Documente Academic
Documente Profesional
Documente Cultură
The purpose of this resource book is to provide cost effective and convenient
access to some of the material produced as part of the CCOPPS project,
including the commercially available learning modules. Hopefully it will also act
as an incentive for individuals to enrol on the work-based learning modules
developed as part of CCOPPS and now offered by the University of Strathclyde
(http://www.mecheng.strath.ac.uk/cpd.asp).
1.1 Introduction
The CCOPPS Educational Base (available at http://www.ccopps.eu/) defines a
set of recommended minimum educational requirements, for users of finite
element analysis systems, in the Power and Pressure Systems Industry. The
requirements provide a transparent, understandable description of the abilities
that should be apparent in staff carrying out analysis and simulation in this
industry sector.
The Educational Base was developed as part of the EU-funded CCOPPS project
(Certification of Competencies in the Power and Pressure Systems Industry) and
it followed from a study of the educational and training needs of industry.
All of the above are extensive areas of study in their own right and many
researchers spend their entire professional lives working in only one or two of
these. It is emphasized therefore that the competencies expressed in the
CCOPPS Educational Base represent a recommended minimum for practising
engineers and analysts using FEM and who will invariably already have a first
degree in engineering or a related discipline. For such engineers and analysts,
moving into a new area of analyses, or perhaps needing to refresh existing
competencies, this educational base and supporting material, should provide a
useful resource. It therefore focuses on aspects of technology that will be
relevant to current engineering practice. The aim is that these should be
sufficient to allow safe and effective use of modern analysis and simulation tools.
They should also provide a basis for further educational development in these
and related areas. They do not however consider the training needs of the
particular application software being used, which is also clearly an
important area of competence for effective use of the technology.
The route for the formation and development of engineers and analysts who use
modern analysis and simulation tools to develop innovative, safe and competitive
products in today’s marketplace are a progression of learning experiences
beginning with an undergraduate degree, continuing through initial professional
development in early employment and thereafter extending throughout their
professional life. Once qualified, professional engineers and analysts today are
expected to keep up to date by continued learning throughout (and in support of)
their career through Continuing Professional Development (CPD). Many of the
competencies covered in the Educational Base would normally be developed as
part of a formal postgraduate course, as part of a series of training courses or as
part of on-the-job learning.
While the CCOPPS Educational Base has some similarities to the above, there
are differences and it may be useful to consider these briefly. Firstly, the
Educational Base statements guiding the design of undergraduate courses and
membership of professional bodies are much broader in their scope and
generally less specific in the statements of learning outcomes. For example the
general learning outcomes will address practical skills in laboratories as well as
transferable skills in working with others, information retrieval and planning self-
learning. The specific learning outcomes will address the underpinning science
(materials, mathematics, mechanics, thermodynamics, fluid mechanics, vibration,
production and manufacture, control etc) as well as industrial context of the
subject being studies. Design, creativity, project management, finance,
environmental issues, sustainability, management, ethics, risk, health and safety
are all indicative of the wider scope of such educational bases. Secondly, the
individual modules forming part of a course of study will generally contain no
more than half a dozen (often less) “higher level” learning outcomes. In this
respect subject or module descriptors in degrees and diplomas may differ, with
the latter containing more learning outcomes and often embodying different
assessment regimes to certify competence. The reasons for this are partly
historical. However, there is no doubt that assessing and re-assessing a large
number of competencies in individual subject areas by the traditional examination
methods inherent in existing University degree courses, would prove problematic.
Assessing competencies using the appropriate tool is clearly important and much
guidance is available for staff involved in this process. However the CCOPPS
project was not concerned with the practicalities and constraints within any
educational system or sector and therefore had the luxury of developing a
detailed list of competencies that can be used in a formative sense by individuals
engaged in continuous professional development in the work place. In this case,
self-assessment or assessment by a mentor or line-manager is likely to be the
order of the day, rather than by means of the invigilated examination systems
typical of University systems. This obviously does not preclude use of such
learning outcomes in a formal assessment system, but in this case the
development of module descriptors for use as part of a formal degree
programme would almost certainly involve a focus on delivery and formal
assessment of a fewer selected number of “higher level” learning outcomes.
