k E x 106 ν α x 10-6
[oF] [Btu/h/in/oF] [psi] [-] [1/oF]
1600 1.21 20.1 0.32 9.42
1800 1.29 17.5 0.35 10.3
2000 1.38 14.2 0.39 11.6
The graphical representation of the tabular data is shown in FIGURE 78 after

the least square fitting was performed.
FIGURE 78 Material properties for Inconel 718. Graphical representation
Heat transfer solution To solve the heat transfer problem, we need to solve the linear problem first
and then perform the nonlinear analysis to account for the changes in the ther-
mal conductivity with temperature. Make sure the theory selector is set to Heat
Transfer, and for the linear solution provide the following input in the solver
interface:

1
tive > Convergence: None > Solve button.
Check the overall quality of the solution by computing the estimated relative error
in energy norm (FIGURE 79). The solution corresponding to p=6 will be used as
the starting point for the nonlinear analysis.
FIGURE 79 Connected pipes: Error estimate for the linear

solution of the heat transfer problem.
Perform a nonlinear analysis with the following settings:

• Nonlinear tab > Linear Sol.: SOL_TH > Run: 6 > Nonlinear ID: NL_TH >
Type: Material (NLMat) > Convergence: Temperature > Technique: Direct
Int. > Tolerance(%): 0.1 > Iteration Limit: 5
The nonlinear solution should be obtained in 4 iterations. The temperature distribu-

tion corresponding to the linear and nonlinear solutions for run #6 (888 DOF) is
shown in FIGURE 80.
Elasticity solution Set the theory selector to Elasticity, and consider the boundary conditions shown in
FIGURE 81. Symmetry constraints are imposed in the back face and along the
symmetry plane. A thermal load is applied to all the elements in the mesh by
importing the nonlinear solution from the heat transfer problem as shown in the
input interface in the figure. The reference temperature is set to zero in this case,

(a)
FIGURE 80 Connected pipes:

Temperature distribution for
run #6,
(a) linear and (b) nonlinear
solutions. (b)
since we are working in degrees Fahrenheit. The reference temperature is

important when radiation boundary conditions are used and the temperature
field is in degrees Kelvin.
To obtain the linear solution, provide the following input in the solver inter-
face:
Iterative > Convergence: None > Solve button.
The results of the analysis are shown in FIGURE 82, which includes the esti-
mated relative error in energy norm, the contour plot of the first principal stress
(S1) and the convergence of the maximum S1 as a function of the number of
degrees of freedom (DOF).

FIGURE 81 Connected pipes: Boundary conditions for

the elasticity problem.

(a)
(b)
(c)
FIGURE 82 Connected pipes: Results from the elasticity solutions. (a) estimated
relative error in energy norm, (b) stress distribution S1, (c) convergence of
maximum S1.

2 Multi-body Contact
A multi-body contact capability has been implemented in StressCheck to account

for frictionless mechanical contact in the context of the p-version of the finite ele-
ment method. The implementation is available for Planar, Extrusion, Axisymmetric
and 3D elasticity problems. Material nonlinearities of the contacting bodies can
also be incorporated in the analysis. The effect of geometric nonlinearities are not
yet available.
Formulation
The approach, using a contact element formulation that does not require mesh con-
formity between contacting surfaces, is based on the principle of minimum poten-
tial energy subjected to the proper kinematic conditions at the contacting surfaces.
The problem must be solved by an iterative procedure to determine the contact
regions and has been treated using the augmented Lagrangian method (Ref. [70]).
The method is briefly described in the following.
Consider two elastic bodies (bodies 1 and 2) with a declared contact region as indi-
cated schematically in FIGURE 83. A normal spring of constant Kc is placed along
the contact surfaces of each body. Let un1 be the normal displacement (positive if
outward) at any point along the declared contact region of body 1, and un2 the cor-
responding normal displacement of body 2. The first step in the contact analysis is
to determine the initial gap between the contacting surfaces. This is accomplished
Advanced Topics Chapter 2: Multi-body Contact

2
FIGURE 83 Schematic of the contact algorithm.
by projecting normal vectors from each control point of one contact zone to the
element faces belonging to the opposite contact zone. The distance between the
point and its projection determines the initial gap (g0). Therefore the gap func-
tion (g) is given by:
g = g 0 – u n1 – u n2
If g < 0, then body 1 is penetrating body 2. The condition to be satisfied for

each contact region is
P×g = 0
102 Chapter 2: Multi-body Contact Advanced Topics

where P is the pressure in the contact zone and g is the corresponding gap.
During each iteration of the contact solution procedure, the values of un1 and un2
are recomputed and the value of the gap updated. For the ith iteration, the contact
pressure (P) is computed as:
(i)
(i)  0 if ( g > 0 )
P =  (i – 1) (i) (i)
P + K c × g if ( g < 0 )
The corresponding normal tractions in the contact region of body 1 are then given
as:
2
(i) (i)
(i)  K c ( g 0 – u n2 ) + P if ( g < 0)
T n1 =  EQ 9
 K c u n1 if ( g ( i ) > 0 )
The traction applied at each point outside the penetration region (corrective traction
when g > 0) is needed to cancel the effect of the contact spring. Similarly, the trac-
tions for body 2 are:
(i) (i)
(i)  K c ( g 0 – u n1 ) + P if ( g < 0)
T n2 =  EQ 10
 K c u n2 if ( g ( i ) > 0 )
After each iteration, the maximum pressure between the contacting surfaces is mea-
sured to determine whether a significant change has occurred since the previous
iteration. If the change is larger than a pre-specified tolerance, then the iterative
procedure continues. If the change in pressure is less than the tolerance, then the
analysis procedure is terminated, and the solution vectors are available for post-
processing.
With this implementation, the stiffness matrix for each body is computed only once
and the effect of the contact is incorporated in the load vector by updating the trac-
tions in the contact zone according to EQ 9 and EQ 10. The solution is very effi-

cient because only the right hand side of the system of equations is updated in
each iteration. However this solution strategy does not account for material or
geometric nonlinearities. Material nonlinearities can be incorporated after the
initial contact solution was obtained. Geometric nonlinearities are not consid-
ered in this implementation.
Contact analysis
To perform a multi-body contact analysis with the bodies having linear or non-
linear material properties, proceed as follows:
• Create (or import) the geometric description of the bodies and create
the mesh for each body. There is no need to have mesh conformity
between contacting surfaces.
2 • Create Contact Zones by selecting surfaces or faces in 3D and curves
or edges in 2D. At least two contact zones must be created for each
potential contact pair.
• Define and assign material properties to all the elements as done for a
standard linear or material nonlinear analysis.
• Apply the mechanical/thermal loads to the bodies as done for any anal-
ysis.
• Specify the constraints for each body and activate the contact zones by
connecting the contact zone pairs. A contact constant (Kc) must be pro-
vided. The contact constant should be selected based on the moduli of
elasticity of the materials in contact, in such a way that its value is
between 0.1 and 1.0 times the value of the smallest modulus of elastic-
ity (0.1Emin < Kc < Emin) when using US units (psi), and between 4 and
40 times the value of the smallest modulus of elasticity (4Emin < Kc <
40Emin) when using SI units (MPa). The units for the contact constant
are [F/L/L2].
• After defining the solution ID record, perform a linear analysis. The
program will recognize the presence of the contact constraints and per-
form the necessary iterations until the variation in the maximum con-
tact pressure is less than 0.01% or the number of iteration reaches 40.
The procedure will be illustrated with an example.
Changing defaults It is possible to override the tolerance and iteration limit values by creating
underscore parameters. To override the tolerance in the variation of the contact
pressure create the parameter _contact_tol (must be a positive number less than

or equal to 10). To override the maximum number of iterations create the parameter
_contact_iter (must be a positive number less than or equal to 100).
Note about constraints Each body of a multi-body contact problem must have enough constraints for
obtaining the solution. The presence of the contact zone is equivalent to having a
normal spring acting on the declared contact region. If no other constraints are
present in the body, it is important to ascertain whether the body can have rigid
body displacement or rotations. If it does, it is necessary to add nodal constraints
records to prevent the rigid body modes.
Note about error estimation When performing a sequence of linear solutions with contact, the estimated rel-
ative error in energy norm may not be reliable. This is because the contact condi-
tion and the load vector are not the same for each p-level. Consequently, a slightly
different problem is solved for each p-level which affects the error estimator. The 2
sequence of solutions should be used to ascertain the convergence of the data of
interest, however.
Model problem 1: Plate in tension with filled hole

Consider the case shown in FIGURE 84 which represents a 0.1-in thick aluminum
plate with a central hole and with a neat fit steel fastener filling the hole. The plate
is loaded in tension, and because of symmetry only half of problem is discretized
using 4 quadrilateral elements for the plate and four triangular elements for the fas-
tener. The symbols around the interface between the fastener and the plate represent
the distributed contact spring.
Model creation Open a new data base and make sure the analysis type is set to Planar Elasticity and
the units selector is set to in/lbf/sec/F. Select the Model Information icon and create
the parameter ‘gap’:
• Name: gap > Description: Fastener gap > Value: 0.05 > Accept. This
parameter is useful for facilitating mesh construction. It will be set to zero
when performing the contact analysis.
Next, select the Create Model icon and then the Geometry tab and construct the
geometric description for the plate and fastener as follows:
• Create > Rectangle > Locate > Input: X: 0.0, Y: 0.0, Z: 0.0, Width: 0.5,
Height: 1.0 > Rot-Z: 0.0 > Accept.
• Create > Circle > Locate > Input: X: 0.0, Y: 0.5, Z: 0.0, radius: 0.15, P1-
Min: 0.0, P1-Max: 360, Rot-Z: 0.0 > Accept.

2
FIGURE 84 Plate with filled hole. Dimensions and boundary conditions.
• Create > Circle > Local > Input: radius: 0.15-gap, P1-Min: 0.0, P1-
Max: 360, Rot-Z: 0.0 > Click on the system of the previous circle.
• Create > System > Locate > Data Type: Cylindrical > Input: X: 0.0, Y:
0.5, Z: 0.0, Rot-X: 0.0, Rot-Y: 0.0, Rot-Z: 0.0 > Accept. This system
will be used for the extraction of stresses around the hole.
Contact zones Next, select the Mesh tab and create the nodes and elements for the plate and
fastener as shown in FIGURE 85a. Then create two contact zones as follows:
• Create > Contact Zone > Edge curve. Select any element edge of the
plate around the hole (all 4 edges will be highlighted) and then click on
Accept.
• Create > Contact Zone > Edge curve. Select any element edge of the
fastener around the hole (all 4 edges will be highlighted) and then click
on Accept (FIGURE 85b).

contact zones
select edge to
create contact zone
(b)
(a)
FIGURE 85 Plate with filled hole. (a) Nodes and elements when gap=0.05,
(b) Contact zones.
Select the Thickness tab and complete as follows:

• Select > All Element > Thickness: 0.1 > Accept.
Select the Material tab and define the material properties for the plate and fastener
with the coefficients shown in FIGURE 84. Assign the corresponding materials to
the elements of the plate and fastener.
• Define > ID: Alum > Material: Linear > Type: Isotropic > Case: Pl. Stress >
E: 10.5e6, v=0.3 > Accept.
• Define > ID: Steel > Material: Linear > Type: Isotropic > Case: Pl. Stress >
E: 29e6, v=0.295 > Accept.
• Assign > Select > Any Element > ID: Alum > Color: Aluminum > Select
the 4 quadrilateral elements and then click on Accept.
• Assign > Select > Any Element > ID: Steel > Color: Fastener > Select the 4
triangular elements and the click on Accept.

After that, select the Load tab and apply a uniform normal traction of magni-
tude 10,000 to the top and bottom of the plate as shown in FIGURE 84.
• Select > Any Curve > Traction > ID: Load > Direction: Normal/Tan. >
Normal: 10000. Select the top and bottom lines of the rectangle and
click on Accept.
Now select the Constraint tab and proceed as follows:

• Select > Any Curve > Symmetry > ID: Const. Select the left line of the
rectangle and click on Accept.
• Select > Contact Zone > Contact > ID: Const > Direction: Normal/Tan
> Normal: 10e6 > With the mouse, left-click on each contact zone and
the corresponding record will be created. The constraint symbols in the
contact zone are shown in FIGURE 84.
2 Select the Solution ID tab and create the solution ID record:
• Solution ID: SOL > Constraint ID: CONST > Load ID: LOAD >
Accept.
Note: This problem can be loaded from the Parts Handbook:

Fastener2D_contact.sci.
Solution To perform the contact analysis, first set the parameter gap=0.0, then select the
Compute Solution icon and the Linear tab. Perform an upward p-extension
from p=6 to 8.
• Linear tab > Upward Extension > p-limits: 6 to 8.
• SOLVE! tab > Initialize > Automatic > Iterative > Solve.
The status window of the solver interface will display the status of execution as
shown in FIGURE 86. A summary of the number of iterations for each run and
the corresponding error during the contact analysis can be found under File >
View Session Log.
Results The deformed shape in scale 20:1 and the contour plot of the first principal
stress for run #3 are shown in FIGURE 87. To plot the deformed shape, select
the View Results icon and then the Plot tab:
• Select > All Elements > SOL > Run: 3 > Shape: Deformed > Midsides:
10 > Scale: Turn switch on and enter 20 > Plot
To plot the contour fringes of the first principal stress:

FIGURE 86 Solver status after analysis completion. 2
• Select > All Elements > SOL > Run: 3 > Contour: Fringe On > Shape:
Undeformed > Function: S1 > Midsides: 10 > Plot.
FIGURE 87 Deformed shape (Scale 20:1) and first principal stress for run # 3.

Note that the maximum principal stress occurs in the plate at the edge of the
hole. To find the magnitude and location of the maximum principal stress,
select the Min/Max tab and complete as follows:
• Select > Edge Curve > Line > SOL > Run: 1 to 3 > Function: S1 >
Midsides: 15 > Maximum > Select any edge of the plate elements
around the hole and click on Accept. You may need to select the Shrink
Elements icon to make the proper edge selection
The results shown in FIGURE 88 include the magnitude and location of the
maximum principal stress S1 for each run as well as the estimated value for the
maximum.
FIGURE 88 Convergence of the maximum principal stress around the hole.
FIGURE 89 shows the bearing stress distribution around the hole in the plate
for all three runs. Because of the coarse mesh used in the region of contact,
there is an oscillation in the stress distribution near the point where the contact
is lost. The pressure distribution was obtained by extracting the stress Sx in a
cylindrical system located at the center of the hole. The point extraction is per-
formed along the quadrilateral element edges located around the hole. To per-
form the extraction, select the Points tab and complete as follows:
Select > EdgeCurve > SOL > runs: 1 to 3 > Function: Sx > # of pts.: 43 > Sys-
tem: SYS3 (the name of the cylindrical system located at the center of the
hole). Select any edge of the plate elements around the hole and click on
Accept. You may need to select the Shrink Elements icon to make the proper
edge selection.

2
FIGURE 89 Pressure distribution
around the hole in the plate for
all 3 runs.
The effect of the neat fit fastener is to limit the free deformation of the hole, and
therefore the stresses are lower than in an open hole analysis. The stress distribution
corresponding to the open hole analysis (no fastener inside the hole) is shown in
FIGURE 90. To perform the open hole analysis, simply delete the contact con-
straint record, add a nodal constraint, delete the contact zones and the triangular
elements that represent the fastener. To do this, select the Create Model icon, and
proceed as follows:
• Constraint tab > Select the record with the contact information > Delete
• Constraint tab > Select > Node > Node > Data Type: Fixed > X off, Y on.
Select any node along the symmetry plane and then click on Accept.
• Mesh tab > Select > Contact Zone > Index subtab > Select both record from
the list and click on Delete.
• Mesh tab > Select > Triangle > Select all the triangular elements and then
click on Delete.
To run the linear analysis, select the Compute Solution icon and the Linear tab and
perform an upward p-extension from p=6 to 8.

2
FIGURE 90 First principal stress distribution from an open hole analysis.
Model Problem 2: Lug-clevis-pin

The next example involves a multi-body contact problem: The contact analysis
between a pin, a lug and a clevis as shown in FIGURE 91a. The 0.625” diame-
ter pin is made out of steel (E=30x106 psi, ν=0.29), while the lug (E=10.3x106
psi, ν=0.33) and clevis (E=10.1x106 psi, ν=0.30) are made out of aluminum.
The lug is 1.2 inches wide with a central hole of the same diameter as the pin
(neat fit pin). A load of 5,000 lb is applied to the free end of the lug and the
opposite end of the clevis is fixed. A contact constant of Kc=10x106 lb/in3 was
used for this problem.
Two contact analyses are considered for this problem: First, the bodies are
hand meshed and the analysis performed at p=8. Second, the bodies are auto
meshed and the analysis is performed at p=5.
The solid model was created in parametric form to facilitate meshing and visu-
alization of each body. Each body was created following the general proce-

Aluminum Lug
Aluminum Clevis
Steel Pin 2
(a) (b)
FIGURE 91 Multi-body contact problem - hand mesh: (a) Lug-pin-clevis

assembly. (b) Deformed configuration scale 10:1.
dures described in the Modeling Guide. The details of model construction are
omitted here.
Hand meshed bodies Each body was hand meshed using hexahedral and pentahedral elements maintain-
ing mesh compatibility across the contact region, that is node-by-node, edge-by-
edge and face-by-face compatibility. A total of 68 solid elements were used (18 for
the lug, 26 for the clevis and 24 for the pin) and the analysis was performed for p-
level=8 (24,698 DOF). Note that no special meshing was used in the contact
regions, the mesh layout follows the standard recommendation for smooth prob-
lems when using the p-version of the finite element method.
The contact zones were defined using the face option for a total of six contact zone
pairs. For example, FIGURE 92 shows two contact zone pairs (four contact zones)
for the clevis-pin. Four element faces of the clevis around the hole form one contact
zone. Similarly, four element faces of the pin around the hole form the other contact
zone. This contact zone pair is connected during the specification of the contact
constraint. Two additional contact zone pairs were defined for the clevis-pin to
cover the other half of the possible contact around the hole. One contact zone pair
between the pin and the lug is shown in FIGURE 93. The elements have been

Four faces on clevis
clevis-pin contact zone pairs Four faces on pin
FIGURE 92 Hand mesh model. Two contact zone pairs for the clevis-pin.
shrunk to facilitate visualization of the contact zones. Four faces of the hexahe-
dral elements in the lug were used for creating one contact zone, while four
faces of the pentahedral elements in the pin were used for creating the other
contact zone. One additional contact zone pair was defined for the lug-pin to
cover the other half of the possible contact around the hole.
Note: This problem can be loaded from the Parts Handbook: LugClevis-
Full.sci.
The deformed configuration in scale 10:1 is shown in FIGURE 91b. The con-
tour plot of the first principal stress on each body is shown in FIGURE 94. The
maximum principal stresses are:
Lug: σ1max = 61.6 ksi; Clevis: σ1max = 47.4 ksi; Pin: σ1max = 24.5 ksi
Auto meshed bodies In the second analysis, each body was meshed using the automatic mesh gener-
ator of StressCheck without mesh compatibility across the contact region. A
total of 1724 tetrahedral elements (181 for the lug, 1387 for the clevis and 156
for the pin) were created, and the contact analysis was performed for p-level=5
(122,079 DOF).

2
contact zone pair
FIGURE 93 Hand mesh model. One lug-pin contact zone pair.
The contact zones were defined using the surface option for a total of three contact
zone pairs as shown in FIGURE 95. The surface option allows the creation of the
contact zone lug
contact zone pin contact zone clevis
FIGURE 95 Auto mesh model. Contact zone pairs.
contact zones directly on the solid model before meshing.

2
FIGURE 94 Multi-body contact problem - hand mesh: Contour plots of the principal stress σ1
in each body.
Note: This problem can be loaded from the Parts Handbook: LugClevisFul-
lAutoMesh.sci.
The results shown in FIGURE 96 include the deformed configuration in scale

10:1 and the contour plots of the first principal stress for each body. The maxi-
mum principal stresses are:
Lug: σ1max = 61.8 ksi; Clevis: σ1max = 47.1 ksi; Pin: σ1max = 24.6 ksi.
Note that these results are very close to those obtained with the hand meshed
configuration, demonstrating the robustness of the algorithm when non-con-
forming meshes are used across the contact interface.

(a)
2
(b)
FIGURE 96 Multi-body contact problem - auto mesh: (a) deformed shape scale
10:1, (b) contour plots the principal stress σ1 in each body.
Creating contact zones

It is important to follow some guidelines when creating contact zone pairs to
improve the efficiency of the solution. These include:
• Create contact zone pairs of similar sizes on each body. Avoid creating con-
tact pairs in which one contact zone is much larger (or smaller) than the

other. There is no intrinsic problem with this, except for the efficiency
of the solution.
Contact pair Contact pair
recommended not recommended
• When creating a contact zone for one of the contact pairs try to avoid
multi intersection of the normal. To recognize the multi intersection
simple consider a point on one of the contact zones and imagine a nor-
mal to the surface of the body at that point. If the normal intersects the
other contact zone in more than one place, then you have a multi inter-
section situation. Again, there is no intrinsic problem with this, except
for the efficiency of the solution.
• Avoid contact zones that span over too many elements. Whenever pos-
sible split large contact regions is smaller groups if overlapping of the
contact pairs can be avoided.

3 Fracture Mechanics
3
This chapter describes the procedures for computing the stress intensity factors and the T-stress in linear
elastic fracture mechanics for two- and three-dimensional problems and the J-integral in elastic-plastic
fracture mechanics for planar and axisymmetric problems. The concept of stress intensity factors asso-
ciated with crack tips in linear elastic fracture mechanics is extended to general singular points or edges
for composites, anisotropic materials, and multi-material interfaces for linear elastostatics and steady
state heat transfer problems in two and three-dimensions.
You should be familiar with the procedures for creating a finite element mesh for planar, axisymmetric
and three-dimensional problems and executing a linear analysis. Additionally, you should be familiar
with Nonlinear Analysis if elastic-plastic computation is of interest.
Computing stress intensity factors

StressCheck has an advanced method implemented for the computation of Mode 1
and Mode 2 stress intensity factors in linear elastic fracture mechanics. The method
is superconvergent. This means that the error in stress intensity factors converges to
zero much faster than the error in energy norm as the number of degrees of freedom
is increased, provided that the number of degrees of freedom is sufficiently large.
Advanced Topics Chapter 3: Fracture Mechanics 119

Mesh design in the presence of cracks
There is an important practical benefit: The mesh does not have to be overly
refined at the crack tip, and the polynomial degree does not have to be high to
obtain accurate results.
The method, called the Contour Integral Method (CIM) is described in the
3 Theoretical Background chapter under Stress intensity factors heading. A
description of the scope and limitations of the implementation is given below.
Scope and limitations The stress intensity factor can be computed for the linear solution of isotropic
material in the absence of body forces and thermal loading. For axisymmetric
problems, the stress intensity factor depends on the radius of the integration
path (r). However, as r → 0 the contour integral becomes path-independent.
Numerical studies (Ref [45]) indicated that for r/rc < 0.1, the values of the
stress intensity factors are practically independent of r (rc = crack tip location).

As described in the Theoretical Background chapter under the Choice of an
extension process heading, geometrically graded meshes are optimal for prob-
lems in Category B. Typically, two layers of geometrically graded elements
towards the crack tip with a common factor of 0.15 will give excellent results.
When two layers of elements are used, the radius of the integration path for the
computation of stress intensity factors (SIF) and J-integral should be selected
such that it runs outside the innermost layer of elements.
It is possible, however, to use coarse meshes for the computation of the frac-
ture mechanics parameters. Care should be exercised in such cases because the
dependence of the results on the size of the integration circle can be more sig-
nificant. Numerical evidence indicates that for most practical cases in two-
dimensions, the coarse mesh approach should give results which will not differ
by more than 4 or 5% from those obtained with 2 layers of graded elements.
The following guidelines should be followed when designing a coarse mesh
near crack tips for planar and axisymmetric analyses:
• Avoid very distorted elements near the crack tip.
• Keep the size of the elements near the crack tip about the same.
• Whenever possible, use quadrilateral elements around the crack tip and
select the product space.
120 Chapter 3: Fracture Mechanics Advanced Topics

• When computing SIF or J-integral, select the radius of the integration path
such that it runs roughly through the middle of the smallest element around
the crack tip.
Example: Through-crack in linkage arm 3
As an application of the Contour Integral Method (CIM) in 2D, consider the prob-
lem shown in FIGURE 97 which represents a linkage arm detail. This problem was
discussed in Planar Elasticity (Analysis Guide) and the mesh shown in FIGURE 97
can be constructed following the procedure described in that chapter. Assume that a
through-crack originated at the edge of the hole is growing in the vertical direction
as shown. We wish to compute the stress intensity factors for crack lengths ranging
from 2.0 to 6.0. A 10 KN bearing load is applied to the inner bore of the linkage
arm, and the material properties of an aluminum alloy 2014-T6 are used: E = 75200
MPa, v = 0.397.
15 crack
10
20
20
100
FIGURE 97 Linkage arm detail. Geometric description and mesh

for uncracked case. All dimensions in mm.

Mesh modification The first task is to modify the mesh shown in FIGURE 97 such that the crack is
represented. Since the crack length is variable, we will use a parameter such
that the mesh can be changed conveniently. The procedure is described in the
following:
3 1. Load the Linkage Arm problem from the Tutorial Handbook (file: Link-
ageArm.sci) or refer to Planar Elasticity in the Analysis Guide to construct the
mesh shown in FIGURE 97. If you load the problem from the handbook
library, make sure to turn ON all the objects (points, nodes, curves) before pro-
ceeding.
2. Define a parameter for the crack length (cl). From the Main Toolbar, select
the Model Info icon and when the corresponding dialog window appears select
the Parameters tab and enter the following information:
• Name: cl, Description: Crack length, Value: 2.0, Limit: > 0.0, Accept
3. Create a node at the crack tip: From the Input dialog box select the following
commands:
• Mesh tab > Create > Node > Locate > Input toggle ON > X: 100, Y:
10+cl, Z: 0 > Accept
4. Create four nodes by the offset method: (FIGURE 98a)

• Create > Node > Offset > offset: 75. Click on the inner and outer cir-
cles.
• Create > Node > Offset > offset: 105. Click on the inner and outer cir-
cles.
5. Re-define the two existing elements in the neighborhood of the crack so as

to make room for the new elements that will be needed for the crack detail. Use
the following commands:
• Mesh tab > Move To > Quadrilateral > Selection.
Now point to one of the quadrilateral elements to be redefined. It will be high-

lighted. Point to and click on the new set of nodes which are to be the vertices
of the new quadrilateral element. Then repeat this procedure for the other
quadrilateral element. When you are done you should see the display as shown
in FIGURE 98b on your screen.
6. Define a node as an offset on the inner circle: (FIGURE 99)

• Create > Node > Offset > offset: 85. Click on the inner circle.

(a) (b)
FIGURE 98 Finite element mesh at the end of step 5.
This node is in a temporary location, the value of 85 was selected arbitrarily to lie
to the right of the crack face. This enables us to define the elements such that the
crack face will not be interpreted as an interelement boundary.
7. Define triangular elements, using:

• Mesh tab > Create > Triangle > Selection
Point to the appropriate nodes to create 6 triangular elements. At the end of this step
you should see the display as shown in FIGURE 99a on your screen. Note that an
open crack is visible.
8. Move the node created in step 6 to the crack face using the following commands:
• Edit > Node > Offset. Point to and click on the node created in step 6.
Change the offset to 90 then click on the Replace button.
The node will move onto the crack face and will overlap with the other node
already on the crack face as shown in FIGURE 99b.
Design study 9. You are now ready to execute a design study. From the main toolbar select the
Handbook icon, then the Analysis tab and then the Design Study tab as shown in
FIGURE 100. When performing a design study, you control which parameter will

(a) (b)
FIGURE 99 Detail of mesh at the end of steps 7 and 8.
remain constant, and which parameter will vary during the analysis. You also
control how many steps will be performed during which the variable parameter
will be increased from its initial to its final value.
FIGURE 100 Design Study interface for model problem.
To make the crack length (cl) variable, simply check the box to the left of the
parameter name and enter the initial (2.0) and the final values for cl (6.0).
Select the analysis type as Linear, supply the number of Steps between the ini-

tial and the final (5 in this case) and the polynomial order for the execution (p=8).
FIGURE 100 shows the settings for the Design Study interface.
In order to visualize how the parametric definition of crack tip works, click on the
toggle switch in front of Scale and click on the up arrow switch. The crack tip will
move and the elements will be automatically re-sized to suit the new crack length. 3
When you are ready to begin the design study, simply click the Solve button. The
model will be updated automatically and the resulting configuration will appear in
the model window.
Stress intensity factors 10. To find the stress intensity factors select the Results tab in the Handbook inter-
face and then the Fracture tab (FIGURE 101). Click on the last solution in the
FIGURE 101 Handbook Results tab for fracture extraction.
scrolling list and the solution name and run numbers will be included in the corre-
sponding fields. Click on the Aux. toggle switch and enter the name of the parame-
ter representing the crack lenth (cl). This will make cl the independent variable of
the tabular data. Choose Method: SIF (Stress Intensity Factors) and the radius of
the integration path. For a design study with variable crack length, the value of
Radius can be entered as a parametric expression (cl/3 in this case).
Now bring the cursor to the node on the crack tip, then click the left button. A circle
will appear which will change in size as the model is updated for each value of the

crack length. If you entered Radius=cl/3 then you will have the results shown
in FIGURE 102.
FIGURE 102 Stress Intensity Factors.Graph

and Table.
The circle represents the integration path used by StressCheck. If, instead of
the finite element solution, the exact solution had been used, the stress inten-
sity factors would be independent of the integration path. Because the finite
element solution is only an approximation, a weak dependence on the size of
the circle can be expected. Variations in the stress intensity factors with respect
to the radius of the integration circle should be well within the acceptable toler-
ance.
Note: The radius of the circle is arbitrary but the circle must be inside the
solution domain and it must cross only interelement boundaries which lie
on radial lines with respect to the crack tip.
Energy release rate 11. To compute the energy release rate (G) which, in the case of linear prob-
lems, is the same as the J-integral, select the Method “J-Integral” and repeat
the procedure described in step 10. If the radius of your integration circle is cl/
3 then you will have the results shown in FIGURE 103. Note that for run #3
and plane stress:
2 2
K +K 2 2
I II 1005 + 15 N
J = ---------------------- = ------------------------------ = 13.47 ---------
E 4 mm
7.5 × 10

Model problem: 2D-Cracked lug
which is very close to the J-value computed directly by StressCheck (J=13.65 for
1502 DOF).
FIGURE 103 J-integral computation.

This section presents a step by step description for the creation, execution and post-
processing of a two-dimensional problem containing a through crack.
Problem description
The lug shown in FIGURE 104 is loaded through the pin hole by a horizontal force
of 60000 lb. The corresponding imposed normal stresses (the tractions) are
assumed to be distributed sinusoidally over one half of the pin hole. The lug has a
uniform thickness of 0.5 inches. It is made of an isotropic elastic material
(E=30x106 psi, v=0.3). There are two symmetric cracks, 0.75 inches long, emanat-
ing from the edge of the hole, perpendicular to the direction of the load application.
Of interest are the values of the stress intensity factors K1 and K2. The problem is
solved using two-dimensional elasticity and assuming plane-stress conditions. Due
to symmetry, only half of the lug needs to be analyzed. The lug is held in equilib-
rium by uniform tensile stress along CD.

AB: Tn = Tt = 0
BC: Tn = Tt = 0
CD: Tn = 20000 psi, Tt = 0
DE: un = Tt = 0
3
EF: Tn = Tt = 0
FG: Tn = Tt = 0
GH: Tn = Tt = 0
HK: Tn =-(192000/π) cosθ, Tt = 0
(0 ≤ θ ≤ π/2)
KA: un = Tt = 0
FIGURE 104 Cracked lug. The dimensions shown are in inches.
Geometry and mesh To generate the finite element mesh shown in FIGURE 105 open a new data-
base, select the Create Model icon and then select the Geometry tab in the
StressCheck Input window. Make sure that the analysis type is Planar Elastic-
ity, the units are in/lbf/sec/F, and proceed as follows:
FIGURE 105 Cracked lug: Finite element mesh.

1. Provide the geometric description of the lug: Create a rectangle, two arcs of cir-
cle, a horizontal line through the symmetry section of the lug and two auxiliary
lines.
• Geometry tab > Create > Rectangle > Locate > X: 0.0, Y: 0.0, Z: 0.0,
Width: 7.5, Height: 3.0, Rot-Z: 0.0 > Accept 3
• Create > Circle > Locate > X: 7.5, Y: 0.0, Z: 0.0, Radius: 1.25, P1-Min: 0.0,
P1-Max: 180, Rot-Z: 0.0 > Accept
• Create > Circle > Local > Radius: 3.0, P1-Min: 0.0, P1-Max: 90 > Click on
the system of the circle created in the previous step. This action creates an
arc of circle of radius 3 inches in the range 0 degrees to 90 degrees.
• Create > Line > Locate > X: 0, Y: 0, Z: 0, Length: 10.5, Angle: 0.0 >
Accept.
• Create > Point > Offset > offset: 75. Click on the inner circle.
• Create > Point > Offset > offset: 105. Click on the inner circle.
• Create > Line > Point > Length: 2.0, Angle: 85. Point to the first point cre-
ated in the previous step.
• Create > Line > Point > length: 2.0, angle: 95. Point to the second point cre-
ated in the previous step.
The Model Window should look as shown in FIGURE 106.
FIGURE 106 Geometric representation of the solution domain.
2. Define the position of the nodes needed for mesh construction. Select the Mesh
tab in the StressCheck Input window and follow these steps:

• Mesh tab > Create > Node > Offset > Repeat #= 5 > offset: 0 + 45.
Make sure that the switch in front of offset is on and with the mouse
cursor select the inner circle.
• Create > Node > Offset > Repeat #= 3 > offset: 0 + 45. Point to the arc
of the outer circle.
3 • Note: After completing the last step, turn the Repeat switch OFF.
• Create > Node > Intersection. Click on the line with an inclination of
85 degrees and on the Circle-Arc. A node is located at the intersection.
Repeat for the inner circle. Then, select the line with an inclination of
95 degrees and the upper horizontal line of the rectangle to locate
another node. Create one more node by intersecting the same line and
the inner circle.
• Create > Node > Locate > Input switch on > X: 4.5, Y: 0, Z: 0. >
Accept
• Create > Node > Locate > X: 4.5, Y: 3, Z: 0. > Accept
• Create > Node > Intersection. Define two nodes at the intersection of
the left vertical line of the rectangle and the two horizontal lines.
In order to introduce the crack into the mesh, create a node at 2.0 inches from
the center of the hole (the crack tip), and another node on the circle close to the
90 degrees offset.
• Create > Node > Locate > X: 7.5, Y: 2, Z: 0. > Accept
• Create > Node > Offset > offset: 95. Click on the inner circle. This
node will be moved to the 90 degrees offset position once the elements
are defined, to close the crack.
The Model window should look as shown in FIGURE 107.
FIGURE 107 Location of nodes.

3. Define all the elements of the mesh.

• Create > Quadrilateral > Selection. Define five quadrilateral elements by
associating the elements with the appropriate nodes as shown in FIGURE
105.
• Create > Triangle > Selection. Define six triangular elements by associating 3
the elements with the appropriate nodes as shown in FIGURE 105 and FIG-
URE 108a.
(a) open crack (b) closed crack
FIGURE 108 Crack definition.
• Edit > Node > Offset. Click on the node created at 95 degrees offset on the
inner circle. When the offset appears on the input area, modify the input for
the offset field to offset= 90. Then select the Replace button. The mesh
should be updated as shown in FIGURE 108b.
Thickness Assign the thickness to all elements by selecting the Thickness tab in the Stress-
Check Input window and completing the required information:
• Thickness tab > Select > All Elements > Thickness: 0.5 > Accept
Material properties Define and assign the material properties to all elements by selecting the Material
tab and completing the required information.
• Material tab > Define > Linear > Selection > Define tab > ID: PROP >
Material: Linear > Units: US > Type: Isotropic > Case: Plane Stress > E:
3.0e+07, v: 0.3 > Accept
• Assign tab > Select > All Elements > Selection > ID: PROP > Accept

Load Equilibrating tractions are applied to the lug. A bearing is applied at the lug pin
location, and a uniform traction is applied at the other end of the lug. To apply
these loads, select the Load tab and supply the following information (see FIG-
URE 105):
• Load tab > Select > Any Curve > Bearing > ID: LOAD > Direction:
3 Mag./Dir. > Magnitude: 60000, Angle: 0.0. Select the lug hole and
then click on the Accept button.
• Select > Any Curve > Traction > ID: LOAD > Direction: Norm./Tan. >
Normal: 20000. With the mouse click on the left line of the rectangle,
and then select the Accept button.
Constraint Select the Constraint tab in the StressCheck Input window. Assign zero normal
displacement to the symmetry line and avoid rigid body motion, by following
the next two steps (FIGURE 105):
• Constraint tab > Select > Any Curve > Symmetry > ID: CONST. With
the mouse point to the symmetry line, and then select Accept.
• Select > Node > Node > ID: CONST > Direction: XY > Data Type:
Fixed > Turn on the switch in front of X and with the mouse point to
the leftmost node on the symmetry line, and then click on the Accept
button.
Solution ID Select the Solution ID tab in the StressCheck Input window and complete as
follows:
CONST > Load ID: LOAD > Accept
Execution
Select the Compute Solution icon from the Main Toolbar and perform a Linear
Analysis, upward from p=1 to p=8 in automatic mode.
• SOLVE! tab > Initialize > Automatic > Iterative > Convergence: None
> Solve.

Results
Error estimation The converging sequence of finite element solutions is displayed in FIGURE 109.
The large error in energy norm is an indication of the influence of the singularity on
the strain energy of the solution. Note however that the objective of the analysis is 3
to compute the stress intensity factors using a superconvergent method, therefore
the quality of the solution is assessed based on the convergence of K1 and K2 rather
than in energy norm, as shown in the following.
Stress intensity factors The stress intensity factors K1 and K2 are computed using superconvergent extrac-
tion by the contour integral method over a circular path around the crack tip. The
main advantage of this implementation is that simply selecting a radius for the cir-
cle which defines the integration path and pointing to the crack tip with the cursor,
accurate results and fast convergence are obtained, even with coarse meshes, such
as the one used in this model problem.

To compute the stress intensity factors, select the Fracture tab from the Results
window. When the input area appears (FIGURE 110), complete the following
information:
FIGURE 110 Fracture input area in the Results tab.
• Fracture tab > Select > Node > SIF (Stress Intensity Factor) > Solution:
SOL > Run: 1 to 8 > Radius: 0.3 (turn on the switch in front of Radius
and enter the value).
Move the cursor (its shape is circular now) into the Model window and click
on the node at the crack tip (FIGURE 111). The values of K1 and K2 will be
displayed on the Graph windows, as shown in FIGURE 112. K1 and K2 have

units of psi-in1/2. Note that the difference between the solution at p=8 and the esti-
mated limit is about 1% for K1 and K2.
FIGURE 111 Integration path around the Crack tip.
If the switch in front of Radius is left unchecked, then the size of the integration
path is determined on the screen. Bring the cursor to the node on the crack tip, then
click the left button and drag the cursor. A circle will appear which will grow in
size as you are dragging the cursor. Once the circle is of the desired size, click the
left button of the mouse to perform the computation. As you drag the cursor, the
current radius of the integration circle appears in the dialog box.
Remark The sensitivity of K1 and K2 to the size of the integration path is very small consid-
ering the coarse mesh used for the analysis. Theoretically it should be path indepen-
dent. The size of the integration path can be chosen arbitrarily, provided that the
circle intersects only radial lines emanating from the crack tip.
1. Compute K1 and K2 for various radius of the integration path. For all possible
integration path radii, the values of K1 and K2 will not vary by more than 5% from
each other.
2. Refine the mesh as shown in FIGURE 113. The two layers of elements around
the crack tip have radii of 0.11 inches and 0.017 inches, respectively. Recompute
the values of K1 and K2. For this case the estimated relative error in energy norm is
below 1% at p=8 and the values of the stress intensity factors are K1=115000 psi-
in1/2 and K2=13000 psi-in1/2.

FIGURE 112 Stress intensity factors K1 and K2.
FIGURE 113 Refined mesh around crack tip.

Model problem:3D-Cracked panel
3. Referring to the mesh shown in FIGURE 105, change the crack length and
recompute the stress intensity factors. In many applications K1 and K2 must be cal-
culated for a range of crack lengths. StressCheck makes it easy to accomplish this.
Select the Mesh tab from the StressCheck Input window and follow the steps indi-
cated below.
3
• Mesh tab > Edit > Node > Locate. Click on the node that define the tip of
the crack. The information about the location of the node is displayed in the
input region. Change the Y-coordinate to its new value and then click on the
Replace button. The new position of the node appears on the screen, and the
mesh is updated.