It may also be useful at this point to introduce some of the educational rationale
behind the use of statements of competence or learning outcomes in an
educational base. Learning outcomes in this context are statements of what an
analyst should be able to do at the end of a programme of learning. The
emphasis on doing is what distinguishes a learning outcome approach from one
based on more intangible ideas related to educational aims, objectives and a list
of syllabus content. For some, the main problem with such a syllabus is that it
can give little or no indication of depth or approach to any particular topic and
also time spent. Many academics in fact value such looseness and academic
freedom, to place emphasis where they see fit. There can also be course
management advantages and disadvantages associated with such freedom.
However, in terms of providing the employer, or the student with details of what
competences they should have at the end of the course, it is argued that it is less
than satisfactory, even when notional hours are provided in such syllabi. A list of
detailed learning outcomes, on the other hand, provides employers and students
with useful information on what competences should be in place at the end of the
learning experience. Learning outcomes also help instructors to design and
select suitable resource material more effectively, to select the appropriate
method of delivery and to select appropriate assessment methods. It might also
be argued that learning outcomes are particularly useful where resource material
and learning activities are going to be designed by many different people, in
order to be used by others, perhaps in a distributed environment. This point is
particularly relevant in today’s diverse and distributed finite element community
and the way that this Educational Base may be used in the development and
selection of supporting resource material. It is concluded therefore that learning
outcomes or statements of competence are the natural way to frame educational
requirements in this environment.
It is also argued that, given the various levels in the cognitive area, discussed
below, the process of identifying suitable material (text books, short courses,
web-based learning modules etc), that can be used in satisfying the learning
outcomes, should involve more than simply identifying textual information aimed
solely at imparting knowledge. Any material specified should allow those using it
to develop competence in the so-called higher cognitive levels (if not also in the
affective and psychomotor areas), even if only in a formative manner. This latter
point emphasises the need for diversity in resource material, including
workbooks, case studies, worked examples, tutorials etc.
Also inherent in today’s holistic and increasingly global view of education is the
concept of level and when used to construct a degree award it is common to
associate learning outcomes with a level of study. A number of generic level
definitions exist across the Higher Education sector and efforts are underway to
produce a standard model. These efforts are often driven by the need to facilitate
mobility (nationally and internationally) amongst students through some form of
Credit Accumulation and Transfer System. The diversity of education systems
throughout the world makes the realisation of this particularly challenging.
It should be recognised that the level of study is different from the level of
performance or standard of attainment achieved in fulfilling the learning outcome.
When used as part of a formal award such as a degree, learning outcomes would
normally be accompanied by a threshold statement and grade indicators. These
would in turn be used by assessors to categorise and rank student performance.
To illustrate this further development of learning outcomes, it may be useful to
consider an example. Consider the learning outcome Employ a range of post-
solution checks to determine the integrity of FEA results. A minimum acceptable
performance or threshold statement associated with this might be Use 2 or 3
post-solutions checks to determine the integrity of FEA results. An “A grade” or
comprehensive performance on the other hand might be to Identify the most
appropriate post-solution checks and use them to specify the integrity of FEA
results, with a full justification of choices. It should be recognised that even with
this level of detail, there is still scope for any assessor to exercise their
professional judgement in setting any examination instrument and in interpreting
level of achievement. It has already been indicated however that the Educational
Base will find use in an individual assessing their own competence level in an
informal manner or in his manager attesting to a member of staff’s competence
informally, as part of an internal system of staff development, or even in the
development of a Register of Suitably Qualified Staff as part of an internal Quality
Assurance System. In these instances, users may not be interested in
performance level indicators. Performance level statements for Threshold and
Comprehensive have however been included in the Educational Base, for
completeness and to assist with adoption and integration into formal educational
programmes. These are included on the Individual Competence Record Sheet
associated with each statement of competence in each topic area, as shown in
Figure 2. These sheets may be printed out to enable individuals to keep a record
of their achievements.
Figure 2: Individual Record Sheet from the CCOPPS Educational Base
Clearly the first three demands more rigour in assessing individual competence
than the latter.
As shown in Figures 1 and 2, the sheets printed out will contain the following
information:
Most of the above information is for the use of the person engaged in personal
development. The latter two items of information are to encourage and facilitate
the development of company staff development schemes and perhaps a future
Registered Analyst Scheme.
The Educational Base has been developed in a manner that allows future
expansion and modification of the learning outcomes and supporting topic areas.
It is anticipated that these modules will also act as exemplars for the
development of web-based learning material in the other topic areas, as future
funding becomes available. The Educational Base however provides
references to conventional text-based resource material that can also be
used to develop competencies in all areas.
This resource book contains two introductory units from these work-based
learning modules.
Before investigating the unit from the CCOPPS learning module, please have a
look at the readme file on the following link:
http://personal.strath.ac.uk/j.wood/CCOPPS_FEA/readme.htm
This unit from the CCOPPS Introduction to FEA of Pressure Systems and Components work-
based learning module is available by simply clicking on the link below:
http://personal.strath.ac.uk/j.wood/CCOPPS_FEA/home.htm
As can be seen, the structure of the full work-based learning module is available, although access
is restricted to this unit only. This enables readers to browse the module content and structure to
some extent, before registering for the full module. Registration also allows access to the course
tutors for 5 months.
The Element Selection self-test quiz from the Basic Modelling unit in this module is available by
clicking on the following link:
http://personal.strath.ac.uk/j.wood/CCOPPS_FEA\quiz\Element_Selection\eleme
nt_selection_quiz.html
3. An Introduction to Pressure Vessel Design by Analysis.
Before investigating this unit from the CCOPPS learning module, please have a
look at the readme file on the following link:
http://personal.strath.ac.uk/j.wood/CCOPPS_FEA/readme.htm
This unit from the CCOPPS Introduction to DBA of Pressure Systems and Components work-
based learning module is available by simply clicking on the link below:
http://personal.strath.ac.uk/j.wood/CCOPPS_DBA/home.htm
As can be seen, the structure of the full work-based learning module is available, although access
is restricted to this unit only. This enables readers to browse the module content and structure to
some extent, before registering for the full module. Registration also allows access to the course
tutors for 5 months.
The DBA Basics self-test quiz from the Introduction to Pressure Vessel DBA unit in this module is
available by clicking on the following link:
http://personal.strath.ac.uk/j.wood/CCOPPS_DBA/Quiz\Quiz_DBA_basics\Quiz_
DBA_basics.html
4. Worked Examples and Tutorials.
The following are a selection of the worked examples and tutorials available in
the work-based learning modules developed as part of CCOPPS and now
offered by the University of Strathclyde. The following link provides further details
on costs and how to enrol: (http://www.mecheng.strath.ac.uk/cpd.asp). The
modules contain a further 55 worked examples.
To access the solid models contained in the following worked examples and
tutorials you have to use the freely available Adobe Reader 8.1 or later:
http://www.adobe.com/downloads/.
WORKED EXAMPLE
DEFINITION Page 1 of 5
Statement of Purpose:
The main purpose of this example is to demonstrate the use of 2D planar elements and
axisymmetric elements to model a long thick cylinder under different loadings: internal
pressure, non-uniform temperature field, rotation about its centre line and a shrink fit.
Geometry:
Ti b
T= ln
b r
ln
a
Case 4: the interference shrinkage, δ=1E-4m, at
the bore of the outer cylinder i.e. inner
radius=0.1999m.
WORKED EXAMPLE
DEFINITION Page 3 of 5
p (b 2 + c 2 )
The hoop stress at this position σ h =
c2 − b2
Idealisations:
The cylinder is long enough and loadings are symmetric so that the cross-section remains
plane during deformation. Due to the symmetry, only a quarter of the cross section need be
modelled as shown in the following figure. An axisymmetric idealisation is also possible as
shown.
WORKED EXAMPLE
DEFINITION Page 4 of 5
Further Considerations:
1. Reduce the model to say a 10 degree sector (Mesh ABFE) and apply suitable
constraints along edge EF. Compare results with previous model.
5. How much would you have to heat the outer cylinder up by so that it just slipped onto
the inner cylinder, for load case 4?