This section presents an example for the computation of stress intensity factors
(SIF) for a three-dimensional problem containing a through crack. This problem
may be loaded from the 3D-Fracture Handbook (file: CrackPanel.sci).
Problem description
Consider the case of a through-thickness crack in a thick isotropic plate under ten-
sion loading as shown in FIGURE 114. The plate is 2.0 inches wide, 2.0 inches
high and 1.0 inch thick with a 1.0 inch edge crack. The plate is subjected to a unit
traction and the material properties are those of an aluminum alloy 2014-T6 with
E=10.9x106 psi and ν=0.397. The finite element mesh was designed to include one
layer of geometrically graded elements towards the singular edge with a common
factor of 0.15. Since the crack length is unity, the circular region around the crack
front has a radius of 0.15. A total of 16 elements (8 hexas and 8 pentas) were
defined as shown in FIGURE 114. Because the body is in equilibrium under the
applied loads, only rigid body constraints were specified.
Execution
Once the model is created or loaded from the Fracture Handbook, select the Com-
pute Solution icon from the Main Toolbar and perform a Linear Analysis in auto-
matic mode.
• Linear tab > Extension: Downward-p > p-limits: 8 to 1.

3
1.0 1.0
1.0
1.0
1.0
FIGURE 114 Through crack in a thick isotropic plate.
• SOLVE! tab > Initialize > Automatic > Iterative > None> Solve.
Results
First, let us assess the overall quality of the solution by computing the esti-
mated relative error in energy norm, and then extract the stress intensity factors
using the contour integral method.
Error estimation The converging sequence of finite element solutions is displayed in FIGURE
115. The rate of convergence and the magnitude of the error in energy norm are
an indication of the influence of the singularity on the strain energy of the solu-
tion. Note, however, that the objective of the analysis is to compute the stress
intensity factors using a superconvergent method, therefore the quality of the
solution is assessed based on the convergence of K1 and K2 rather than in
energy norm, as shown in the following.

Stress intensity factors The stress intensity factors K1 and K2 are computed using superconvergent extrac-
tion by the contour integral method over a circular path around the crack front. The
procedure implemented in StressCheck determines a cutting plane normal to the
tangent to the crack edge at the pick point and extracts the global components of the
stresses and displacements along a circular path contained in the cutting plane.
These stresses and displacements are projected into the cutting plane and integrated
with the extraction function to compute the contour integral as described in the
Theoretical Background chapter under Stress intensity factors heading.
To compute the stress intensity factors, select the Fracture tab from the Results win-
dow. When the input area appears (FIGURE 110), complete the following informa-
tion:
• Fracture tab > Select > Edge > SIF > Solution: SOL > Run: 1 to 8 > Radius:
0.25 (turn on the switch in front of Radius and enter the value).
Move the cursor into the Model window and click on the crack front at the location
you want the computation to be performed. FIGURE 116 shows the results of the
stress intensity factors (SIFs) obtained from the set of finite element solutions for
polynomial orders ranging from p=1 to 8 at three locations along the crack front.
Z=0.5 represents the center of the plate and Z=0.0 is one of the free faces.

FIGURE 116 Stress intensity factors at

three locations along the crack front.
Note the variation of the SIFs along the crack front, with the largest value of
K1 at the center of the plate. The estimated limits for the SIFs are also included
which indicate the strong convergence characteristics of the implemented
extraction procedure.

In order to visualize the crack front, you need to turn off the shade mode by select-
ing the wire frame icon from the Display options toolbar.
SIF along crack front The post-processing of StressCheck allows the extraction of K1 along the element
edges of the crack front at any number of points along each edge. To compute K1 at
3
several points along the crack front, select the Points tab from the Results interface
and proceed as follows:
• Select the edge(s) where the extraction of K1 is required (the option Edge
Curve can be used to select all edges that are slope continuous).
• Select K1 from the function list.
• Enter the number of points along the edge(s)
• Select the coordinate system for the computation of the angle at the extrac-
tion point and enter the value of the radius of the integration path.
The rule for the computation of the angle at the extraction point (P) is based on the
associativity of the edge to the underlying geometry as follows (FIGURE 117):
• If the selected edge lies on an ellipse, then the parametric angle of the
ellipse (φ) is computed in the local system of the ellipse, regardless of the
system selected.
• If the selected edge lies on a circle, then the polar angle (θ) is computed in
the local system of the circle, regardless of the system selected.
• If the selected edge lies on any other type of curve or surface, or it is
straight, then a polar angle (θ) is computed relative to the system selected in
the interface.
FIGURE 117 Angle at extraction point. Notation.

Computing the T-stress
If, during the extraction of K1, the radius of the integration path is too large
and falls outside the domain, the program will reduce its size by half and try
again. This operation will be repeated up to five times, each time dividing the
previous radius by 2. If, after five attempts, the radius is still too large, the
computation of K1 will fail and a warning message will be issued.
3
To illustrate the procedure, consider the extraction of the stress intensity factor
along the crack front of the 3D-cracked panel. Select the Points tab from the
Results window and complete the following information:
• Points tab > Select > Edge > Solution: SOL > Run: 1 to 1 > Func: K1 >
# of points: 10 > Rad: 0.25, select the edge along the crack front and
click on Accept.
The value of K1 is computed for run #1 at 12 points along the crack front (10
midpoints plus the two end points). The setting of the input box and the graph
of the results are shown in FIGURE 118.

The second term in the asymptotic expansion of the elasticity solution in the
neighborhood of a crack tip, called the T-stress, has significant influence on
crack growth direction and in the stability of the crack path. Cracks tend to turn
sharply in the presence of positive T-stress. The accurate computation of the T-
stress is therefore important for fracture analysis assessment of complex struc-
tures subjected to arbitrary loading. The combination of the p-version of the
finite element method and the extraction procedure based on a path indepen-
dent integral described in this section provide for the accurate and reliable
computation of the T-stress for cracks in two- and three-dimensions. A descrip-
tion of the implementation and examples are included in the following.
Implementation
The implementation of the T-stress extraction utilizes a path-independent inte-
gral based on the Betti-Rayleigh reciprocal theorem as described in Ref. [69].
Referring to Figure 119, in two-dimensions the T-stress can be computed from
the following contour integral:

FIGURE 118 SIF along crack front for the 3D-cracked panel.
°∫
* FE FE *
T = Ẽ ( σ ij u i – σ ij u i )n j dS
Γ EQ 11
°∫ °∫
* FE * FE FE * FE *
= Ẽ ( Tx ux + T y u y )Rdθ – ( Tx ux + T y u y )Rdθ
Γ Γ

where ni (i = 1, 2) are the components of the outward normal to the integration

path, and:
* cos 2θ + cos 4θ * cos 2θ – cos 4θ * sin 4θ

σ x = ------------------------------------
2
-, σ y = ------------------------------------
2
, τ xy = -------------2-
2πR 2πR 2πR
3
are the extraction stress functions in the local Cartesian coordinate system
located at the crack tip and:
* κ cos θ + cos 3θ * κ sin θ – sin 3θ

u x = – -------------------------------------, u x = -----------------------------------
8πGR 8πGR
are the extraction displacements in the same coordinate system, with κ=κ1=3-
4ν for plane strain and κ=κ2=(3-ν)/(1+ν) for plane stress (ν is the Poisson's
ratio of the material).
FE FE
σ ij and u i are the displacements and stresses on Γ obtained from the finite
element solution, and for plane-strain:,
E
Ẽ = E 1 = -------------2-
1–ν
and for plane-stress: Ẽ = E 2 = E (E is the modulus of elasticity).

y
Ty n
dS
σx
τxy Tx
σy
R
θ
crack x
Γ FIGURE 119 Contour integral near the crack tip.

In three-dimensions, the T-stress is computed using the same path-independent

integral over a circular path around the crack front at the pick point. The procedure
determines a cutting plane normal to the tangent to the crack edge at the pick point
and extracts the global components of the stresses and displacements along a circu-
lar path contained in the cutting plane. These stresses and displacements are pro-
jected into the cutting plane and integrated with the extraction function to compute 3
the T-stress as described above.
In general, a three-dimensional problem is neither a plane-stress nor a plane-strain

problem. Therefore, the question is how to determine the values of κ and Ẽ for the
extraction functions. The procedure to determine which values to use is based on
computing the stress components σx, σy, σz at a point in front of the crack and to
define γ as the ratio:
σz
γ = -----------------
σx + σy
If γ is close to Poisson’s ratio (ν), then the condition approaches that of plane-strain.
If is close to zero, then is plane-stress. The approach implemented in StressCheck
was to compute the value of γ at a point along the integration path (r = R, θ = 0),
and apply the following rule to determine κ and Ẽ :
 κ if ( γ > γ1 )  E if ( γ > γ 1 )
κ =  1 Ẽ =  1
 κ ( γ ) if ( γ < γ1 )  E ( γ ) if ( γ < γ 1 )
where γ1 = 0.1 and κ(γ) and E(γ) are determined from a cubic spline fitting between
the values of plane-strain and plane-stress:
2γ γ 2
κ ( γ ) = κ 1 + ( κ 2 – κ 1 )  ----- + 1  1 – ----
 γ1  γ 1
2γ γ 2
E ( γ ) = E 1 + ( E 2 – E 1 )  ----- + 1  1 – ----
 γ1  γ 1
With this procedure, the path-independent integral was extended to compute the T-
stress at any point along curved crack fronts.
The path-independent integral given by Eq. EQ 11 is computed along a circular

path around the crack tip/front. Once a sequence of finite elements solutions of

varying number of degrees of freedom is available, simply select the Results

interface to determine the value(s) of the T-stress. In Planar and 3D Elasticity,
the extraction of the T-stress is provided in the Fracture tab of the Results inter-
face when the method SIF is selected. Enter the radius of the circle for the inte-
gration path and selects the node located at the crack tip in 2D or the edge of
3 the crack front in 3D and the program provides the values of K1, K2 and T-
stress as shown in Figure 120.
FIGURE 120 Extraction of the T-stress from the Fracture tab in the Results interface.
For 3D-problems it is also possible to extract the T-stress at any number of

points along the crack front using the Points tab of the Results interface. In this
case, select the element edge(s) that correspond to the crack front, enter the
radius of the integration circle and the number of points along the crack front,
and select the function T-str from the list of available extraction functions to
obtain the distribution of the T-stress at a number of equally spaced points.
(Figure 121).
Benchmark problem
The benchmark problem analyzed in Ref. [69] is considered in this section and
consists of an edge crack on a square plate with a/L=0.5. Each side was loaded

FIGURE 121 Extraction of the T-stress from the Points tab in the Results interface.
with the tractions obtained from the K1, K2 and T related terms of the asymptotic
expansion of the stress field around the crack tip. The crack surface is traction free.
Thus, the problem represents an internal crack in an infinite plate under remote
loading such that K1, K2 and T can be varied independently. The tractions are
derived from the following expansions:
K1 θ θ K2 θ θ
- cos ---  1 – sin --- sin 3θ
σ x = ------------ - sin ---  2 + cos --- cos 3θ
------ + T – ------------ ------
2πr 2  2 2  2πr 2  2 2
K1 K2
- cos θ
σ y = ------------ ---  1 + sin θ
--- sin -----
- + - sin θ
3θ ------------ --- cos θ 3θ
--- cos -----
-
2πr 2  2 2  2πr 2 2 2
K1 θ θ K2
3θ- + ------------ θ θ
τ xy = ------------
- sin --- cos --- cos ----- - cos ---  1 – sin --- sin 3θ
------
2πr 2 2 2 2πr 2  2 2

Because there are no body forces, the system is in equilibrium under the
applied tractions and only nodal constraints are required to prevent rigid body
displacements and rotation. A finite element mesh was constructed with two
layers of refinement in geometric progression towards the crack tip with com-
mon factor of 0.15. The finite element mesh consisting of 24 elements is
3 shown in Figure 122. The following material properties were considered:
E=1.0, ν=0.3, plane-strain.
The values of K1, K2, and T were set to 1.0 in the definition of the tractions,
and the 2D problem was solved by p-extension for polynomial orders ranging
from 1 to 8 The estimated relative error in energy norm for the sequence of
0.15a
0.152a
FIGURE 122 Benchmark problem. 2D Mesh and boundary conditions.
solutions is shown in Figure 123. For p=8 (1583 DOF), the estimated global
error is less than 1%. The figure also shows the values obtained during the
extraction of K1, K2 and T for each run, indicating the excellent convergence
characteristics of the results. The exact value are K1=K2=T=1.0.
Table 3summarizes the results of the computation using the same mesh but for
different exact values of the fracture mechanics parameters. As can be seen

(a)
(b)
FIGURE 123 Benchmark problem. Plane-strain results: (a) Estimated relative error in
energy norm, (b) Convergence of K1, K2, and T-stress.
from the results, the values extracted from the finite element solution correspond-
ing to p=8 are very close to the exact values.
Note: This problem can be loaded from the 2D Fracture Handbook: TstressBench-
mark.sci.
TABLE 3. Summary of fracture mechanics parameters for various

combinations of exact values for K1, K2 and T-stress. Results from 2D
plane-strain at p=8.
Exact Values FEA at p=8 (1583 DOF)
K1 K2 T-stress K1 K2 T-stress
1.00 1.00 1.00 0.9997 0.9999 0.9998

1.00 0.00 1.00 0.9997 0.0 0.9998
10.0 0.00 1.00 9.9971 0.0 0.9983
1.00 0.00 10.0 0.9997 0.0 9.9998

TABLE 3. Summary of fracture mechanics parameters for various

combinations of exact values for K1, K2 and T-stress. Results from 2D
plane-strain at p=8.
Exact Values FEA at p=8 (1583 DOF)

3
K1 K2 T-stress K1 K2 T-stress
0.00 1.00 1.00 0.0 0.9999 1.0000

0.00 10.0 1.00 0.0 9.9990 1.0000
0.00 1.00 10.0 0.0 0.9999 10.000
The problem was also solved as a 3D elasticity problem, and to simulate plane-
strain conditions in 3D, symmetry constraints were imposed on element faces
with normals in the direction of the positive or negative z-axis. The finite ele-
ment mesh was graded in geometric progression towards the singular edge
with two layers of refinement as done for the 2D case (Figure 124). The equili-
Uz=0
FIGURE 124 Benchmark problem. 3D Mesh and boundary conditions.

brated tractions from the asymptotic expansion were applied to the 3D benchmark
problem and the solution was obtained by p-extension. The estimated relative error
in energy norm and the convergence of the values of K1, K2 and T-stress at one
point along the crack front are shown in Figure 125. The exact values are
K1=K2=T=1.0.
3
(a)
(b)
FIGURE 125 Benchmark problem. 3D results: (a) Estimated relative error in energy
norm, (b) Convergence of K1, K2, and T-stress at a point along the crack front.
Finally, Figure 126 shows the extraction of the T-stress from the Points tab of the
Results interface at 7 equally spaced points along the crack front. As expected, all
the values of the T-stress are practically the same, since plane-strain conditions
were enforced in the 3D problem.
Note: This problem can be loaded from the 3D Fracture Handbook:

TstressBenchmark3D.sci.

FIGURE 126 Benchmark problem. 3D results: T-stress along the crack front for p=8 (7010 DOF)
Model Problem: Fracture specimen

Double cantilever beam (DCB) specimens have a large positive T-stress that
may cause crack path instability under pure mode I loading. A DCB configura-
tion with a/w=0.5 and h/w=0.2 is considered for the analysis. The 44-element
mesh and boundary conditions are shown in Figure 127. Again, two layers of
refinement in geometric progression towards the crack tip were used. The load
was applied as a sinusoidal traction distribution along half of each hole with a
resultant value P=100. The following values were used for the dimensions:
a=1.0, w=2.0, h=0.4, th=0.5. No exact solution is available for the DCB speci-

3
0.092w
2h
w/12
P
w/12 a
w
FIGURE 127 Double cantilever beam specimen. 2D Mesh and boundary conditions.
men, but the results obtained with StressCheck are compared with those presented
in Ref. [69]. Since no values for the material parameters are given in the reference,
the following properties were considered for the analysis: E=29x106, ν=0.295,
plane-strain.
To compare the results of the implementation with those of Ref. [69], a normalized
biaxial stress parameter is defined as follows:
T πa
B = --------------
K1
A sequence of finite element solutions was obtained by p-extension (p=1 to 8) and

the results are shown in Figure 128. The estimated relative error in energy norm is
1.46% for p=8 (2995 DOF), and the values of K1 and the T-stress are: K1=3475.6
and T=5796.5. The corresponding biaxial stress parameter is:
5796.5 π × 1.0
B = -------------------------------------- = 2.956
3475.6
The value reported in Ref. [69] is B=2.951, a difference of less than 0.2%.

(a)
(b)
FIGURE 128 Double cantilever beam specimen. 2D results: (a) Estimated

relative error in energy norm, (b) Convergence of K1, K2, and T-stress.

DoubleCantileverBeam2D.sci.
The problem was also solved as a 3D elasticity problem with symmetry con-
straint imposed on element faces with normals in the direction of the positive
or negative z-axis in order to simulate plane-strain conditions. The same 2D
finite element mesh was used changing the reference to Extrusion. The prob-
lem was solved by p-extension and the results are shown in Figure 129. For
p=8 (13388 DOF), the global estimated error measured in energy norm is
1.41%, K1=3475.6 and T=5784, from where B=2.950.

DoubleCantileverBeam3D.sci.
Finally, the same problem was analyzed again in 3D without enforcing plane-
strain conditions. The objective is to assess the influence of the free surface
condition in the computation of the T-stress. Taking advantage of symmetry,
only 1/4 of the problem was discretized to reduce computational time. The 66-

(a)
(b)
(c)
FIGURE 129 Double cantilever beam specimen. 3D results: (a) Estimated relative error in energ
norm, (b) Convergence of K1, K2, and T-stress, (c) T-stress along crack front.
element mesh was refined in two directions: (1) two layers of elements graded in
geometric progression towards the singular edge with a common factor of 0.15, and
(2) two layers of elements graded in geometric progression towards the free surface
of the beam. The problem was solved by uniform p-extension and the results are
shown in Figure 130.
The T-stress is fairly constant through half the thickness of the beam, with the min-
imum of T=5832 at the symmetry plane and a maximum of T=6143 near the free
surface.

(a)
(b)
Symmetry
Free surface
FIGURE 130 Double Cantilever beam specimen. 3D results for the case of free
surface condition. (a) Global error estimate, (b) T-stress along crack front.

DCB_3D_Refined.sci.

Computing the J-integral
Computing the J-integral

StressCheck computes the J-integral along a circular path around the crack tip as
described under the heading Fracture mechanics in the Theoretical Background
chapter. The current implementation is applicable for two-dimensional problems 3
only (plane-stress, plane-strain and axisymmetric).
The J-integral can be computed for linear and elastic-plastic solutions of isotropic
materials and for linear solutions of orthotropic/laminate materials. For elastic-
plastic solutions, the integration path should be selected in such a way that it does
not cut through the plastic zone around the crack tip. The procedures for computing
J-integral for elastic and elastic-plastic solutions will be illustrated with an example
in the following.
Model problem: Panel with multi-site damage
Problem Description
Consider a 2024-T3 aluminum alloy panel with six equally spaced holes and a
crack emanating from each hole (FIGURE 131). The cracks are of different sizes
and the panel is subjected to a remote tensile load. The panel was described in
Ref. [43], and it corresponds to a multi-site damage (MSD) test coupon for constant
amplitude crack growth testing. Numerical and experimental crack growth data are
available for the panel.
The objective of the analysis is to compute the stress intensity factors and the J-
integral for each crack near the middle of the crack propagation life (the panel
failed during test after about 13500 cycles). An elastic-plastic analysis is performed
using a Ramberg-Osgood description for the material and the results are compared
with the experimental values. The length of each crack (in mm, measured from the
edge of the hole) is the experimentally measured value after 6,000 loading cycles:
a1=4.35; a2=3.48; a3=3.70; a4=3.70; a5=3.26; a6=3.26.

Material 2024-T3:
Sy = 330 MPa
3 E = 71 GPa
υ = 0.33
Dimensions (mm):
e = 12.5
s = 15.0
D = 5.0
t = 2.5 (thickness)
W = 100.0
L = 300.0
Loading:
σ = 96.5 MPa
FIGURE 131 Panel configuration for the MSD problem.
Discretization
In designing the finite element mesh and selecting the other discretization
parameters, the smoothness of the exact solution of the mathematical model
representing the physical system should be considered. In this case the follow-
ing modeling aspects are relevant:
• Given the thickness and loading of the panel, a two-dimensional (plane
stress) analysis is considered appropriate for this problem.

• Due to symmetry conditions, only half of the panel needs to be analyzed.

• Elastic-plastic material properties are used to assess the influence of plas-
ticity on the fracture mechanics parameters.
• The solution of the mathematical model just described is smooth inside the
domain and on the boundaries, except at the crack tips where the stresses 3
are unbounded in the linear case. For the material nonlinear case, the
stresses are bounded, but the strains are not. The design of the finite ele-
ment mesh should account for the nature of the exact solution. Geometri-
cally graded meshes toward the singular points are known to be optimal in
conjunction with p-extension for problems of this type. However, the
design of geometrically graded meshes for domains with multiple cracks
emanating from holes with sizes comparable to the hole diameters, is very
cumbersome and in many cases impractical. For that reason, the use of very
simple meshes combined with the product space is used.
The finite element mesh consisting of 38 quadrilateral elements, is shown in FIG-

URE 132. Note that no mesh refinement around the crack tips have been used. Due
to symmetry, only half the panel was discretized. The material was specified as
Ramberg-Osgood with the following properties: E=71,000 MPa, υ=0.33, S70E=320
MPa, n=12. A uniform traction of magnitude 96.5 MPa was specified at the top
boundary of the panel, and symmetry constraints were specified on each uncracked
ligament along the symmetry line. The leftmost node was fixed in the x-direction to
prevent rigid body translation (see detail in FIGURE 132). This problem can be
loaded from the Tutorial Handbook (file: Panel.sci).
Product space was assigned to all the elements in the mesh and the linear solution
was computed for polynomial degree ranging from 1 to 8, downward-p. Four mate-
rial nonlinear solutions were computed in order to obtain an assessment of the dis-
cretization error for the nonlinear analysis. The linear runs corresponding to runs 1,
2, 3 and 4 were used to start the nonlinear iterations. The number of iterations
required to complete any given analysis was based on a specified tolerance of
0.25% in energy. For additional details about performing a nonlinear analysis, refer
to “Executing material nonlinear analyses” .
Results
FIGURE 133 shows the estimated relative error in energy norm corresponding to 8
linear solutions.

Detail
FIGURE 132 Mesh and boundary conditions for the MSD problem.
The fracture mechanics parameters can be computed for the available sequence
of linear and nonlinear solutions. FIGURE 134 shows the convergence data
obtained for the linear and nonlinear fracture mechanics parameters for Crack
1. To compute K1, select the Fracture tab from the Results window and com-
plete the input area as follows:
• Fracture tab > Select > Node > SIF > Solution: SOL, Run: 1 to 8 >
Radius: 2.0. With the mouse cursor click on the node at crack tip 1.
To compute the J-integral for the linear and nonlinear solutions simply select
the corresponding solution records and option “J-integral” from the Fracture
input combo box and click on the same node again. Note that the size of the

FIGURE 133 MSD panel: Estimated relative error in

energy norm for the linear solution.
integration path was increased for the nonlinear computation. The following set-
tings of the input area were used to obtain the results shown in FIGURE 134 for the
linear and nonlinear J-integrals, respectively:
• Fracture tab > Select > Node > J-Integral > Solution: SOL, Run: 1 to 8 >
• Fracture tab > Select > Node > J-Integral > Solution: NLSOL, Run: 1 to 4 >
Table 4 summarizes the values of the fracture mechanics parameters for each crack.
They include the stress intensity factor (KI) and the J-integral computed from the
linear (Je) and nonlinear (Jp) solutions. The relative difference between the elastic
and plastic values of J are also included. All values included in the table correspond
to the linear or nonlinear solutions for p=8.
These results clearly indicate that the stress intensity factors are not substantially
affected by crack tip plasticity, which is in agreement with the experimental obser-
vation: Only towards the end of the fatigue life, the actual crack propagation rate is
larger than that predicted by using the linear stress intensity factors.
FIGURE 135 shows the extent of the plastic zone at each crack tip obtained from
the nonlinear run corresponding to p=8. The plastic zone is the set of all points for
which the total equivalent strain is greater than or equal to the uniaxial yield strain
ε ≥ εY, where εY=Sy/E=330/71000=0.0046. Note however, that for a Ramberg-

K1 from Linear solution
J from Linear solution
J from Nonlinear solution
FIGURE 134 MSD panel: Fracture mechanics parameters for Crack 1.

Osgood material the yield strength is not at the end of the linear region in the stress-
strain diagram.
FIGURE 135 Plastic zone sizes at each crack tip.
This example demonstrates that advanced analysis methods for the computation of
stress intensity factors and the J-integral under conditions of small scale plasticity
makes it possible for practicing engineers to assess damage in structural compo-
nents under elastic-plastic conditions with guaranteed reliability.

Crack Path Analysis
TABLE 4. Fracture mechanics parameters for each crack tip.
Value of the parameter for crack #

3
Parameter 1 2 3 4 5 6
KI [MPa mm 1/2
] 450 448 448 444 430 418
Je [N mm-1] 2.83 2.81 2.81 2.76 2.59 2.44
Jp [N mm-1] 2.93 2.95 2.94 2.89 2.68 2.52
(Jp-Je)/Je [%] 3.5 5.0 4.6 4.7 3.5 3.3
Crack Path Analysis

Crack Path Analysis is limited to use with 2D planar domains and requires that
a set of model construction rules be carefully followed. In contrast to other
analysis modules in StressCheck, Crack Path Analysis does not require that the
user construct a finite element mesh in advance. Rather, the analysis procedure
will automatically mesh the domain during the solution process. The following
guidelines will help you create a proper geometric representation of your 2D
problem domain:
1. The domain analyzed must be a single trimmed sheet body. There
may be additional unattached curves that may be used to locate crack
objects, but no other trimmed sheet bodies or surfaces other than the
one to be analyzed.
2. At least one crack object must be defined that lies within the prob-
lem domain. The crack object may be defined by any available
method, and may or may not have an initial length and direction. Crack
objects may not be attached to points, curves or surfaces that are con-
tained within the sheet body.
3. When defining a crack object, you may assign an initial crack length
of zero. If you do, the orientation of the crack will be determined auto-
matically after the first solution is computed, and an initial crack added
automatically during the second iteration. The orientation of the crack

Crack Path Analysis
will be determined normal to the direction of the first principal stress at the
given crack location.
4. When defining a crack object, you may use the automatic method
(AUTO), to attach the crack to a given surface or curve at an arbitrary off-
set. In this case, the initial location of the crack will be determined after the 3
first solution iteration by computing the maximum first principal stress on
the curve/surface that the crack was attached to, and locating the initial
crack at the computed location. The orientation will be determined as
described in item (3) above.
5. You may define an automesh record (method=AUTO) that will control
the global meshing behavior during the crack path analysis. If you do not
supply this record, it will be created automatically during the analysis using
default values. If you do supply an AUTO record, you must be sure not to
supply a transition value (TRANS). Supplying a transition value will cause
the local refinement around the crack tips to be removed during the mesh-
ing process.
6. All thickness, material, boundary condition, and p-discretization assign-
ments must be described as ALL records, or must be assigned using
“Region” sets. This is because the geometric and mesh objects are changing
during the solution process, and therefore will not be the same ones that
were defined at the time the attributes were assigned.
7. If you need to define node constraints or rigid body constraints, these
constraints must be attached to point objects that are not imbedded in the
sheet body representing the domain, rather than to node objects. This is
because nodes are being dynamically created during the analysis. The
points will be used only to identify locations in the model where nodes will
eventually appear, so that they may be selected automatically at the appro-
priate time during the analysis. Therefore, be sure to create these reference
points in exactly the same location as other points that are associated with
the sheet body. Define these points parametrically if necessary. The points
may not be attached to the sheet body, but may be attached to other geomet-
ric objects that are not associated with the problem domain.
8. In general, it is not advisable to have any sheet bodies (surfaces) defined,
other than the problem domain. The only exception to this rule is if you
want to define a crack object using the AUTO method, and attach the crack
to an independent surface in order for the crack location to be determined
by the location of the maximum first principal stress within that area of the
surface it is attached to.

Crack Path Analysis
After defining the model domain, locating the crack objects, and assigning all
attributes, you are ready to initiate the crack path analysis. Go to the Solver
interface and select the Crack Path tab. In this interface you will see a list of all
crack objects that are currently defined in the model, with the default status set
to not active. If you want the crack to actually be included in the analysis, you
3 must check the Active box. Next supply the crack increment to be added at
each end of each crack after each iteration. One or both values may be zero.
When a crack has an initial length of zero and if the crack object lies on an
external boundary, you should supply a non-zero crack increment for both tips.
This is because the direction of growth can only be determined plus or minus
180 degrees. If only one tip is assigned a non-zero length, this tip may immedi-
ately grow outside the problem domain. After the first iteration, the crack will
grow only inside the domain. If the crack is initially internal, it is not possible
to predict which tip will be A and which will be B when starting from a zero
length crack, so assigning only one tip increment may or may not give the
desired result.
A few other guidelines are important:

1. Since the fracture mechanics extraction methods are only valid for
isotropic materials, and in regions of constant thickness, there may be
situations in which K1 and K2 cannot be computed. In this case, the
crack may stop growing.
2. When a crack gets close to an external boundary, such that the
default contour radius used to compute K1 and K2 goes outside the
problem domain, the radius will automatically be reduced in an effort
to complete the computation successfully. Similarly, if the crack
reaches boundary in the model where there is a material or thickness
transition, the radius will also be reduced automatically in order to stay
within a region of constant material and thickness. When the crack is
incremented, it will then cross the boundary so that the crack can con-
tinue to grow. If the crack tip location moves outside the domain, the
crack will stop growing from that tip.
3. Choosing a crack increment that is too large will result in a crack
path that is wavy. To obtain a smoother crack path, choose a smaller
crack increment.
4. When there are multiple cracks in the domain, all active cracks with
non-zero crack tip increments will grow simultaneously. As each crack
reaches its individual maximum crack length, it will stop growing.
When all cracks have reached their maximum, or when the run limit is
reached, the analysis will stop.

Crack Path Analysis
5. If you wish to update crack increment values or make new cracks active
during the analysis, you may either run in STEPWISE mode, or set the run
limit to a small number of step. Return to the Crack Path tab and change
values as desired, then continue the analysis using the RESTART option.
6. Choosing INITIALIZE will reset the problem to the original configura- 3
tion and start the analysis from the beginning.
NOTE: It is not advisable to interrupt a crack path analysis using the ABORT fea-
ture (Esc key), unless you intend to INITIALIZE prior to the next analysis.
Although you can RESTART after an abort, the results obtained from the analysis
will be INCORRECT. This is because the abort request may have come at a time
after the mesh was regenerated, but before a new solution is computed and there-
fore the mesh may not match the solution information maintained in the program.
Under these circumstances, all extraction operations would be invalid. StressCheck
will display a warning message anytime you attempt to post-process a crack path
solution that was aborted.
NOTE: It is possible to force a crack path analysis with fixed crack orientation. To
do that, define the paramenter _fix_crack_angle and the initial value of the crack
orientation will be used during the analysis.
Model problem 1: External crack

Consider the case of a clamped rectangular panel in shear. The panel is 7 inches by
16 inches with a unit thickness. The material is isotropic, the modulus of elasticity
is E=30x106 psi and Poisson’s ratio is v=0.25, plane-strain. A 3.5 inch edge crack is
located as shown in FIGURE 136. This problem can be loaded from the 2D-Frac-
ture Handbook, file ShearPanelCrackPath.sci.
The steps to create the solution domain are as follows. Make sure the analysis type
is set to Planar Elasticity and the units selector is set to in/lbf/sec/F before proceed-
ing.
• Geometry tab > Action: Create > Object: Plane > Method: Locate > Data
tab > Input on > X: 0.0, Y: 0.0, Z: 0.0, Width: 7.0, Height: 16.0; P1-Min: -
0.5, P1-Max: 0.5, P2-Min: -0.5, P2-Max: 0.5, Rot-X: 0.0, Rot-Y: 0.0, Rot-
Z: 0.0 > Accept
• Mesh tab > Create > Crack > Locate > Input on > X: -3.5 > Y: 0.0 > Z: 0.0
> Length: 3.5 > Angle: 0.0 > Accept
Accept

Crack Path Analysis
Tx=1.0
16.0
3.5
7.0
FIGURE 136 Panel in shear. US units.
• Material tab > Define tab > Linear > Selection > Define tab > ID: mat
> Material : Linear > Units: US > Type: Isotropic > Case: Plane Strain
> E: 3.0e+07 > v: 0.25 > Accept
mat > Accept
• Sets tab > Select > Any Curve > Locate > ID: botedge > X: 0.0, Y: -
8.0, Z: 0.0, X-Width: 0.1, Y-Height: 0.1, Z-Depth: 0.1 > Accept. This
creates a set that will be used for the assignment of constraints.
• Sets tab > Select > Any Curve > Locate > ID: topedge > X: 0.0, Y: 8.0,
Z: 0.0, X-Width: 0.1, Y-Height: 0.1, Z-Depth: 0.1 > Accept. This cre-
ates a set that will be used for the assignment of the load.
• Load tab > Select > Any Curve > Traction > ID: load > Set: topedge >
Direction: XY > System: Global > X: 1.0 > Accept
• Constraint tab > Select > Any Curve > Built-In > ID: const > Set: bot-
edge > Accept

Crack Path Analysis
• Solution ID tab > Define > Name > Selection > Solution ID: sol > Con-
straint ID: const > Load ID: load > Accept
Discretization: Variable > Level: 4 > Accept
After defining the model domain, locating the crack object, and assigning all
3
attributes, you are ready to initiate the crack path analysis. Go to the Solver inter-
face and select the Crack Path tab. In this interface you will see the record of the
crack object with the default status set to not active. If you want the crack to be
included in the analysis, you must check the Active box. Next supply the crack
increment (0.0 for a-tip and 0.2 for b-tip) to be added at the end of the crack after
each iteration and all other information as shown in FIGURE 137 (initial direction:
0.0, initial length: 3.5, max length=7.0, p-level=4, Run limit=15).
FIGURE 137 Solver interface, crack path tab. External crack.
tion:
• SOLVE! tab > Initialize > Automatic > Iterative > Click on the Solve but-
ton.
The top portion of the box displays the status of the execution. The message bar
located on the main window displays which operation is being performed for each
run. Once the fifteenth run is completed, a summary of the crack path will be dis-
played in the Edit window (FIGURE 138). To perform post-processing operations
select the View Results icon from the Main Toolbar. To display the deformed shape

Crack Path Analysis
FIGURE 138 Crack path summary for panel in shear.
as shown in FIGURE 139, select the Plot tab and complete the required infor-
mation as follows:
> Shape: Deform > Click on the Plot button.
Model problem 2: Internal crack

Consider the rectangular panel shown in FIGURE 140 loaded by a constant
traction Ty=30000 at one end and fixed at the other. The panel is of unit thick-
ness, a length to width ratio of 1/4, the modulus of elasticity is 16x106 and
Poisson’s ratio v=0.29. The initial internal crack has a length of 0.56 inches
forming an angle of 30 degrees from the vertical. This problem can be loaded
from the 2D-Fracture Handbook, file TensionPanelCrackPath.sci.
Before creating the solution domain, define two parameters to facilitate the
change in length and orientation of the crack. Select the Model Info icon from
the Main Toolbar and when the dialog box appears select the Parameters tab.
Enter the following information:
• Name: ac > Description: Total crack length > Value: 0.56 > Accept

Crack Path Analysis
FIGURE 139
Deformed shape of
cracked panel in
shear.
• Name: Ang > Description: Crack orientation from vertical > Value: 30 >
Accept
Next, provide the geometric description of the panel by selecting the Creacte Model
icon and completing the information as follows:
Z: 0.0 > Accept
• Mesh tab > Create > Crack > Locate > Input on > X: -ac/2*cos(rad(90-
Ang)) > Y: -ac/2*sin(rad(90-Ang)) > Z: 0.0 > Length: ac > Angle: 90-Ang
> Accept. The starting position of a crack object is at one of the crack tips.

Crack Path Analysis
Ty=30000
30o
12.0
0.56
3.0
FIGURE 140 Panel in tension. US units.
Accept
• Material tab > Define tab > Define > Linear > Selection > ID: mat >
Material : Linear > Units: US > Type: Isotropic > Case: Plane Stress >
E: 1.6e+07 > v: 0.29 > Accept
mat > Accept
• Sets tab > Select > Any Curve > Locate > ID: botedge > X: 0.0, Y: -

Crack Path Analysis
• Sets tab > Select > Any Curve > Locate > ID: topedge > X: 0.0, Y: 6.0, Z:
0.0, X-Width: 0.5, Y-Height: 0.5, Z-Depth: 0.5 > Accept. This creates a set
that will be used for the assignment of the load.
• Load tab > Select > Any Curve > Traction > ID: load > Set: topedge >
Direction: Norm/Tan > System: Global > Normal: 30000 > Accept 3
• Constraint tab > Select > Any Curve > Built-In > ID: const > Set: botedge >
Accept
Discretization: Variable > Level: 4 > Accept
face and select the Crack Path tab. Check the Active box in order for the crack to be
included in the analysis. Supply the crack increment to be added at the ends of the
crack after each iteration and all other pertinent information as shown in FIGURE
141.
FIGURE 141 Solver interface, crack path tab. Internal crack.
tion:

Crack Path Analysis
• SOLVE! tab > Initialize > Automatic > Iterative > Click on the Solve
button.
The top portion of the box displays the status of the execution. Once the crack
length is reached, a summary of the crack path will be displayed in the Edit
3 window (FIGURE 142). To perform post-processing operations select the
FIGURE 142 Crack path summary for panel in tension.
View Results icon from the Main Toolbar. To display the deformed shape as
shown in FIGURE 143, select the Plot tab and complete the required informa-
tion as follows:
Model problem 3: No initial crack

Consider a clamped rectangular plate perforated with three circular holes, and
loaded by normal traction. The dimensions are as shown in FIGURE 144. This
problem can be loaded from the 2D-Fracture Handbook, file
Plate3HolesPath.sci.
Steps to create the solution domain:

Crack Path Analysis
FIGURE 143 Deformed

shape of cracked panel
in tension.
Z: 0.0 > Accept
• Geometry tab > Create > Circle > Local > Radius: 0.5. Click on the
plane’s system.
• Geometry tab > Create > Circle > Locate > X: -0.9, Y: -0.9, Z: 0.0,
Radius: 0.2 > Click Accept.
• Geometry tab > Create > Circle > Locate > X: 0.9, Y: 0.9, Z: 0.0,
Radius: 0.2 > Click Accept.
• Geometry tab > Create > Body > Bool-Subtract. > Click on the plane and
then on the three circles. > Accept
• Geometry tab > Create > Circle > Locate > X: 0.0, Y: 0.0, Z: 0.0,
Radius: 0.5 > Accept. This circle will be used for attaching the crack.

Crack Path Analysis
Crack location and direction are

selected by StressCheck based on the
location and direction of the maximum
principal stress around the center hole.
3 0.2
0.5
0.9
4.0
0.2 0.9
0.9 0.9 T=1000
10.0
FIGURE 144 Perforated plate.
• Mesh tab > Create > Crack > Auto. Leave all input fields off, turn off
Display Surfaces, and with the mouse select the last circle created. A
symbol representing the crack will be shown on the screen. If you do
not see a crack symbol, click on the Display Objects button and enable
the display of cracks.
Accept
• Material tab > Define > Linear > Selection > Define tab > ID: 7075-T6
> Material : Linear > Units: US > Type: Isotropic > Case: Plane Stress
> E: 1.05e+07 > v: 0.3 > Accept
7075-T6 > Accept
• Sets tab > Select > Any Curve > Locate > ID: fixededge > X: -5.0, Y:
• Sets tab > Select > Any Curve > Locate > ID: loadedge > X: 5.0, Y:
creates a set that will be used for the assignment of the load.
• Load tab > Select > Any Curve > Traction > ID: load > Set: loadedge >
Direction: Norm/Tan > System: Global > Normal: 1000 > Accept

Crack Path Analysis
• Constraint tab > Select > Any Curve > Built-In > ID: const > Set: fixededge
> Accept
• p-Discretization > Select > All Elements > Selection > Space: Trunk > p- 3
Discretization: fixed > Level: 4 > Accept
face and select the Crack Path tab. Complete the information as shown below:
• Crack Path tab > Crack #: 1 > Active switch on > a Tip Incr: 0.1 > b Tip
Incr: 0.1 > Init. Dir: 0.0 > Init. Length: 0 > Max. Length: 1 > p-level: 4 >
Run limit: 10
tion:
• SOLVE! tab > Initialize > Automatic > Iterative > Click on the Solve but-
ton.
The top portion of the box displays the status of the execution. Once the crack
length is reached, a summary of the crack path will be displayed in the Edit window
(FIGURE 145). To perform post-processing operations select the View Results icon
FIGURE 145 Crack path summary for perforated plate.

Eigenpairs and generalized intensity factors
from the Main Toolbar. To display the deformed shape as shown in FIGURE
146, select the Plot tab and complete the required information as follows:
FIGURE 146 Deformed shape of perforated plate.