6. What rotational speed would cause loss of contact at the interface in load case 4.
Useful references:
1. S. Timoshenko, Strength of Material, Part II, Advanced Theory and Problems, 3rd
Edition, D. Van Nostrand Co., Inc., New York, NY, 1956, pp. 208, loading case 1, pp.
217, loading case 2, pp. 231, loading case 3, pp. 211, loading case 4.
SOLUTION
Page 1 of 13
Idealisation:
The cylinder is long enough and loadings are symmetric so that the cross-section remains
plane during deformation. Due to the symmetry, only a quarter of the cross section need
modelled ABCD as shown in the following figure. A smaller sector model is possible, although
this would involve the imposition of constraints in a non-global axis set. Both 2D plane stress
and plane strain elements may be used for the 90 degree 2D solid model, although only the
plane strain idealization is equivalent to the axisymmetric model shown. An axisymmetric
idealisation is also possible as shown. Again a single element wide idealisation should be
possibly in the absence of end effects.
SOLUTION
Page 2 of 13
Mesh:
Axisymmetric model
2D planar model
(Plane Stress or Plane Strain)
SOLUTION
Page 3 of 13
Axisymmetric model
2D planar model
Axisymmetric model
2D planar model: An 8-node bi-quadratic plane stress quadrilateral element with reduced
integration, CPS8R (stress in normal direction is zero) and an 8-node bi-quadratic plane strain,
quadrilateral, reduced integration element, CPE8R (strain in normal direction is zero).
Figure 1 Mesh convergence study for the axisymmetric model, load case 1
Figure 2 Hoop stress fringe plot for the axisymmetric model, load case 1
Figure 3 Radial stress fringe plot for the axisymmetric model, load case 1
SOLUTION
Page 5 of 13
From the previous convergence study, it was found that 2 elements provided a reasonable
estimate of maximum values, The model with six elements in the radial direction is used to
compare results with theory. For the 2D planar model, a mesh convergence study was also
carried out by fixing the number of elements in the radial direction at 6 and increasing the
elements in the hoop direction. This study is however providing more of an indication of the
effects of element distortion than mesh refinement.
Figure 4 Mesh convergence study for the 2D planar model, load case 1
Figure 5 Hoop stress fringe plot for the 2D planar model, load case 1
It is clear from the above figure that the stress fringe plot is not quite smooth. This may be due
to the typical variation observed between corner and midside node results. This difference will
reduce with mesh refinement in the hoop direction. Such variations are also sometimes a
result of fringe plotting algorithms. A check of corner and nodal values will help to confirm the
cause.
SOLUTION
Page 6 of 13
Figure 6 Radial stress fringe plot for the 2D planar model, load case 1
As the same results were obtained for plane strain and plane stress elements, in the Figure 7,
only one set of the results are plotted with the name “2Dplanar”.
Figure 8 Mesh convergence study for the axisymmetric model, load case 2
Figure 9 Hoop stress fringe plot for the axisymmetric model, load case 2
Figure 10 Radial stress fringe plot for the axisymmetric model, load case 2
SOLUTION
Page 8 of 13
From the axisymmetric model convergence study, it is found that six elements along radial
direction are able to provide excellent results. For the 2D planar model, a mesh convergence
study was also carried out by fixing the number of elements in the radial direction at 6 and
increasing the elements in the hoop direction.
A coarse mesh with six elements in the hoop direction is used in the following analysis.
Figure 11 Hoop stress fringe plot for the 2D planar model, load case 2
Figure 12 Radial stress fringe plot for the 2D planar model, load case 2
SOLUTION
Page 9 of 13
3.00E+08
2.50E+08
2.00E+08
Theory_radial
2
) Theory_hoopl
m 1.50E+08 Axisym_hoop
/
N
( Axisym_radial
ss
e plane_strain_hoop
rt 1.00E+08 plane_strain_radial
S
plane_stress_hoop
5.00E+07 plane_stress_radial
0.00E+00
0.08 0.1 0.12 0.14 0.16 0.18 0.2 0.22
‐5.00E+07
Radial distance (m)
Figure 13 Comparison with theoretical results for load case 2
By comparing the analytical stress solutions for load case 1, 2 and 3 and mesh convergence
studies for load case 1 and 2, it is reasonable to say that 6 elements along radial direction is
able to produce a satisfactory solution. The figure 14 proves our judgement. The numerical
hoop stress results show a large discrepancy from the analytical solution. A 10 percent error
was predicted by the axisymmetric model. Hence a finer mesh with 10 elements in the radial
direction was created. In the fine planar mesh, 10 elements in the hoop direction were used.