The concept of stress intensity factors associated with crack tips in linear elas-
tic fracture mechanics is extended to general singular points for composites,
anisotropic materials, and multi-material interfaces for two- and three-dimen-
sional linear elastostatic and steady state heat transfer problems.
In the neighborhood of singular points, the exact solution of two-dimensional

elastostatic problems for example, can be expanded in the form:
∞ αi
{u} = ∑ Ai r { Φi ( θ ) } (12)
i=1

where {u} is the displacement vector with components ux(x,y), uy(x,y), r and θ are
polar coordinates centered on the singular point; αi are called eigenvalues and φi(θ)
are called eigenfunctions. These eigenpairs (αi, φi) depend on the material proper-
ties, the geometry and the boundary conditions [φi(θ) are smooth vector functions].
The Ai are coefficients which depend on the loading. Because of their close analogy
to stress intensity factors in linear elastic fracture mechanics, Ai are called general- 3
ized stress intensity factors (GSIFs). In the case of heat transfer problems they are
called generalized flux intensity factors (GFIFs).
The objective of the implementation is to provide computable data for establishing

reliable quantitative failure initiation criteria in electronic packages, laminated
composites and adhesively bonded joints. The implementation is based on the
methods presented in references [44], [45] and [46] for the computation of eigen-
pairs and the corresponding generalized stress/flux intensity factors (GSIFs/GFIFs)
for any multi-material interface problem involving isotropic or anisotropic materi-
als, subject either to mechanical or thermal loading, in a two-dimensional setting.
The implementation for edge singularities in three-dimensions is based on the
methods described in references [56], [57], [59].
The key to successful failure analysis is to compute reliably both the eigenpairs and
the GSIFs. The eigenvalues characterize the strength of the singularity, the eigen-
functions characterize the straining modes, and their amplitudes (the GSIFs/GFIFs)
quantify the amount of energy residing in particular straining modes.
A general method for computing the solution in the vicinity of any singular point in
2D and singular edges in 3D is available in StressCheck, which first determines the
eigenpairs, followed by the computation of the GSIFs/GFIFs. In two-dimensions,
the modified Steklov method is employed for computing the eigenpairs, which are
subsequently used in conjunction with the complementary energy principle for
computing the generalized flux and stress intensity factors from the finite element
solution. A similar procedure is used in three-dimensions. See Eigenpairs and gen-
eralized intensity factors in the Theretical Background chapter for additional
details of the implementation.
Computation of eigenpairs
The steps necessary to compute only the eigenpairs (αi, φi) can be summarized as
follows:
• The problem is loaded from an existing StressCheck input file, or it is cre-
ated inside the program using the existing pre-processing tools. The mesh-

ing rules for problems with singularities in heat transfer are the same as
described in the previous section for problems with cracks in isotropic
materials.
• Perform a linear analysis for a low polynomial order (typically, p=1).
After the execution is completed, select the Fracture tab from the
3 Results window and the Fracture input area shown in FIGURE 110
will be displayed.
• To compute the eigenvalues and eigenvectors, select the number of
eigenpairs to be computed in the “# terms” region of the input area and
then select the singular point/edge by pointing to it with the cursor,
clicking the left button of the mouse and dragging the mouse cursor. A
circle will appear which will grow in size as the cursor is dragged.
Once the circle is of the desired size, click the left button of the mouse
to perform the computation. The results are displayed in tabular and/or
graphical format and are independent on the size of the circle.
If one of the eigenvalues is a complex number, then the conjugate of the com-
plex eigenvalue is also an eigenvalue.
Examples of the computation of eigenpairs (αi, φi)
The computation of eigenvalues and corresponding eigenvectors in two- and

three-dimensions is illustrated by solving four model problems: Two heat
transfer problems and two elasticity problems.
2D-Heat transfer Consider the problem shown in FIGURE 148, corresponding to a circular
domain of unit radius with a crack along the positive x-axis. This problem can
be loaded from the Tutorial Handbook (file: GFIF1.sci). The governing differ-
ential equation is:
2
∇u = 0
where u(x,y) is the temperature field. One face of the crack (Γ1) has zero tem-
perature boundary condition and the other face (Γ2) is flux free. The outside
boundary (ΓR) has an imposed flux. The disc is of unit thickness, and the coef-
ficient of thermal conduction (k) is also unity. The exact solution for this prob-
lem is available in Ref. [47] and the values of the first three eigenvalues are:
α1=0.25; α2=0.75; α3=1.25.

y
ΓR
Boundary Conditions:
r=1
u = 0 on Γ1 r 3
∂u/∂θ = 0 on Γ2 θ Γ1
∂u/∂r = y on ΓR Γ2 x
FIGURE 148 2D-heat transfer problem: Clamped-free crack.
To create the finite element mesh for this problem, follow the procedures described
in the User’s Guide and the guidelines for mesh design in the presence of cracks
given at the beginning of this chapter. FIGURE 149 shows the recommended mesh
design and the boundary conditions. Note that two layers of geometrically graded
elements have been used at the crack tip. The elements were graded with a common
factor of 0.15.
Once the mesh has been defined, assign unit thickness to all elements. Select the
Thickness tab from the StressCheck Input dialog window and complete the input
area as follows:
Make sure that the analysis type is set to Planar Heat Transfer before defining and
assigning the material properties. Since the material is isotropic, the coefficients of
thermal conduction are: kxx=kyy=1.0, kxy=0.0. Select the Material tab from the
StressCheck Input window, and when the Material input area appears on the screen,
complete as follows:
• Material tab > Define tab > Define > Linear > Selection > ID: mat1 > Mate-
rial: Linear > Units: US > kxx=1 > kxy=0 > kyy=1 > Accept
• Assign tab > Select > All Elements > Selection > ID: mat1 > Accept.
Two types of boundary conditions should be specified: temperature and flux. Select
the Flux tab and complete the Flux input area as follows:

1.0 0.15
3
0.0225
FIGURE 149 Mesh and boundary conditions for the 2D-heat transfer problem.
• Click on the Formulae icon and when the Formula dialog window
appears, supply the following information: Name: flx1 > System
Option: Cartesian > Formula: y > Accept.
• Flux tab > Select > Any Curve > Flux > ID: FLX > Value: flx1. Select
the outside circle of the disc and then click on the Accept button in the
StressCheck Input dialog window. The case of zero flux is a natural
boundary condition and hence does not have to be explicitly pre-
scribed.
Now select the Temperature tab in the StressCheck Input window and com-
plete the input area as follows:
• Temperature tab > Select > Edge > Temperature > ID: TEMP > Value:
0.0. With the mouse cursor select all element edges along the upper
side of the crack and then click on the Accept button. The flux and
temperature symbols are shown in FIGURE 149.
Note: Use the Shrink Element from the Display Option Toolbar to facilitate the
edge selection on one side of the crack only.

Identify the combination of flux and temperature boundary conditions with a

unique name, select the Solution ID tab and complete the information as follows:
• Solution ID tab > Define > Name > Selection > Solutions tab > Solution
ID: SOL > Temperature ID: TEMP > Flux ID: FLX > Click on the Accept
button. 3
To execute the linear analysis, select the Linear tab from the Solver dialog window
and complete the input area as follows:
• SOLVE tab > Execute: Initialize > Run Mode: Automatic > Method: Itera-
This will result in one solution at p-level=1. To extract the eigenvalues and the cor-
responding eigenvectors, select the Fracture tab from the Results window, and the
input area shown in FIGURE 150 will appear. Complete the following information:
• Fracture tab > Input tab > Select > Node > General FIF (General Flux
Intensity Factor) > Solution: SOL > Run: 1 to 1 > Radius: 0.6 > # terms: 3.
With the mouse cursor click on the crack tip node.
To display the eigenvalues and corresponding eigenfunctions, go to the Graph win-

dow and click on any of the tabs appearing between the table and the graph. Select-
ing the tab E-vec 1 will automatically change the graph to show the eigenvector
associated with the first eigenvalue. FIGURE 151 shows the first, second and third
eigenfunctions as displayed on the graph window. Note that the numerically com-
puted eigenvalues are identical to the exact values.
3D-Heat transfer Consider an anisotropic material whose boundary consists of a 90o reentrant corner
generated by two flux-free faces Γ1 and Γ2, which meet along an edge as shown in
FIGURE 152.
The cylindrical boundary of the domain (ΓR) is loaded by flux boundary condition
which corresponds to the first symmetric eigenfunction of the asymptotic expan-
sion of u(x,y,z) about the reentrant edge as given in Ref. [67]. The exact values of
the first and second eigenvalues are α1=2/3, and α2=4/3. This problem can be
loaded from the Tutorial Handbook (file: GFIF2.sci). To execute a linear analysis,
select the Linear tab from the Solver dialog window and complete the input area as
follows:

FIGURE 150 Fracture mechanics tab

for heat transfer.
This will result in one solution at p-level=1. To extract the eigenvalues and the
corresponding eigenvectors, select the Fracture tab from the Results window,
and the input area shown in FIGURE 150 will appear. Complete the following
information:
• Fracture tab > Input tab > Select > Edge > General FIF (General Flux
Intensity Factor) > Solution: SOL > Run: 1 to 1 > Radius: 0.6 > #
terms: 2. With the mouse cursor click at any location along the singular
edge.
The results will be shown in the Graph window, FIGURE 153. Selecting the
tab E-vec 1 will automatically show the eigenvector associated with the first
eigenvalue. In this case the eigenpairs are constant along the entire edge.
2D-Elasticity Consider the traction-free isotropic L-shaped domain shown in FIGURE 154.
The plane elastic body is loaded by the Mode 1 and Mode 2 stress components
obtained from the asymptotic expansion of the displacement field about the

First eigenpair
Third eigenpair
Second eigenpair FIGURE 151 Eigenfunctions for the 2D-

heat-transfer problem.
vertex as given in Ref. [3]. The exact values of the first two eigenvalues are
α1=0.54448374 and α2=0.90852919. This problem can be loaded from the Tutorial
Handbook (file: GSIF1.sci).
To create the finite element mesh for this problem, follow the procedures described
in the User’s Guide and the guidelines for mesh design in the presence of cracks
given at the beginning of this chapter. FIGURE 155 shows the 18-element mesh
designed for this problem, with two layers of geometrically graded elements
towards the reentrant corner. The elements were graded in geometric progression
towards the singular point with a common factor of 0.15. A local coordinate sys-
tem, located at the reentrant corner and rotated -135 degrees with respect to the pos-
itive global x-axis, was defined for the specification of the tractions (see FIGURE
154 and FIGURE 155).

y Material properties:
ΓR kxx=4.0
3
kyy=kzz=1.0
r
r=1 θ
Γ1 x
Γ2 Boundary Conditions:
u(0,0,0) = 0.0
∂u/∂θ = 0 on Γ1, Γ2
FIGURE 152 3D-heat transfer problem. Anisotropic reentrant corner.
1
y
x
1 1
FIGURE 154 2D-Elasticity problem: L-shape domain.
Once the mesh has been defined, assign unit thickness to all the elements.
Make sure that the analysis type is set to Planar Elasticity before defining and

FIGURE 153 Eigenvalues and eigenfunctions

for the 3D-heat transfer problem.
assigning the material properties. Select the Material tab from the StressCheck
Input window and when the input area appears on the screen, complete as follows:
• Material tab > Define tab > Define > Linear > Selection > ID: mat > Mate-
rial: Linear > Type: Isotropic > Case: Pl. Strain > E: 1.0 > v: 0.3 > Accept
• Assign tab > Select > All Elements > Selection > ID: mat > Accept.
To apply tractions, select the Load tab from the StressCheck Input dialog window

2
FIGURE 155 Mesh design for the 2D-elasticity problem.
• Complete the formula form as shown in FIGURE 156 for TX1 and
TY1..
• Load tab > Select > Any Curve > Traction > ID: TRAC > Direction:
XY > System: SYS1 > X:TX1 > Y: TY1. With the mouse cursor,
select the boundary identified with 1 in FIGURE 155 and then click on
the Accept button
Note that the formulas are defined with the System Option: Cylindrical and the
Angle Option: -pi to pi. The parameters MODE1 and MODE2 have been
defined in the Parameters interface of the Model Information dialog window.
Both parameters have the value of unity. The constants shown in FIGURE 156
are the same for all formulas.
The expressions for the other formulas are entered in a similar way. A total of 8
formulas are needed, two per boundary. Load the Tutorial Handbook problem
(file: GSIF1.sci) to find the definitions for all the formulas used in this prob-
lem. Refer to the User’s Guide on how to load a problem from the Handbook
Framework.

FIGURE 156 Formula form for TX1.
Now select the Constraint tab from the StressCheck Input window and specify rigid
body constraints by selecting any two nodes on the mesh. Since the body is in equi-
librium under the applied tractions, only rigid body constraints needs to be speci-
fied.
• Constraint tab > Select > Node > Rigid Body > ID: SPC. With the mouse
cursor select any 2 nodes on the mesh.
Next, define the solution ID’s as described in previous examples and execute a lin-
ear analysis for p=1 only. To extract the eigenvalues and the corresponding eigen-
vectors, select the Fracture tab from the Results window, and when the input area
appears, complete the following information:
• Fracture tab > Input tab > Select > Node > General SIF > Solution: SOL >
Run: 1 to 1 > Radius: 0.1 > # of terms: 2. With the mouse cursor click on
the node located at the reentrant corner.
The results are shown in FIGURE 157. The reported values for the eigenvalues
coincide with the exact values to 8 significant digits. Selecting the tab E-vec 1 the
graph will show the eigenfunctions associated with the first eigenvalue as shown in
FIGURE 157a. Two eigenfunctions (Ux, Uy) and three eigenstresses (Sx, Sy, Txy)
are included in the plot. The second set of eigenfunctions can be visualize by select-
ing the E-vec2 tab, FIGURE 157b.

(a) First set of eigenfunctions (b) Second set of eigenfunctions
FIGURE 157 Set of eigenfuctions for the 2D-elasticity problem.
3D-Elasticity Consider a composite body consisting of two dissimilar, isotropic, homoge-

neous, and elastic wedges that are perfectly bonded along the interfaces. The
body is loaded on the circular boundary by the stress field corresponding to the
exact solution of the asymptotic expansion about the singular point as given in
Ref. [68]. The normal displacement is set to zero in the two lateral faces. FIG-
URE 158 shows the description of the problem and the finite element mesh
consisting of 10 solid elements. The first three eigenvalues characterizing the
stress singularity along the edge are: λ1=0.5124722, λ2 = 0.6001175 and λ3 =
0.7309757.
This problem can be loaded from the Tutorial Handbook (file: GSIF2.sci). To
execute a linear analysis, select the Linear tab from the Solver dialog window
• SOLVE tab > Execute: Initialize > Run Mode: Automatic > Method:

E1=10000 y
ν=0.3
3
r
60o θ
x
r=1.0
E2=1
ν=0.3
FIGURE 158 3D-Elasticity problem: Isotropic inclusion.
This will result in one solution at p-level=1. To extract the eigenvalues and the cor-
responding eigenvectors, select the Fracture tab from the Results window, and the
input area shown in FIGURE 150 will appear. Complete the following information:
• Fracture tab > Input tab > Select > Edge > General SIF (General Stress
Intensity Factor) > Solution: SOL > Run: 1 to 1 > Radius: 0.2 > # terms: 4.
With the mouse cursor click at any location along the singular edge.
The results are shown in FIGURE 159 The reported values coincide with the exact
values to 7 significant digits. Selecting the tab E-vec 1 will automatically show the
eigenfunctions associated with the first eigenvalue. The eigenfunctions associated
with the first four eigenvalues are also shown in the figure.
Computation of GFIFs and GSIFs

StressCheck has an advanced method implemented for the computation of “gener-
alized stress intensity factors” (GSIFs) for linear elastostatic problems and “gener-
alized flux intensity factors” (GFIFs) for steady state heat transfer problems in two-

FIGURE 159 Eigenvalues and eigenfunctions for the 3D-elasticityt problem.
and three-dimensions. The method is superconvergent. This means that errors

in the intensity factors converge to zero much faster than the error in energy
norm, as the number of degrees of freedom is increased.
The GFIFs/GSIFs are computed as a post-solution operation on the finite ele-

ment solution vector. The method is applicable to isotropic materials, anisotro-
pic materials and multi-material interfaces under mechanical load, and for
cases where the singularities are characterized by complex eigenpairs. The
generalized stress intensity factors are the coefficients Ai of the asymptotic
expansion in equation (12) on page 178, which are computed once the eigen-
pairs are available. The case of thermal load is discussed in the next section.

The necessary steps to compute the GFIFs /GSIFs can be summarized as follows:
• The problem is entered into StressCheck from an existing neutral file or it is
created inside the program using the existing pre-processing tools.
• After executing the corresponding linear analysis (elasticity or heat trans-
fer, depending on the problem), select the Fracture tab from the Results 3
window.
• To extract the GFIFs/GSIFs, enter the radius of the extraction circle, the
number of terms to be computed in the “# of terms” region of the input
area, and then select the singular point/edge by pointing to it with the cursor
and clicking the left button of the mouse. The results are displayed in tabu-
lar and/or graphical format depending on the display selection in the dialog
box. The influence of the size of the extraction circle in the results is gener-
ally very small if proper meshing is utilized. The size of the circle should be
such that it runs outside the innermost layers of element around the singu-
larity.
Examples
In this section we consider the computation of the GFIFs and GSIFs for the same
four problems described in the previous section.
2D-Heat transfer Consider once again the circular domain of unit radius with a crack along the posi-
tive x-axis shown in FIGURE 149. The objective is to compute the first two gener-
alized flux intensity factors. The exact solution for this problem is available in
Ref. [46] and the values of the first two GFIFs are: Α1=1.358097, Α2=0.970087.
Using the same input data described before (Tutorial Handbook: GFIF1.sci), exe-
cute a sequence of finite element solutions from p=1 to 8. To execute the linear
analysis, click on the Compute Solution icon and when the Solver dialog window
appears complete the input area as follows:
This will result in a set of solutions for p-level=1 to 8. To extract the GFIFs, select
the Fracture tab from the Results window, and supply the following information:
• Fracture tab > Select > Node > General FIF (Flux Intensity Factor) > Solu-
tion: SOL > Run: 1 to 8 > Radius: 0.6 > # of terms: 2. With the mouse cur-
sor click on the crack tip node.

The results in tabular and graphical form are shown in FIGURE 160. Note that
FIGURE 160 GFIFs for the 2D-heat transfer problem.
a sequence of GFIFs are computed corresponding to each available finite ele-

ment solution. When computing intensity factors from a sequence of at least 3
solutions, StressCheck performs an estimation of the true value of the func-
tions by projecting the results to an infinite number of degrees of freedom. The
results of this projection is reported as the “Estimated Limit” together with the
percent deviation from the value corresponding to the solution with the highest
number of degrees of freedom in the graph window. The computed values of
the GFIFs for run #8 are very close to the exact values indicated above. The
extrapolated limit for the first GFIF is even closer to the corresponding exact
value.
3D-Heat transfer Consider once again the anisotropic reentrant corner shown in FIGURE 152.
The objective is to compute the first two generalized flux intensity factors. The
exact values of the first two GSIFs are: Α1=1.0, Α2=0.0.
Using the same input data described before (Tutorial Handbook: GFIF2.sci),
execute a sequence of finite element solutions from p=1 to 8. To execute the
linear analysis, click on the Compute Solution icon and when the Solver dialog
window appears complete the input area as follows:

This will result in a set of solutions for p-level=1 to 8. To extract the GFIFs, select
the Fracture tab from the Results window, and supply the following information:
• Fracture tab > Select > Edge > General FIF (Flux Intensity Factor) > Solu-
tion: SOL > Run: 1 to 8 > Radius: 0.6 > # of terms: 2. With the mouse cur-
sor click on the crack front. 3
The results will include convergence information as shown in FIGURE 161. Note
that GFIFs for p=8 are very close to the exact solution
FIGURE 161 GFIFs for the 3D-heat transfer problem.
2D-Elasticity Consider once again the traction-free L-shaped domain shown in FIGURE 155.
The exact values of the GSIFs for this problems are A1=A2=1.0.
Use the same input data described above (Tutorial Handbook: GSIF1.sci), but
change the value of the parameters MODE1 and MODE 2 to the following:
MODE1 = 0.312578; MODE2 = 0.281231
This change is necessary to account for the difference between the maximum value
of the exact eigenfunction and the computed numerical eigenfunction associated
with each eigenvalue. Execute a sequence of finite element solutions from p=1 to 8
by selecting the Linear tab from the Solver interface and completing the input area
as follows:
tive > Convergence: None > Solve

This will result in a set of solutions for p-level=1 to 8. To extract the GSIFs,
select the Fracture tab from the Results window, and complete the following
information:
• Fracture tab > Action: Select > Object: Node > Method: General SIF >
Solution: SOL > Run: 1 to 8 > Radius: 0.1 > # of terms: 2. With the
3 mouse cursor click on the node located at the reentrant corner.
As before, the results will be included in the graph window as shown in FIG-
URE 162. Again, the GSIFs computed for p=8 are very close to the exact val-
ues.
FIGURE 162 GSIFs for the 2D-elasticity problem.
3D-Elasticity Consider once again the isotropic inclusion problem shown in FIGURE 158.
Using the same input data described before (Tutorial Handbook: GSIF2.sci),
execute a sequence of finite element solutions from p=1 to 8. To execute the
linear analysis, click on the Compute Solution icon and when the Solver dialog
window appears complete the input area as follows:
Iterative > None > Click on the Solve button.
This will result in a set of solutions for p-level=1 to 8. To extract the GFIFs,
select the Fracture tab from the Results window, and supply the following
information:
• Fracture tab > Select > Edge > General SIF (Stress Intensity Factor) >
Solution: SOL > Run: 1 to 8 > Radius: 0.2 > # of terms: 4. With the
mouse cursor click on the edge of the material interface.

The results will be included in the graph window as shown in FIGURE 163.
FIGURE 163 GSIFs for the 3D-elasticity problem.
Computation of TSIFs
StressCheck has a numerical algorithm for the computation of the thermal stress
intensity factor (TSIF) for thermal loading in two-dimensions. The implementation
involves the following steps:
Step 1. Solve the thermal problem: Find the temperature distribution over the entire
domain given the flux and temperature boundary conditions as described in the
Conduction Heat Transfer chapter. Compute the smallest eigenvalue (β1) associated
with the flux singularity by the modified Steklov method.
Step 2. Solve the thermoelastic problem: Using the temperature distribution

obtained from the solution of the thermal problem, obtain the corresponding dis-
placement field of the elasticity problem. Refer to Conduction Heat Transfer (Anal-
ysis Guide) on how to execute an elasticity solution after a thermal analysis.
Step 3. Compute eigenvalues and SIFs: Extract the first two eigenvalues (α1, α2)
and the first two stress intensity factors (KI, KII) from the thermoelastic problem for
a given radius (R) of the integration circle. The program will automatically com-
pute the stress intensity factors for three values of the integration circle: 0.9R, R and
1.1R.
Step 4. Compute TSIFs: Using Richardson’s extrapolation, the program will project
the three values of each stress intensity factor computed in Step 3 to R=0. The error

in the computation of the stress intensity factors is of the order (R)q, where
q=β1-α1+1 for the first intensity factor (KI) and q=β1-α2+1 for the second
intensity factor (KII).
Example Central crack in a rectangular plate. Consider a rectangular isotropic plate

3 subjected to two different thermal loadings for which numerical results are
available in the literature. A plate of width 2W, length 2L and crack of length
2a=2.0 with L/W=1.0 and a/W=0.2 is solved for two thermal loadings repre-
senting modes I and II respectively (see FIGURE 164). Since the results are
A B
Mode I Loading
AB, BC, CD, DA: T=100
EF: T=0
E F 2L Mode II Loading
2a
AB: T=100
CD: T=-100
BC, DA, EF: qn=0
D C
2W
FIGURE 164 Plate with central crack. Notation.
independent of the thermal conductivity for isotropic materials, the heat con-
duction coefficients are taken as k11=k22=1.0, k12=0. The mechanical proper-
ties of the materials are: Modulus of Elasticity E=1.0; Poisson’s ratio ν=0.3;
coefficient of thermal expansion α=0.01. Plane-strain conditions are assumed.
Because of symmetry, only one half of the plate is analyzed. FIGURE 165
shows the finite element mesh used to solve the problem for Mode I and Mode
II loadings. Only one layer of geometrically graded elements towards the crack
tip was used.

3
Detail
crack
tip
FIGURE 165 Mesh for the cracked plate problem.
The finite element mesh with the boundary conditions for Mode I is available in the
Tutorial Handbook by selecting the file: TSIF1.sci. To execute the thermal analysis
make sure the Reference/Theory Selectors in the Main Menu Bar is Planar Heat
Transfer. Then select the Linear tab from the Solver interface and perform a
sequence of solutions fro p=1 to 8, upward. After the execution is completed, select
the Fracture tab from the Results window and compute the first eigenvalue (β1)
associated with the flux singularity. Complete the following information in the
Fracture input area:
• Fracture tab > Select > Node > General FIF > Solution: SOL_TH > Run: 1
to 8 > Radius: 0.3 > # of terms: 1. With the mouse cursor click on the crack
tip node. The result will be displayed in the graph window under the E-vec1
tab: E-value (β1) = 0.50.
Next, change the Planar Heat Transfer in the Reference and Theory Selectors to
Planar Elasticity and select the Linear tab from the Solver interface to perform an
Upward-p extension from p = 1 to 8. After the execution is completed, select the
Fracture tab from the Results window (FIGURE 166a) and compute the stress
intensity factors. Complete the following information in the Fracture input area:

• Fracture tab > Select > Node > General SIF > Solution: SOL_EL >
Run: 1 to 8 > Radius: 0.5 > # of terms: 2 > EV_th: 0.5. Note that the
eigenvalue associated with the flux singularity (0.5) is entered in the
EV_th field. With the mouse cursor click on the crack tip node.
3 A warning message will be issued that the computation is being performed by

extrapolation.
The program will automatically compute the stress intensity factors for three
values of the integration circle: 0.9R=0.45, R=0.5 and 1.1R=0.55, and perform
the extrapolation to R=0. The extrapolated values that correspond to the Ther-
mal Stress Intensity Factors (TSIFs) are included in the graph windows as
shown in FIGURE 166 for Mode I loading.
To obtain the solution for the Mode II loading, set the Reference and Theory
Selectors back to Planar Heat Transfer and update the boundary conditions to
those indicated in FIGURE 164. Rerun the heat transfer analysis and the elas-
ticity analysis and perform the post-processing operations indicated for mode I
loading. The extrapolated values of the TSIFs for mode II are shown in FIG-
URE 167.

FIGURE 166 Convergence of TSIFs for Model I loading.

FIGURE 167 Convergence of TSIFs for Model II loading.

4 Material Property
Fitting
4
This chapter describes the material property fitting capabilities of StressCheck. By property fitting, we
refer to the possibility of describing the orientation of the material 1-axis of an orthotropic or trans-
versely isotropic material as smooth polynomial functions. These polynomial functions are computed
using a least square fitting on data points provided by the user. This capability is available for planar and
3D elasticity.
Fiber fitting procedure

The elastic properties for a transversely isotropic or orthotropic material are known
in the material axes. By default, the material 1-direction (the fiber-direction) is
aligned with the x-direction of the global or a local coordinate system, and remains
constant within the elements. To have the material 1-direction defined as a variable
orientation within the mesh, tabular data must be provided with the values of the
direction cosines of the fiber angle with respect to the global coordinate system (x,
y, z). These discrete values of the fiber orientation are then fitted with smooth poly-
nomials over the entire domain using least square fitting.
For each data point (xi, yi, zi) there is an associated direction cosine vector ni(xi, yi,
zi), where ni={n1, n2, n3} and the superscript i refers to the ith data point. In order
to have a smooth and continuous representation of the fiber orientation within the
Advanced Topics Chapter 4: Material Property Fitting 203

Fiber fitting procedure
body, each direction component nj of n can be approximated by polynomials

over each element as follows:
∑b
(j)
n j ( x, y, z ) = k Φ k (x,y, z)
k=1
where {b} are N unknown coefficients multiplying the known polynomial

functions {Φ}. Next, we define the error between the data points and the fitted
values as:
4 NDP N
2
∑ ∑
(j)
e0 ( bk ) = n̂ j ( x l, y l, z l ) – b k Φ k (x l,y l, z l)
l=1 k=1
where NDP is the number of data points, and n̂ j ( x l, y l, z l ) is the given jth-direc-
tion cosine component at point l. Next, consider the derivatives of the approxi-
mating functions:
2 2 2
 ∂n j +  ∂n j +  ∂n j
e1 ( bk ) = P ∫∫ ∫ ∂x  ∂y  ∂z 
dx dy dz
Vol
where P is a penalty factor. This term is added to minimize the jumps in the
derivative across element boundaries. The vector of unknown coefficients {b}
is computed by minimizing the error in the least square sense:
∂
( e + e1 ) = 0 k = 1 ,2 ,…N
∂ bk 0
The minimum number of data points (NDP) per element to perform the fitting
depends on the polynomial order of the approximating functions and on the
element type. If p represents the polynomial order of the fitting functions, at
least 8p points per planar element (quadrilateral and triangular) and 40p points
per solid element (hexa, penta or tetra) should be used if the penalty P=0. The
use of a penalty factor can reduce the minimum number of points needed.
204 Chapter 4: Material Property Fitting Advanced Topics

Fitting in 2D
The fiber angle data must be entered in the Tables dialog window of the Input form
by providing the global coordinates of the point under the X, Y, Z fields, and the
direction cosines (n1, n2, n3) of the fiber angle with respect to the global x-axis,
under the Value fields. It is possible to automatically generate fiber angle data in
StressCheck along boundaries (curves or surfaces) or element edges in the direction
normal or tangent to them. Once the tabular data is available, the material coeffi-
cients must be entered together with the reference to the tabular data for the fiber
orientation. The user then selects the p-level and a penalty factor to perform the
fiber orientation fitting. After the fitting is performed, the fiber distribution is dis-
played on the screen for verification. The procedure will be illustrated with two
examples.
4
The fitting may be performed over the entire model or within one or more sub-
domains of the model. When a fitting is performed, it is necessary to identify which
elements will be included in the fitting, and at what polynomial order the fitting
will be done. In addition, a penalty factor may be supplied. If the penalty factor is
zero, StressCheck will automatically determine the minimum number of sample
points required to perform the fitting. If insufficient sample values are provided for
an individual element, that element will automatically be excluded from the fitting
procedure.
Fitting in 2D
Consider the 2D domain shown in FIGURE 168 consisting of a semicircular arc, a
semi-elliptical arc and two lines. The material 1-axis (the fiber direction) of a trans-
versely isotropic material has to remain tangent to the circular and elliptical bound-
aries and normal to the straight edges. Once the boundaries and elements have been
created as shown in FIGURE 168, create the fiber tabular data needed for the anal-
ysis by selecting the Create Model icon and then clicking on the Tables tab from the

Fitting in 2D
Input dialog window. This problem can be loaded from the Tutorial Handbook
(file: Fitting2D.sci).
Ellipse
Circle
3.0
2.0
4
1 Line 1 2 3 Line 2 4
5.0
FIGURE 168 Domain for the 2D example problem.
The input area should be set as shown in FIGURE 169. The Action, Object,
Method combo boxes are set to Select > Any Curve > Pos. Tangent. With the
mouse cursor select the ellipse and the circle shown in FIGURE 168. The tan-
gent refers to the direction of the fiber. In this case, the fibers run tangent
(counterclockwise is positive) to the boundaries. The type of table was selected
as Material, and the number of points to be generated per boundary was set to
100. Provide a table ID (in this case T1) and click on the Accept button. The
program will generate 100 equally spaced points along each boundary and
compute the direction cosines of the positive tangent for each point on the
boundary. The newly created records are displayed in the scroll window of the
Tables dialog window.
Next, select line 1 indicated in FIGURE 168 to create 20 records using the nor-
mal direction. Use the negative normal if the line was created from point 1 to 2
or the positive normal otherwise. The arrows representing the input fiber orien-
tation can be visualized by turning on the ID toggle switch shown in FIGURE
169. Use the DeLast button to delete the last set of data records if the arrows do
not point in the indicated direction. Repeat for line 2 using the corresponding

Fitting in 2D
normal. In this way the ‘flow’ of the fibers is from right to left (counterclockwise)
FIGURE 169 Tables input area.
Now select the Material tab to define and assign the material properties. Enter the
following information (FIGURE 170):
• Material tab > Define > Linear > Selection > Define tab > ID: MAT > Lin-
ear > Trans. Iso. > Units: US > Pl. Stress > E11=1000, E22=100, v12=0.3,
v23=0.3, G12=200.
To assign the material MAT to all elements in the mesh:

• Material tab > Assign Tab > Select > All Elements > Selection > ID: MAT
> Type: Fitting > Angle: T1 (Table name) > Accept.
Next, go to the Fitting tab and enter
• Select > All Elements > Selection > ID: MAT (any name can be used to
identify the fitting record) > Fit order: 2 > Penalty: 0.15 > Click on the
Accept button.
Note that the p-level of the fitting was set to 2 and the penalty to 0.15. The selection
of the p-level and penalty is problem dependent and different values should be

Fitting in 2D
FIGURE 170 Material input area for fiber fitting for the 2D example problem.
explored until the desired results are obtained. Refer to the explanation of
Thickness by Fitting in the User’s Guide for additional details.
To display the input fiber orientation turn on the toggle switch in front of mate-
rial ID shown in FIGURE 170. To perform the fitting of the fiber angle, simply
click on the Fitting button in the Material dialog window. After the fitting oper-
ation is completed, click on the Fit switch next to Scale to display the fitted
fiber orientation over the mesh. The results should look as shown in FIGURE
171. At the time of the analysis, the solver will utilize the fitted fiber orienta-

Fitting in 3D
tion to determine the material stiffness matrix at each integration point within an
element when computing the stiffness matrix.
4
FIGURE 171 Fitted fiber orientation for the 2D example problem.
Fitting in 3D
Consider the 3D domain shown in FIGURE 172 consisting of four planar sections
connected by a Spline surface. Each section lies in a plane parallel to the xy-plane
and consists of a semicircular arc, a semi-elliptical arc, and two lines. The circular
arcs for the two outer sections are of radius 4.0, and the two central section are of
radius 3.5. The corresponding semi-elliptical arcs are of radii r1=10.0, r2=7.0 for
the outer sections and of radii r1=9.0, r2=6.0 for the central sections, respectively.
Each section is separated 4.0 in the z-direction. The four hexahedral element mesh
is also shown in the figure. This problem can be loaded from the Tutorial Handbook
(file: TunelFiberSTD.sci).
The material 1-axis (the fiber direction) of a transversely isotropic material has to
remain tangent to both Splines in the u-direction of the surfaces. To create the fiber
tabular data needed for the analysis, select the Create Model icon from the Main
Menu Bar and then click on the Tables tab from the Input dialog window as
explained for the 2D example problem.
The Action, Object, Method combo boxes should be set to Select > Any Surface >
Pos. U. With the mouse cursor select the two Spline surfaces shown in FIGURE
172. In this case, the fibers run following the positive surface coordinate along the
P1-direction (refer to the Modeling Guide for the definition of the surface coordi-
nates for splines). The type of table was selected as Material, and the number of
points to be generated per boundary was set to 40. Provide a table ID (in this case

Fitting in 3D
Spline surface
geometry mesh
4
FIGURE 172 Geometry and mesh for the 3D example problem.
T1) and click on the Accept button. The program will generate a grid of 40 by
40 equally spaced points on each surface and compute the direction cosines for
each point. The newly created records are displayed in the scroll window of the
Tables dialog window. The arrows representing the input fiber orientation
should be as shown in FIGURE 173.
FIGURE 173 Input fiber data for the

3D example problem.
Now select the Material tab to define the material properties. Enter the follow-
ing information (FIGURE 170):

Fitting in 3D
• Material tab > Define tab > Define > Linear > Selection > ID: MAT > Lin-
ear > Trans. Iso. > Units: US > E11=1000, E22=100, v12=0.3, v23=0.3,
G12=200 >
To assign the material MAT to all the elements in the mesh:

• Material tab > Assign tab > Select > All Elements > Selection > ID: MAT >
Type: Fitting > Angle: T1 (table name) > Accept.
Next, go to Fitting tab and enter
• Select > All Elements > Selection > ID: MAT (any name can be used to
identify the fitting record) > Fit order: 1, Penalty: 0.1 > Click on the Accept
button. 4
Note that the p-level of the fitting was set to 1 and the penalty to 0.1. The selection
of the p-level and penalty is problem dependent and different values should be
explored until the desired results are obtained. Refer to the explanation of Thick-
ness by Fitting in the User’s Guide for additional details.
To display the input fiber orientation turn on the toggle switch in front of material
ID and make sure that Arrows is checked in the Display Materials pulldown menu
of the Attributes toolbar. To perform the fitting of the fiber angle, simply click on
the Fitting button in the material dialog window. After the fitting operation is com-
pleted, click on the Fitting switch next to Scale to display the fitted fiber orientation
over the mesh and then on the Display Reset button. The results should look as
In three-dimensions, the fitting operation can be very time consuming because

there is an inverse mapping operation necessary to find the standard coordinates of
the tabular data, that is the location of the points in the element natural coordinate
system. Once the data has been fitted, it is possible to save the information in stan-
dard coordinates so that additional fitting operations will not require the expensive
inverse mapping. To save the data in standard coordinates into a file, create the
parameter _dump_table (note the ‘_’ in front of the parameter name) before fitting
for the first time. This will instruct the program to write two output files in the
working directory with the names: table.tb1 and table.tb2. The first one contains all
the data points that were successfully inverse mapped and the second the ones that
fail the inverse mapping operation. To utilize the new data in standard coordinates,
replace the contents of all *802 records in the input file of the model (*.sci) with the
*802 records found in table.tb1. Both files are in ASCII format and therefore any
Windows editor (like Notepad) can be used. After replacing the original data in glo-
bal coordinates by the new data in standard coordinates, read the input file into

Fitting in 3D
FIGURE 174 Fitted fiber direction for the 3D example problem.
StressCheck and select Tools > Table Reset from the Main Menu before click-
ing on the Fitting button.
One important advantage of using the standard coordinates is that the direction
cosines are also converted to the element natural coordinates. If the size of the
element changes because the geometry to which it is attached is modified, then
the fiber direction will “follow” the element. After each topology modification,
remember to select Tools > Table Reset before performing the fitting.

5 Theoretical
Background
5
This chapter contains an outline of the theoretical basis of the procedures incorporated in StressCheck.
Frequently asked questions, such as: what are h-, p- and hp-extensions; when and why should one use
one or the other; what are discretization and modeling errors; how do we estimate and control these
errors, are answered. For additional information and details Ref. [3] is recommended.
Mathematical models and their approximate solutions

Let us look at the process by which engineers obtain quantitative information about
some real physical system. FIGURE 175 represents the main links in the chain
which connects a real physical object or system, such as a pressure vessel or a sus-
pension arm, with quantitative information generated by computers.
Our understanding of a real physical object or system is through the physical laws
which govern its response to external stimuli, such as pressure, temperature, accel-
eration, etc. Some laws are well known, others are not. As new physical laws are
discovered, our ability to describe a real system or process improves. Nevertheless,
at any given time a state-of-the-art error may be incurred, indicated in FIGURE 175
as ‘error A’. In structural and mechanical engineering practice errors of type A are
generally negligible. There are important exceptions, such as crash dynamics and
biomechanics, where errors of type A are dominant. We will be concerned in the
Advanced Topics Chapter 5: Theoretical Background 213

Quantitative
Information
Physical System State-of-the-Art Working Numerical

“Reality” Model Model Solution
Error Error Error

A B C
FIGURE 175 The main elements of mathematical models.
5
following discussion only with problems for which errors of type A are negli-
gible.
State-of-the-art models are generally very complicated. In order to make

progress possible, it is necessary to construct a simplified mathematical model,
called working model, which forces engineers to decide which phenomena are
important enough to be considered and which are not. In doing so, a modeling
error, also called “error of idealization”, indicated as ‘error B’, is incurred. The
working model is then solved numerically, using some approximate method,
most commonly the finite element method. The numerical solution introduces
approximation errors, shown as ‘error C’. Errors of type C are called “discreti-
zation errors”.
Control of errors of type C in engineering analyses was a focal point of aca-

demic research on the finite element method during the 1970’s and 1980’s. The
very substantial interest in adaptive methods among analysts is motivated by
the desire to control type C errors.
In general, large discrepancies between the predicted and actual responses of

real physical systems or objects are not caused by errors of type C, however.
Rather, they are caused by improper formulation of models, or by something
essential having been omitted from the model, or misrepresented in the model,
that is, by errors of type B.
214 Chapter 5: Theoretical Background Advanced Topics

Some examples of type B errors are: improper representation of constraint condi-

tions, especially in the case of structural connections; using plate models where
three-dimensional representation is required; representation of laminated plates by
models which do not represent the transverse variation of displacements properly;
failing to account for nonlinear effects, etc.
The commonly used phrase: “finite element model” is generally understood to

mean that the solution of a mathematical problem, representing some real physical
system or object, is to be approximated by the finite element method. The mathe-
matical problem is called the mathematical model and its exact solution is denoted
by u EX . The finite element solution, denoted by u FE , is an approximation to u EX .
The usefulness of information generated by the means of “finite element models”
depends on: how well the data of interest (such as stress, displacement, natural fre-
quency, etc.) corresponding to u EX , denoted by Φ i ( u EX ) i=1,2,..., represent their
real physical counterparts, and how well the data of interest computed from the
finite element solution Φ i ( u FE ) i=1,2,..., approximate Φi ( u EX ) .
The errors between Φ i ( u EX ) and their real physical counterparts are the combina- 5
tion of errors of types A and B indicated in FIGURE 175. We will assume in the
following that type A errors are negligible and will refer to these errors as the mod-
eling errors, or the errors of idealization. The errors Φ i ( u FE ) – Φi ( u EX ) , i=1,2,...
are called the errors of discretization, or approximation errors.
The sizes of modeling errors depend on the degree by which the assumptions incor-
porated into a simplified mathematical model are justified in the particular case
under consideration. For example, the linear theory of elasticity incorporates the
assumptions that the strain values are much smaller than unity; the displacements
are so small that the equilibrium equations can be written for the original configura-
tion rather than the deformed configuration; the stresses are proportional to the
strains, etc. Additional modeling assumptions are often introduced by modifying
the solution domain (e.g., omitting fillets), imposing zero displacement conditions
on boundaries which is impossible to achieve in practice, etc.
The validity of assumptions incorporated in the model can be tested once a solution
is available. Since u EX is generally not known, u FE can be used for testing the
assumptions, provided that u FE is reasonably close to u EX . The simplified mathe-
matical model is acceptable if the data of interest Φ i ( u EX ) i=1,2,... are substantially
independent of the modeling assumptions.
StressCheck was designed to provide means for controlling both type B and type C
errors. For example, it introduces methods by which modeling errors for structural
plates can be effectively controlled, including modeling errors associated with lam-

Control of discretization errors
inated plates. StressCheck is the first finite element software product to pro-
vide hierarchic modeling capabilities for structural plates. The essential
property of hierarchic plate models is that the exact solutions corresponding to
the hierarchic sequence of models converge to the exact solution of the fully
three-dimensional problem. StressCheck is also the first finite element soft-
ware product to provide for consideration of the effects of material and geo-
metric nonlinearities, within the context of the p-version of the finite element
method and convenient means for computing the J-integral.