Figure 15 presents the comparison with theoretical results using fine meshes.
SOLUTION
Page 10 of 13
2.00E+06
1.50E+06
1.00E+06
5.00E+05
) 0.00E+00
2
m
/ 0.08 0.1 0.12 0.14 0.16 0.18 0.2 0.22
(N
s s ‐5.00E+05
re
tS
‐1.00E+06
‐3.00E+06
Radial distance (m)
Figure 14 Comparison with theoretical results for load case 3, 6 elements in radial direction.
2.00E+06
1.50E+06
1.00E+06
5.00E+05
2
) 0.00E+00
m
/ 0.08 0.1 0.12 0.14 0.16 0.18 0.2 0.22
N
( ‐5.00E+05
ss
e
rt
S ‐1.00E+06
‐3.00E+06
Radial distance (m)
Figure 15 Comparison with theoretical results for load case 3, 10 elements in radial direction.
SOLUTION
Page 11 of 13
Figure 16 Radial stress fringe plot for the axisymmetric model, load case 3
Figure 17 Hoop stress fringe plot for the axisymmetric model, load case 3
Figure 18 Radial stress fringe plot for the axisymmetric model, load case 4
Figure 19 Hoop stress fringe plot for the axisymmetric model, load case 4
24.7 MPa
24.49 MPa 24.48 MPa (0.04%) 27.2 MPa (11%)
(0.8%)
Conclusion(s):
In this example, a thick cylinder was modelled under four different loadings: internal pressure,
non-uniform temperature field, rotation about its centre line and a shrink fit.
SOLUTION
Page 13 of 13
The axisymmetric idealisation, with Plane Strain constraints, provides good agreement with
theory.
Both 2D plane stress or plane strain elements also provide excellent results, as the hoop
stress and radial stress do not depend on the elongation εy.
The axisymmetric idealisation, with Plane Strain constraints, provides good agreement with
theory.
The radial stress results from all the 2D planar models are in good agreement with the
theoretical solution. For the hoop stress distribution, the plane strain element model shows a
little deviation (6.6%), for the mesh used.
The axisymmetric idealisation, with Plane Strain constraints, provides good agreement with
theory.
Not surprisingly, a large discrepancy occurs between the Plane Strain analytical solution and
results predicted by the Plane Stress model, indicating that the assumption that the axial stress
is planar and zero, is not suitable for this case.
The axisymmetric idealisation, with Plane Strain constraints, provides good agreement with
theory.
If you are interested in analysing this type of structure and component, it is highly
recommended that you repeat this exercise with your own FE system and elements therein.
WORKED EXAMPLE
DEFINITION Page 1 of 3
Statement of Purpose:
The main purpose of this example is to demonstrate the use of thin shell elements to model a small
pipeline with an elbow. In addition, simple beam elements with a “flexibility factor” and “stress
intensification factor” are used to quantify the global effect of ovalization.
Geometry:
Idealisations:
Since the pipe mean radius to thickness ratio is 19, thin shell element would be appropriate. Due to
geometry symmetry, a quarter of the pipeline is modelled in a shell element model. The schematic
representation of the FE idealisation is shown as below.
An alternative method of modelling the pipeline is to use beam elements. To include the effects of
ovalization, reduced bending stiffness should be implemented in elements of the elbow. These elements
are highlighted in green colour in the following figure.
The value of the reduced stiffness was obtained from the equation:
1.66 Rt
k= λ=
λ where
r 2 1 −ν 2
WORKED EXAMPLE
DEFINITION Page 3 of 3
R is the radius of the curved section, r is the mean radius of the pipe, t is the wall thickness of
the pipe, and v is Poisson’s ratio. The reference for the above equation is Dodge and
Moore [1].