In the finite element method the domain of interest is subdivided into subdo-
mains (finite elements) over which the unknown quantities (displacements, for
example) are represented by polynomial functions (elemental shape functions).
5 The errors of discretization depend on the choice of the finite element mesh,
the polynomial degree of elements, and the mapping functions used. Associ-
ated with a particular choice of mesh (∆), polynomial degree of elements (p)
and mappings (Q) are a family of functions S=S(∆, p, Q). After enforcing the
displacement boundary conditions on S we have a subset of S which we denote
by S̃ . The number of linearly independent functions in S̃ is called the number
of degrees of freedom (N). The finite element solution is that function in S̃
which minimizes the potential energy Π ( u ) . Thus:
π ( u FE ) = min ( Π ( u ) )
u ∈ S̃
If we enlarge S̃ either by mesh refinement, or increase of the polynomial

degree of elements, or both, then the finite element solution converges to the
exact solution in the following sense:
lim Π ( u FE ) = Π ( u EX ) .
N→∞
Since u EX is independent of the discretization (i.e., the mesh ∆ , the polyno-

mial degree p and the mapping Q), u FE cannot be close to uEX unless u FE is
substantially independent of ∆ , p and Q. The error associated with a finite ele-

ment solution is generally measured by the difference between the exact value of
the potential energy and the approximate value:
e = Π ( u FE ) – Π ( u EX ) .
One of the reasons for taking the square root is that the potential energy is propor-
tional to the square of loading. This is called the energy norm measure of the error.
Denoting the strain energy of a displacement field u by U(u),
e = U ( u EX – u FE ) ≡ u EX – u FE E .
This error measure has a good engineering sense: It is closely related to the root-
mean-square error in stresses. For proof we refer to Ref. [3].
From these expressions it is seen that the discretization error is reduced by increas-
ing the number of degrees of freedom N, which is controlled by the polynomial 5
degree of the approximating elemental shape functions and by the number of ele-
ments in the mesh. Orderly sequences of discretization by mesh refinement,
increase of the polynomial degree of elements, or a combination of both, are
respectively called h-, p-, hp-extensions.
h-extension The size of a finite element is the diameter of the smallest circle (or sphere) that
contains the element. This diameter is denoted by h and the diameter of the largest
finite element in the mesh is denoted by hmax. h-Extension involves letting
hmax→0. In the conventional h-version of the finite element method, the shape
functions are constructed from polynomials of low degree: linear (p=1), or qua-
dratic (p=2), and the error of discretization is controlled by mesh refinement.
p-extension Alternatively, we can hold the number of elements constant and increase the poly-
nomial degree of elements. The polynomial degree of elements is a vector p, the
size of which is the number of elements in the mesh. p-Extension involves letting
the smallest polynomial degree pmin→∞. In the p-version of the finite element
method a hierarchic sequence of finite element spaces S1 ⊂ S2 ⊂ S3... is created.
The corresponding finite element solutions, u FE 1, u FE 2, u FE 3 ... have the property
that the potential energies are monotonically decreasing:
Π ( uFE1 ) > Π ( uFE2 ) > Π ( u FE3 ) ...
In StressCheck the polynomial order can be varied from 1 to 8.

hp-extension In hp-extensions mesh refinement is combined with an increase of the polyno-

mial degree of elements. In these processes the number of equations that has to
be solved, the number of degrees of freedom, is progressively increased, hence
the name ‘extension’. Note that h- and p-extensions can be viewed as special
cases of hp-extension, which is the general discretization strategy of the finite
element method.
Choice of an extension process

An important question is: which is the most efficient method of extension with
respect to reducing discretization errors. This question can be answered on the
basis of a simple classification of the exact solution of the problem one wishes
to solve. The exact solutions have been classified into three main categories:
Category A: uEX is analytic. A function is analytic in a point if it can be

expanded into a Taylor series about that point on the entire solution domain,
including the boundaries of the solution domain. Alternatively, the domain can
5 be divided into subdomains and uEX is analytic on each subdomain, including
the boundary of each subdomain. The finite element mesh is so constructed
that the boundaries of the subdomains are coincident with element boundaries.
For example, consider a 45-degree sector of a pipe, loaded by an internal pres-

sure p. The pipe is made of two materials, as shown in FIGURE 176. The
mathematical problem is to solve the problem of elasticity on the domain
ABCD, with symmetry boundary conditions imposed on AB and CD, constant
negative normal stress and zero shear stress on AD and zero normal and shear
stress on BC. This problem is in Category A, provided that the material inter-
face is an element boundary.
Category B: uEX is analytic on the entire domain, including the boundaries,

with the exception of a finite number of points, (in three dimensions uEX needs
not be analytic along a finite number of lines). The mesh is so constructed that
the points where uEX is not analytic are nodal points (in three dimensions the
lines where uEX is not analytic are coincident with element edges). An exam-
ple of problems in Category B is a cracked elastic body. The crack tip (or edge)
is a singular point (or line).
Category C: Problems for which the exact solution is neither in Category A

nor in Category B are in Category C.

C material 2
D
material 1
A B
FIGURE 176 Example of a problem in Category A.
Most problems in mechanical and structural design belong in either Category A or

B. Most problems in fluid dynamics belong in Category C. StressCheck was
designed for solving problems in Categories A and B very effectively. 5
In the neighborhood of singular points the exact solution of two-dimensional prob-
lems is typically of the form:
∑A r
λi
u EX = i ϕ i ( θ ) + smooth
smooth terms
– terms
i=1
where r and θ are polar coordinates centered on the singular point; λi > 0 and ϕ i ( θ )
are eigenpairs which depend on the mathematical model and the boundary condi-
tions ( ϕ i ( θ ) are smooth vector functions); Ai are coefficients which depend on the
loading. Because of their close analogy to stress intensity factors in linear elastic
fracture mechanics, Ai are called generalized stress intensity factors. The minimum
value of λi, denoted by λ, is a convenient measure of smoothness of uEX in the
neighborhood of a singular point.
Knowing the classification of a problem, and the value of λ, it is possible to com-

pare h- p- and hp-extensions on the basis of rate of convergence, that is the rate of
change of the error measured in energy norm with respect to the number of degrees
of freedom. The rate of convergence is either algebraic:
C-
u EX – uFE E ≤ --------
2β
N

or exponential:
C
uEX – uFE E ≤ --------------------------
θ
-
exp ( γN )
where C represents some constant which is independent of N. The symbols β,

γ, θ denote positive constants. The relationship among type of extension, prob-
lem class, and asymptotic rates of convergence, are summarized in Table 5. For
h-extensions, category B, uniform or quasiuniform mesh refinement is
assumed. For optimal meshes β=p/2, independently of λ.
TABLE 5. Asymptotic rates of convergence in two dimensions for h-, p- and hp-extensions
by problem class.
Type of Extension
5 Category h p hp
A Algebraic Exponential Exponential
β = p/2 θ ≥ 1/2 θ ≥ 1/2
B Algebraic Algebraic Exponential
β = 1/2 min(p,λ) β=λ θ ≥ 1/3

C Algebraic Algebraic ____________
β>0 β>0
For models in Category B the discretization errors are most efficiently con-
trolled by designing a mesh such that when the polynomial degree of elements
is uniformly increased then the discretization error decreases exponentially
until the required degree of accuracy has been reached. This usually involves
grading the mesh in geometric progression in the vicinity of singular points.
Criteria for the design of discretization for problems in this category will be
discussed later.
The importance of controlling discretization errors

Intrinsic control of discretization errors by finite element programs is very
important for properly correlating computed information with experimentally

observed data and for using computed data in engineering decisions which cannot
be measured directly, and/or exploring a range of loadings and constraint condi-
tions, not covered by the experiment. This point is discussed in the following.
Let Φ EXP be some experimental information, for example a measured displace-

ment or strain, Φ MOD the same information predicted by a mathematical model,
for example the equations of elasticity together with the appropriate material prop-
erties, boundary conditions and loading data; Φ FEA the same information pre-
dicted by a finite element approximation of the mathematical model. Φ MOD is
generally not known, nevertheless the purpose of an experiment is to determine
whether the mathematical model correctly describes the physical system being
modeled, that is, whether Φ MOD is sufficiently close to Φ EXP . Writing:
ΦEXP – Φ MOD ≡ Φ EXP – Φ FEA + Φ FEA – Φ MOD
≤ Φ EXP – Φ FEA + Φ FEA – Φ MOD

5
it is seen that if the quality of a mathematical model is to be assured then it is neces-
sary to ensure that Φ FEA – Φ MOD is not larger, and preferably much smaller,
than Φ EXP – Φ FEA , which is known from measurement and computation. The
quality control procedures incorporated in StressCheck make it possible to ascer-
tain that Φ FEA – Φ MOD is smaller than Φ EXP – Φ FEA .
Unfortunately, there are many examples in industrial practice where close correla-
tion with experimental results is achieved through skillful manipulation of the dis-
cretization parameters. The sensitivity of the approximate solution to discretization
parameters is a strong indication that the numerical error is large, or the model is
improperly defined.
In the absence of intrinsic procedures designed for controlling approximation

errors, it is possible to correlate computed data with experimental observations
through near cancellation of two large errors:
Φ EXP – Φ FEA ≡ Φ EXP – Φ MOD + Φ MOD – Φ FEA ≈ 0















−
+M −
+M
where M is some big number. Such models are unreliable and can be very mislead-
ing when the model is used for drawing general conclusions from experimental
information.

Control of modeling errors

The control of modeling errors is based on two important considerations: First,
each working model must be so formulated that it has a non-trivial exact solu-
tion. More precisely, it is necessary to have 0 < u EX < ∞ where • rep-
resents the natural norm of the formulation. StressCheck is based on the
displacement formulation, the natural norm of which is the energy norm. Fur-
thermore, the data of interest corresponding to u EX must be finite. Second, the
data of interest corresponding to u EX must be substantially independent of the
assumptions incorporated into the working model. In current modeling practice
these fundamental rules are often violated. Some examples are given in the fol-
lowing section.
Common errors in formulating working models

The most common modeling errors are caused by intuitively plausible but con-
5 ceptually wrong formulation of working models and extracting data from finite
element solutions the limiting value of which is not finite. Typical cases are
illustrated by simple examples in the following.
Example 1 Improper working model, u EX E

= 0.
Consider a rectangular plate of constant thickness constrained by four fasteners

or spot welds F1, F2, F3, F4, shown in FIGURE 177. Assume that the sides
D C
• F4 • F1
• F3 • F2
A B
FIGURE 177 Rectangular plate constrained by four fasteners.
AB, CD, DA are stress free and a small uniform normal displacement δ is
imposed on side BC. Of interest is to determine the fastener forces correspond-

ing to this displacement. The plate is of an elastic material and of uniform thick-
ness. All dimensions are given.
It would be intuitively plausible to argue that the diameters of the fasteners are
small in relation to the plate and hence we should be able to represent this plate rea-
sonably well by a working model based on the two-dimensional theory of elasticity
with the displacement components in the center of each fastener hole set to zero.
This argument is conceptually wrong, however, because a simple analysis will
show that in this case u EX E = 0 and the force necessary to impose the displace-
ment is zero. Of course, any finite element solution u FE corresponding to a partic-
ular mesh ∆ and polynomial degree of elements p would have non-zero energy and
non-zero force. However if we increased the number of degrees of freedom either
by mesh refinement or by increasing p, we would have:
lim u FE = 0
N→∞ E
and the force would converge to zero. It would be seen that increasingly large
strains are concentrated in a neighborhood of the point constraints the size of which 5
is decreasing as N → ∞ . In the limit the strains will be infinity on an area which is
zero.
In general, point constraints and multi-point constraints lead to improper model

definitions. The only exception is when the body is in equilibrium and point con-
straints are used to prevent rigid body displacements and rotations only.
Example 2 Improper working model, u EX E

does not exist.
Consider now the same problem as in Example 1 but impose uniform normal trac-
tion on side BC. Simple analysis will show that in this case u EX E does not exist.
For any finite element mesh ∆ and polynomial degree of elements p, u FE E
would be a finite number and the normal displacement along side BC would be
finite. However, as we increase the number of degrees of freedom N by mesh
refinement or by increasing p, we would have:
lim u FE = ∞
N→∞ E
and the normal displacement along side BC would converge to infinity.
Proper formulation requires that each fastener be represented by constraints

imposed along the perimeter of the fastener holes. Characterization of the con-
straints is a modeling decision.

Example 3 Proper working model improperly used.
Consider now a rigid frame subjected to some loading such as the load shown
in FIGURE 178. The displacement components along AB and EF are zero. For
C D
A B E F
FIGURE 178 One-bay frame.

5
the sake of simplicity, assume uniform thickness. The exact solution of a work-
ing model based on the two-dimensional theory of elasticity will have finite
strain energy. The model is properly formulated if the data of interest are dis-
placements, or stress resultants (bending moments, axial forces and shear
forces) but it would not make sense to try to compute the maximum stress. The
stresses corresponding to the exact solution will be infinitely large in points A,
B, C, D, E and F. Of course, for any finite element mesh and polynomial
degree of elements the stresses computed from a finite element solution in
these points are finite, however as the number of degrees of freedom (N) is
increased, either by mesh refinement or by increasing p, the absolute value of
the stresses will increase and converge to infinity. The maximum stress would
be strongly discretization dependent.
Hierarchic sequences of working models

Working models must be so constructed that the data of interest corresponding
to the exact solution are substantially independent of restrictive assumptions
incorporated in the model. Consider a rectangular plate constrained by fasten-
ers, such as the plate shown in FIGURE 177. Assume that the plate is subjected
to compressive and shear loading. Consider now four working models:
Model 1: Plane elasticity, the fasteners represented by elastic springs.

Model 2: Plane elasticity, the fasteners represented by nonlinear springs, the

properties of which are determined by coupon tests. Note that Model 1 is a
special case of Model 2.
Model 3: Effects of bending and stability are considered in addition to the in-
plane forces, as well as the nonlinearity of fastened connections. Note that
Model 1 and Model 2 are special cases of Model 3.
Model 4: Same as Model 3, but local failure (elastic-plastic) of the panel is

considered also. The first three models are special cases of Model 4.
The answer to the question: “Which of these models should be used?”, depends on
the range of loading and the desired accuracy. Model 1 generally does not represent
reality well, but is much easier to use than the other models. Model 2 is generally
accurate until buckling or local failure occurs. Model 3 is quite sophisticated (and
expensive), and Model 4 involves a choice of failure theories, thereby introducing
another set of modeling problems and requiring information which is not readily
available. 5
The essential point is that a single model cannot provide sufficient information for
deciding whether the model itself is adequate or not. A particular mathematical
model must be viewed as a member of a hierarchic sequence of models. The proper
choice of a model from the hierarchic sequence is such that the data of interest are
independent, up to the desired level of accuracy, from restrictive assumptions incor-
porated into the model.
Treatment of uncertainties
In many practical problems it is necessary to take into consideration uncertainties in
the information available about the physical system modeled. In some cases uncer-
tainties are negligible, in other cases they may be the dominant consideration.
There may be substantial uncertainties in material properties, loading, constraints,
and the initial state of stress. StressCheck was designed so as to allow for efficient
evaluation of the effects of uncertainties on the data of interest.
Uncertainties in loading are best handled through multiple load cases, providing
bounds on the data of interest for a range of loading conditions. StressCheck has
provisions for handling multiple load cases.

Hierarchic models for structural plates
Uncertainties in constraint conditions are best handled through the use of

spring constraints, with the spring coefficients representing the elastic compli-
ance of connected parts. StressCheck has provisions for spring boundary con-
ditions.
Stress distributions are generally not sensitive to small variations in elastic

properties. Local yielding, on the other hand, can alter stress distributions very
substantially. StressCheck provides capabilities for evaluating the effects of
local yielding.

In analyzing structural plates generally two types of information are sought:
The structural response: Load-deflection relationships, shear forces, bending

5 moments, buckling loads, natural frequencies, etc., are of interest. For lami-
nated plates, the relevant characteristic is that composite materials typically
have very large bending modulus to shear modulus ratios.
The strength response: Detailed stress distribution near external boundaries

and at interfaces, the conditions under which delamination occurs, crack prop-
agation problems, etc. For laminated plates the relevant characteristic is that at
the laminar interfaces the transverse normal and shear stresses are continuous,
but the strains are discontinuous.
Usually structural and strength models are required in different phases of

design. Many writers refer to alternative mathematical representations of plates
and shells as theories. Thus, in the literature one encounters references to
membrane theory, Kirchhoff theory, Reissner-Mindlin theory, etc. It is better
and more descriptive to use the word ‘model’ however, since one wishes to
model the mechanical response of various solid objects, one dimension of
which happens to be much smaller than the other two, by mathematical means.
The quality of a plate or shell model must be judged on the basis of how well
its exact solution approximates the corresponding three-dimensional problem.
Of course, the exact solution depends not only on the choice of the model but
also on the topology, material properties, loading and constraints. The desired
degree of approximation depends on the analyst’s goals in performing the anal-
ysis and cannot be determined reliably a priori. For these reasons models have

to be chosen adaptively. The hierarchic system of plate models incorporated in

StressCheck was developed for this purpose.
The basic requirement of a simplified two-dimensional approximation is that it

should include all essential features that characterize the three-dimensional nature
of the plate problem. In the case of laminated composites for example, the essential
features include variable material properties through the thickness (generally,
piecewise constant properties), substantial influence of transverse shear deforma-
tion (normally neglected for homogeneous materials), and discontinuity of the
transverse shear strains at the interfaces between laminae.
Shear deformation effects are of critical importance in the analysis of laminated

composite plates and shells. For thick laminated plates, or in the presence of local
discontinuities, such as holes, reinforcements, etc., and at the boundaries, the trans-
verse components of stress and strain have a substantial influence on its strength.
At the boundaries “boundary layer effects” occur, that is the stress distribution is
substantially different from the stress distribution in the interior.
5
Conventional models for plates
Most of the available methods of analysis for multilayered anisotropic plates and
shells are extensions of the methods originally developed for homogeneous isotro-
pic plates. First-order and higher-order shear-deformation models were developed
to account for the effects of transverse shear strains. The terminology used in con-
nection with high-order models refers to the level of truncation of terms in a power
series expansion for the displacements.
First-order shear-deformation models, also known as generalized Reissner-Mindlin

models (Ref. [7], [8], [9]), are simple but adequate only for predicting the gross
response characteristics of the laminate for large length-to-thickness ratios. They
yield poor strength response approximations for thick plates and near boundaries.
For homogeneous plates they provide excellent gross response and reasonable
strength response approximations. Shear locking generally develops in h-version
finite elements based on this model. The use of lower-order interpolation functions
causes the shear stiffness to dominate when the element thickness is reduced, and
hence shear locking occurs. The performance of these models is dependent on the
factors used for adjusting the transverse shear stiffness. The main limitation of the
first-order shear-deformation model for analyzing laminated composites is that the
transverse strain components are described by continuous functions across inter-
faces between dissimilar materials, and therefore the transverse shear stresses are
discontinuous at the interfaces between laminae. This violates the action-reaction

principle, according to which the normal and tangential stress components

must be continuous.
To overcome the limitations of the first-order shear-deformation model,

higher-order models that involve higher order terms of the transverse displace-
ments were developed. A significant amount of research has been conducted in
this field. See, for example, Ref. [10] and [11]. Higher-order models give more
accurate results than first-order models at the expense of greater computational
cost and more difficult implementation. The main limitation of these models is
that they use the incorrect assumption that displacements do not account for
possible discontinuities in the derivatives of the displacements at the interfaces
of adjoining laminae. This limitation is more evident for thick laminates, or in
the presence of complex loading and geometric discontinuities.
The discrete-layer models were derived to overcome the limitation of shear-

deformation models (Refs. [12], [13], [14]). They are based on assuming a dis-
placement field which allows piecewise linear variation of the in-plane dis-
placements through the thickness and constant transverse displacement
5
through the thickness. They give better results than shear-deformation models,
and yield more accurate interlaminar stress distribution, even for very thick
laminates. However, since the normal strain is neglected, the interlaminar
stress distribution near discontinuities is not represented accurately by these
models. Another disadvantage is that the complexity of the model increases
with the number of layers in the laminate, making this approach impractical for
large problems.
Several survey papers are available which describe in detail the advantages and
limitations of the various models for homogeneous and laminated plates and
shells (Refs. [15], [16], [17], [18]). None of these approaches address boundary
layer effects which are most critical for laminated plates, however.
Remark The problem of shear locking. Conventional finite element software products
typically offer plate and shell models based on the Reissner-Mindlin theory.
However, these models exhibit shear locking. Shear locking occurs because the
solution of the Reissner-Mindlin model converges (as it should) to the solution
of the classical Kirchhoff model as the thickness of the plate decreases. The
solution of the Kirchhoff model has a higher level of continuity than the Reiss-
ner-Mindlin model. Such higher continuity can be well approximated only
with elements based on polynomials of degree higher than four. To overcome
this difficulty various techniques, such as reduced integration, mixed formula-
tions, etc. have been employed. Unfortunately, in many cases these techniques

lack a theoretical basis and hence the character of the exact solution and its relation-
ship to the underlying problem of elasticity is unknown. The hierarchic models
incorporated into StressCheck are not sensitive to thickness variations. Neverthe-
less, for plates of small thickness, use of the product space is recommended.
Hierarchic models for plates

The formulation of hierarchic models for structural plates was developed in the
period 1985 to 1992. StressCheck is the first software product to make this technol-
ogy available in professional practice.
Hierarchic models combine the advantages of both, the shear-deformation models

and the discrete-layer models. They allow for discontinuities in the slopes of the
deflections at interfaces, and the number of degrees of freedom do not depend on
the number of layers in the laminate. Unlike any other model, they also allow the
construction of a sequence of models to satisfy the equilibrium equations to the
desired degree of accuracy. In the limit the solutions corresponding to a hierarchic 5
sequence of models converge to the fully three-dimensional solution. Depending on
the goals of computation, the analyst can select the model that best fits the goals.
Choosing progressively higher models, the computational effort increases, but of
course, the accuracy in the results is improved also. When only the structural
response is required, low-order models are generally sufficient.
The construction of hierarchic models for homogeneous isotropic plates and shells
was discussed by Szabo and Sahrmann in 1988 (Refs.[19], [20]). The optimality
conditions for the construction of the hierarchic models for homogeneous plates
was first discussed by Schwab in 1989 (Ref. [22]). The extensions of these concepts
for laminated composites was first addressed in Refs. [22] and [23] for plates in
cylindrical bending, and in Refs. [24] and [25] for general plates with mid-plane
symmetry.
The goals of the computation may require different approximations in different

regions of the domain. This can only be accomplished with a hierarchic sequence of
models. StressCheck is currently supporting hierarchic models for the analysis of
homogeneous and laminated plates in bending. The laminae must be symmetrically
distributed with respect to the mid-surface. The hierarchic models implemented in
StressCheck are suitable for obtaining both the structural and detailed stress
response of laminated composites without remeshing.
As noted earlier, structural plates and shells behave very differently in the smooth
interior regions, where representation by models of low order is possible, from the

boundaries, where the exact solution of low order models often differs by a
substantial margin from the exact solution of the corresponding fully three-
dimensional problem. Of course, there is a transition region as well. Reliability
in numerical simulation can be achieved at a reasonable cost only if the model
is properly selected for each subregion of the domain. Model convergence tests
are implemented in StressCheck to make it possible to evaluate the level of
approximation to the three-dimensional problem.
Implementation
In the following it is assumed that the middle surface of the plate lies in the x-y
plane. The two-dimensional domain occupied by the planform of the plate is
denoted by ω and the plate thickness is denoted by h. The three-dimensional
domain is denoted by Ω :
Ω = { x, y, z x, y ∈ ω, –-----
2
h h
- < z < --- }
2
5
The formulation is based on the principle of virtual work which is stated as fol-
lows:
Find u EX ∈ Ẽ ( ω ) such that
F( v)
( 3D )
B ( u EX , v ) = for all v ∈ Ẽ o ( Ω )
( 3D )
where u EX is the exact solution in terms of the displacement vector compo-
nents, v is a virtual displacement, B ( u (EX
3D )
, v ) is the virtual work of the inter-
nal stresses and F(v) is the virtual work of external forces. Ẽ ( Ω ) and
Ẽ ( Ω ) are function spaces, defined as follows. First define E(Ω) as the set of
o
functions which have finite strain energy:
E(Ω ) = { u B ( u, u ) < ∞ } .
Ẽ ( Ω ) , the space of admissible functions, that is, the set of all functions in
E(Ω)which satisfy the kinematic boundary conditions. Ẽ o ( Ω ) is the set of
functions in Ẽ ( Ω ) which vanish on those boundary segment where kinematic

boundary conditions are prescribed. We associate the energy norm with the space
E(Ω). By definition, the energy norm is:
def
u E(Ω) = B ( u, u ) .
Consider a sequence of subspaces of E(Ω), to be denoted by En(Ω):
 n 
 1

E (Ω) = u
n

u
x
( x, y, z ) = ∑
j=1
u
x j
( x, y ) φ j ( z ), 

 
n
2
u
y
( x, y, z ) = ∑
j=1
u
y j
( x, y ) ψ j ( z ),
5
n 
3

u
z
( x, y, z ) = ∑
j=1
u
z j
( x, y ) ρ j ( z ) 


(13)
where φ j ( z ) , ψ j ( z ) , ρ j ( z ) are given functions which represent the transverse

variation of the displacement components. The subspaces are hierarchic, that is,
E ( Ω ) ⊂ E ( Ω ) ⊂ .... ⊂ E ( Ω ). The sum n f = n 1 + n 2 + n 3 is called the
1 2 n
number of fields. Different ni are possible for each displacement component and in
different regions of the domain.
( HM i )
The exact solution corresponding to the ith subspace Ei(Ω) is denoted by u EX .
A hierarchic sequence of models is properly formulated if it satisfies the condition
that the corresponding exact solutions converge to the exact solution of the fully
3D-problem for a fixed plate thickness:
( 3D ) ( HM i )
lim u EX – u EX
E(Ω)
= 0.
i →∞
In addition, they should posses two highly desirable features:

1. The exact solution of each model converges to the same limit as the exact
solution of the corresponding 3D-problem with respect to the plate thickness
(h) approaching zero:
( 3D ) ( HM i )
u EX – u EX E(Ω)
lim ------------------------------------------------------
( 3D )
= 0, i=1, 2,...
h→0 u EX E ( Ω )
( 3D )
This requirement is important because, typically, u EX in the interior regions
of the domain behaves as if h were close to zero, assuming that the loading is
smooth.
( 3D )
2. Optimality of convergence: When the exact solution u EX is sufficiently
smooth:
5 ( 3D ) ( HM i )
u EX – u EX E( Ω) γi
------------------------------------------------------
( 3D ) ≈ Ch
u EX E ( Ω )
where C is a constant, independent of i; γ i , the rate of convergence, is a con-

stant which depends on i, and γ i + 1 > γ i .
The various plate and shell models differ in the way the transverse variation of
the displacement components is represented. Once the transverse variation of
displacement components and the number of fields are determined, the prob-
lem is to find the solution for the in-plane components of the displacement
field. The resulting two-dimensional problem is solved by the p-version of the
finite element method.
The procedure for the selection of the transverse shape functions in (13) clearly
differentiates the approach implemented in StressCheck from other displace-
ment formulations for plates. φ j ( z ) , ψ j ( z ) , ρ j ( z ) are derived on the basis of
the degree to which the equilibrium equations of the three-dimensional elastic-
ity problem are satisfied (Refs. [24] and [25]). In the case of homogeneous
plates, these functions are simply the terms of the power series expansion in
the transverse variable z. For the laminated plates, they are piecewise continu-
ous functions in z which depend on the lamina material, the ply layup and the
stacking sequence.

Six models have been implemented in StressCheck for the bending of homoge-
neous plates and five models for the bending of symmetric laminates (no membrane
terms are included). They are characterized by the number of fields nf and by the
transverse shape functions as follows:
Model 1. Reissner-Mindlin model characterized by n1=n2=n3=1 → nf =3.
u x ( x, y, z ) = u x 1 ( x, y ) φ 1 ( z )
u y ( x, y, z ) = u y 1 ( x, y ) ψ 1 ( z )
u z ( x, y, z ) = u z 1 ( x, y ) ρ 1 ( z )
where the transverse shape functions for homogeneous and laminated plates are the
5
same: φ 1 ( z ) = ψ 1 ( z ) = z and ρ 1 ( z ) = 1 . Plane-stress constitutive relations
are used for this model.
Model 2. Characterized by n1=n2= 1, n3=2 → nf =3.
u x ( x, y, z ) = u x 1 ( x, y ) φ 1 ( z )
u y ( x, y, z ) = u y 1 ( x, y ) ψ 1 ( z )
u z ( x, y, z ) = u z 1 ( x, y ) ρ 1 ( z ) + u z 2 ( x, y ) ρ 2 ( z )
where ρ 2 ( z ) = 2
z for homogeneous plates, and
 2z + ( Q 13 + Q 36 ) φ 2 ( z ) + ( Q 23 + Q 36 ) ψ 2 ( z )
ρ
2
(z) = ∫ 
----------------------------------------------------------------------------------------
Q
33
- dz

0

for laminated plates. The Qij are the coefficients of the transformed 3D-lamina
material matrix. Note that if the Qij are constant within each lamina, but differ-
ent from lamina to lamina, the transverse variation of the uz displacement com-
ponent is piecewise quadratic. The definition of φ2(z) and ψ2(z) are as given
below.
Models 3 to 6. Characterized by:
n
1
= n
2
= n
3
= 2 → n
f
= 6 for Model 3
n
1
= n
2
= 2 , n3 = 3 → n
f
= 7 for Model 4
n
1
= n
2
= n
3
= 3 → n
f
= 9 for Model 5
n
1
= n
2
= 3, n
3
= 4 → n
f
= 10 for Model 6.
5
u ( x, y, z ) = u ( x, y )φ ( z ) + u ( x, y )φ ( z ) + u ( x, y )φ ( z )
x x 1 1 x 2 2 x 3 3
u ( x, y, z ) = u ( x, y )ψ ( z ) + u ( x, y )ψ ( z ) + u ( x, y )ψ ( z )
y y 1 1 y 2 2 y 3 3
u ( x, y, z ) = u ( x, y )ρ ( z ) + u ( x, y )ρ ( z ) + u ( x, y )ρ ( z ) + u ( x, y )ρ ( z )
z z 1 1 z 2 2 z 3 3 z 4 4
where: φ2(z)=ψ2(z)=z3, φ3(z)=ψ3(z)=z5, and ρ3(z)=z4, ρ4(z)=z6 for homoge-

neous plates. For laminated plates the transverse functions are:
z
 Q 44 – Q 45 
φ2 ( z ) =
∫  Q-------------------------------
44 Q –Q 
55
2
45
- dz
z
 Q 55 – Q 45 
ψ2 ( z ) = ∫  -------------------------------
Q Q –Q 44 55
2
45
- dz

In this case, if the Qij are constant within each lamina but different from lamina to
lamina, the transverse variation of the in-plane displacement components are piece-
wise linear.
z
 Q 13 + 2Q 36 + Q 23
ρ3( z ) =
∫  ---------------------------------------------- zdz
 Q 33 
0
z
 M 0 ( z )Q 44 – N 0 ( z )Q 45 
φ3 ( z ) =
∫  -------------------------------------------------------- – ρ 2 ( z ) dz
 Q 44 Q 55 – Q 45
2 
0
z
 N 0 ( z )Q 55 – M 0 ( z )Q 45 
ψ3 ( z ) =
∫  -------------------------------------------------------- – ρ 2 ( z ) dz
 Q 44 Q 55 – Q 45
2 
5
0
with
z
 Q 13 + Q 36
M0 ( z ) =
∫  Q 11 + 2Q 16 + Q 66 – ( Q 13 + Q 36 ) -------------------------- φ 2 ( z )
 Q 33 
0
 Q 13 + Q 36
+  2Q 26 + Q 12 + Q 66 – ( Q 23 + Q 36 ) -------------------------- ψ 2 ( z )
 Q 33 
Q 13 + Q 36
– 2 -------------------------- z dz
Q 33
The construction of the models follows a clear pattern. The first model has the min-
imum number of fields for bending (nf = 3). After that, the models of even order (2,
4, etc.) are created by adding one more field to the transverse displacement compo-

z
+ Q 36
 Q + Q + Q + Q – ( Q + Q ) Q-------------------
∫
23
N (z) = - φ2 ( z )
0  26 12 66 16 13 36 Q
33 
0
Q +Q
+  2Q 26 + Q 22 + Q 66 – ( Q 23 + Q 36 ) -------------------
23 36
- ψ2 ( z )
 Q
33 
Q +Q
23 36
– 2 -------------------
Q
-z dz
33
5
nent of the previous model. The odd numbered models (3, 5, etc.) are created
by adding one more field to each one of the in-plane displacement components
of the previous model.
Remark Optimal shear correction factors
The results of recent research established some interesting and important facts:
By properly choosing the shear correction factor for the Reissner-Mindlin
model, either the energy or the average mid-surface displacement can be opti-
mized with respect to the fully three-dimensional model of the plate. For the
next model in the hierarchy there is only one shear correction factor. For higher
models no shear correction factor exists.
Shear correction factors (κ) are used for homogeneous isotropic plate models 1
and 2. They are included to compensate for the additional shear stiffness incor-
porated into these models due to the assumed displacement field. They are
computed as a function of the Poisson’s ratio (ν) according to the following
criteria (Ref. [26]):
5
For model 1: For optimal strain energy of the solution, κ = ----------------
- .
6(1 – ν)
20
For optimal plate deflection, κ = ------------------
- .
3 ( 8 – 3ν )

For model 2: If ν = 0, κ = 5/6.
2
If ν >0 then: κ = 12 – 2ν- – 1 +
------------- 1
20ν
+ --------------------
-2 .
2
ν ( 12 – 2ν )
Recommended mesh design for plates

As discussed earlier, both the errors of discretization and the errors of modeling
must be controlled. The finite element implementation in StressCheck is oriented
toward resolving each type of error independently from each other. An essential
prerequisite to proper assessment of the errors of idealization is the control of the
errors of discretization.
5
Discretization errors are controlled by users of finite element computer programs,
most commonly by the design of the finite element mesh. Proper mesh design
depends on the characteristics of the exact solution of the mathematical model. In
the case of plate models one of those characteristics is the boundary layer.
Boundary layer effects occur at the plate boundaries, and are characterized by the
fact that the solution “near” the boundary is substantially different from the solution
in the interior. All hierarchic plate models (as the fully three dimensional model)
exhibit boundary layers, and an important part of the energy of the solution is con-
tained in them. For further information on boundary layer effects we refer to [27],
[28]. Therefore, the mesh design necessary to obtain accurate solutions for any
given member of the hierarchic sequence of models should properly account for the
boundary layers. Extensive numerical experimentation clearly showed that the hier-
archic models implemented in StressCheck are very capable of resolving the
boundary layer effects when proper meshing is used.
Based on the numerical evidence, guidelines for mesh design to be used with the
hierarchic models that will provide optimal or near-optimal meshes in energy norm
were developed. These guidelines are summarized as follows:
(a) The first step is to design a finite element mesh that provides optimal rate of
convergence for the exact solution of the plate problem in the interior of the domain
without consideration of the boundaries. For smooth problems and p-convergence

this typically involves the use of uniform or quasiuniform meshes. This will be
referred to as the “coarse mesh”.
(b) Once the coarse mesh is available, the boundary layers should be accounted
for by the use of graded meshes. The largest size of the elements of the coarse
mesh near the boundary of the plate (L) is used as the reference (FIGURE
179). The relevant element size is the dimension in the direction perpendicular
to the boundary. For most practical problems one or two layers of graded ele-
ments towards the edges are sufficient to account for boundary layer effects:
For one layer of graded elements, the “empirical optimal” grading depends on
the material type. For homogeneous plates r1 should be selected closest to the
lower bound in (14) and for laminated plates closest to the upper bound value.
0.08L < r < 0.12L (14)

1
For two layers of graded elements, the values of r1, r2 in (15) are nearly inde-
5 pendent of the material type.
r ≈ 0.04L
1
r ≈ 0.24L
2 (15)
Strategy for model selection

The modeling error is defined as the difference between the exact solution of
the three-dimensional problem and the exact solution of a given hierarchic
model. The finite element solution provides an approximation to the exact
solution of any one model and therefore, the target tolerance for the modeling
error should not be tighter than for the discretization error.
The discretization errors are controlled by proper meshing and p-extension as

discussed earlier. The objective is to obtain an approximated solution which is
“close” to the exact solution of the particular model:
HM i HM i
u EX – u FE E(Ω) ≤ τd
where τd is a prescribed tolerance for the discretization error of the ith model.

D C
AB and AD are external boundaries.
BC and CD are symmetry lines.
FIGURE 179 Graded mesh for plate bending.
A B
r1
r2 5
L
Reliability in numerical simulation can be achieved at a reasonable cost only if the

model is properly selected. For plates of homogeneous isotropic materials the use
of model 1 gives, in general, excellent approximation to the structural response
(natural frequencies, deflection, bending moments, etc.) and adequate approxima-
tion for the in-plane stress distribution (normal stresses σ x, σ y, τ xy ). If transverse
stresses ( τ xz, τ yz, σ z ) are of interest, higher order models will be required. In
those cases, model 4 or 5 should be generally sufficient.
For laminated plates proper model selection strongly depends on the material
anisotropy (number of layers, fiber orientation, and stacking sequence). In general,
model 1 will be adequate for predicting the structural response of “thin” plates, but
higher order models are needed for the computation of in-plane and transverse
stresses, specially at the interfaces between layers.
The strategy for model selection can be summarized as follows:
1. Compute the finite element solution using a low-order model for all elements of
the mesh. Select model 1 for example, and control the quality of the solution by
designing a mesh such that τd is a small number, typically about 1% or less.

2. Fix the value of the polynomial degree equal to the smallest p-level for
which the estimated error in energy norm is less than or equal to the tolerance
τd. With fixed polynomial degree run a sequence of three models, say from 1 to
3.
3. Examine the convergence of the quantities of interest (deflection, bending

moment, shear force, directional stress, etc.) in the critical regions of the plate.
If convergence within an expected modeling tolerance (τm) is not realized,
increase the model order and use the restart option in the linear analysis status
box.
The target tolerance for the modeling error should be larger than the tolerance
for the discretization error ( τ m > τ d ).
Example Consider the problem shown in FIGURE 180a which corresponds to a simply-
supported square plate with a transverse load acting on a centrally located
small circular area. Because of symmetry, only one quarter of the plate is
shown. The plate is made of a graphite-epoxy laminated composite material
5
with a symmetric 3-ply orthotropic layup with respect to the middle plane (90/
0/90). The length-to-thickness ratio of the plate is 10:1. This model can be
loaded from the Tutorial Handbook, file: PlateLaminate.sci.
FIGURE 180 (b) and (c) show the bending moment distributions (Mx and My)
along one of the symmetry lines for three hierarchic models. The bending
moment distributions are almost identical for the three plate models which
would indicate that the first order model (Model 1) is adequate for predicting
the structural response. FIGURE 181 (a) and (b) show the normal stress distri-
bution through the plate thickness corresponding to point A in FIGURE 180a,
where the absolute values of Mx and My are maximal. Since the stress distri-
bution is antisymmetric with respect to the middle plane, only the top half of
the stress distributions are shown in the figures. These results clearly show that
while the Sx stress distribution is very close for all models, the distribution of
Sy is substantially different for Model 1 than for the other two models. Model
1 underestimates the maximum absolute value of Sy at the surface of the plate
(Z=0.5) by 40%.
This simple example clearly shows the benefits of the hierarchic models in the
computation of both the structural and strength responses of laminated com-
posites.

(b) Bending moment Mx along the

indicated symmetry line.
(a) Mesh and boundary condi-

tions.
A
5
(c) Bending moment My along the

indicated symmetry line.
FIGURE 180 Laminated plate example problem.

Meshing and boundary conditions
(a) Through-thickness Sx stress

distribution.
(b) Through-thickness Sy stress

distribution.
5
FIGURE 181 Laminated plate problem. Through the thickness

stress distribution at point A.

The control of the errors of discretization for two- and three-dimensional prob-
lems in elastostatics is a very simple task when p-extensions are used in con-
nection with properly designed meshes. It is very important to produce
sufficiently accurate representation of the domain and boundary conditions
(loads and constraints), so that these potential error sources do not significantly
influence the accuracy of the functionals of interest.

Designing the mesh

A general rule is that finite element meshes should be so constructed that the vertex
angles of triangular, pentahedral and tetrahedral elements are as close to 60 degrees
as possible, and the vertex angles of quadrilateral and hexahedral elements are as
close to 90 degrees as possible. The p-version is much more ‘forgiving’ with
respect to deviation from the optimal vertex angles than the h-version, nevertheless
vertex angles should not be less than 10 degrees or greater than 150 degrees. There
are simple rules for designing finite element meshes for problems in Categories A
and B. These rules are outlined in the following.
Consider, for example, the linkage arm detail shown in FIGURE 182. Assuming
FIGURE 182 Part of a linkage arm.
that the applied loading is smooth and the constraints do not introduce singular
points, the solution of this problem will be very nearly in Category A. Strictly
speaking, the solution will not be perfectly in Category A because the curvature of
the boundary changes abruptly at the endpoints of the fillets. The solution is not
analytic in these points which therefore must be nodal points. Note that Stress-
Check automatically inserts nodes at the endpoints of fillets. The abrupt changes in
curvature will not introduce high stresses, nor will they cause the computed stresses
to oscillate. Therefore no mesh refinement is necessary around these points.
For problems in Category A, uniform meshes are optimal. The number of elements
is controlled by the geometric complexity of the object being modeled and the use
of nearly optimal vertex angles. An example is the mesh for a torque arm shown in
FIGURE 183. This problem is discussed in references [1] and [2]. The mesh shown
here is simpler than in the references. StressCheck makes it easy for the user to
describe boundaries with smoothly connected fillets.

FIGURE 183 Typical mesh for problems in Category A. Torque arm.
For problems in Category B geometrically graded meshes are optimal. The

diameter of elements decreases in geometric progression toward singular
points, with a common factor of approximately 0.15. Elsewhere the size of the
elements is controlled by the geometric complexity of the object being mod-
eled. An example of mesh design for the p-version is shown in FIGURE 184.
5
In FIGURE 184 two layers of geometrically graded elements are shown for the
crack tip. The number of layers depends on the desired degree of accuracy and
the amount of strain energy contained in the singular terms, which depends on
the loading. The number of layers is independent of the strength of singularity
λ. One or two strongly graded layers is generally sufficient for practical com-
putations. In some cases refinement is not necessary.
For problems involving a few thousands degrees of freedom adaptivity does

not provide appreciable benefits because selective assignment of the polyno-
mial degree and/or adaptive mesh refinement carry an overhead which would
very likely exceed the reduction in the number of arithmetic operations made
possible by the adaptive process. It is recommended therefore that for small
problems a good initial mesh should be selected, based on the expected
smoothness of the solution, and uniform p-extension performed.
In most cases the first mesh, designed in accordance with the guidelines pro-
vided here, will yield the desired accuracy at p ≤ 8 , trunk space. If the accu-
racy is not sufficient at p=8 then the product space should be tried. If the
accuracy is still not sufficient for the product space at p=8 then a new mesh is
needed.
Designing a new mesh for use in conjuction with p-extensions is based on sim-
ple considerations: p-Extensions are very effective in those parts of the domain

Crack tip region. Deformed

configuration.
5
FIGURE 184 Typical mesh detail for problem in Category B.
where the solution is smooth. Therefore attention must be focussed on those parts
of the domain where singular points, notches, fillets with small radii or other “stress
risers” lie. Introducing an additional layer of geometrically graded elements, simi-
lar to those in FIGURE 184, is usually sufficient.
In some cases there are many singularities and it is possible that an unacceptable
discretization error is caused by a few singular points only. In such cases mesh
refinement, confined to the vicinity of those singular points, is sufficient.
Selection of underrefined regions is best performed visually. In plotting stress val-

ues, StressCheck generates a data mesh, the fineness of which is defined by the user
through the selection of the “midsides”, which are the number of grid points of the
data mesh, not counting the vertices, along the side of an element. For example, a
data mesh for five midsides is shown in FIGURE 185.
The desired data are evaluated at the grid points of the data mesh separately for
each element and are displayed with smoothing within each element only. Thus the
results are displayed as two sets of values at interelement boundaries, one from
each element. Any discontinuity in stress is therefore clearly visible. The sources of
discretization error are made visible by the discontinuities. There is an interesting

FIGURE 185 Typical data mesh corresponding to five “midsides”.
pattern: The discretization errors are much more strongly manifested by the
stress discontinues at odd p-values. Therefore the stresses should be plotted for
p=3, p=5 or p=7 and the mesh refined in those parts of the solution domain
5 where the stress discontinuities are the greatest.
Adaptivity makes good sense for large problems when the data of interest are
directly related to the potential energy. When the data of interest are not
directly related to the energy then care must be exercised. This is because
adaptivity seeks to minimize the error in energy norm for a given number of
degrees of freedom. Smallness of error in energy norm does not guarantee that
the errors in the data of interest are small. For this reason, the following proce-
dure is recommended:
• In the case of large problems it is usually possible to distinguish
between two regions, the region of primary interest and the region of
secondary interest. The region of primary interest is that part of the
solution domain where the data of interest, for example, the highest
stress or the largest stress intensity factor, is expected to be found. The
region of secondary interest is included in the solution domain only
because the boundary conditions on the region of primary interest are
not known. The region of secondary interest acts as a sophisticated
spring, conveying information about constraints and loads to the region
of primary interest. Since only the traction-displacement relations
along the boundary between the two regions are of interest, which are
directly related to the energy, adaptivity is sensible for the region of
secondary interest only.
• For large problems execute runs from p=3 to 5 and obtain an estimate
of the error in energy norm. If this error is under 2 percent then restrict

the p-level to p=5 everywhere except in the region of primary interest. Con-
tinue to increase the p-level in the regions of primary interest until the data
you wish to compute are clearly converging to a limit. If the global error is
larger than 2 percent then caution is advised. It may be necessary to revise
the mesh or perform uniform mesh refinement using h-Discretization.
Remark Abrupt changes in thickness. If the thickness changes abruptly then the continuity
of the membrane force, the stress times the thickness should be checked.
Applying loading
Several type of loadings such as traction, spring, body forces or point are available
in StressCheck to describe common situations in engineering practice. The defini-
tions are as follows:
Traction loading: means that a distributed load (in force per unit area) is imposed
on a boundary. Traction is a vector quantity. Thus, in 2D analyses two vector com- 5
ponents must be given (in 3D analyses three vector components are required).
These may be in the normal-tangent reference frame, in the global system, or an
arbitrary local system.
Spring loading: means that load is applied through a displacement imposed on a

distributed spring. Specifically, the traction vector components in the normal-tan-
gent reference frame, for 2D elasticity for example, are dependent on the imposed
spring displacements:
=
where Tn, Tt are traction vector components in the normal and tangent directions,
respectively, ann, ant, att are spring coefficients (in units of force per length cubed),
δn, δt are the imposed spring displacement components in the normal-tangent direc-
tions. These are the loading terms. The symbols un, ut represent the normal-tangent
displacement vector components.
Spring loading can be applied in global or local coordinates as well. The only
restriction is that the spring coefficients and the imposed spring displacements must
be in the same reference frame. When the spring stiffness is small in comparison
with the stiffness of the elastic body, then the tractions are nearly equal to the spring
stiffness coefficients multiplied by the imposed displacements. When the spring

stiffness is large in comparison with the stiffness of the elastic body, then the
displacements imposed on the spring are nearly equal to the displacement
imposed on the elastic body.
Remark Uses for spring loading. This type of loading is very useful for the following
reason: In many cases the kinematic boundary conditions are known only
approximately. In such cases it is advisable to investigate the sensitivity of the
data of interest to changes in boundary conditions. By changing the spring
stiffness and the imposed spring displacement, it is possible to investigate con-
ditions from nearly rigid conditions to very soft conditions.
Body force loading: The body is loaded by some distributed load, such as by
gravity or inertia, due to acceleration. The body force is in terms of force per
unit volume. If the acceleration vector a = {a x, a y, a z} and the mass density
(ρ) are known then, X = ax ρ, Y = ay ρ, Z = az ρ are the body force compo-
nents.
Thermal loading: means that a temperature distribution will be specified on

5
elements. Temperature distribution obtained from a heat transfer analysis are
also available.
Point loads: The elastic body is loaded by a concentrated force

F = {F x, F y, F z} . You may specify Fx, Fy, Fz directly in the global or in a
local coordinate system. For 2D problems it is also possible to specify the
absolute value of F and the direction of F in terms of the angle measured
counterclockwise from the positive x-axis.
Remark Using concentrated forces. In the linear theory of elasticity the displacement in
the point of application of a concentrated force is infinitely large. Since the
strain energy is proportional to the force times the displacement, the strain
energy is also infinitely large. StressCheck is based on the displacement formu-
lation which is applicable only to problems for which the exact solutions have
finite strain energy. For this reason concentrated forces are inadmissible.
Applying a concentrated force would yield finite strain energy for any choice
of mesh and any p-distribution, however the energy would increase with
respect to mesh refinement, or increase of p, and the strain energy would keep
on increasing, rather than converging to a limiting value. The error for any p
and any mesh is infinitely large.
Nevertheless, concentrated forces are useful short-hand in finite element analy-

ses. In applying a concentrated force the analyst indicates that the precise dis-

tribution of displacements, strains and stresses in the neighborhood of the point of

application of the concentrated force is unimportant. Any local stress distribution
which is statically equivalent to the applied concentrated force is acceptable. Inter-
preted in this way, concentrated forces make sense because the underlying assump-
tion, namely that Saint-Venant’s principle is applicable, is valid.
StressCheck allows using concentrated forces. Concentrated forces may be applied

to a node, side, or in the interior of the element. The polynomial degree of the ele-
ments to which concentrated forces are applied should be fixed at some low p value
(typically p=2 or 3). The precise distribution of the statically equivalent stresses
will depend on the polynomial degree selected for the element and the local mesh
layout.
In axisymmetric analysis, the equivalent of a point load is a line load over a one
radian sector. Therefore, if a line load of magnitude W (force/length) needs to be
specified acting at a position r0, then the entry in the point force field should be the
product of W times r0.
5
Applying constraints
Three types of constraints: General, Rigid Body, and Spring Coefficient are avail-
able in StressCheck. An explanation follows:
General: Means that you will prescribe one or more displacement components to
one or more boundaries or edges. For example, if one of the boundaries of the solu-
tion domain lies on a plane of symmetry, which in 2D elasticity appears as a line of
symmetry, then you will prescribe the normal displacement to be zero. StressCheck
also includes the symmetry, antisymmetry and built-in constrains as particular
cases of the general constraint case.
Rigid Body: StressCheck provides a convenient way for specifying rigid body con-
straints. In the case of two-dimensional elasticity the user selects two nodes (3
nodes for 3D analysis). StressCheck will then create a right-handed local coordinate
system, the x-axis of which is directed from the first node to the second. In this
local coordinate system both displacement components are constrained for the first
node; only the displacement component in the direction of the local y-axis is con-
strained for the second node (similarly for 3D with three nodes instead).
Remark Using point constraint. Point constraints are admissible only for preventing rigid
body displacement and rotation. The body must be in equilibrium under the action
of external forces. Unlike the errors introduced by concentrated forces, errors intro-

duced by improper point constraints are not localized. The reactions at point
constraints are dependent on the mesh and the polynomial degree of elements.
Saint-Venant’s principle does not apply in the case of point constraints. For this
reason multipoint constraints should not be used in StressCheck.
Spring Coefficients: When an elastic body is constrained by means of distrib-

uted springs, or loaded by displacements imposed by distributed springs, then
the spring coefficients modify the stiffness matrix. For this reason the spring
coefficients are entered as constraints.
Remark Prescribing displacements. When constant normal or tangential displacements

are prescribed on curved boundaries, or displacements are prescribed by for-
mulae, then the prescribed displacement conditions cannot be satisfied exactly
in general. Consider, for example, a circular boundary with a displacement
un(t) prescribed in the normal direction.
Referring to FIGURE 186, we have:

5 un = un(t) = - (ux(t) cos t + uy(t) sin t)
y
uy
un ux
t
x
FIGURE 186 Normal and directional displacements.
If un is not a constant, ux and uy are linear combinations of basis functions

which are polynomials on the standard element but are not polynomials once
they are mapped into an element with a circular boundary. The displacement
components ux and uy can be written as:
p+1 p+1
ux ( t ) = ∑ α i φi ( t ), uy ( t ) = ∑ βi φi ( t )
i=1
i=1

p-discretization: The trunk and product spaces
where αi, βi are arbitrary constants, φ i ( t ) are the basis functions along the circular
boundary and p is the polynomial degree of the element. Unless it so happens that
the prescribed displacement û n ( t ) can be exactly represented by some particular
choice of αi and βi, it is necessary to approximate û n ( t ) . StressCheck performs this
task by enforcing constraint equations on the coefficients αi, βi:
π
---
2
∫ ( û ( t ) – u ( t ) ) φ ( t ) r dt = 0
π
n n i i=1,2,..., p+1
---
–2
that is, the difference between the user-specified displacement and the displacement
enforced by StressCheck is orthogonal to the basis functions. This condition leads
to p+1 algebraic equations in 2(p+1) unknowns. Note that different constraint
equations are enforced for different p-levels. The prescribed constraints are
enforced exactly in the limit as p → ∞ . Consequently, as p is increased, a slightly
different problem is approximated at each p-level and the convergence of the poten- 5
tial energy is not necessarily monotonic.
In the special cases when un = ut = 0 on a curved boundary or ux = constant, uy =

constant the constraints are enforced exactly.

StressCheck supports the use of trunk and product spaces for quadrilateral and
(q)
hexahedral elements. By definition, the standard quadrilateral element Ω st is the
set of points (ξ,η) which satisfy ξ ≤ 1 , η ≤ 1 , see FIGURE 187.
The trunk space of degree p is defined as the set of polynomials which can be
expressed as linear combinations of ξi ηj, i, j =0,1,...p; i + j ≤ p and supplemented
by the monomial term ξ η for p=1, and the monomial terms ξp η, ξ ηp for p ≥ 2 .
For example, the set of monomials which span the trunk space for p=4 is shown in
Table 6.

η
(-1,1) (1,1)
ξ
(q)
Ω st
(-1,-1) (1,-1)
FIGURE 187 The standard quadrilateral element.
TABLE 6. Span of the trunk space, p=4.

5
1
ξ η
ξ2 ξη η2
ξ3 ξ2η ξη2 η3
ξ4 ξ3η ξ2η2 ξη3 η4
ξ5 ξ4η ξ3η2 ξ2η3 ξη4 η5
ξ6 ξ5η ξ4η2 ξ3η3 ξ2η4 ξη5 η6
ξ7 ξ 6η ξ5η2 ξ4η3 ξ3η4 ξ2η5 ξη6 η7
ξ8 ξ7η ξ6η2 ξ5η3 ξ4η4 ξ3η5 ξ2η6 ξη7 η8
The product space of degree p is the set of polynomials which can be

expressed as linear combinations of ξi ηj, i, j =0,1,...p. For example, the set of
monomials which span the product space for p=4 is shown in Table 7.
StressCheck supports both spaces for quadrilateral elements. It is possible to

mix product and trunk spaces for elements within the same mesh. The shape
functions used by StressCheck for quadrilateral and hexahedral elements are
described in Ref. [3]. For hexahedral elements the use of the mixed spaces
should be avoided.

TABLE 7. Span of the product space, p = 4.
1
ξ η
ξ2 ξη η2
3 2 2
ξ ξ η ξη η3
4 3 2 2 3
ξ ξ η ξ η ξη η4
ξ5 ξ4η ξ3η2 ξ2η3 ξη4 η5
6 5 4 2 3 3 2 4 5
ξ ξ η ξ η ξ η ξ η ξη η6
ξ7 ξ6η ξ5η2 ξ4η3 ξ3η4 ξ2η5 ξη6 η7
ξ8 ξ7η ξ6η2 ξ5η3 ξ4η4 ξ3η5 ξ2η6 ξη7 η8
The answer to the question: “Which space should be used in a given application?”
depends on the accuracy requirements. The following points should be taken into
consideration:
5
1. The number of degrees of freedom increases much more rapidly with respect to p
when the product space is used. For example, in the case of p=8 there are 47 (resp.
81) degrees of freedom per field for the trunk space, (resp. product space) in two
dimensions. In 3D there are 192 degrees of freedom per field for the trunk space
and 729 for the product space.
2. For a given polynomial degree (p>1) the error, measured in energy norm, is
never greater for the product space than for the trunk space. This is because the
trunk space is a subset of the product space.
3. For a given number of degrees of freedom the error, measured in energy norm,
may be smaller or larger. There appears to be no way of predicting which space per-
forms better for specific cases.
4. For static analyses, work performed by the computer increases slower with
respect to increasing p for the product space than for the trunk space. This is
because in the case of the product space the increase affects the internal modes only
and those modes are eliminated by the solver locally, i.e. no assembly is involved
and the front width is unaffected.
Remark Standard polynomial space for triangular elements.

Generalized Hooke’s law in three-dimensions
On the standard triangular element shown in FIGURE 188, the standard poly-
nomial space of degree p is comprised of those functions which can be
expressed as linear combinations of the monomials ξi ηj, i, j =0,1,...,p;
i+j≤p.
η
(0, 3)
(t)
Ω st
(– 1,0) (1,0) ξ
FIGURE 188 The standard triangular element.
5
The shape functions used by StressCheck for triangles and pentahedral ele-
ments are described in Ref. [3].

The following convention is used in connection with the material properties:
1. The elastic properties are assumed to be known in the material axes

( x, y, z ). The material 1-direction is aligned with the x – direction.
2. The elastic properties in a rotated coordinate system (x,y,z) are

obtained by rotation of the properties given in the material axes.
Under these conditions the three-dimensional (3D) constitutive equations for

an anisotropic materials can be written in the following matrix form in terms of
the material axes:

σx C 11 C 12 C 13 C 14 C 15 C 16 εx
σy C 22 C 23 C 24 C 25 C 26 εy
σz C 33 C 34 C 35 C 36 εz
.
= (16)
τ C 44 C 45 C 46 γ
yz yz
τ sym C 55 C 56 γ
xz xz
τ C 66 γ
xy xy
The material stiffness matrix in 3D-space contains 21 independent components. If

there are any planes of material symmetry then the number of independent compo-
nents will be fewer.
Assume that z = 0 is a plane of elastic symmetry. Then (see, for example, 5

Ref. [39], pp. 5-6):
C 14 = C 15 = C 24 = C 25 = C 34 = C 35 = C 46 = C 56 = 0
and the stiffness matrix in (16) becomes:
C 11 C 12 C 13 0 0 C 16
C 22 C 23 0 0 C 26
C 33 0 0 C 36
[C] = (17)
C 44 C 45 0
sym C 55 0
C 66
In this case there are 13 independent stiffness coefficients and the material is
referred to as a monoclinic material.

If there are three mutually perpendicular planes of elastic symmetry then:

C 16 = C 26 = C 36 = C 45 = 0
and (17) becomes:
C 11 C 12 C 13 0 0 0
C 22 C 23 0 0 0
C 33 0 0 0
[C] = (18)
C 44 0 0
sym C 55 0
C 66
5
There are 9 independent stiffness coefficients and the material is called ortho-
tropic.
Further reduction takes place if one of the planes is isotropic, that is the proper-
ties in the isotropic plane are independent of direction. Consider, for example,
the case where x = 0 (i.e., the 2-3 plane) is a plane of elastic isotropy. Then,
C 22 – C 23
C 33 = C 22, C 13 = C 12, C 44 = ------------------- , C 55 = C 66
2
and (18) becomes:
C C C 0 0 0
11 12 12
C C 0 0 0
22 23
C 0 0 0
22
[C] = C –C (19)
22 23-
------------------------- 0 0
2
sym C 0
66
C
66

Such material is referred to as transversely isotropic and its stiffness properties are
determined by 5 independent constants.
For completely isotropic materials:
C 11 – C 12
C 22 = C 11, C 23 = C 12, C 66 = -------------------
2
and the number of independent constants is 2:
C C C 0 0 0
11 12 12
C C 0 0 0
11 12
C 0 0 0
11
C –C
[C] = 11 12-
------------------------- 0 0 (20) 5
2
C –C
11 12-
sym ------------------------- 0
2
C –C
11 12
--------------------------
2
Engineering constants
The material properties are usually given in terms of Young’s moduli, Poisson’s
ratios and shear moduli. The relationship among these constants and the compo-
nents of the material stiffness matrix are described in this section.
In the case of orthotropic materials nine engineering constants are required (E11,
E22, E33, G12, G23, G31, v12, v23, v13), and the relation among these constants and
the Cij in (18) are (Ref. [35], page 41):

C = (1 – ν ν )VE C = (1 – ν ν )VE
11 23 32 11 22 13 31 22
C = (1 – ν ν )VE C = (ν +ν ν )VE
33 12 21 33 12 12 13 32 22
C = (ν +ν ν )VE C = (ν +ν ν )VE
13 31 32 21 11 23 32 12 31 22
C = G C = G C = G
44 23 55 31 66 12
where
E E E
22 33 , 11
ν
21
= ν ---------
12 E
, ν
32
= ν ---------
23 E ν
13
= ν ---------
31 E
11 22 33
5 V = (1 – ν ν – ν ν – ν ν – 2ν ν ν )
–1
12 21 23 32 13 31 12 23 31
In the case of transversely isotropic materials five engineering constants are

required (E11, E22, G12, v12, v23), and the relation among the Cij in (19) and the
Young’s moduli, Poisson’s ratios and shear moduli are (Ref. [40], page 42):
2 C = (1 – ν ν )VE
C =  1 – ν  VE 22 12 21 22
11 23 11
C = ν (1 + ν )VE C = (ν +ν ν )VE
12 12 23 22 23 23 12 21 22
C = G
66 12
where E11 is the stiffness in the material 1-axis (the direction of the fibers), E22
is the stiffness in the material 2-axis (the direction transverse to the fibers), v12
is the major Poisson’s ratio (transverse contraction due to an axial extension),
v21= v12 E22/ E11 is the minor Poisson’s ratio (axial contraction due to a trans-
verse extension), v23 is the Poisson’s ratio for a transverse contraction due to
transverse extension), and G12 is the shear modulus. In this case V is defined as

2 –1
V =  1 – 2ν ν – ν – 2ν ν ν  .
12 21 23 12 21 23
In the case of isotropic materials only two engineering constants are required (E, v)
and their relationship with the cij in (20) are
( 1 – ν )E νE
C = -------------------------------------- C = --------------------------------------
11 ( 1 + ν ) ( 1 – 2ν ) 12 ( 1 + ν ) ( 1 – 2ν )
Transformation of the material stiffness matrix

In engineering applications it is often necessary to determine the material stiffness
matrix with respect to a rotated coordinate system x, y, z when the material stiffness
matrix is known in terms of another coordinate system x, y ,z . Consider the case of
a rotation β about the y –axis followed by a rotation α about the z – axis. The 5
stress-strain relation in the x, y ,z coordinate system can be written as:
{σ} = [C]{ε} (21)
where [C] is the stiffness matrix defined in (17), (18), or (19) and { σ } , { ε } are the
stress and strain components as given in (16). For isotropic materials, the stiffness
matrix (20) is not affected by the rotation of the coordinate system. The strain
energy per unit volume is an invariant, that is, independent of the coordinate sys-
tem, and therefore we can write:
1
= -2 { σ } { ε }
T
U = U
or using (21):
= 12- { ε } [ C ] { ε }
T
U (22)
The strains in the x, y ,z system are written in terms of the strains in the rotated sys-
tem as:
{ ε } = [ Tβ ] [ Tα ] { ε }

where [Tβ] is the transformation matrix due to the β rotation about the y – axis:
2 2
r 0 s 0 rs 0
0 1 0 0 0 0
2 2
[ Tβ ] = s 0 r 0 – rs 0 (23)
0 0 0 r 0 –s
2 2
– 2rs 0 2rs 0 r –s 0
0 0 0 s 0 r
with r=cos β, s = sin β. [Tα] is the transformation matrix due to the α-rotation
about the z – axis (Ref. [40], page 3):
5 2 2
m n 0 0 0 mn
2 2
n m 0 0 0 – mn
[ Tα ] = 0 0 1 0 0 0 (24)
0 0 0 m –n 0
0 0 0 n m 0
2
– 2mn 2mn 0 0 0 m – n2
with m=cos α, n=sin α. Therefore, (22) can be written as:
= 12- { ε } [ Q ] { ε }
T
U
and the stress-strain relations in the global coordinate system will be:

σ Q 11 Q 12 Q 13 Q 14 Q 15 Q 16 ε
x x
σ Q 22 Q 23 Q 24 Q 25 Q 26 ε
y y
σ Q 33 Q 34 Q 35 Q 36 ε
z z
= (25)
τ Q 44 Q 45 Q 46 γ
yz yz
τ sym Q 55 Q 56 γ
xz xz
τ Q 66 γ
xy xy
where [Q] is the material stiffness matrix in the global (rotated) coordinate system
which is computed from
T
[ Q ] = ( [ Tβ ] [ Tα ] ) [ C ] ( [ Tβ ] [ Tα ] ) 5
using (23) and (24). In the general case (25) contains 21 independent coefficients.
{σ}and {ε} are the stress and strain components in the global system. For orthotro-
pic and transversely isotropic materials when the rotation is performed about the z-
axis only, the [Q] matrix (25) reduces to:
Q 11 Q 12 Q 13 0 0 Q 16
Q 22 Q 23 0 0 Q 26
Q 33 0 0 Q 36
Q = (26)
Q 44 Q 45 0
sym Q 55 0
Q 66

Laminate composites for planar and 3D problems

A capability is available to facilitate the input of orthotropic material proper-
ties for individual plies and for sub-laminate properties for 2D plane-strain and
3D analyses.
• Individual plies: When the ply-angles are not contained in the standard
2D working plane (the plane-strain XY plane), the material matrix
needed for the 2D analysis are extracted from the 3D material matrix.
The material coefficients are entered in the material directions, and
then assigned to the elements together with the ply angle information.
The program performs all necessary transformations to compute the
equivalent 2D properties in the XY work plane. The appropriate 3D
properties are be applied when the planar model is extruded into 3D as
well.
• Sub-laminates: When a set of plies need to be combined in a single
layer (sub-laminate), the properties of the sub-laminate are obtained by
homogenization. Again, the 3D material coefficients in the material
5
axes of the composite are entered together with the stacking sequence
of the sub-laminate, and the program computes the equivalent 2D
properties in the Stress Check XY work plane for the stack. The appro-
priate 3D properties are applied when the planar model is extruded into
3D as well.
The procedures implemented in Stress Check to account for these two types of
material properties include the following:
• The material coefficients of a single orthotropic ply are entered in the
direction of the material axes (FIGURE 189a). Nine engineering coef-
ficients (E11, E22, E33, G12, G23, G31, ν12, ν23, ν13) in the direction of
the material axes and the ply thickness must be provided. The three
coefficients of thermal expansion (α11, α22, α33) and the mass density
can also be entered, but they are not required unless there is thermal
loading or a modal analysis is needed.
• After the material properties for a single ply are entered, the ply group
information must be provided. This includes the angular orientation of
each ply in the group, in accordance with the following convention: A
positive ply angle (θ) is measured as a counterclockwise rotation about
the z-axis of a local coordinate system (xyz). The z-axis of the local is
aligned with the material 3-axis as shown in FIGURE 189b.
• The ply groups are assembled into a stack by providing the total num-
ber of layers in the laminate, defining whether the stack is symmetric

z≡3 1
θ x
y
3 Y
1 X 2
2
‘plane-strain’
plane
Z
5
or not and entering the stacking sequence based on the ply group names.
This procedure provides a great flexibility, since various ply groups of dif-
ferent material properties can be assembled together to represent a single
sub-lamina.
• Finally, the stack is assigned to the elements in the mesh identifying the
local coordinate system, the z-axis of which is assumed to coincide with the
material 3-axis and must be perpendicular to the plane of the ply (FIGURE
189b). The case of a single ply assigned to an element is treated as a partic-
ular case of a sublaminate with one orthotropic layer.
The two-dimensional material stiffness matrix needed for the plane-strain analysis
is then extracted from the three-dimensional material matrix in global coordinates
(XYZ). When a group of laminae is assigned to a single element, the three-dimen-
sional effective properties are computed by homogenization, following the proce-
dure described in Ref. [55] and discussed in detail below.
The formulation is based on the following: The 3D stress-strain relations for an ortho-
tropic material in the material coordinate system (FIGURE 189a) can be written as:
{σ} = [C]{ε} (27)

where,
σ 11 ε 11 C 11 C 12 C 13 0 0 0
σ 22 ε 22 C 22 C 23 0 0 0
σ 33 ε 33 C 33 0 0 0
{σ} = , {ε} = , [C] = (28)
τ 12 γ 12 sym. C 66 0 0
τ 23 γ 23 C 44 0
τ 31 γ 31 C 55
[C] is the material stiffness matrix with nine independent constants. The rela-
tion between the Cij and the engineering coefficients is given by (Ref. [35],
page 41):
5
C 11 = ( 1 – ν 23 ν 32 )VE 11 , C 22 = ( 1 – ν 31 ν 13 )VE 22 , C 33 = ( 1 – ν 12 ν 21 )VE 33
C 12 = ( ν 21 + ν 23 ν 31 )VE 11 , C 13 = ( ν 13 + ν 23 ν 12 )VE 33 , C 23 = ( ν 32 + ν 12 ν 31 )VE 22
C 44 = G 23 , C 55 = G 31 , C 66 = G 12
–1
V = ( 1 – ν 12 ν 21 – ν 23 ν 32 – ν 31 ν 13 – 2ν 12 ν 23 ν 31 )
E 22 E 33 E 33
ν 21 = ν 12 -------- , ν 32 = ν 23 -------- , ν 31 = ν 13 --------
E 11 E 22 E 11
Next consider the situation shown in FIGURE 189b, in which a rotation θ

about the z-axis (coincident with the material 3-axis) is performed for the kth
layer in the laminate. The stress-strain relations in the xyz local system can be
written as:
(k)
{ σ } xyz = [ Q ] { ε } xyz (29)
where,

(k)
σx εx Q 11 Q 12 Q 13 Q 16 0 0
σy εy Q 22 Q 23 Q 26 0 0
σz εz Q 33 Q 36 0 0
{ σ } xyz = , { ε } xyz = , [ Q(k) ] = (30)
τ xy γ xy sym. Q 66 0 0
τ yz γ yz Q 44 Q 45
τ zx γ zx Q 55
(k)
The material matrix for the kth layer in the local system [ Q ] is obtained by the
following transformation (Ref. [35], page 31):
(k) T
[ Q ] = [T ] [C ][T ]
2 2 5
m n 0 0 0 mn
2 2
n m 0 0 0 – mn
[T] = 0 0 1 0 0 0
0 0 0 m –n 0
0 0 0 n m 0
2 2
– 2mn 2mn 0 0 0 m – n
where [C] is given in Eq. (28), m = cosθ and n = sinθ. Equations (29) and (30) rep-
resent the stress-strain relations for a lamina in a local coordinate system after a
rotation θ about the z-axis.
Next we consider a laminate of constant thickness h composed of N-layers of thick-

ness tk:.
N
h =
∑t k
k=1
Each lamina can have an arbitrary orientation (z-rotation) with respect to the local
coordinate system (xyz). The properties of this laminate can be effectively repre-
sented by an homogeneous anisotropic solid. Consider the stress-strain relations for
the homogeneous solid in the local system of the form:

{ σ } xyz = [ Q ] { ε } xyz
Q 11 Q 12 Q 13 Q 16 0 0
Q 22 Q 23 Q 26 0 0 (31)
Q 33 Q 36 0 0
[Q] =
sym. Q 66 0 0
Q 44 Q 45
Q
55
where the effective elastic constants of the solid Q ij are computed from the
material matrices of the individual plies by the transformations (Ref. [55]):

 N  h
1
N Q 33 = ---------------------------
= --- ∑ ∑ ( Q 13 – λ 13 )t k ( Q 13 – Q 13 ) ⁄ Q 33 
(k) (k) (1) (k) (k) N
Q t k Q 11 +
h 
∑t
11 (k)
k = 1  k ⁄ Q 33
k=2
k=1
 N N 
1
= --- ∑ ∑ ( Q 13 – λ 13 )t k ( Q 23 – Q 23 ) ⁄ Q 33 
(k) (k) (1) (k) (k)
Q t k Q 12 +  N ( k )
12 h  t k Q 44 
Q 44 = 
k = 1 k=2 

k = 1
∑
------------- ⁄ ∆
h∆ k 

 N N 
1
= --- ∑ ∑ ( Q 33 – λ 33 )t k ( Q 13 – Q 13 ) ⁄ Q 33 
(k) (k) (1) (k) (k)
Q t k Q 13 +
13 h 
k = 1 k=2   N ( k )
t k Q 45 
Q 45 = 

k = 1
∑
------------- ⁄ ∆
h∆ k 

 N N 
1
Q 22 = ---  ∑ ( Q 23 – λ 23 )t k ( Q 23 – Q 23 ) ⁄ Q 33 
∑
(k) (k) (1) (k) (k)
t k Q 22 +
h 
k = 1 k=2   N ( k )
t k Q 55 
Q 55 = 

k = 1
∑
------------- ⁄ ∆
h∆ k 

5
 N N 
1
Q 23 = ---  ( Q 33 – λ 33 )t k ( Q 23 – Q 23 ) ⁄ Q 33 
∑ ∑
(k) (k) (1) (k) (k)
t k Q 23 +
h 
k = 1 k=2  λ 13 = Q 13 λ 23 = Q 23
 N N 
1
Q 16 = ---  ∑ ( Q 13 – λ 13 )t k ( Q 36 – Q 36 ) ⁄ Q 33 
∑
(k) (k) (1) (k) (k)
t k Q 16 + λ 33 = Q 33 λ 36 = Q 36
h 
k = 1 k=2 
 N N 
1
Q 26 = ---  ∑ ( Q 23 – λ 23 )t k ( Q 36 – Q 36 ) ⁄ Q 33 
∑
(k) (k) (1) (k) (k) (k) (k) (k) 2
t k Q 26 + ∆ k = Q 44 Q 55 – ( Q 45 )
h 
k = 1 k=2 
 N N   N ( k )  N ( k )  N ( k )
2
1  ( k )
∑ ∑ t Q t Q t Q
(k) (k) (1) (k)
( Q 33 – λ 33 )t k ( Q 36 – Q 36 ) ⁄ Q 33 ∆ =  k 44   k 55 
– k 45 
Q 36 = ---
h
k = 1
t k Q 36 +
k=2



k = 1
∑
-------------
h∆ k  
 k = 1
∑
-------------
h∆ k  
 k = 1
h∆ k  ∑
-------------

 N N 
1
Q 66 = ---  ( Q 36 – λ 36 )t k ( Q 36 – Q 36 ) ⁄ Q 33 
∑ ∑
(k) (k) (1) (k) (k)
t k Q 66 +
h 
k = 1 k=2 
(32)

Once the material matrix for the homogenized solid is computed in the local
system (xyz), we need to determine the material matrix in the global system
(XYZ) shown in FIGURE 189b, and then extract the material matrix associated
with the ‘plane-strain’ plane (XY-plane).
In the global system, the three-dimensional stress-strain relations can be writ-

ten in the following form:
{ σ } XYZ = [ E 3D ] { ε } XYZ (33)
where the global material matrix [E3D] of the homogenized solid is computed
from the local matrix [ Q ] as follows:
T
[ E 3D ] = [ T 0 ] [ Q ] [ T 0 ] (34)
The transformation matrix [T0] is given by:
5
2 2 2
l1 l2 l3 l1 l2 l2 l3 l3 l1
2 2 2
m1 m2 m3 m1 m2 m2 m3 m3 m1
2 2 2n
[ T0 ] = n1 n2 n3 n1 n2 n2 n3 n3 n1 (35)
2l 1 m 1 2l 2 m 2 2l 3 m 3 l 1 m 2 + l 2 m 1 l 2 m 3 + l 3 m 2 l 3 m 1 + l 1 m 3
2m 1 n 1 2m 2 n 2 2m 3 n 3 m 1 n 2 + m 2 n 1 m 2 n 3 + m 3 n 2 m 3 n 1 + m 1 n 3
2n 1 l 1 2n 2 l 2 2n 3 l 3 n1 l2 + n2 l1 n2 l3 + n3 l2 n3 l1 + n1 l3
where li, mi, ni are the direction cosines that relate a point in the global (XYZ)
coordinate system to a point in the local (xyz) coordinate systems:
 x  l1 l2 l3  X 
   
 y  = m1 m2 m3  Y  (36)
   
z
  n1 n2 n3  Z 
Since the transformation indicated by Eq. (34) is very general, the 3D global
material matrix can be fully populated. Therefore, the material matrix in Eq.
(34) can be written as (using a single index notation for the 21 independent
coefficients):

E 1 E 2 E 4 E 7 E 11 E 16
E 3 E 5 E 8 E 12 E 17
E 6 E 9 E 13 E 16
[ E 3D ] = (37)
sym. E 10 E 14 E 19
E 15 E 20
E 21
The 2D plane-strain material matrix is extracted for Eq. (37) by pulling out the
rows and columns associated with the XY-plane of the global system:
σX E1 E2 E7 εX
σY = E3 E8 εY (38)
τ XY sym E 10 γ XY 5
The material matrix in Eq. (38) is used for the computation of the stiffness matrices
for 2D-elasticity under plane-strain conditions, and the material matrix in Eq. (37)
is used for the computation of the stiffness matrices for 3D-elasticity for those ele-
ments with laminate property assignments.
To compute the effective coefficients of thermal expansion for a group of plies the
homogenization procedure is based on averaging the thermal stresses {σth} over
the thickness of the stack and multiplying the inverse of the homogenized material
matrix [ Q ] by {σth}.
The three coefficients of thermal expansion (α11, α22, α33) and the nine engineer-
ing coefficients (E11, E22, E33, G12, G23, G31, ν12, ν23, ν13) for each ply are given
in the direction of the material axes. The CTEs are rotated θ about the material 3-
axis using the rules for strain transformation. As a result of this transformation, the
CTE vector in the local coordinate system xyz for the kth layer is:
(k)
{ α } = α xx α yy α zz α xy 0 0

(k)
The material matrix for the kth layer in the local system [ Q ] is obtained by
the following transformation:
(k) T
[ Q ] = [T ] [C ][T ]
2 2
m n 0 0 0 mn
2 2
n m 0 0 0 – mn
[T] = 0 0 1 0 0 0
0 0 0 m –n 0
0 0 0 n m 0
2 2
– 2mn 2mn 0 0 0 m – n
where [C] is the material matrix in the material coordinate system, m = cosθ
and n = sinθ. The thermal stress per unit temperature change for the kth layer is
5 defined by:
(k) (k) (k)
{ σ th } = [ Q ] { α }
Next we consider a laminate of constant thickness h composed of N-layers of

thickness tk ( h =
∑N t k ). Each lamina can have an arbitrary orientation (z-
rotation) with respect kto= 1the local coordinate system (xyz). The average thermal
stress is computed as:
N
1
∑ t {σ
(k)
{ σ th } = --- k th }
h
k=1
The properties of this laminate are represented by an homogeneous anisotropic

solid with an homogenized material matrix [ Q ] as described in the Stress-
Check documentation, and the homogenized CTEs are then computed from:
–1
{ α } = [ Q ] { σ th }
where { α } is the vector of the homogenized coefficients of thermal expansion

for the equivalent anisotropic solid.

Cylindrical coordinates StressCheck makes it possible to define single-ply material properties in a cylindri-
cal coordinate system, where the ply angle is directed at an arbitrary angle. At any
given point within an element, the material stiffness matrix is determined from the
material properties defined in the lamina coordinate system and the location of the
point in local cylindrical coordinate system. The material axes are aligned with the
local cylindrical system in such a way that a 0o ply is in the θ-direction and a 90o
ply is in the z-direction. If the user requests an off-axis ply, a rotation angle would
be input, and the code performs the necessary transformation about the r-axis at
every integration point within the curved element. Since StressCheck requires the
material matrix in the global system (Cartesian), the material matrix at every inte-
gration point is transformed back to the global system. The procedure works seam-
lessly for both the 2D-plane strain and 3D extruded problems.
Sub-laminate homogenization can be defined in a cylindrical coordinate system,

such as would be required around a fillet region. The same homogenization rules
utilized to determine the integrated sub-laminate material properties for a group of
“flat” plies were used for determining the homogenized properties in the local
cylindrical system. The local cylindrical system is treated as Cartesian for the appli- 5
cation of the homogenization rules. The coordinate transformation rules for the
cylindrical system are then utilized for 2D-plane strain and extruded 3D problems.
Formulation: The transformation from material coordinates to global coordinates

can be visualized in FIGURE 190. Let XYZ represent the global Cartesian coordi-
nate system with unit vectors eX, eY, eZ and let xyz be a local Cartesian coordinate
system with unit vectors ex, ey, ez. The relation between the two sets of unit vectors
is given by:
e eX
x
e = [ TL ] eY
y
e eZ
z
[TL] is the direction-cosine matrix that defines the orientation of the local system
relative to the global system.
Consider a cylindrical coordinate system associated with the local system, and let
er, eθ, eZ be the unit vectors in the radial, circumferential and axial directions,

FIGURE 190 Notation for cylindrical coordinate system tranformation.
respectively. The relation between the cylindrical and local Cartesian unit vec-
tors depends on the angle θ shown in FIGURE 190 and is given by:
eθ e – sin θ cos θ 0
x
ez = [ T ] e , [ Tθ ] = 0 0 1
θ y
cos θ sin θ 0
er e
z
Next, consider the unit vectors e1, e2, e3 in the direction of the material axes.
The material 1-axis forms an angle α with the unit vector eθ and the material 3-
axis is in the direction of er as shown in FIGURE 190. Note that when α=0, the
material 1-axis is in the θ−direction and when α=90o, the material 1-axis is in

the z–direction. The relation between the material and cylindrical unit vectors
depends on the angle α shown in FIGURE 190 and is given by:
e1 eθ cos α sin α 0
e2 = [ Tα ] ez , [ T α ] = – sin α cos α 0
0 0 1
e3 er
Finally, the relation between the material directions and the global coordinates is
given by:
e1 eX
e2 = [ Tα ] [ Tθ ] [ TL ] eY
e3 eZ
5
The transformation given by this last equation correlates the material directions rel-
ative to a cylindrical coordinate system with the global coordinate system. The pro-
cedure to compute the material stiffness matrix in global coordinates during the
computation of the element stiffness matrix and during post-processing operations
in StressCheck can be summarized as follows:
1) Compute the 3D material matrix [C]k for the kth ply in the material directions.
2) Construct the matrix [Tα]k for the kth ply by rotating an angle α about the mate-
rial 3-axis and compute the rotated material matrix [Q]k.
3) Apply homogenization rules if more than one-ply laminate (k > 1) is assigned to
a single element, and determine the homogenized material matrix [Q].
4) Pass homogenized matrix [Q] to stiffness matrix generation. For each Gauss
point (see FIGURE 190):
• Compute the Cartesian coordinates of the point in the local system xyz.
• Compute the angle θ of the point and construct the matrix [Tθ].
• Multiply [Tθ] with the direction cosine matrix of the cylindrical system
[TL].
• Compute global material matrix [E] using [Q] and the product [Tθ]•[TL].
• If the problem is plane-strain extract [E]2D from [E].

5) Pass [Q] to post-processing functions to compute [E] at each extraction

point before computing stress components in global coordinates.
Material property update for geometric nonlinear analysis

In the solution of problem with large deformations, as the structure deforms,
the original material orientation definition for orthotropic/laminate materials
must be updated to conform to the deformed state. For material properties
defined as orthotropic or laminate orthotropic and assigned in Cartesian or
cylindrical systems, the material stiffness matrix [E] in global coordinates is
determined at each point within an element based on the undeformed configu-
ration. During the nonlinear iteration, this material matrix is updated to
account for the relative change in orientation between the undeformed and the
current configuration.
Formulation The 2D or 3D material stiffness matrix [E] in global coordinates corresponding

to the undeformed configuration for flat laminates and for laminates in cylin-
5
drical systems is computed as described above. For each finite element, [E] can
be constant or variable, depending on whether the material properties are
assigned in Cartesian or cylindrical systems.
During a geometric nonlinear analysis, the shape of the element is updated to

account for the deformed configuration corresponding to the current iteration.
The [E] matrix at each point must be updated to account for this change in
(i)
topology. Let [ E k ] be the global material matrix at point k during the ith iter-
ation, with i=0 representing the undeformed state. A non-orthogonal curvilin-
ear coordinate system is associated with point k as shown in Figure 191 for 2D,
which depends on the Jacobian matrix derived from the mapping of the ele-
ment.
The material matrix at point k in the deformed configuration can be written as:
(i) T ( 0)
[ Ek ] = [ Tk ] [ Ek ] [ Tk ] (39)
where [Tk] is a transformation matrix accounting for the relative rotation of the
local systems between Ω(i) and Ω(0). To determine the transformation matrix,
consider orthogonal unit vectors in the global system and at point k before and
after deformation as shown in Figure 191. These orthogonal vectors are com-
puted by normalization and orthogonalization of the vectors associated with
the Jacobian transformation at point k.

(i)
(o) e2
v e2
(o) e1 v
u
.k . k e( i) u
Y 1
Ω(0) deformation Ω(i)

ey
X
ex
FIGURE 191 Finite element before and after deformation.
5
The relation between the undeformed/deformed unit vectors and the global unit
vectors can be written as:
( 0) (i)
 e1   ex   e1   ex 
  (0)     (i)  
 e2  = [ R k ]  e y ,  e2  = [ Rk ]  ey  (40)
       
 e 3 k  ez   e 3 k  ez 
where e3 is the unit vector orthogonal to the XY-plane and [Rk] is the rotation
matrix between the orthogonal unit vectors in the undeformed/deformed configura-
tion and unit vectors in the global system. Therefore the relation between the
deformed and undeformed configurations will be given by:
(i) (0)
 e1   e1 
(i)   (i) (0) T   (0)
{ e }k =  e 2  = [ R k ] [ R k ]  e 2  = [ R *k ] { e } k
   
 e 3 k  e 3 k
Denoting,

l1 l2 l3
[ R *k ] = m1 m2 m3 (41)
n1 n2 n3 k
the transformation matrix [Tk] in Eq. (39) is given by (Ref. [59]):
2 2 2
l1 l2 l3 l1 l2 l2 l3 l3 l1
2 2 2
m1 m2 m3 m1 m2 m2 m3 m3 m1
2 2 2
[ Tk ] = n1 n2 n3 n1 n2 n2 n3 n3 n1 (42)
2l 1 m 1 2l 2 m 2 2l 3 m 3 l 1 m 2 + l 2 m 1 l 2 m 3 + l 3 m 2 l 3 m 1 + l 1 m 3
2m 1 n 1 2m 2 n 2 2m 3 n 3 m 1 n 2 + m 2 n 1 m 2 n 3 + m 3 n 2 m 3 n 1 + m 1 n 3
2n 1 l 1 2n 2 l 2 2n 3 l 3 n1 l2 + n2 l1 n2 l3 + n3 l2 n3 l1 + n1 l3
5 Because every point within an element undergoes a different rotation, the

material matrix at each point has to be transformed as indicated by Eq. (39)
whenever the material properties are orthotropic or laminate.
Implementation The implementation into StressCheck of the transformation of the global mate-
rial matrix to account for the deformed configuration, requires two consider-
ations:
• During the nonlinear solution, the transformation has to be performed
for each integration (Gauss) point within each element. The location of
each point is known a priori, and depends only on the type of element
and on the polynomial order of the element.
• During post-processing operations, the updated material matrix has to
be computed at any specified extraction location.
In the first case, the strategy used was to store the material matrix for each
point for the (i-1) iteration, and use that matrix to compute the updated matrix
for the next iteration as follows:
(i) T (i – 1)
[ Ek ] = [ Tk ] [ Ek ] [ Tk ] (43)
where [Tk] is computed as in Eq. (47) and Eq. (41) is determined from:

(i) (i – 1) T
[ R *k ] = [ R k ] [ R k ] (44)
(i)
To compute the rotation matrix [ R k ] , the vectors tangent to the curvilinear coordi-
nates at point k are computed first, based on the current mapping for the element
(see Figure 192). The Jacobian matrix, which expresses the relation between the
b
v e2
e3 . e1 a u
Y k
w Ω(i) 5
ey
X
ex
Z ez
FIGURE 192 Tangent vectors in the deformed configuration.
curvilinear and global coordinates, is related to the tangent vectors as follows:
∂x ∂y ∂z
∂u ∂u ∂u  a   ex 
(i) ∂x ∂y ∂z ,   (i) 
[ Jk ] =  b  = [ J ]k  ey  (45)
∂v ∂v ∂v    
 c k  ez 
∂x ∂y ∂z
∂w ∂w ∂w
The tangent vector a is normalized and the orthogonal vector e 3 is then deter-
mined by cross product of the normalized vector and the b vector. The resulting
vector is then normalized and a third orthogonal vector is determined by cross prod-

uct of the other two. The components of the these three normalized vectors are
the terms in the rotation matrix in Eq. (40). This is summarized in the follow-
ing:
a e1 × b
e 1 = ----- , e 3 = ----------------- , e2 = e3 × e1
a e1 × b
(i)
e 1x e 1y e 1z
(i)
[ R k ] = e 2x e 2y e 2z
e 3x e 3y e 3z
(46)
(i)
Once the rotation matrix [ R k ] is known for two iterations, it is possible to
compute the updated material matrix at point k using Eq. (43) and (44).
5
A similar procedure is used during the post-processing operation, except that
the updated material matrix is computed from the transformation between the
nonlinear solution and the undeformed configuration, and has to be computed
every time an extraction is requested.
The material stiffness matrix update is performed automatically whenever a

general nonlinear analysis is selected and the material properties for the ele-
ment are orthotropic, transversely isotropic or laminate.
Two-dimensional relations
Two-dimensional formulations are frequently used in the solution of engineer-
ing problems. In the following the stress-strain relationships are given explic-
itly for the cases of plane strain, plane stress for isotropic and orthotropic
materials, and the Reissner-Mindlin model for isotropic plates.
Plane-strain A state of plane-strain exists when
ux = u x ( x, y ), uy = u y ( x, y ) , uz = 0

where ux, uy, uz are the displacement components along the global x, y, z axes. From
the three-dimensional strain-displacement relations of the small strain, small dis-
placement theory we have:
εz = γ yz = γ xz = 0 .
For orthotropic and transversely isotropic materials,
τ xz = τ yz = 0
σz = Q 13 ε x + Q 23 ε y + Q 36 γ xy
and
σx Q 11 Q 12 Q 16 εx
5
σy = Q 22 Q 26 εy
τ xy sym Q 66 γ xy
Only rotation about the z-axis were considered in the above. For plane strain condi-
tions, only 7 engineering constants are required for orthotropic materials: E11, E22,
E33, v12, v23, v13 and G12. For additional information see Ref. [8], page 88.
For the case of isotropic materials
τ xz = τ yz = 0
σz = ν ( σx + σy )
and
where E is the Young’s modulus and v is the Poisson’s ratio.
Plane-stress A state of plane-stress exists when

σx 1 –ν ν 0 εx
E –ν
σy = --------------------------------- 1 0 εy
( 1 + ν ) ( 1 – 2ν )
1 – 2ν
τ xy sym ------------ γ xy
2
σz = τ yz = τ xz = 0.
The transverse normal strain εz is obtained from (25) with the stiffness matrix
(26). See Ref. [39], page 16:
13 Q 23 Q 36 Q
εz = – ------
- ε x – ------ - γ xy .
- ε y – ------ (47)
Q 33Q Q33 33
5
Substituting equation (47) into the generalized Hooke’s law in terms of Qij, we
have for orthotropic and transversely isotropic materials:
2
Q 13 Q
13 23 Q Q13 36Q
Q 11 – ------
Q
- Q 12 – -------------
Q
- Q 16 – -------------
Q
-
33 33 33
σx εx
2
Q 23 Q Q
σy = Q 22 – ------
- Q 26
23 36
– -------------
- εy (48)
Q 33 Q 33
τ xy γ xy
2
Q 36
sym Q 66 – ------
Q
-
33
Only rotation about the z-axis was considered above. For plane-stress condi-
tions, only 4 engineering constants are required for orthotropic materials: E11,
E22, G12 and v12.

For isotropic materials
ν
εz = – E-- ( σ x + σ y )
and
σx 1 ν 0 εx
E 1 0
σy = ----------------
2
- εy .
(1 – ν ) 1–ν
τ xy sym ---------
- γ xy
2
Reissner-Mindlin plate The Reissner-Mindlin plate has a special stress-strain relationship which is similar
to the case of plane stress, yet the assumed mode of the deformation is such that the
strain is zero in the transverse direction. This relationship ensures that as the plate 5
thickness approaches zero the solution of the Reissner-Mindlin plate converges in
energy norm to the solution of the Kirchoff plate model. This point is discussed in
Section 17.3 of Ref. [3]. The material stiffness matrix of the Reissner-Mindlin
plate, also known as the Reissner-Mindlin matrix, is displayed below for isotropic
materials:
1 ν 0 0 0
σx 1 0 0 0 εx
σy 1–ν εy
E ---------
- 0 0
τ xy = ----------------
- 2 γ xy
( 1 – ν2 ) 1–ν
τ yz sym κ ( ---------- ) 0 γ yz
2
τ zx γ zx
1–ν
κ ( ---------- )
2
The symbol κ represents the shear correction factor. When the thickness of the plate
is small, as compared with other dimensions, the solution is virtually independent
of the shear correction factor. In general, the shear correction factor should be cho-

Strains and stresses for planar elasticity
sen in accordance with the discussion in “Hierarchic models for plates” on

page 229 (See “Remark” on page 236: Optimal shear correction factors).
Strains and stresses for planar elasticity

The strain values are computed directly from the finite element solution. That
is, the strains are obtained from the displacement vector components by differ-
entiation.
∂
0
εx ∂x
ux
{ε} = εy = 0
∂
.
∂y uy
γ xy ∂ ∂
5
∂y ∂x
For plane-strain: ε z = 0 . For plane-stress and isotropic materials: .

–ν
ε z = ------------ ( ε x + ε y )
1–ν
The stresses are computed from:
σ E1 E2 E4 ε
x x
σ
y = E2 E3 E5 ε
y
τ E4 E5 E6 γ
xy xy
In contracted notation:
{ σ} = [E]{ε}

Fracture mechanics
where [E] is the material stiffness matrix. The principal stresses σ1, σ2, equivalent
stress σe (or von Mises stress) and related quantities are computed from the follow-
ing expressions:
2
x – σ y
 σ--------------
τ max = - + τ2xy
 2 
σx + σy
σ 1, 2 = ------------------ ± τ max
2
1 2 2 2
σe = --- [ ( σ 1 – σ 2 ) + ( σ 2 – σ 3 ) + ( σ 3 – σ 1 ) ]
2
For plane-stress: 5
σz = σ3 = 0.
For plane-strain and isotropic materials:
σz = σ3 = ν ( σ1 + σ2 ) .
Fracture mechanics
In the crack tip the solution of the linear problem is singular and the stress values
are infinity. Whether or not a crack will propagate, and at what rate it will propa-
gate, depends on the energy available to drive crack extension. The available
energy is characterized by the stress intensity factors and the J-integral which
depend on the geometry of the body, the configuration of the crack, the boundary
conditions and the loading. This section describes the procedures implemented in
StressCheck for the computation of fracture mechanics parameters for two- and
three-dimensional problems.

Fracture mechanics
The J-integral
The J integral is a measure of the energy release rate for cases where plasticity
effects are not negligible. In the case of plane-stress and plane-strain the J-inte-
gral is defined by:
∫  Wdy – T∂x ds

∂u
J = (49)
Γ
where Γ is a contour around the crack tip, W is the strain energy density:
1
W = W ( x, y ) = --- ( σ x ε x + σ y ε y + τ xy γ xy )
2
T is the tension vector (with components Tx, Ty) perpendicular to Γ , u is the

5 displacement vector (ux, uy) and ds is an element of Γ . The coordinate system
is located such that the origin is in the crack tip, the x-direction is parallel to the
crack face. (See FIGURE 193).
If the integration is performed along a circular path of radius ρ centered at the

crack tip, equation (49) can be written as:
π
 ∂ux ∂uy 
J =
∫  W cos θ – Tx
 ∂x
– Ty
∂x 
 ρ dθ
–π

Fracture mechanics
y
Ty
ds Tx
ρ
dθ
σy
θ σx
x
crack
FIGURE 193 Crack tip region

5
Scope and limitations The J integral can be computed for linear and elastic-plastic solution of isotropic
materials and for linear solution of orthotropic materials. For elastic-plastic solu-
tions, the integration path should be selected in such a way that it does not cut
through the plastic zone around the crack tip. In the plastic region the strain energy
density is not defined.
For planar analysis (plane-stress and plane-strain) the crack can have any orienta-
tion with respect to the xy-coordinate system. For axisymmetric problems, the cur-
rent implementation (based on Ref. [44]) requires the crack to be parallel to the
global R-axis. The J-integral is not implemented for three-dimensional problems.
The computation of the J integral is not valid when thermal loading and/or body
forces have been specified for the elements around the crack tip.
Stress intensity factors

The stress intensity factor can be computed from the finite element solution for iso-
tropic materials using the contour integral method as discussed in Ref. [3], page
227. In the case of plane-stress and plane-strain, the stress intensity factors in mode
I (opening) and mode II (shear) are given by:

Fracture mechanics
(1) (2)
KI = 2πA 1 , K II = 2πA 1
(m)
where A 1 , m = 1, 2, is the first term in the asymptotic expansion of the solu-
tion in the neighborhood of the crack tip.
Let Γ be a circle of radius ρ centered on the crack tip, and assume that ρ is suf-
ficiently close to the crack tip. Then we can write:
∫
(W )
A1
(m)
≈ ( Wm T FE – u FE T m
) ds (50)
where Wm is an extraction function, TFE is the traction vector along Γ com-

5 puted from the finite element solution uFE, and T(Wm) is the traction vector
along Γ due to the extraction function.
The terms in (50) obtained from the finite element solution are:
T σ
x
cos θ + τ sin θ
x xy
T FE = = , along Γ
T
y FE τ cos θ + σ y sin θ
xy FE
and
u
x
u FE = , along Γ .
u
y FE
The terms due to the extraction function can be computed as follows:

1
– ---
2
ρ (m)
W m = ----------
(m)
{Ψ (θ)}
D

Fracture mechanics
3
– ---
2
( Wm ) Gρ
T (m)
- { ϒ(m) ( θ ) }
= – -----------
D
(m)
where G is the modulus of rigidity, D is given by
D
(1)
= π ( 2κ – 1 ) , D
(2)
= π ( 2κ + 3 )
and κ depends on the Poisson’s ratio. For plane-strain:
κ = κ 1 = ( 3 – 4ν )
and for plane-stress:

5
– ν-
κ = κ 2 = --------- 3
1+ν
Also
--- cos θ
κ – 1 1 3θ
--- – --- cos ------
(1)  2 2 2 2
{Ψ (θ)} =
--- sin θ
κ + 1 1 3θ
--- – --- sin ------
 2 2 2 2
( κ + -32 ) sin θ- + 1-2 sin 3θ

----
2 2
{Ψ (θ)} =
(2)
– ( κ – 32- ) cos θ- – 1-2 cos 3θ

----
2 2

Fracture mechanics
 3--- cos θ 1 5θ 1 5θ- 1

- cos θ +  --- sin -----
--- + --- cos ----- θ
– --- sin --- sin θ
(1) 2 2 2 2 2 2 2 2
{ϒ (θ)} =
 1--- sin 5θ 1 θ 5 θ 1 5θ
------ – --- sin --- cos θ +  --- cos --- – --- cos ------ sin θ
2 2 2 2   2 2 2 2
Extension to 3D In three-dimensions, the stress intensity factors KI and KII are computed using
the contour integral method over a circular path around the crack front at the
pick point. The procedure determines a cutting plane normal to the tangent to
the crack edge at the pick point and extract the global components of the
stresses and displacements along a circular path contained in the cutting plane.
These stresses and displacements are projected into the cutting plane and inte-
grated with the extraction function to compute the contour integral as
described above.
5 In general, a three-dimensional problem is neither a plane-stress nor a plane-
strain problem. Therefore the question is how to determine the value of κ for
the extraction functions. The procedure to determine which value of κ to use is
based on computing the stress components σx, σy, σz at a point in front of the
crack (see FIGURE 193). Next define γ as the ratio:
σz
γ = -----------------
σx + σy
If γ is close to Poisson’s ratio (ν), then the condition approaches that of plane-
strain. If is close to zero, then is plane-stress. The approach implemented in
Stress Check was to compute the value of γ at a point along the integration path
(r = ρ, θ = 0), and apply the following rule to determine κ:
 κ if ( γ > γ 1 )
κ =  1
 κ ( γ ) if ( γ < γ 1 )
where γ1 = 0.1 and κ(γ) is determined from a cubic spline fitting between the
values of plane-strain and plane-stress:
2γ γ 2
κ ( γ ) = κ 1 + ( κ 2 + κ 1 )  ----- + 1  1 – ----
 γ1  γ 1

Fracture mechanics
7 1 1 3
( – -2 sin θ- – -2 sin 5θ ---- + -2 cos θ- ) sin θ
---- ) cos θ + ( -2 cos 5θ
2 2 2 2
{ ϒ(2) ( θ ) } =
1 3 1 1
---- + -2 cos θ- ) cos θ + ( – -2 sin θ- + -2 sin 5θ
( -2 sin 5θ ---- ) sin θ
2 2 2 2
With this procedure the CIM was extended to compute the stress intensity factors
along any point along curved crack fronts.

Scope and limitations The stress intensity factor can be computed for the linear solution of isotropic
material in the absence of body forces and thermal loading. For axisymmetric
problems, the stress intensity factor depends on the radius of the integration
path (r). However, as r → 0 the contour integral becomes path independent.
Numerical studies (Ref [45]) indicated that for r/rc < 0.1, the values of the
stress intensity factors are practically independent of r (rc = crack tip location).
Note: When extracting stress intensity factors for tetrahedral meshes generated
with the automatic mesh generator for embeded cracks, it may be required to
create the parameter _crack_360 (no value is required) to overcome some diffi-
culties in computing the solid angle when elements are very distorted.

The concept of stress intensity factors associated with crack tips in linear elas-
5 tic fracture mechanics can be extended to general singular points in two-
dimensions and singular edges in three-dimensions for composites, anisotropic
materials, and multi-material interfaces for linear elastostatics and steady state
heat transfer problems.
The mismatch of the physical properties cause flux and stress intensification at
the corners of interfaces and can lead to mechanical failures. These material
interfaces, as well as crack tips, are called singular points/edges because the
temperature flux is infinity in the linear theory of steady-state heat conduction
and so are the stresses in the linear theory of elasticity. Typical singular points
where failures initiate and propagate in an electronic device are shown in FIG-
URE 194.
New approaches to predicting the initiation and extension of delamination in

plastic encapsulated LSI (Large Scale Integrated Circuit) devices, for example,
are based on the computation of certain functionals, called the generalized
flux/stress intensity factors (GFIFs/GSIFs), the strength of the stress singular-
ity, and in thermo-elastic problems the thermal stress intensity factors (TSIFs).
Since the stress and strain components are generally not bounded, it is not pos-
sible to construct failure initiation events with them. A key requirement for for-
mulating failure laws is a software-tool which computes reliably the GFIFs/
GSIFs and the strength of the singularity. The reliability and accuracy of
numerical results is an essential requirement for proper evaluation and formu-

Crack
Encapsulant
Si Chip Singular points
Solders
Substrate Printed Circuit Board
FIGURE 194 Typical sites of failure initiation in an electronic device.
lation of failure theories through experimental observation of failure initiation

5
events.
Introduction and Notation

In the neighborhood of singular points the exact solution of two-dimensional elas-
tostatic problems for example, can be expanded in the form:
∞ αi
{u} = ∑ Ai r { Φi ( θ ) } (51)
i=1
where {u} is the displacement vector with components ux(x,y), uy(x,y), r and θ are
polar coordinates centered on the singular point; αi are called eigenvalues and φi(θ)
are called eigenfunctions. These eigenpairs (αi, φi) depend on the material proper-
ties, the geometry and the boundary conditions [φi(θ) are smooth vector functions].
The Ai are coefficients which depend on the loading. Because of their close analogy
to stress intensity factors in linear elastic fracture mechanics, Ai are called general-
ized stress intensity factors (GSIFs). In the case of heat transfer problems they are
called generalized flux intensity factors (GFIFs).
The stresses in the same neighborhood can be computed from the displacements
given by Eq. (51) and the material properties as:

∞ αi – 1
{σ} = ∑ Ai r { ψi ( θ ) } (52)
i=1
where ψi(θ) depend on the eigenfunctions φi(θ) in Eq. (51) and the material
coefficients. It is clear from Eq. (52) that when αi < 1, the stresses become sin-
gular for r=0. For additional details see Ref. [58] and [59].
Three-dimensional singularities are considerably more difficult to analyze than

two-dimensional ones, where only one type of singularity exists. In 3D in a
neighborhood of the edges and the vertices the singular behavior is different.
Edge Singularities: If a coordinate system (x,y,z) is located at an edge, with the

y-axis along the edge, then there are three edge GSIFs which are y-dependent:
AI(y), AII(y) and AIII(y). These edge GSIFs are analytic along the edge, how-
ever they become themselves singular when this edge intersects with a free
5 plane, at a vertex. In the neighborhood of an edge-vertex type geometry the
GSIFs can be represented once again by vertex and vertex-edge stress intensity
γ
factors. For example, A I ( y ) =
∑
SI, j y I, j + smoother terms.
j
Vertex Singularities: In the neighborhood of a vertex, and away from edge-ver-
tex geometry, the displacement field can be represented by only one vertex
intensity factor and the corresponding eigenpairs. Investigating the mathemati-
cal behavior of the singularities in 3-D is an active field of research in the
mathematical community, and the decomposition of the displacement field into
singular and regular terms is documented in some recent papers.
Formulation
The solution of second order elliptic boundary value problems (BVP) in three-
dimensions in the vicinity of any singular point, can be decomposed into three
different forms, depending whether the singular point is in the neighborhood of
an edge, a vertex or an intersection of the edge and the vertex. Mathematical
details of the decomposition can be found e.g. in Ref. [60]-[63] and the refer-
ences therein. A representative three-dimensional domain denoted by Ω, which
contains typical 3D singular points is shown in FIGURE 195. Vertex singulari-
ties arise in the neighborhood of the vertices Ai, and the edge singularities arise
in the neighborhood of the edges singularities Λij. Close to the vertex/edge
intersection, vertex-edge singularities arise.

A9 A10
A2 A5
Λ12
A8 A7 Λ13
A1
Λ14 A3
A11
A4 A6
Ai - Vertex i
Λij - Edge between Ai and Aj
A13 A12
FIGURE 195 Typical 3D singularities.
Consider the simplest elliptic BVP, the Laplace equation:
∇ 2u = 0 in Ω (53)
u = 0 on ΓD ⊂ ∂Ω, ∂u on ΓN ⊂ ∂Ω (54)
= 0
∂n
where u(x1, x2, x3) denotes the temperature field (in the following x1, x2 and x3 will
be either Cartesian, cylindrical or spherical coordinates), and Γ D ∪ Γ N = ∂Ω . It
shall be assumed that curved edges that intersect at vertices do not exist, at that
crack faces, if any, lie in a plane.
Edge Singularities: We first examine the edges denoted by Λij which connect the
vertices Ai and Aj. Moving away from the vertex a distance δ/2, and creating a
cylindrical subdomain of radius r = R with the edge Λij as its axis, we define a sub-
domain in the vicinity of the edge denoted by ε δ, R ( Λ ij ) . FIGURE 196 shows the

edge singularity subdomain ε δ, R ( Λ 12 ) . We restrict our attention to domains

having straight edges.
A2
R
δ/2 z
ω12
Λ12
y
δ/2 r
5 θ
A1
x
FIGURE 196 The edge neighborhood ε δ, ε ( Λ 12 ) .
The solution in εδ,R can be decomposed as follows:
K S
α s
u (r,θ, z) =
∑ ∑a ks ( z ) r k ( ln r ) f ks ( θ ) + v ( r, θ, z ) (55)
k = 1s = 0
where S ≥ 0 is an integer which is zero unless ak is an integer, αk+1 ≥ αk are

called edge eigenvalues; aks(z) are called the edge flux intensity functions
(EFIFs), are analytic in z but can become very large as they approach one of
the vertices; and fks(θ), called eigenfunctions, are analytic in θ. The function
v(r,θ,z) belongs to Hq(ε), the Sobolev space, where q can be as large as
required and depends on K. We shall assume that αk for k ≤ K is not an integer,
therefore Eq. (55) becomes:

K
αk
u (r,θ, z) =
∑ a (z) k r f k ( θ ) + v ( r, θ , z ) (56)
k=1
Vertex Singularities: A sphere of radius ρ = δ, centered in the vertex A1 for exam-

ple, is constructed and intersected by the domain Ω. Then, a cone having an open-
ing angle φ = σ is constructed along every edge intersecting at A1, and removed
from the previously constructed subdomain, as shown in FIGURE 197.
δ
Λ12
ρ
Λ13
A1 5
Λ14
σ
FIGURE 197 The vertex neighborhood Vδ(A1).
The resulting vertex subdomain is denoted by Vδ(A1), and the solution u can be
decomposed in Vδ(A1) using a spherical coordinate system by:
L Q
γ q
u (ρ,φ, θ) =
∑∑b lq ρ l ( ln ρ ) h lq ( φ, θ ) + v ( ρ, φ, θ ) (57)
l = 1q = 0
where Q ≥ 0 is an integer which is zero unless γl is an integer, γl+1 ≥ γl are called

vertex eigenvalues, and hlq(φ,θ), called the eigenfunctions, are analytic in φ and θ.
The blq are called vertex flux intensity factors (VFIFs). The function v(r,θ,z)
belongs to Hq(V), where q depends on L. We shall assume that γl for l ≤ L is not an
integer, therefore, Eq. (57) becomes:

L
γl
u (ρ,φ, θ) =
∑b l ρ h l ( φ , θ ) + v ( ρ , φ, θ ) (58)
l=1
Vertex-Edge Singularities: The most complicated decomposition of the solu-

tion arises in case of vertex-edge intersections. For example, let us consider the
neighborhood where the edge Λ12 approaches the vertex A1. A spherical coor-
dinate system is located in the vertex A1, and a cone having an opening angle
φ=σ with its vertex coinciding with A1 is constructed with Λ12 being its center
axis. This cone is terminated by a ball-shaped basis having a radius ρ=δ, as
z
θ
5
φ ω12
δ Λ12 σ
A1
FIGURE 198 The vertex-edge neighborhood Vεδ,ε(A1, Λ12).
The resulting vertex-edge subdomain is denoted by Vεδ,ε(A1,Λ12), and the

solution u can be decomposed in Vεδ,ε(A1,Λ12):

K S  L 
 γ α
a ksl ρ l + m ks ( ρ ) ( sin φ ) k [ ln ( sin Φ ) ] g ks ( θ ) +
∑∑ ∑
s
u (ρ,φ, θ) =
 
k = 1 s = 0 l = 1 
(59)
L Q
γl
∑∑c
q
lq ρ ( ln ρ ) h lq (φ,θ) + v (ρ,φ, θ)
l = 1q = 0
where mks(ρ) is analytic in ρ; gks(θ) is analytic in θ, and hlq(φ,θ) is analytic in φ and

θ. The function v(r,θ,z) belongs to Hq(Vε) where q can be as large as required
depending on L and K. Again we shall assume that γl for l ≤ L is not an integer, and
αk for k ≤ K is not an integer, therefore, Eq. (59) becomes:
K  L  L
 γ α γ
a kl ρ l + m k ( ρ ) ( sin φ ) k g k ( θ ) +
u (ρ,φ, θ) =
∑ ∑   ∑ c l ρ l h l (φ,θ) + v (ρ,φ, θ) (60)
5
k = 1 l = 1  l=1
The eigenvalues and the eigenfunctions are associated pairs (eigenpairs) which
depend on the material properties, the geometry, and the boundary conditions in the
vicinity of the singular point/edge only. Similarly, the solution for problems in lin-
ear elasticity, in the neighborhood of singular points/edges is analogous to Eq. (55)-
Eq. (60), the differences are that the equations are in vector form and the eigenpairs
may be complex. For general singular points the exact solution uEX is generally not
known explicitly, i.e., neither the exact eigenpairs nor the exact EFIFs, VFIFs are
known, therefore numerical approximations must be found.
Implementation
Eigenpairs for edge singularities in heat transfer problems Separation of variables was utilized for the
Laplace equation in the neighborhood of the edge singularity as shown in Eq. (56).
Each eigenpair rαkfk(θ) is independent of z and satisfies the Laplace equation over
the plane (r,θ) which is perpendicular to the edge. This is exactly a two-dimen-
sional problem and therefore the computation of eigenpairs was performed using
the modified Steklov method described in Ref. [44] and summarized below.
The algorithm developed for two-dimensional heat-transfer problems was extended

to three-dimension by considering a slice contained in the plane normal to the edge

singularity. The region denoted by Ω∗R in FIGURE 199, represents a cross sec-
y
Material 2 Material 1
Γ3
Material 3
Ω* R
Γ1
θ1 R* R
ω
Γ2
Γ4 x
Singular Edge
5
FIGURE 199 Cross section of an edge singularity

for the modified Steklov formulation.
tion along the edge singularity. The edge is assumed to be along the local z
axis, and (r, θ) are the coordinates of a cylindrical system located at the inter-
section of the normal plane and the singular edge. The area around the singu-
larity is internally divided into sub-domains.
This ‘2D-internal mesh’ is arranged in a circular ring around the singularity in

such a way that the element boundaries coincide with the material interfaces as
The number of elements of the internal mesh is controlled by the number of

material interfaces around the singularity. The largest solid angle for a single
element is limited to 120o. Therefore, if the partition of a single material is
larger than 120o, the element is divided in two or more partitions. For each ele-
ment of the ‘internal mesh’ the corresponding stiffness and trace matrices are
computed, and after assembly, the following eigenvalue problem is obtained:
[ KS ] { uR } = α [ M ] { uR }

Material 2 Material 1
Elem. 2
Material 3
Elem. 1
Solid Angle
Elem. 3
Singularity
FIGURE 200 Typical ‘internal mesh’ around the singularity. 5
where [KS] is the condensed stiffness matrix, [M] is the trace matrix computed by
integration on the circular boundary segments on Ω∗R, and {uR} is the vector of
coefficients that correspond to the degrees of freedom associated with the circular
boundaries of the ‘internal mesh’. The solution of the eigen-problem given by the
above equation yields approximation for the eigenvalues αi and the corresponding
eigenvectors. The steps and fundamentals for obtaining the system described by Eq.
() are described in the following.
By formulating the weak form over Ω∗R, the singularity is excluded from the
domain of interest such that the accuracy of the finite element solution does not
deteriorate in its vicinity. On the boundaries Γ1 and Γ2 consider either zero temper-
ature or zero flux boundary conditions:
∂u
u = 0 or = 0 on Γi , i = 1, 2. (61)
∂n
In Ω∗R, the temperature field u can be represented as follows:
α
u = r f(θ) (62)

Differentiating Eq. (62), and considering the boundary Γ3 we can write:
∂u
= ( α ⁄ R )u (63)
∂r
and a similar condition on Γ4.
Multiplying the Laplace equation by a test function v in H1(Ω*R), integrating

over the domain Ω∗R and using Green's theorem, the modified Steklov weak
form is obtained:
1 *
Seek α ∈ ℜ, 0 ≠ u ∈ H ( ΩR )
1 *
B (u,v) – ( N R (u,v) + N * (u,v) ) = α ( M R (u,v) + M * (u,v) ), ∀ v ∈ H ( ΩR ) (64)
R R
where
5
∫ ∫ ( [ D ]v ) [ E ] ( [ D ]u ) dΩ,
T
B (u,v) = (65)
*
ΩR
∫ [ v [ Q ] [ E ] [ Q ]u ]
T
M R (u,v) = r=R dθ, (66)
θ
∫ [ v[ Q] [ E]( [D
T (θ)
N R (u,v) = ]u ) ] r = R dθ, (67)
θ
and [D], [D(θ)], [Q] and [R] are given as follows:
∂ ∂
( – sin θ )
[D] = ∂ x , [Q] = cos θ , [D
(θ)
] = ∂θ
(68)
∂ sin θ ∂
cos θ
∂y ∂θ
[E] is the 2D material matrix reduced from the 3D material matrix.
Remark 1. The domain Ω∗R does not include singular edge, hence no special
refinement of the finite element mesh is required.

Remark 2. The formulation of the weak form was not based on the assumption that
the material is isotropic, and in fact can be applied to multi-material anisotropic
interface.
The domain Ω∗R is divided into finite elements through a meshing process, as
described before. The polynomial basis and trial functions, {Ψj}, are defined on a
standard element in the ξ, η space such that -1< ξ <1, -1 < η <1. The temperature
field is then expanded in terms of the known polynomial shape functions multiplied
by a set of unknown coefficients {utot}:
u = u tot { Ψ } (69)
The entries of the unconstrained stiffness matrix corresponding to B(u,v) are given
by (see Ref. [3]):
∫ ∫ ( [ D ] { Ψ } ) [ E ] [ D ] { Ψ } dΩ
T
K ij = i j (70)
*
ΩR 5
Similar expressions are obtained for the matrices [NR], [NR*] and [MR], [MR*] asso-
ciated with the bilinear forms NR(u,v) and MR(u,v).
Denoting the set of all unknown coefficients by {utot}, and the set of coefficients
associated with Γ3 and Γ4 by {uR}, the following eigen-problem is obtained:
( [ K ] – [ N R ] – [ N R * ] ) { u tot } = α ( [ M R ] + [ M R * ] ) { u R } = α [ M ] { u R } (71)
The vector which represents the total number of nodal values in Ω∗R can be divided
into two vectors such that one contains the coefficients {uR}, the other contains the
remaining coefficients: {utot}T = {{uR}T, {uin}T}. By eliminating {uin}, the reduced
eigen-problem is obtained:
[ KS ] { uR } = α [ M ] { uR } (72)
The solution of the eigen-problem given by Eq. (72) yields approximations for
eigenpairs with high accuracy, efficiency and robustness.
The procedure implemented in Stress Check for the computation of the eigenpairs
requires the user only to click with the mouse cursor along the singular edge. The
program then determines a cutting plane normal to the singular edge at the pick

point and creates the partitions indicated in FIGURE 200 based on the material
properties of the elements intersecting the cutting plane. Once these partitions
are identified, the elemental matrices are computed and the system of equa-
tions is assembled and solved.
Generalized flux intensity factors for edge singularities in heat transfer problems The algorithm for the
computation of the GFIFs is based on L2 projection of the finite element solu-
tion into the space of functions characterized by the asymptotic expansion in
terms of the eigenpairs. The algorithm can be summarized as follows:
Consider the domain around a singular edge for a heat transfer problem shown
in FIGURE 201. The temperature field can be expanded around the singular
edge in terms of the eigenpairs and the generalized flux intensity factors:
N
αi
u (r,θ) =
∑A r i fi ( θ ) = φ { A } (73)
i=1
5
where Ai are the GFIFs, αi are the eigenvalues and fi(θ) are the corresponding
eigenfunctions, with φi=rαifi(θ).
Let uFE(r,θ) be the finite element solution of the temperature distribution

around the singular edge. Then the L2 projection of uFE into the domain ΩR is
characterized by the following:
∫ (u – u
ΩR
FE )v dA = 0, for all v ∈ S (74)
where S is the space of eigenfunctions, and v(r,θ) is a test function given by:
N
αi
v (r,θ) =
∑B r i fi ( θ ) = B {φ} (75)
i=1
Substituting Eq. (73) and Eq. (75) into Eq. (74) and rearranging we have:
B
∫ {φ}
ΩR
φ dA { A } = B
∫ { φ }u
ΩR
FE dA (76)

Ω
mat 2 θ2 r
ΓR
mat 1
P θ
R1 X
ΩR R2
θ1
mat 3
5
FIGURE 201 Typical cross-section of an edge singularity through point P.
Eliminating Bi from Eq. (76) and noting that dA=rdrdθ, the following system is
obtained:
R2 θ2 R2 θ2
∫ ∫ {φ} φ r dr dθ { A } =
∫ ∫ { φ }u FE r dr dθ (77)
R1 θ1 R1 θ1
From the definition of {φ} in Eq. (73) and integrating in the radial direction, the
system of equation reduces to:

K ij A j = R j
( αi + αj + 2 ) ( αi + αj + 2 ) θ2
R2 – R1
2 + αi + αj ∫
K ij = --------------------------------------------------------- f i ( θ )f j ( θ ) dθ
θ1
(78)
R2 θ2
( αj + 1 )
Rj =
∫ ∫r f j ( θ )u FE dr dθ
R1 θ1
Solving the system of equations represented by Eq. (78) over the domain ΩR
gives the values of Ai (the GFIFs). The L2 projection provides excellent
approximation when the generalized intensity factor along the singular edge is
either constant or linear, but the approximation deteriorates for cases in which
the variation is of higher order. Nevertheless the current implementation pro-
vides a very robust algorithm for a large class of practical problems.
5
Eigenpairs for edge singularities in elasticity problems The elastostatic displacement field in three-
dimensions in the vicinity of an edge singularity can be decomposed in terms
of the eigenpairs and the corresponding stress intensity functions. Mathemati-
cal details on the decomposition are found in Ref. [60], [62] and the references
therein. Elastic edge singularities associated with anisotropic materials and
multi-material interfaces have been less investigated, however. Analytical
methods as in Ref. [64], [65] provide procedures for the computation of eigen-
pairs, but require very involved mathematical operations.
In the following, a description of the numerical procedure implemented in

Stress Check for the computation of eigenpairs for edge singularities in elastic-
ity is addressed. In the neighborhood of a typical edge singularity the displace-
ment fields can be written as:
K S
α s
u (r,θ, z) = ∑ ∑a ks ( z ) r k ( ln r ) f ks ( θ ) + w ( r, θ, z ) (79)
k = 1s = 0
where S ≥ 0 is an integer which is zero for most practical problems, and aks(z)
are analytic in z and are called the generalized stress intensity functions. The

vector function w(r,θ,z) belongs to [Hq(εδ,R)]3, where q depends on K. Considering

the case S = 0, Eq. (79) becomes:
K
αk
u (r,θ, z) =
∑ a (z)
k r f k ( θ ) + w ( r, θ , z ) (80)
k=1
T
The tractions on the boundaries are denoted by T = ( T x T y T z ) , and the Carte-
sian stress tensor by σ = σ x σ y σ z τ xy τ yz τ zx . Body forces are not considered
in the vicinity of the singular edge.
For computing the eigenpairs, a two-dimensional sub-domain ΩR is constructed in

a plane perpendicular to the edge bounded by r=R and r=R* as shown in FIGURE
202. On the boundary θ=0 and θ=ω12 of the sub-domain ΩR homogenous tractions,
A2
R 5
z R*
ω12
Λ12
r ΩR
A1
FIGURE 202 The modified Steklov domain ΩR.
homogeneous displacement constraints or a combinations of these are prescribed.

In view of Eq. (80) the displacement vector in ΩR has the following functional
representation:
fx ( θ )
λ λ
u = a ( z )r f y ( θ ) = a ( z )r f ( θ ) (81)
fz ( θ )
and the in-plane variation of the displacements is written as:
u
u ( r, θ ) = ---------- (82)
˜ a ( z)
Following the steps presented in detail in Ref. [56], an eigenvalue problem is

cast in a weak form over a two dimensional domain involving all three dis-
placement components:
5 1 3 1 3
Seek α ∈ C, 0 ≠ u ∈ [ H ( ΩR ) ] , such that, ∀v ∈ [ H ( Ω R ) ]
˜ ˜
B ( u, v ) – [ N R ( u, v ) – N R* ( u, v ) ] = λ [ M R ( u, v ) – M R* ( u, v ) ]
˜ ˜ ˜ ˜ ˜ ˜ ˜ ˜ ˜ ˜ (83)
with
ω 12 T
R
 ∂ θ   ∂ 
- v  [ E ]   [ A r ]∂ r + [ A θ ] ----θ- u r dr dθ
B ( u, v ) =
˜ ˜ ∫ ∫
R* 0
  [ A r ]∂ r + [ A θ ] ----
 r ˜
 
 r ˜

ω 12
∫
T T
N R ( u, v ) = ( v ) [ A r ] [ E ] [ A θ ]∂ θ u r=R
dθ
˜ ˜ 0 ˜ ˜
ω 12
∫
T T
M R ( u, v ) = ( v ) [ A r ] [ E ] [ A r ]u r=R
dθ
˜ ˜ 0 ˜ ˜
(84)
where [E] is a 6x6 material matrix with up to 21 independent coefficient for a

completely anisotropic material; ∂r= ∂/∂r, ∂θ= ∂/∂θ; and

cos θ 0 0 – sin θ 0 0
0 sin θ 0 0 cos θ 0
[ Ar ] = 0 0 0 [ Aθ ] = 0 0 0 (85)
sin θ cos θ 0 cos θ – sin θ 0
0 0 sin θ 0 0 cos θ
0 0 cos θ 0 0 – sin θ
Remark 1. The test and trial functions have three components, but the domain over
which the weak eigen-formulation is defined is two-dimensional and excludes any
singularity. Therefore the application of the p-version of the finite element method
(FEM) for solving Eq. (84) is very efficient.
Remark 2. The bilinear forms NR and NR* are non-symmetric with respect to
u and v thus they are not self-adjoint. As a consequence the ‘minimum princi-
˜ ˜
ple’ does not hold, and any approximation of the eigenvalues obtained using a finite
1 3
dimension subspace of [ H ( Ω R ) ] cannot be considered as an upper bound of the 5
exact ones, and the monotonic behavior of the error is lost as well. Convergence is
assured under a general proof provided in Ref. [66], however.
Next consider the discretization of the weak formulation given by Eq. (83) using
1 3
the p-version FEM over a finite dimensional sub-space of [ H ( Ω R ) ] . Assuming
that the domain ΩR consist of three different materials as shown in FIGURE 203,
let us divide it into three finite elements as shown. Let us consider a typical element
bounded by θ1 and θ2 (element 1 in FIGURE 203). A standard element is consid-
ered in the ξ, η plane such that -1< ξ <1, -1 < η <1 over which the polynomial basis
and trial functions are defined. These standard elements are mapped into the real
elements by proper mapping functions (for details, see Ref. [3] chapters 5 and 6).
The functions u and v are expressed in terms of the basis functions Ψi(ξ,η) in
˜ ˜
the standard plane:
ψ1 … ψN 0 … 0 0 … 0 c1
u = 0 … 0 ψ1 … ψN 0 … 0 … = [Ψ]{C} (86)
˜
0 … 0 0 … 0 ψ1 … ψN c 3N
where Ci are the amplitudes of the basis functions and ψi are the products of the
integral of the Legendre polynomials in ξ and η. Since u and v lie in the same
˜ ˜
space, we define similarly v = [ Ψ ] { B } .
˜

Γ3
1 2
ΩR
Γ1
3
R*
θ2 θ1
R
Γ2
Γ4 x
FIGURE 203 Discretization of the domain ΩR.

5
The unconstrained stiffness matrix corresponding to the bilinear form B in Eq.

(84) is given by:
θ2 T
R
 ∂ θ   ∂ 
- [ Ψ ]  [ E ]   [ A r ]∂ r + [ A θ ] ----θ- [ Ψ ] r dr dθ (87)
[K] =
∫ ∫
R*
  [ A r ]∂ r + [ A θ ] ----
θ1  r 
 
 r 

Next we define the matrices:
P 1 cos θ … P N cos θ 0 … 0 0 … 0
0 … 0 P 1 sin θ … P N sin θ 0 … 0
0 … 0 0 … 0 0 … 0
[ P̃ ] =
P 1 sin θ … P N sin θ P 1 cos θ … P N cos θ 0 … 0
0 … 0 0 … 0 P 1 sin θ … P N sin θ
0 … 0 0 … 0 P 1 cos θ … P N cos θ

′ ′
– P 1 sin θ … – P N sin θ 0 … 0 0 … 0
′ ′
0 … 0 P 1 cos θ … P N cos θ 0 … 0
0 … 0 0 … 0 0 … 0
[ ∂P ] = ′ ′ ′ ′
P 1 cos θ … P N cos θ – P 1 sin θ … – P N sin θ 0 … 0
′ ′
0 … 0 0 … 0 P 1 cos θ … P N cos θ
′ ′
0 … 0 0 … 0 – P 1 sin θ … – P N sin θ
where Pi(ξ) for i>3 are the integrals of the Legendre polynomials, and P1(ξ)=(1-ξ)/ 5
2, P2(ξ)=(1+ξ)/2. Therefore, the bilinear form N can be written as:
T 
1
∫
T T
N R ( u, v ) = [ B ]  [ P̃ ] [ E ] [ ∂P ] η = –1
dξ [ C ] = [ B ] [ N R ] [ C ] (88)
˜ ˜  –1 
The entries in [NR] are computed using Gauss quadrature. Similarly, the expression
for the bilinear form M is evaluated as follows:
T  θ2 – θ1 
1
∫
T T
M R ( u, v ) = [ B ]  ---------------- [ P̃ ] [ E ] [ P̃ ] η = –1
dξ [ C ] = [ B ] [ M R ] [ C ] (89)
˜ ˜  2 –1 
The matrices [NR*] and [MR*] have the same values as those of [NR] and [MR], but
of opposite sign. This is because the shape functions on boundaries Γ3 and Γ4 are
the same (except for a sign change), and so is the mapping to the standard element.
Denoting the set of amplitudes of the basis functions associated with the boundary
Γ3 by [CR], and those associated with the boundary Γ4 by [CR*], the eigenpairs can
be obtained by solving the following generalized eigenvalue problem:
[ K ] [ C ] – [ N R ] [ C R ] + [ N R* ] [ C R* ] = λ ( [ M R ] [ C R ] – [ M R* ] [ C R* ] ) (90)

Augmenting the coefficients of the basis functions associated with Γ3 with

those associated with Γ4 and denoting them by the vector [CRR*], Eq. (90)
becomes:
[ K ] [ C ] – [ N RR* ] [ C RR* ] = λ [ M RR* ] [ C RR* ] (91)
The vector that represents the total number of nodal variables in ΩR is divided
into two vectors, one containing the variables [CRR*] and the other the rest of
the variables: [C]T = {[CRR*]T, [Cin]T}. By partitioning the stiffness matrix [K],
Eq. (91) can be written as:
[ K ] – [ N RR* ] [ K RR* – in ] C RR* [ M RR* ] [ 0 ] C RR*

= λ (92)
[ K in – RR* ] [ K in ] C in [ 0 ] [ 0 ] C in
The relation in Eq. (92) is used to eliminate [Cin] by static condensation, thus
obtaining the reduced eigenvalue problem:
5
[ K S ] [ C RR* ] = λ [ M RR* ] [ C RR* ] (93)
where
–1
[ K S ] = [ K ] – [ N RR* ] – [ K RR* – in ] [ K in ] [ K in – RR* ]
In Eq. (93) [KS] is a full matrix. However the order of the matrices is relatively
small, so the solution of the eigenvalue problem (using Cholesky factorization
to compute [Kin]-1) is not expensive. Note that the derivation was performed
for a single element. For multiple elements along Γ3 and Γ4 the formulation is
identical and the matrices [K], [NR] and [MR] are obtained by an assembly pro-
cedure.
Generalized stress intensity factors for edge singularities in elasticity problems As in the case of the
generalized flux intensity factors, the algorithm for the computation of the
GSIFs is based on a L2 projection of the finite element solution into the space
of functions characterized by the asymptotic expansion in terms of the eigen-
pairs. The algorithm can be summarized as follows: Consider a section perpen-
dicular to a singular edge of an elasticity problem as shown in FIGURE 204.

r
ΓR
P θ
R1 X
ΩR R2
FIGURE 204 Typical cross-section of an edge singularity through point P. 5
The displacement field corresponding to the exact solution can be expanded around
the singular edge at point P in terms of the eigenpairs and the generalized stress
intensity factors:
 u x ( r, θ )  N
  λi
u (r,θ) =  u y ( r, θ )

 =

∑A r i fi ( θ ) (94)
 u z ( r, θ )  i=1
where Ai are the GSIFs, λi are the eigenvalues and {f(θ)}={fx(θ), fy(θ), fz(θ)} are
the corresponding eigenfunctions. Let {uFE(r,θ)} be the finite element solution of
the displacement field around the singular edge. Then the L2 projection of {uFE}
into the domain ΩR is characterized by the following:
∫ (u – u FE
)v dA = 0, for all v ∈ S (95)
ΩR

where S is the space of eigenfunctions, and {v(r,θ)}, is given by:

N
λi
v (r,θ) =
∑B r i fi ( θ )
i=1
Substituting {u} and {v} into Eq. (95) and rearranging we have:
B
∫ ({φ } x φ x + { φ y } φ y + { φ z } φ z ) dA { A }=
ΩR (96)
B
∫ ( { φ }u
x FE x + { φ y }u FE y + { φ z }u FE z ) dA
ΩR
where:
5 λ λ λ
{ φ x } = r i { f x }, { φ y } = r i { f y }, { φz } = r i { fz } (97)
Eliminating Bi from Eq. (96) and noting that dA=rdrdθ, the following system is
obtained:
R2 θ2
∫ ∫ {φ } x φ x + { φ y } φ y + { φ z } φ z )r dr dθ { A }=
R1 θ1 (98)
R2 θ2
∫ ∫ { φ }u
x FE x + { φ y }u FE y + { φ z }u FE z )r dr dθ
R1 θ1
From the definition of {φ} in Eq. (97) and integrating in the radial direction,
the system of equation reduces to:

Summary of the main points
K ij A j = R j
( λi + λj + 2 ) ( λi + λj + 2 ) θ 2
R2 – R1
∫
K ij = -------------------------------------------------------
2 + λi + λj
θ1
[ f xi ( θ )f xj ( θ ) + f yi ( θ )f yj ( θ ) + f zi ( θ )f zj ( θ ) ] dθ
(99)
R2 θ2
( λj + 1 )
Rj =
∫ ∫r ( f xj u FE x + f yj u FE y + f zj u FE z ) dr dθ
R1 θ1
Solving the system of equations represented by Eq. (99) over the domain ΩR gives
the values of Ai (the GSIFs). Note that only a line integral is required to compute Kij
and an area integral to compute Rj. The displacement vector from the finite element
solution {uFE} has to be evaluated relative to the local coordinate system located at
the extraction point P.

The assurance of reliability of computed information is a systematic process which
involves the use of hierarchic extensions and hierarchic working models. Discreti-
zation errors are controlled through the creation of hierarchic sequences of finite
element spaces; estimation of the overall error, and showing that the data of interest
are clearly converging to a limiting value.
If only one finite element solution is available then it is impossible to tell how large
the errors of approximation are. In order to ascertain that the errors of approxima-
tion are small, it is necessary to obtain a converging sequence of finite element
solutions and demonstrate that the corresponding data of interest are substantially
independent of the mesh and/or the polynomial degree of elements. StressCheck
was designed with this objective in mind. It is convenient to generate a strongly
converging sequence of finite element solutions and the convergence of the data of
interest is readily ascertained. Specifically:
1. StressCheck provides an estimate of the relative error in energy norm. With the
exception of some special cases, this can be taken to indicate the root-mean-square
error in stress. The reliability of this estimate is very high. Exhaustive investigation
of benchmark problems, for which exact solutions are available, have demonstrated

that the ratio of the predicted error to the true error (the effectivity index) is
very close to unity in all cases.
2. The convergence of the data of interest to a limiting value, with respect to

progressively increasing degrees of freedom, is displayed and tabulated. While
there are no methods for reliably estimating the relative error for data other
than those directly related to the potential energy, in practice the important
question is whether the errors in the data of interest are within an acceptable
range or not. This question can be readily answered on the basis of information
provided by StressCheck.
3. For problems of linear elastic fracture mechanics StressCheck provides a

superconvergent method of computation for the stress intensity factors in a
very user-friendly setting.
Modeling errors are controlled by showing that the data of interest are not sen-
sitive to restrictive assumptions incorporated into the model. StressCheck pro-
vides an unique capability for the control of modeling errors in the case of
5 structural plates. The control of modeling errors usually involves consideration
of nonlinear phenomena and sensitivity of the data to changes in boundary
conditions and loading.
StressCheck is the first software product designed to provide means for effec-
tive control of both the errors of idealization and discretization. Properly for-
mulated hierarchic working models and effective control of discretization
errors make it possible to approximate any real system or process up to the
limit of our ability to represent by mathematical relationships the natural laws
which control that process.

6 Solvers
Two solvers are supported in StressCheck: a sequential linear system solver and a sequential eigen-
solver. These are described in the following.
6
Sequential linear system solver
StressCheck uses an iterative solver by default and a direct solver as back-up for the
solution of the linear system of equations. The iterative solver is based on a precon-
ditioned conjugate gradient method and the direct solver is a sparse solver. In the
case where it is determined that the iterative solver will be less efficient than the
direct solver, StressCheck will switch to the direct solver. When switching happens,
if a parameter “_notes’’ with a non-zero value has been supplied, a message
“Switched to direct solver’’ will appear in the session log file. Users may also man-
ually select the direct solver by choosing the Direct Method in the Solver Interface.
The default tolerance for the iterative solver is 10-6. This tolerance value can be
changed by defining the parameter _tol with the required tolerance value.
In the p-version a large number of variables are “local” variables which are con-
densed at the element level. Condensation is one way to achieve efficiency.
Another way to achieve efficiency is through the use of Linear Algebra PACKage
Advanced Topics Chapter 6: Solvers 315

Sequential linear system solver
(LAPACK) Ref. [43] whose subroutines are written so that as much of the
computation as possible is performed by calls to Basic Linear Algebra Subpro-
grams (BLAS).
In addition to the solution vector, some other engineering quantities, such as

the potential energy, must be computed. StressCheck computes the potential
energy in a such way that only the solution and load vectors are needed and
only vector-vector products are performed. Thus both the I/O and CPU time
are reduced.
Another important characteristic of the p-version is that the stiffness matrices

arising from the p-version are hierarchic in the sense that the matrices associ-
ated with the lower p-levels are submatrices of the matrices associated with the
higher p-levels. When a sequence of solutions is sought, the Cholesky factors
of the stiffness matrices associated with the lower (resp. higher) p-levels can be
reused at higher (resp. lower) p-levels to improve the performance, if the p-
extension is upward (resp. downward).
StressCheck uses the hierarchic solution procedure for the downward solution
sequence. For the downward solution sequence, the element stiffness matrices
and the Cholesky factors of internal variables are stored and reused for the
lower p-levels. Hence the downward sequence is faster than the upward, but
requires more disk space.
6 The solvers of StressCheck have two running modes: in-core or out-of-core.
However, users usually do not need to decide which mode to choose, for it is
selected automatically by StressCheck, based on the nature and size of the
problem, and the amount of random access memory (RAM) available. If the
out-of-core mode is activated, a message such as “NOTE: out-of-core solution.
Disk space required (estimated): XXXX bytes’’ will appear in the log-file. If
there is no such message, the in-core mode is used. This decision by Stress-
Check is made prior to the actual solving stage. Therefore, for large problems,
there may be a situation where StressCheck decides to use the in-core mode but
later on, during the actual solution, there is not enough swap space due to some
unpredictable reason, such as another user having started running a storage
intensive job after the decision has been made. Should the in-core mode fail for
a large problem, users may select the out-of-core mode manually by defining
the parameter “_out_core’’.
Users can find out how much disk space is needed before starting a solution.
This can be accomplished by using the Check feature in the Solution interface.
316 Chapter 6: Solvers Advanced Topics

Eigensolver
Eigensolver
There are two eigensolvers in StressCheck. The first one is based on the QR
method. The second one is of the Lanczos type.
The QR eigensolver first transforms the generalized eigenproblem into a standard

eigenproblem and the LAPACK eigensolver DSPEVX is then applied. The QR
solver is a dense in-core solver, and hence large in-core memory requirements
make it unsuitable for medium to large problems.
The Lanczos solver transforms the generalized eigenproblem into a standard eigen-
problem and then applies the Lanczos iteration. The Lanczos solver is implemented
as an out-of-core solver and is capable of handling large problems. This is the
default solver in StressCheck, except for small problems. Should the default solver
fail for any reason, the QR solver is activated automatically. When the parameter
_notes is set greater than 0, the information of the solver actually used will be found
in the session file as follows:
“NOTE: solver = LAPACK, split = YES’’ means the direct eigensolver is used and
the mass matrix is a block diagonal matrix;
“NOTE: solver = LAPACK, split = NO’’ means the direct eigensolver is used and
the mass matrix is not a block diagonal matrix;
6
“NOTE: solver = LASO, runs = 10’’ means the Lanczos solver is used and the
number of iterations is 10.
The parameters _laso and _qr may be used to specify a solver, i.e., to override the
default solver.
Advanced Topics Chapter 6: Solvers 317

Eigensolver
318 Chapter 6: Solvers Advanced Topics

Network License
Management
StressCheck uses a network license server to control access to the various modules available in the pro-
gram. The license management facility provided by StressCheck is based on a third party toolkit from
Globetrotter Software called FlexLM.
Each customer licenses one or more of the modules contained in StressCheck, and may license one or
more “keys” for each module. License keys may be locked to a specific computer, or may “float” on the
network to give any network connected user access to the program. The license manager also provides
several utilities which can be used to monitor program usage, and to troubleshoot network licensing
problems.
What is License Management?

License management is the control of user access to an application program which
has been licensed from a software vendor. Each license key issued to a customer
corresponds to one concurrent user of a licensed module within an application pro- L
gram. Each module has an associated code which is embedded in the password(s)
issued to and installed by the application user. When the user activates the program
and uses one of the licensed modules, the license server checks to see whether or
not the module has been licensed, and whether or not the maximum number of con-
current users of the module has been reached (i.e. there is a key available). If a key
Master Guide License Manager

What is License Management?
is available, it is checked out to the current user, and the module is enabled for
use. If no key is currently available (either the module was not licensed, or all
keys are currently checked out) an error message is issued and access to the
licensed module is prevented. When the user finishes using the module, it is
automatically checked back in to the server so that it may be used by another
user.
There are two basic types of licenses which are available for StressCheck, a
node-locked license and a floating license. The node-locked license limits
usage of the program to a single computer. This means that a user must login to
a specific computer in order to use StressCheck. A floating license means that
the program is accessible to anyone whose computer is network connected to
the computer where StressCheck is installed. As long as the user's computer is
one for which the program is supported, and the appropriate version of Stress-
Check has been installed somewhere on the network, then the user may use
StressCheck if a license key is available. Since the type of license is controlled
by the password which is issued to the user, the license type may be changed at
any time by simply issuing a new password.
At the present time, the StressCheck license manager controls access to the fol-
lowing features:
1. Graphic User Interface (Base module)

2. Linear Elasticity Analysis
3. Nonlinear Elasticity Analysis
4. Thermal Analysis
5. Modal/Buckling Analysis
6. Measurement Analysis
7. Margin Check Analysis
8. Cold Working Analysis.
9. Fracture Mechanics Extraction
L
Once a password has been installed and the license server has been activated,
there is little need for further direct interaction between the license server and
the StressCheck user. However, if problems should occur, or if the user is sim-
ply interested in monitoring the usage of StressCheck, a variety of utilities are
320 License Manager Master Guide

License Server and Password Installation
provided for the purpose of trouble shooting, and monitoring license usage. These
utilities are summarized in the sections which follow.
License Server and Password Installation

Installation of the StressCheck password and activation of the license server are
usually performed in one step. You simply place a copy of the file (license.dat or
license_demo.dat) in the top level StressCheck installation directory.
The license server is actually two separate programs which are running in the back-
ground. The vendor (server) daemon monitors the number of concurrent users and
usually runs on only one system in the network. The second program is called a
local daemon, and runs on each user's computer. The local daemon validates the
password for the requested program module and communicates with the server dae-
mon to find out whether a license key for the requested module is available. In large
networks, where there is some probability that the server may go down or the net-
work connection between the user's computer and the license server may go down,
it may be necessary to have the license server running on more than one machine.
The vendor daemon (esrd.exe) is usually activated at the time StressCheck is

installed and should run continuously. If the system goes down, it should automati-
cally be reactivated when the system comes back up. If for any reason the license
server stops, it may be reactivated using FlexLM control panel applet described in a
later section. The server daemon is automatically started when StressCheck is
installed.
The local daemon (Imgrd.exe) is automatically activated when the StressCheck

program is started. It should not be necessary to manually start the local daemon.
License Management Trouble-shooting

L
A number of problems can occur during license management activation, and during
the day to day use of applications licensed in a network environment. These prob-
lems usually result in an error message issued by StressCheck when a user attempts
to use the program or attempts to access a specific module of the program.
Master Guide License Manager

License Management Trouble-shooting
The message: “Cannot access license file” usually means that the license file is
not installed the correct path (i.e. .../esrd), or the file has the wrong access priv-
ilege for the user who is trying to run the program, or perhaps the license ven-
dor daemon (esrd.exe) is not running. You may also check the status of the
license management daemons by running the utility /mtools.exe.
Documentation describing the functionality of FlexLM is found in the Stress-

Check installation folder named /flexuser.
322 License Manager Master Guide

Error and Warning
Messages
This Appendix contains a brief discussion of important error and warning messages which may be
issued by StressCheck.
There are many other messages which can occur but are very unlikely. If you encounter a message not
listed here that you do not understand, please contact your software supplier for a complete explanation.
Error 13 Unable to allocate XXX bytes of dynamic memory in “NAME”.
A software error has occurred. Please notify your software supplier.
Error 14 Database version mismatch (Program=XXX, Database=YYY) Do you wish to con-

tinue? (Answer: [Y]es to continue, No to abort)
The program version does not match the database file version. There may be
incompatibility between the database and the program. If you have an older version
of the program, you may wish to use it to open the database file and dump a batch
file of your model input data before proceeding to use the new program.
M
Error 15 Could not obtain a license for feature = NAME.
All license keys for the named feature are in use by other StressCheck users. You
must wait for a key to be returned to the license server. To find out who has the
Master Guide Messages 323

license keys checked out, activate the license administration tool “xnetAd-
min”.
Error 74 DatabasInput string has { } character limit. Truncating input string from
{< >} to {< >}.
Your input string is too long for the current input field.
Error 75 Parameter: { } may not be used in a formula.

Parameter name must not begin with a digit, or contain a special symbol.
Parameter name must be distinct from intrinsic functions (i.e. sin,cos).
Parameter name must not be constant or coord. name (X,Y,Z,R,T,C1,C2,...).
Parameter names have restrictions so that the name will not be confused with
intrinsic functions or spatial variable names that might appear in a formula.
Please avoid use of the restricted names.
Error 76 Parameter: { } is invalid or conflicts with a reserved name.

Parameter name must not be constant or coord. name (X,Y,Z,R,T,C1,C2,...).
Parameter name must be distinct from intrinsic functions (i.e. sin,cos).
Parameter name must not begin with a digit, or contain a special symbol.
Parameter names have restrictions so that the name will not be confused with
intrinsic functions or spatial variable names that might appear in a formula.
Please avoid use of the restricted names.
Error 113 Invalid file name = XXX.
The file name shown is not valid, or cannot be opened for some reason. Check
the pathname for mistakes, or check file ownership, protection, etc.
Error 115 Bad file: XXX.

M
The file named does not exist, is already open, or is not the proper format.
Error 116 A data base file must be open to perform the selected operation.
324 Messages Master Guide

Use the OPEN command to open a data base file, then issue the desired command
to UPDATE, LOAD, DUMP, CLOSE, etc.
Error 117 A data base file must be open before executing a module.
Use the OPEN command to open a data base file, then issue *EXECUTE to invoke
the appropriate module execution.
Error 118 Cannot open file: XXX.
The file named could not be opened NEW. Perhaps it is already open.
Error 119 The current data base must be CLOSED before opening a new one.
Use the CLOSE command to close the current data base file, then issue the OPEN
command.
Error 300 Function translation failed.
Illegal logical operator: x.
Error 360 Multiple regions cannot be auto-meshed.
The automesher can only mesh one simple closed region at a time.
Warning 361 Mesh may be invalid or not satisfy min and max angle requirement. If you wish to
improve the quality of mesh, make the Repeat # greater than { }.
Warning 362 Some boundary loop cannot be closed.
The automesher has detected an open loop. There must be a gap in the boundaries
defining the region to be meshed.
Error 363 The automesher requires that the user select boundaries to indicate regions which
are to be meshed.
M
No boundaries are selected.
Error 364 Boundaries { } and { } overlap.

Some boundaries are overlapping, and creating a redundant area of the bound-
ary defining the region. Remove the overlap condition and try again.
Error 1002 The requested feature has not been password enabled. Please contact your soft-
ware supplier for a valid password to activate this feature.
Error 1009 Invalid neutral format record type = XXX.
An invalid format specifier was encountered in your batch input file. Look for
*XXX in the batch file and remove or correct the record.
Error 1023 Elements { }, { } and { } are redundant.
More than 2 elements are sharing the same face.
Error 1024 Element { } has redundant nodes.
Element references the same node more than once.
Error 1025 Element { } is too distorted. It should be reconstructed.
Element may have edges which intersect opposite face, which crossover
another edge of the element, or has some other pathologic condition which
leads to a negative area/volume.
Error 1067 Element XXX has no property orientation points in table NAME.
When performing a fitting of property data from a table, there must be data
specified for every element in the model. Please supply tabular data for ele-
ment XXX.
Error 1068 E matrix for property NAME is not positive definite.
Property NAME has coefficients which cause the material stiffness matrix to
be singular.
M
Warning 1069 E matrix for property NAME has a condition number > 20.0.
Property NAME has coefficients which cause the ratio between the largest and
the smallest eigenvalue of the material stiffness matrix to be > 20.

Error 1070 E matrix for property NAME cannot be computed.
E matrix has negative determinant or material type may not be implemented.
Error 1071 Invalid value of Poisson's ratio (must be >= 0.0 and < 0.5).
The Poisson’s ratio for Isotropic material has to be positive and less than 0.5.
Error 1073 Plate extraction failed. Z coordinate may be out of range.
The provided Z-coordinate is out of range. Valid values of Z-coordinate are

between zero and half the plate thickness (0 <= Z <= h/2).
Error 1075 Zero density detected during modal analysis.
Check if density was provided in the property definition.
Error 1076 Assignment of mixed PRODUCT and TRUNK space is not supported in 3D.
StressCheck does not currently support mixing elements which have different basis
function spaces in 3 dimensions. You must make all elements either trunk space or
all product space. If you wish to use product space in 3D all elements must be hexa-
hedrons. Product space in 3D is not recommended.
Error 1077 Model contains elements or combinations of elements which are not currently sup-
ported.
Solids or shells are not allowed in any planar model. Solids and shells may not be
mixed in the 3D reference system.
Error 1078 Composite curve { } contains segments which are not piecewise continuous.
This composite curve should not be referenced by any element definition. The
slope discontinuity will not be properly represented by an element which contains
the discontinuity on its edge.
M
Warning 1079 Composite curve { } contains segments which are not slope continuous.
This curve may introduce errors in attached elements.

This composite curve should not be referenced by any element definition. The
slope discontinuity will not be properly represented by an element which con-
tains the discontinuity on its edge.
Error 1096 OBJECT number XXX may not be copied.
The copy operation is not yet supported for this object type. Some objects
which are created in StressCheck may not be copied because the associativity
relationships are particularly complex. When the copy operation encounters an
object of this type, the operation is aborted.
Error 1097 Object X cannot be associated with Object Y.
Associative objects are evaluated in order of creation. The object being moved
was created before the referenced object. To avoid this limitation, you may first
delete the object you wish to move, and then re-create it.
Error 1099 The operation is not supported for the object(s) selected.
You have selected a combination of ACTION, OBJECT, and METHOD which

is not supported by the program.
Error 1100 Object cannot be created tangent to curves XXX and YYY.
The tangency method is currently supported only for circles and points.
Warn 1101 The location of NODE X is difficult to compute.
The node is probably an intersection or tangency in a region where an iterative

algorithm has failed to isolate the location accurately within a preset iteration
limit. The computed location is only approximate.
Error 1102 The location of NODE Y could not be computed.
One of the following conditions may apply:

M
1. Intersection or tangency cannot be computed.
2. Offset on a curve which does not exist.
3. Specification in a local system which does not exist.

4. Equality with a point which does not exist.
Warning 1105 Nodes were reordered for element XXX.
The nodes of an element were shifted to a position which caused the element to
become illegal. The program has automatically renumbered the nodes so that the
element will be legal. This may cause boundary conditions attached directly to ele-
ment edges or faces to be different than what it was intended.
Error 1106 OBJECT “NAME” cannot be deleted. Object is part of an associative relationship.
You have attempted to purge an object which is still being referenced by some other
object in the model. You must remove this reference before you will be able to
purge the object. This reference may be from a SET definition, or some other asso-
ciative geometry definition.
Error 1128 Zero or negative thickness detected in element XXX.
Formula or fitted thickness has resulted in a zero or negative thickness at some

point in element XXX.
Error 1130 Element EEE edge nodes YYY & ZZZ do not form a valid edge.
A set definition probably references an element edge by explicitly listing the edge
nodes (element: node1-node2). It is advisable to convert the reference to a direct
edge reference (element: edge).
Error 1131 Element EEE face nodes WWW,XXX,YYY,ZZZ do not form a valid face.
A set definition probably references an element face by explicitly listing the face
nodes (element: n1-n2-n3-n4). It is advisable to convert the reference to a direct
face reference (element: face).
Error 1138 XXX number YYY references an object which is not selected. The copy operation
cannot be performed.
M
When objects are selected for a copy operation, the program checks to see that all
associative relationships can be preserved in the new copy. If any object referenced
in an associative relationship from an object selected for copying, is not itself
selected, the copy operation is aborted. Make sure that all appropriate objects are
selected before attempting to copy.

Error 1139 XXX number yyy is not selected but is referenced by a copied object.
This is the object which is referenced by the object selected for the copy opera-
tion. You must select this object in order for the copy to be performed. This is
the first object encountered which is referenced by an object being copied, but
may not be the only such object.
Error 1140 Redundant or conflicting boundary conditions on the same element edge.
Name = NAME, element = XXX, edge =YYY.
Edge YYY of element XXX has been assigned redundant or conflicting

boundary conditions in case NAME. This can happen under the following con-
ditions:
1. The user assigned more than one load or constraint to the same edge
(face).
2. The user assigned more than one load or constraint to the same
boundary which was then inherited by edge (face) YYY.
3. The user assigned a load or constraint to a boundary and to an edge

(face) which inherits from the boundary.
An edge (face) can inherit a boundary condition from a boundary if the ele-
ment is explicitly attached to the boundary or if the edge is collinear (face is
coplanar) with a boundary. Sometimes you may want some element edges
(faces) which are collinear with a LINE (coplanar with a PLANE) to be loaded
or constrained. In this case you should apply the boundary condition explicitly
to the edge (face) instead of applying it to the boundary. One particular situa-
tion where this might occur is when attempting to load or constrain the faces of
a crack. In this case, the element edges (faces) which lie on the crack face
should be loaded or constrained explicitly.
Error 1141 Redundant body force or Poisson conditions on the same element. Name =
NAME, element = XXX already loaded.
M The same element probably appears in more than one load specification for the
same load case (i.e. the element is contained in more than one SET definition).
Error 1142 Redundant or conflicting boundary conditions on the same element face. Name
= NAME, element = XXX, face = YYY.

See discussion above for Error 1140. Applies to 3D surfaces and faces.
Error 1150 Inconsistent constraints found in element = XXX, case = NAME.
Displacement and temperature boundary conditions specified for intersecting

boundaries, edges or faces must be consistent. Consistency tests are performed for
nodes and edges. Check the constraint or temperature data.
Error 1151 Constraint error found in element = XXX, case = YYY. Symmetry or anti-symmetry
condition applied to a face which is not flat. Tolerance may be controlled with
parameter _SURFACE_TOL. Number of points checked on each face controlled
with parameter _SURFACE_MIDPT.
The tolerance parameter (_SURFACE_TOL) may be set to a value which repre-

sents the difference in distance between a point which is in the plane of the con-
strained surface, and the point which lies on the face of the element. There can
sometimes be some numerical error in representing faces which have curved edges
but which are supposed to be flat.
It is sometimes useful to specify the number of points to be sampled from the face
for purposes of determining whether the face is planar. The parameter
_SURFACE_MIDPT is the number of points along an edge of an element. The
actual number of points used to check planarity of a quadrilateral face will be
(n+2)*(n+2).
Error 1163 Parameter { } expression may not be recursive.
A parameter has been defined using an expression which references itself. For
example a = b + c; b = a + d;
Warning 1206 Model changed since last solution. Post-processing results may not be reliable.
The program has detected a change in the model definition due to the addition or
modification of an element, node, boundary, etc. This may or may not affect your
post-processing results, depending on what was changed and what is being post-
processed. Use caution when interpreting results. M
Error 1208 Eigenvectors are not available. Rerun with mode shapes selected.
In the Execution Status interface for modal and buckling analysis, there is a choice
between computing only the eigenvalues (load factors), or both the eigenvalues

(load factors) and mode shapes. If you select to compute only the eigenvalues
(load factors), then the mode shapes are not available and you will not be able
to display the shapes. You must rerun the solution with the appropriate option
selected.
Error 1210 Matrix assembly FAILED. No elements to assemble, check to see whether nick-
names have been provided.
Either there are no elements in the model, or the Parts and Assembly combo
boxes are set such that no elements are being displayed on the main window.
Error 1215 Degrees of freedom exceeds program limit: { }. Reduce element

count or revise p-level assignment.
The version of the program has imposed a limit on the number of degrees of
freedom to be solved. You may use bounded or fixed p-level assignment to
restrict the p-level in regions which are not of interest in order to minimize the
degrees of freedom.
Error 1218 No elements were assigned non-linear material properties.
If no elements have non-linear material properties, you cannot perform a non-

linear analysis. Change material definitions.
Error 1220 Element XXX stiffness or load generation failed.
This is an internal error which should be reported to your software supplier.

The element shape may be too distorted. Use the Edit/ Geometry/ Mesh/
Check/ Distort feature to test element distortion.
Error 1222 Element XXX load generation failed.

The element shape may be too distorted. Use the Edit/ Geometry/ Mesh/
Check/ Distort feature to test element distortion.
M
Error 1223 Element XXX stiffness generation failed.

The element shape may be too distorted. Use the Check/ Mesh/ Distortion fea-

ture to test element distortion or the Check/ Mesh/ Properties feature to test if ele-
ment volume can be computed.
Error 1224 Element XXX constraint generation failed.
This is an internal error which should be reported to your software supplier. The
element shape may be too distorted. Use the Edit/ Geometry/ Mesh/ Check/ Distort
feature to test element distortion.
Error 1225 Element XXX is too distorted. (see Geometry/Check/Distort).
StressCheck makes sure that all elements in the model satisfy basic minimum and
maximum interior angle conditions. The interior angle of an element must not be
less than 1 degree or greater than 179 degrees. The distortion of elements may be
checked in the Geometry/Mesh dialog window. Choose Action: Check, Object:
Element, Method: Distort.
Error 1226 Element { } is too distorted, mapping failed during nonlinear analysis.
The shape change is too large to be properly simulated. You may need to remesh
the model in the distorted area in order to properly represent the deformed domain.
Error 1259 The global matrix and vector assembly failed.
Error 1260 The solver failed.
Error 1261 The eigensolver failed.
Note 1310 XXX redundant nodes have been trimmed from the model.
M
During H-refinement, the program will trim nodes from the model which appear to
be redundant. The tolerance which is used to determine if two nodes are close to
each other may be controlled by the parameter _TRIM_TOL.
Error 1311 Boundary XXX construction failed.

This usually indicates that an object such as a spline curve or surface which is
constructed from a set of points or curves, cannot be constructed because one
of the underlying objects can no longer be evaluated. This can happen if an
underlying object was itself computed as an intersection which is no longer
valid.
Error 1312 Intersection could not be computed.
The program does not yet support all possible combinations of surface inter-
sections. At the present time, it is possible only to intersect two planes and two
properly aligned cylinders.
Warning 1315 The fillet/intersection { } may not be accurate enough for your purposes.
If you need more accuracy, it may help to increase the resolution to a value
greater than { }.
The primary intent of this message is to inform the user that the fillet surface or
intersection curve may not have the desired shape. This usually means that the
fillet/intersection is not closed, but perhaps should be. If the resulting surface/
curve is satisfactory for your modeling purposes, do not increase the resolu-
tion.
Increasing the resolution may have undesired consequences. The computation

time will be significantly increased, and the robustness of the shape to changes
in geometric dimensions may be sacrificed. It is not recommended to increase
the resolution above 25 unless there are extreme circumstances that require
such a high resolution. Do not increase resolution if your only objective is to
suppress this message.
Error 1700 Contour integral requires all element edges intersected by the integration cir-
cle to be radial from the crack tip.
The radius of the integration circle can be arbitrary but the circle must be
inside the solution domain and it must cross only interelement boundaries
which lie on radial lines with respect to the crack tip.
M
Warning 1701 Contour integral may not be valid if material is not LINEAR and ISOTROPIC.
The computation of the stress intensity factors by the contour integral method
is valid only for linear-elastic isotropic materials. See Appendix B.

Error 1702 Node X is not a crack tip.
There may be redundant nodes at the same location as the actual crack tip, or there
is no free face at the node.

M

References
[1] Bennett, J. A. and Botkin, M. E. 1985. “Structural shape optimization with geo-
metric description and adaptive mesh refinement”, AIAA Journal 23, pp. 458-
462.
[2] Schiermeier, J. E. and Szabó, B. A. 1987. “Interactive Design Based on the p-

Version of the Finite Element Method”, Finite Elements in Analysis and Design
3: 93-107. North-Holland.
[3] Szabó, B. A. and Babuska, I. 1991. Finite Element Analysis, John Wiley and
Sons, Inc. New York.
[4] Bortman, J. and Szabó, B. A. 1992. “Analysis of Fastened Structural Connec-

tions”, AIAA Journal, Vol. 30, No. 11, pp. 2758-2764.
[5] Bortman, J. and Szabó, B. A. 1992. “Nonlinear Models for Fastened Structural R
Connections”, Computers and Structures, Vol. 43, No. 5, pp. 909-923.
[6] Gui, W. and Babuska, I. 1986. “The h-, p- and hp-Versions of the Finite Element
Method in One Dimension. Part I: The Error Analysis of the p-Version. Part II:
Master Guide References 337

The Error Analysis of the h and hp-Versions. Part III: The Adaptive hp-Ver-
sion.” Numerische Mathematik, Vol. 49, pp. 577-612; 613-657; 659-683.
[7] J. M. Whitney and J. N. Pagano. 1970. “Shear Deformation in Heteroge-

neous Anisotropic Plates”, Journal of Applied Mechanics, Vol. 37, 1031-
1036.
[8] J. M. Whitney. 1973. “Shear Correction Factors for Orthotropic Laminates

Under Static Load”, Journal of Applied Mechanics, Vol 40, 302-304.
[9] A. K. Noor and J. M. Peters. 1989. “A Posteriori Estimates for Shear Correc-
tion Factors in Multi-Layered Composite Cylinders”, Journal of Engineering
Mechanics, ASCE, Vol. 115, No. 6, 1225-1244.
[10] J. M. Whitney and C. T. Sun. 1973. “A Higher Order Theory for Extensional
Motion of Laminated Composites”, Journal of Sound and Vibration, Vol. 30,
85-97.
[11] J. N. Reddy and C. F. Liu. 1985. “A Higher-Order Shear Deformation Theory

of Laminated Elastic Shells”, International Journal of Eng. Sci., Vol. 23, No.
3, 319-330.
[12] M. Di Sciuva. 1987. “An Improved Shear-Deformation Theory for Moder-

ately Thick Multilayered Anisotropic Shells and Plates”, Journal of Applied
Mechanics, Vol. 54, 589-596.
[13] E. J. Barbero and J. N. Reddy. 1990. “An Accurate Determination of Stresses

in Thick Laminates Using a Generalized Plate Theory”, International Jour-
nal for Numerical Methods in Engineering, Vol. 29, 1-14.
[14] D. H. Robbins Jr. and J. N. Reddy. 1993. “Modelling of Thick Composites

R Using a Layerwise Laminate Theory”, International Journal for Numerical
Methods in Engineering, Vol. 36, 655-677.
[15] R. K. Kapania. 1989. “A Review on the Analysis of Laminated Shells”,

Journal of Pressure Vessel Technology, Vol. 111, No. 2, 88-96.
338 References Master Guide

[16] J. N. Reddy. 1990. A Review of Refined Theories for Laminated Composite
Plates”, Shock and Vibration Digest, Vol. 22, 3-17.
[17] A. K. Noor and W. S. Burton. 1990. “Assessment of Computational Models for

Multilayered Composite Shells”, Applied Mechanics Review, Vol. 43, 67-97.
[18] W. Gilewski and M. Radwandska. 1991. “A Survey of Finite Elements Models

for the Analysis of Moderately Thick Shells”, Finite Elements in Analysis and
Design, Vol. 9, 1-21.
[19] B. A. Szabo and G. J. Sahrmann. 1988. “Hierarchic Plate and Shell Models
Based on p-Extension”, International Journal for Numerical Methods in Engi-
neering, Vol. 26, 1855-1881.
[20] B. A. Szabo. 1989. “Hierarchic Plate and Shell Models Based on p-Extension”,
Analytical and Computational Models of Shells, A. K. Noor, T. Belytschko and J.
C. Simo, Society of Mechanical Engineers, New York”, Vol. 3, 317-331.
[21] C. Schwab. 1989. “Dimensional Reduction for Elliptic Boundary Value Prob-
lems”, Ph.D. Thesis, University of Maryland, Department of Mathematics, Col-
lege Park.
[22] I. Babuska, B. Szabo and R. Actis. 1992. “Hierarchic Models for Laminated
Composites”, International Journal for Numerical Methods in Engineering, Vol.
33, 503-535.
[23] R. Actis and B. Szabo. 1993. “Hierarchic Models for Bidirectional Composites”,
Finite Elements in Analysis and Design, Vol. 13, 149-168.
[24] R. L. Actis. 1991. “Hierarchic Models for Laminated Plates”, Ph.D. Thesis,
Sever Institute of Technology, Washington University, St. Louis.
R
[25] B. A. Szabo and R. L. Actis. 1992. “Hierarchic Models for Laminated Plates”,
Adaptive, Multilevel and Hierarchical Computational Strategies, edited by A. K.
Noor, The American Society of Mechanical Engineers, AMD-Vol. 157, 69-94.

[26] I. Babuska, J. M. d’Harcourt and C. Schwab. 1991. “Optimal Shear Correc-
tion Factors in Hierarchic Plate Modelling”, Institute for Physical Science
and Technology, University of Maryland, College Park, Technical Note BN-
1129.
[27] Arnold, D. N. and Falk, R. S. 1990. “The Boundary Layer of the Reissner-
Mindlin Plate Model: Soft Simply Supported, Soft Clamped and Free
Plates”, SIAM J. Math. Anal., Vol. 21, pp. 281-312.
[28] Schwab, C. and Wright, S. 1992. “Boundary Layer Approximation in Hierar-

chical Beam and Plate Models”, Department of Mathematics and Statistics,
University of Maryland, Baltimore County, Baltimore, MD.
[29] Hodge, P. G., Jr. and White, G. N., Jr. 1950. “A Quantitative Comparison of
Flow and Deformation Theories of Plasticity”, Journal of Applied Mechan-
ics, Vol. 17, pp. 180-184.
[30] Budiansky, B. 1959. “A Reassessment of Deformation Theory of Plasticity”,

Journal of Applied Mechanics, Trans. Am. Soc. Mech. Engrs., Vol 81E, pp.
259-264.
[31] Rice, J. R. 1968. “A Path Independent Integral and the Approximate Analy-
sis of Strain Concentration by Notches and Cracks”, Journal of Applied
Mechanics, Trans. Am. Soc. Mech. Engrs., Vol. 35, pp. 379-386.
[32] Hutchinson, J. W. 1968. “Singular Behavior at the End of a Tensile Crack in

a Hardening Material”, J. Mech. Phys. Solids, Vol. 16, pp. 13-31.
[33] Nagtegaal, J. C., Parks, D. M. and Rice, J. R. 1974. “On Numerically Accu-
rate Finite Element Solutions in the Fully Plastic Range”, Computer Methods
in Applied Mechanics and Engineering, Vol. 4, pp. 153-177.
R
[34] J. E. Ashton and J. M. Whitney. 1970. Theory of Laminated Plates. Progress
in Material Science Series, Volume IV. Technomic Publishing Co., Inc.

[35] S. W. Tsai. 1966. Mechanics of Composite Materials. Part II-Theoretical
Aspects. Technical Report AFML-TR-66-149, Air Force Material Laboratory,
Research and Technology Division, Air Force System Command, Wright-Pater-
son Air Force Base, Ohio.
[36] L. P. Pook. 1988. ‘‘Unacceptable Differences in Published Stress Intensity Fac-

tors Solutions”. Technical Note, Fatigue Fract. Engng. Mater. Struct., Vol. 12,
No. 1, pp 67-69.
[37] Pvm 3.0 User’s Guide and Reference Manual. Technical Report ORNL/TM-
12187, Engineering Physics and Mathematics Division, Mathematical Science
Section, Oak Ridge National Laboratory, February 1993.
[38] Anderson, E., et al. 1992. LAPACK User’s Guide, Society for Industrial and
Applied Mathematics.
[39] Kuo, An-Yu. 1987. “On the Path-Independent Line Integral for Axisymmetric
Cracks with Nonaxisymmetric Loading”, Journal of Applied Mechanics, Vol. 54,
833-837.
[40] Yosibash, Z. and Szabó, B.A. 1995. “The Solution of Axisymmetric Problems
Near Singular Points and Computation of Stress Intensity Factors”, Finite Ele-
ments in Analysis and Design., Vol. 19, 115-129.
[41] Beer, F. P. and Johnston, E.R. 1992. Mechanics of Materials, Second Edition,
McGraw-Hill, Inc.
[42] Szabó, B. A., Actis, R. L. and Holzer S. M. 1995. “Solution of Elastic-Plastic

Stress Analysis Problems by the P-Version of the Finite Element Method”, The
IMA Volumes in Mathematics and its Applications, Vol. 75, pp. 395-416.
[43] Nathan, A. and Brot, A. 1993. “An Analytical Approach to Multi-Site Damage”, R
17th Symposium of the International Committee on Aeronautical Fatigue, Stock-
holm, Sweden, June 10.

[44] Yosibash, Z. and Szabo, B. A. Numerical Analysis of Singularities. Part 1:
Computation of Eigenpairs. International Journal for Numerical Methods in
Engineering, 38, 2055-2082 (1995).
[45] Szabo, B. A. and Yosibash, Z. 1996. Numerical Analysis of Singularities.

Part 2: Computation of Generalized Flux/Stress Intensity Factors. Interna-
tional Journal for Numerical Methods in Engineering, Vol. 39, 409-434.
[46] Yosibash, Z. Numerical Thermo-Elastic Analysis of Singularities in Two-

Dimensions. International Journal of Fracture, 74, 341-361, 1996.
[47] Babuska, I. and Miller, A. The Post-processing Approach in Finite Element

Method - Part 2: The Calculation of Stress Intensity Factors. Int Jour. Num.
Meth. Engrg., 20: 1111-1129, 1984.
[48] Lee, K. and Cho, Y. H. Boundary Element Analysis of Thermal Stress Inten-
sity Factors for Cusp Cracks. Engineering Fracture Mechanics, 37 (4): 787-
798, 1990.
[49] Prased, N. N. V., Aliabadi, M. H., and Rooke, D. P. The Dual Boundary Ele-
ment Method for Thermoelastic Crack Problems. International Journal of
Fracture, 66: 255-272, 1994.
[50] Sumi, N. and Katayama, T. Thermal Stress Singularities at tips of Griffith

Crack in a Finite Rectangular Plate. Nuclear Eng. Desgn., 60: 389-394, 1980.
[51] W.C. Young, Roark’s Formulas for Stress and Strain, McGraw Hill Book
Company, Sixth Edition, 1988.
[52] Tuma, J. J. Structural Analysis, Schaum’s Outline Series, McGraw Hill Book
Company, 1969.
R
[53] Blevins, R. D. Formulas For Natural Frequencies And Mode Shape, R. E.
Krieger Publishing Co., Inc., 1984.

[54] Ramberg W. and Osgood W. Description of Stress-Strain Curves by Three
Parameters. NACA Technical Note 902, 1943.
[55] Sun, C.T. and Li, S. Three-Dimensional Effective Elastic Constants for Thick
Laminates. Journal of Composite Materials, Vol. 22, pp. 629-639 (1988).
[56] Yosibash, Z. Numerical Analysis of Edge Singularities in Three-Dimensional

Elasticity. Int. Jour. Numer. Meth. Engrg., Vol. 40, pp. 4611-4632 (1997).
[57] Yosibash, Z. Computing Edge Singularities in Elastic Anisotropic Three-Dimen-

sional Domains. Int. Jour. of Fracture., Vol. 86, pp. 221-245 (1997).
[58] Actis, R.L. Stress/Failure Analysis Software for Multi-Material Interfaces. Air
Force Office of Scientific Research, Phase I Final Technical Report, Project No.
FQ8671-9501469 STTR/TS, December 1996.
[59] Actis, R.L. Stress/Failure Analysis Software for Multi-Material Interfaces. Air
Force Office of Scientific Research, Phase II Final Technical Report, Project No.
FQ8671-9701107 STTR/TS, October 1999.
[60] Anderson, B., Falk, U., Babuska, I. and Von-Petersdorff, T. Reliable Stress and
Fracture Mechanics Analysis of Complex Components Using a h-p Version of
FEM. Int. Jour. Numer. Meth. Engrg., Vol. 38, pp. 2135-2163 (1995).
[61] Babuska, I., Von-Petersdorff, T. and Anderson, B. Numerical Treatment of Ver-

tex Singularities and Intensity Factors for Mixed Boundary Value Problems for
the Laplace Equation in R3. SIAM. Jour. Numer. Analysis, Vol. 31(5), pp. 1265-
1288 (1994).
[62] Dauge, M. Elliptic Boundary Value Problems in Corner Domains - Smoothness

and Asymptotics of Solutions. Lecture Notes in Mathematics 1341, Springer-Ver-
lag, Heidelberg, 1988. R
[63] Guo, B. and Oh, H.S. The Method of Auxiliary Mapping for the Finite Element
Solutions of Elliptic Partial Differential Equations on Nonsmooth Domains in
R3. Preprint submitted for publication to Mathematics of Computations (1996).

[64] Ting, T.C.T., and Chou, S.C. Edge Singularities in Anisotropic Composites.
Int. Jour. Composites and Structures, Vol. 17 (11), pp. 1057-1068 (1981).
[65] Wang, S.S. and Choi, S.C. Boundary Layer Effects in Composite Laminates:
Part 1 - Free Edge Stress Singularities. Trans. ASME, Jour. Appl. Mech., Vol
49, pp. 541-548 (1982).
[66] Babuska, I. and Aziz, A.K. Survey Lectures on the Mathematical Founda-
tions of the Finite Element Method. In A.K. Aziz, editor, The Mathematical
Foundations of the Finite Element Method with Applications to Partial Dif-
ferential Equations, pp. 3-343, Academic Press, New York, NY, USA, 1972.
[67] Yosibash, Z. Numerical Analysis of Singularities and First Derivatives for

Elliptic Boundary Value Problems in Two-Dimensions. D. Sc. dissertation,
Sever Institute of Technology, Washington University, St. Louis, Missouri,
1994.
[68] Chen, D. H. Analysis of Singular Stress Field Around the Inclusion Corner
Tip. Engineering Fracture Mechanics, Vol. 49 (4), pp. 553-546 (1994).
[69] Chuin-Shan Chen, Roland Krause, Richard G. Pettit, Leslie Banks-Sills, and
Anthony R. Ingraffea, Numerical Assessment of T-stress Computation using
a p-version Finite Element Method, International Journal of Fracture, 107:
177-199, 2001.
[70] I. Páczelt, B. A. Szabó and T. Szabó. “Solution of Contact Problem using the
hp-Version of the Finite Element Method”, Computers and Mathematics with
Applications, Vol. 38, 49-69 (1999).
[71] Hill, H. N. Determination of Stress-Strain Relations from “Offset” Yield

Strength Values, NACA Technical Note 927, 1944.
R

INDEX
(first digit indicates volume; second digit indicates page number)
A-C Blending operation

Abort process 1.54 definition 2.3
Accelerator keys 1.39 edge 2.75
Accuracy 1.195 face 2.81
Action 1.55 Block 1.161
Adaptivity 1.104, 1.119, 3.32, 3.163 Body
Analysis copy 2.73
buckling 3.267 definition 2.3
coldwork 4.44 general 2.3
contact 4.104 imprint 2.65
crack path 1.114, 4.164 sheet 2.2, 2.44, 2.66
linear 1.103, 3.30, 3.78, 3.113, 3.162 solid 2.2, 2.44, 2.66
margin check 1.113, 4.68 stitch 2.63
measurement 1.115 trim 2.74
modal 1.110, 3.253 wire 2.3
nonlinear 1.108, 4.2 Body-Solid
SOLVE! 1.105 extrude 2.66
Animation 1.156 extrude face 2.69
Arc 2.17 spin 2.70
Area/Volume check 1.171 spin face 2.71
Assembly 1.92 sweep 2.72
Assignment thicken 2.72
by region 1.57 Boolean
by surface matching 1.58 operations 2.3, 2.60
Associativity 2.15 subtraction 2.61
explicit 2.3 union 2.85
implicit 2.3 Box 2.51
Attributes 1.35, 1.40 Browser 1.75, 1.197
Auto 2.129 Buckling analysis
Automated mesh generation 2.122, executing 3.269
2.135, 2.140, 2.141 introduction 1.111, 3.253
Automesher load numbers 1.112
MeshSim 2.128 Buttons
Average 1.166 Accept 2.7, 2.8, 2.11
Axial force 3.67 Blank 2.20
Axisymmetric elasticity 3.71 Browse 1.107
Backspace 1.51 Cancel 2.21 I
Batch file Copy 2.22
read 1.21 DeLast 2.8, 2.10
write 1.20 Delete 2.20
Beam 3.55, 3.244, 3.263 Grow/Shrink 1.31
Bending moment 3.64 Merge 2.105
Blank 2.20 Move 2.22
Master Guide Index 1

Plot 1.207 antisymmetry 1.82, 3.25, 3.108, 3.159
Replace 2.22 assign 3.24, 3.77, 3.157, 3.180
Solve 1.107, 1.207, 1.212 built-in 1.82, 3.25, 3.108, 3.158
Update 1.206 contact zone 1.83, 3.26, 3.160
CAD export 1.22 dialog box 3.180
CAD import 1.21 dialog window 3.24
Calculator 1.175 direction 1.83
Cancel 2.21 display 1.40, 3.180
selection 2.61 extrusion 3.182
whole operation 2.61 fastener connection 3.26
Chamfer 2.76 fastener displacement 3.26
Check 1.107, 2.22 fit-auto 3.159, 3.220
equilibrium 3.45, 3.52 fit-manual 3.159, 3.220
load 3.23 general 3.24, 3.158
mesh 3.8 hinge 3.26, 3.160, 3.247
Circle 2.31, 2.63 importation of point
Class constraints 3.215
constraint 1.81 point 1.83, 3.26, 3.109, 3.230, 4.249
geometry 1.63, 2.6 prescribed displacement 4.250
h-discretization 1.87, 2.122 rigid body 1.82, 3.26, 3.109, 3.159,
load 1.77 4.249
material 1.72 single node 1.83, 3.26, 3.160
menu 1.31 soft-simple support 3.108
mesh 1.64, 2.91 spring coefficient 3.159
parts 1.92 spring coefficients 1.82, 3.26, 4.250
p-discretization 1.85 symmetry 1.82, 3.25, 3.27, 3.108,
section properties 1.65 3.158, 3.175, 3.180
selection 1.55 types of 4.249
sets 1.91 Contact 3.26, 4.101
solution ID 1.84 analysis 4.104
tables 1.89 example 4.105, 4.112
thickness 1.68 formulation 4.101
Clipping operations 2.75 Continuity 1.153
Cold working 4.1, 4.44 Convective boundary conditions 3.279,
Color 3.283
editting 1.41 Convective temperature 3.282
objects 1.55 Convergence 1.106, 1.109, 1.122, 1.128,
tables 1.40 1.189, 3.33, 3.164
COM interface 1.58 estimated limit 1.162
I Coordinate system 2.26
Command File 1.49
Condition number 3.179 Copy 2.22, 4.59
Cone 2.49 Crack path analysis
Configuration description 1.114, 4.164
solution 1.85, 1.98 external crack 4.167
Constraints internal crack 4.170
2 Index Master Guide

no initial crack 4.174 options 1.40, 1.42
Cracks ranges 1.42
2D-model problem 4.127 refresh 1.54
3D-model problem 4.137 window 1.42
mesh design 4.120 Display Menu
Create assign colors 1.40
associative object 2.15 attributes 1.40
mesh 2.96 material summary 1.45
methods 2.16 model icon 1.42
parametric object 2.11 model summary 1.45
repeated object 2.10 move 1.37
repetitive parametric object 2.13 objects 1.42
simple object 2.7 reset 1.37
Cursors 1.55 selection 1.40
Curve 2.132 view controls 1.43
definition 2.2 Display Objects
Cutting Plane 1.36 toolbar 1.35
Cylinder 2.49, 2.55 Display Options
Cylindrical coordinates 4.271 toolbar 1.36
Distortion check 1.171
D-F Dynamic operations 1.58
Edge
Database
edge curve 2.93
autosave 1.54
close 1.20 move 2.98
object 2.93
erase 1.21
Edit
new 1.20, 1.25
object 2.21
recover 1.21
Edit Menu
Deformed shape
comment 1.26
plot 1.154, 3.62
formulae 1.29
DeLast 2.8, 2.23
graph 1.30
Delete 1.51, 2.20
Delta 2.19 handbook 1.29
input 1.28
Design Study 4.123
parameters 1.26
Discretization
p- 1.85, 3.78, 4.251 redo 1.25
report 1.30
q- 3.198, 3.205
results 1.29
Discretization errors
control of 4.216 solution 1.29
toolbar 1.34 I
Display
undo 1.25
attributes 1.40
Eigenpairs 4.179
colors 1.40, 1.42
material summary 1.45 Eigensolvers 3.273
Elasticity
model summary 1.45
3D 3.137
objects 1.55
planar 3.1

Element 2.97, 4.243 nonlinear material 4.9
beam 2.94, 3.5 planar 3.46
distortion 2.103 plate bending 3.128
editing 2.95 radiation boundary conditions 4.89
fastener 2.94, 3.5 temperature-dependent
hexahedral (brick) 3.143 properties 4.86
Link 3.5 thermal analysis 3.292
link 2.94, 4.64 Execute
mesh revision 2.102 buckling analysis 3.269
object 2.92 coldwork analysis 4.44
pentahedral (wedge) 3.143 fastened connection analysis 4.53
pentahedron 2.94, 2.95 general nonlinear analysis 4.26
quadrilateral 2.94, 3.5 linear analysis 3.30, 3.78, 3.113
tetrahedral 3.143 margin check analysis 4.74
tetrahedron 2.94 material nonlinear analysis 4.6
thin solid 3.196 modal analysis 3.256
triangle 2.93, 3.5 nonlinear heat transfer analysis 4.83
vanishing angle 2.103, 3.8, 3.144 Exit 1.24
Ellipse 2.32 Expression
Enter 1.50 evaluation 1.51
Equilibrium 3.52 parameter 1.27, 1.28
check 1.148, 3.284 relational 1.203
Equivalent strain 1.177 Extension 1.86, 3.29, 3.30, 3.162
Error adaptive-p 1.104
discretization 1.119, 4.214, 4.216, choice of an 4.218
4.220 downward-p 1.104
energy norm 1.145 h- 4.217
estimator 1.122, 1.146, 3.34, 3.79, hp- 4.218
3.166 p- 4.217
indicator 1.123, 1.147 uniform-h 1.104
logfile 1.23 upward-p 1.103
modeling 4.214, 4.222 Extraction
Example problem animation 1.156
3D-elasticity 3.168 area/volume properties 1.170
adaptivity 1.125 average 1.166
axisymmetric 3.89 contour plot 1.152
coldwork 4.49 data at points 1.163
eigenvalues and eigenvectors deformed shape 1.154
computation 4.180 equilibrium 1.149
I fastener connection 4.56 error estimate 1.146
fitting in 2D 4.205 fracture mechanics 1.172
fitting in 3D 4.209 minimal and maximal values 1.159
fracture mechanics 4.127 resultant 1.148
margin check 4.69 using formulas 1.174
nonlinear geometry 4.30 using the calculator option 1.175

z-shape function 1.156 object 2.37
Extrusion 3.176 parameters 1.28
Face plotting 1.155
face surface 2.93 sub-expressions 1.30
object 2.93 system option 1.29
Face to Face 2.133 Fracture mechanics
Failure Criteria option 1.177 example 4.127, 4.180, 4.184
Fastener 4.1, 4.53 generalized SIF 1.172, 4.178, 4.290
Fiber integration path 4.126, 4.284
angle 4.203, 4.205, 4.208, 4.211 J-integral 1.172, 4.126, 4.157, 4.284
fitting 4.203, 4.208 mesh design 4.120
File stress intensity factors 4.133, 4.139,
extensions 1.18 4.285
menu 1.20, 1.25 thermal SIF 4.197
File Menu T-stress 4.142
delete database 1.21 Frequency 1.110, 3.258
edit ASCII file 1.24 Fringe color 1.41
erase database 1.21 Function keys 1.38
new 1.20
open 1.20 G-J
print 1.22 Generalized intensity factors 4.178
read input 1.21 Geometry
recover database 1.21 associativity 2.15
save 1.20 check 2.22
save input 1.20 class 1.63, 2.6
view error log 1.23 create 1.55, 2.7, 2.8, 2.10, 2.11, 2.14,
view session log 1.23 2.21, 3.170
Fillet 2.32 delete 1.55
curve 2.32 edit 1.55, 2.21
surface 2.57 hierarchy 2.15
Fitting move 2.21
2D-model problem 4.205 revise 2.21, 2.102
3D-model problem 4.209 select 1.55, 2.21, 2.22
Flux 3.284 Graded refinement 1.87, 2.123
Font 1.49 Graphic
Force resultant 1.149 input 1.56
Format 1.43 output 1.155
output 1.161 Grow 1.31
Formula Handbook
I
angle option 1.29 accuracy of solutions 1.195
constants 1.29 analysis 1.205
define 1.174, 3.21 create problem 1.220
extraction 1.174 design Study 1.211
interface 1.29, 1.30 example 1.208
name 1.29

interface 1.15, 1.195 definition 4.284
library 1.15, 1.193 limitations 4.285
model preparation 1.204 nonlinear 4.160
parameter 1.206
results 1.213 K-O
Handbook results interface Key
error tab 1.214
accelerator 1.39
fracture tab 1.216
arrow 1.38
min/max tab 1.216 backspace 1.51
plot tab 1.214
delete 1.51
points tab 1.215
enter 1.50
properties tab 1.217 escape 1.51, 1.107
resultant tab 1.217
function 1.27
h-Discretization 1.87, 2.122
tab 1.50
Heat Transfer analysis Label 2.38
convective boundary
Laminate properties 3.13, 3.101, 3.188
condition 3.286
Cartesian 3.238, 4.262
flux boundary conditions 3.282 cylindrical 3.241, 4.271
heat source boundary
example 3.15, 3.199
condition 3.284
handbook update 1.219
material properties 3.280 nonlinear update 4.274
temperature boundary License
condition 3.284 features 4.320
verification example 3.292
floating 4.320
Help 1.50 local daemon 4.321
Hoops 1.23
management 4.319
Icon
management trouble-shooting 4.321
compute solution 1.13
node-locked 4.320
create model 1.13 server daemon 4.321
handbook library 1.15
Line
view results 1.14 create 2.28, 3.177
Independent Variable 1.189 object 2.27
Input evaluation 1.51
types 1.55
Input file 3.29 Linear analysis
Installation introduction 1.103
password 4.321
Load
Integration 1.167 assign 3.19, 3.75, 3.105, 3.150, 3.175,
Interface 3.179
I conventions 1.9, 1.49, 1.101, 2.25 beam traction 3.246
overall 1.9, 1.101 bearing 1.79, 3.21, 3.152, 4.34
Intersection curve 2.59 body force 1.78, 3.20, 3.76, 3.151,
Intrinsic Functions 1.51 3.246, 4.248
Iteration Limit 1.109 check 1.80, 3.153
J-integral
computation 1.172, 4.126, 4.157

concentrated force 1.79, 3.21, 3.153, library 3.10
3.246, 4.248 orthotropic 3.146
concentrated moment 3.246 Ramberg-Osgood 4.3
convergence 3.272 stiffners 3.12
dialog box 1.77 summary 1.45
dialog window 3.19 temperature dependent 3.147, 4.79
direction 1.79 temperature dependent (formula
display 1.40 option) 3.148
formula 3.21, 3.76 temperature dependent (table
imposed displacement 4.19 option) 3.148
interference 1.79, 3.21 temperature-dependent 4.79
moment 1.79, 3.21 types 4.254
point 3.107 Mathematical models 4.213
spring 1.78, 3.20, 3.151, 4.247 Measurement analysis
thermal 1.79, 3.21, 3.151, 3.299, 4.248 introduction 1.115
traction 1.78, 3.20, 3.151, 3.179, 4.15, Menu Bar 1.10
4.247 Merge 2.105
types 1.77, 3.19, 3.20, 3.75, 3.105, Mesh
4.247 2D automeshing 2.135
Local 2.17, 2.26 3D automeshing 2.140
Locate 2.16 automeshing 2.127
Mapped mesh generation 2.120 check 2.22, 3.8
Margin class 1.64
check 1.113, 4.1 create 2.21, 2.96, 3.7, 3.141, 3.178
criteria 4.72 design 3.5, 4.128, 4.237, 4.243
Margin Check analysis edit 2.21
execution 4.74 face to face 2.118
introduction 1.113, 4.68 mapped 2.99, 2.102, 2.106, 2.110,
Material orientation 4.203 2.120
Material properties move to 2.21
5-parameter 4.3 refinement 1.88, 2.123
assign 1.75, 3.75, 3.145 revise 2.21, 2.102
bilinear 4.3 select 2.21
browser 1.75 Mesh generation (Automesh)
check 1.76, 3.179 auto 2.129, 2.135, 2.142
coefficients 4.257 boundary layer 2.131, 2.137, 2.143
condition number 1.76, 3.179 crack edge 2.132, 2.137
database 1.45 crack face 2.132, 2.143
define 1.73, 3.10, 3.75, 3.145, 4.3 curve 2.132, 2.138, 2.145 I
elastic-plastic 4.3 extrude 2.129, 2.146
homogeneization 4.263 local 2.133, 2.139, 2.146
isotropic exponential 4.24 PAM 2.148
laminate composites 3.13, 3.103, PAM extrusion mesh 2.157
3.188, 4.262 PAM Input 2.152
laminate orthotropic 3.188 PAM quality control 2.155

PAM surface mesh 2.156 Multi-layer toggle 1.75
simple graded 2.133, 2.147 Multiple solution 1.160, 1.163
Metafile 1.23 Node
Method 1.55 constraint 1.83
2-Point 2.17 create 3.178
3-Point 2.17 editing 2.95
By Sampling 2.17 merging 2.105
Delta 2.19 object 2.92
Intersection 2.19 Nonlinear analysis
Local 2.17 coldwork 4.44
Location 2.16 convergence option 4.7, 4.27, 4.48,
Midpoint 2.17 4.84
Normal 2.20 example 4.9, 4.30, 4.37, 4.157
Offset 2.17 execute 4.26
Offset Parallel 2.19 executing 4.6, 4.83
Point 2.17 fastened structural connections 4.53
Projection 2.17 general nonlinear 4.25
Reverse 2.20 geometric nonlinearity 4.25
Tangent 2.20 heat transfer 4.78
Trim 2.135 introduction 1.108
Working plane 2.16 load step 4.8, 4.28
MidOffset 2.17 margin check 4.68
Midpoint 2.17 material nonlinearity 4.2
Midsides 1.161 plastic zone 4.12
Min/Max residual stresses 4.14
extraction 1.159, 3.43, 3.68, 3.82, springs 4.26
3.125 Normal 2.20
Modal analysis Object
beams 3.263 color 1.55
eigensolver 3.273 creation 2.7, 2.14
execute 3.275 Offset 2.17
frequency convergence 3.260 Offset surface 2.57
introduction 1.110 Operation
plates 3.255 blend 2.2, 2.3
Mode shape 3.260 boolean 2.3, 2.60
Model clip 2.2
creation 1.61, 2.5, 2.25, 2.43 clipping 2.75
summary 1.45 edge blending 2.75
I Model Info face blending 2.81
interface 1.11, 1.196
tab 1.26 P-T
Modeling errors
Parallel 2.19
control of 4.222
Parameters
Moment 1.79, 3.21, 3.106
class 1.27
Move To 2.21

description 1.26 Plotting
expression 1.27, 1.28 buckling shape 3.272
limit 1.27 contour 1.152, 3.41
name 1.26, 1.201 deformation 1.154
special reserved 1.54 deformed shape 1.154, 3.41, 3.82,
tab 1.26 3.122
value 1.27 fringes 1.153
Parametric expression 1.28 mode shape 3.260
Parts 1.92 reference function 1.155
Path 1.165 standard functions 1.154, 3.41
P-discretization Z-function shape 1.156
class 1.85 Ply Group 3.102
options 1.86 Point
product space 4.252 definition 2.2
trunk space 4.251 extraction 1.163, 1.164, 3.83
Penalty 1.119 extraction in a local system 1.168
Planar elasticity extraction on a boundary 1.169
plane strain 4.278 load 3.107
plane stress 4.279 method 2.17
Plane 2.18, 2.49 object 2.26
strain 4.278 Polyline 2.33
stress 4.279 Post-processing
Plate bending axisymmetric 3.93
constraint 3.108 buckling 3.272
deformed shape 3.122 fracture 4.133
discretization error 3.116 heat transfer 3.287
distributed load 3.105 modal 3.260
edge moment 3.106 operations 1.133
edge resultants 3.126 planar 3.34
edge shear 3.106 plates 3.116
example 3.128 Preferences 1.49
hierarchic models for 3.111, 4.226, Pre-stress 1.111, 3.258
4.238 Principal directions 3.53
laminated material 3.101 Print 1.22
load 3.105 Product space 1.86
material properties 3.100 Projection 2.17
modeling error 3.117 Properties 1.170
point load 3.107 P-version 1.3
Reissner-Mindlin model 4.233, 4.281 Quad mesh 2.120 I
shear correction factors 4.236 Quality Assessment 1.8, 1.145, 1.153
Plate model 1.105 Radiation 4.82
Plates Ratio 2.141
hierarchic models 4.226 Recover database 1.21
mesh design 4.237 Rectangle 2.38
Reissner-Mindlin 4.281 Redo 1.25, 2.9

Refinement Shrink 1.31
graded 2.124 Solid
uniform 2.123 body 2.2
Repeat booleans 2.2
object creation 2.10, 3.177 primitives 2.1
Replace 2.22 Solution
Report class 1.84
input data 1.143 execution 1.13
template 1.137 Solution ID’s
window 1.135 specification 3.28
Reset 1.37 SOLVE! 1.105
Residual stresses 4.14, 4.21 Solvers 4.315
Resultants check 1.148 eigensolver 4.317
Results 1.14, 1.107, 1.133 iterative 4.315
Reverse 2.20, 2.52 sequential 4.315
Revision 2.21 Sphere 2.48
Right mouse button 1.58 Spline
Rigid Body 1.82 curve 2.35
Rolling ball 2.81 surface 2.51
Ruled surface 2.56 Spring constant
Rules linear 3.26
tab 1.28, 1.203 Stacking Sequence 3.103
Run Mode 1.106 Standard functions
Sample 2.17 3D beams 3.247
Scratch directory 3.33, 3.115, 3.165 axisymmetric elasticity 3.80
Section Property heat transfer 3.290
class 1.65 planar elasticity 3.37
database 1.67 plate bending 3.119
definition 3.58 Strain
functions 1.53 energy 1.147
in load formulas 3.154 invariants 1.177
library 1.66 Strains
Select 2.21 components 3.37, 3.80, 4.282
Session 1.23 equivalent 3.39, 3.82
log 1.23 material directions 3.238
Set browser 1.56 principal 3.38, 3.81
Sets 1.91 thermal 3.234
Settings total/mechanical 3.234
I convergence criteria 1.189 Stress
results 1.134 components 3.39, 3.81, 4.282
solution 1.107 concentration factors 3.51
Shadow database 1.18 equivalent 3.40, 3.82, 4.283
Shear correction factor 3.100, 4.236 material directions 3.238
Shear force 3.66, 3.106 principal 3.40, 3.81
Sheet body 2.2 resultants 3.44, 3.85, 3.120

thermal 3.299 Thickness
Stress intensity factors assign 3.9, 3.99, 3.178
2D-extraction 4.125, 4.133 check 1.71, 4.247
3D-extraction 4.139, 4.141 constant 1.68
computation 4.119, 4.133 display 1.40
definition 4.285 fitting 1.69
limitations 4.120, 4.290 formula 1.69
Subexpression 1.30 Thin solids 3.196
Surface degrees of freedom 3.197
curve 2.58 example 3.199
definition 2.2 multi-layer 3.199
grid 2.16 selecting q-level 3.205
of revolution 2.54 stress/strains in material
orientation 2.46 directions 3.238
parametric 2.45 Tolerance 1.109
types 2.47 Tool 2.3
Symbol Toolbar
equal sign 1.51 animation 1.157
pipe 1.28 attributes 1.35
System display object 1.35
object 2.26 display options 1.36
Tab edit 1.34
navigation 1.50 input grid 1.36
settings 1.189 views 1.31
Tables Tools Menu
class 1.89 command file 1.49
option 3.148, 4.80 mode 1.48
Tabulated cylinder 2.55 set font 1.49
Tangent 2.20 user preferences 1.49
Target 2.3 Torus 2.49
Technique 1.109 Tri mesh 2.120
Temperature Trunk space 1.86
boundary condition 3.286, 3.299 T-stress 4.142
temperature dependent material
properties 4.79 U-Z
temperature dependent
U,V coordinate space 2.120
properties 3.147
Unblank 2.20
units 3.282 Undo 1.25, 2.8
Template I
Uniform refinement 1.88
file 1.138
Units 1.47, 1.62, 1.73, 2.5, 3.2, 3.254,
report 1.137 3.282
Text input 1.56 User Preferences 1.49
Thermal stress intensity factor Vanishing angle 2.103, 3.144
(TSIF) 4.197 View Controls
Thermo-elasticity 4.93

interface 1.43
View Menu
attributes toolbar 1.35
display object toolbar 1.35
display options toolbar 1.36
edit toolbar 1.34
input grid 1.36
views toolbar 1.31
Wet Surface 1.161
Window
dialog 1.11
handbook library 1.15
input 1.13
model 1.11
results 1.14
sizing 1.54
solution 1.13
Wire body 2.3
Work plane 2.16
XY Graph
edit menu 1.188
file menu 1.187
use of 1.182
view menu 1.188
Z-shape 1.156

Master Guide

Încărcat de

Informații document

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Master Guide

Încărcat de

Drepturi de autor:

Formate disponibile



Engineering Software Research & Development, Inc.

4 StressCheck Solutions 101

5 Post-processing Operations 133

6 The Handbook Framework 193

2 Basic Geometry Construction 5

4 Three-Dimensional Elasticity 137

5 Modal and Buckling Analyses 253

6 Conduction Heat Transfer 277

3 Fracture Mechanics 119

4 Material Property Fitting 203

5 Theoretical Background 213

Network License Management 319

Error and Warning Messages 323

For Windows Operating Systems

StressCheck includes portions of FLEXlm license manager version 6.1. Copyright 

Tech Soft America (www.hoops3d.com) supplied the following core technology:

StressCheck incorporates MeshSim™ a product of Simmetrix Inc.

Engineering Software Research & Development, Inc. makes no representations or warran-

From the perspective of managers, StressCheck is a tool for increased productivity

StressCheck is the first finite element analysis program to emphasize both

User’s Guide Chapter 1: Preface 1

StressCheck improves the reliability of computed information while increasing

There is an immediate visual feedback confirming that the data is properly

Why use StressCheck?

By incorporating your proprietary technology into an everyday handbook style

2 Chapter 1: Preface User’s Guide

StressCheck’s unique and advanced post-processing capability allows detailed

StressCheck incorporates a Component Object Model (COM) programming inter-

Who should use StressCheck?

User’s Guide Chapter 1: Preface 3

Frequently asked questions about the p-version

Why is the p-version important?

When was the p-version developed?

4 Chapter 1: Preface User’s Guide

StressCheck features - Elasticity

Model Materials Boundary Solution Output

User’s Guide Chapter 1: Preface 5

Are error estimation procedures available in h-version programs as well?

Does the p-version have clear advantages over the h-version?

Are there significant differences in p-version FEA programs?

6 Chapter 1: Preface User’s Guide

What are the advantages of StressCheck over other FEA programs?

User’s Guide Chapter 1: Preface 7

What are the recommended quality control procedures in FEA?

a) Linear analysis: Control of the errors of discretization.

b) Nonlinear analysis: Control of the errors of idealization.

8 Chapter 1: Preface User’s Guide

User’s Guide Chapter 2: StressCheck Interface 9

10 Chapter 2: StressCheck Interface User’s Guide

“Model Info” Icon

FIGURE 2 Model Info interface.

User’s Guide Chapter 2: StressCheck Interface 11

FIGURE 3 Input interface.

12 Chapter 2: StressCheck Interface User’s Guide

Some dialog windows contain a summary of the data records corresponding to a

“Compute Solution” Icon

Solver Option Interface Solver Execution Interface

FIGURE 4 Solver interface.

User’s Guide Chapter 2: StressCheck Interface 13

“View Results” Icon

Command Buttons FIGURE 5 Results interface.

14 Chapter 2: StressCheck Interface User’s Guide

supported by StressCheck, i.e. Error Estimation, Plot, Point Extraction, Min/Max