This gives a value of k = 3.43, so the bending stiffness was reduced by a factor of 3.43. This
was done by reducing the thickness of the bend.
The other approach to reducing the bending stiffness is to reduce the Young’s modulus. The
flexibility characteristic, flexibility factor and stress intensification factor are calculated as below
according to ASME B31.1-2007.
Rt
Flexibility characteristic, λ =
r2
1.65
Flexibility factor, k =
λ
0 .9
Stress intensification factor, i= 2
h 3
Further Considerations:
(1) Study convergence
(2) Compare results with models using elbow elements. How do these elements include the effects
of ovalization, enhanced flexibility and increased stress levels?
(3) Compare results with those from a specialized pipework stress analysis system.
(4) Will any warping at the ends of the structure affect the region around the bend?
(6) Re-run with end loads rather than prescribed displacements and note differences in results.
(7) Do you think large displacement effects will make any difference?
(8) How will the ovalization affect the opening of the bend angle?
(9) Forming pipe bends can result in a thinning of the extrados and thickening of the intrados
regions. How would you model this?
Useful references:
1. Dodge, W. G., and S. E. Moore, “Stress Indices and Flexibility Factors for Moment Loadings on
Elbows and Curved Pipes,” Welding Research Council Bulletin, no. 179, 1972
SOLUTION
Page 1 of 7
Idealization:
Since the pipe mean radius to thickness ratio is 19, thin shell elements would be appropriate.
Due to problem symmetry, a quarter of the pipeline is modelled in a shell element model.
The unit displacement constraint is applied to the node highlighted by the blue point.
1.66 Rt
k= where λ=
λ r 2 1 −ν 2
R is the radius of the curved section, r is the mean radius of the pipe, t is the wall thickness of
the pipe, and v is Poisson’s ratio. The reference for the above equation is Dodge and
Moore [1].
This gives a value of k = 3.43, so the bending stiffness was reduced by a factor of 3.43. This
was done by reducing the thickness of the bend.
The other method to reduce the bending stiffness is to reduce the Young’s modulus. The
flexibility characteristic, flexibility factor and stress intensification factor are calculated as
bellow according to ASME B31.1-2007.
SOLUTION
Page 2 of 7
Rt
Flexibility characteristic, λ =
r2
1.65
Flexibility factor, k =
λ
0.9
Stress intensification factor, i= 2
h 3
Mesh:
The mesh shown contains 20 elements along the straight section, 18 along the bend as
modelled and 16 elements circumferentially as modelled. It should be possible to obtain
satisfactory results with a coarser mesh.
SOLUTION
Page 3 of 7
The mesh shown contains 20 elements along each straight section and 32 elements in the
entire bend. It should be possible to obtain satisfactory results with a coarser mesh.
The shell model was meshed using 8-noded elements, in this case ANSYS element SHELL93.
The deformation shapes are quadratic in both in-plane directions.
The beam model was meshed using 3-node quadratic beam elements, in this case ANSYS
element BEAM 189.
Fig 5. Stress plot of shell element model. Small displacement assumption used.
Fig 7. Axial strain plot of beam element model without reduced bending stiffness.
Fig 8. Axial strain plot of beam element model with reduced bending stiffness.
SOLUTION
Page 6 of 7
Fig 9. Plot of equivalent stress around 180 degrees of the pipe at the mid span of the bend.
Beam 2 includes reduced bending stiffness in the bend and Beam 1 does not include any
reduced bending stiffness. The shell plot has results from the top and bottom of the shell
element (outer and inner surfaces).
N/A
Conclusion(s):
The effect of ovalization in a pipe bend is to enhance bend (and pipeline) flexibility. This in turn
will reduce terminal reactions at the nozzles on vessels connected by the pipeline. The
ovalization will however result in an increase in the stresses local to the bend.
The shell model displays ovalisation effects with the highest stresses occurring at the sides of
the bend (not at the top and bottom outer fibres if treated as a beam). Beam models cannot
include ovalisation effects directly, which is why a reduced bending stiffness model was
created to simulate this effect. The stress plot on figure 10 shows the difference between the
beam element models and the shell element model. The two beam element models show a
stress distribution which follows
σ = My/I and the shell element model has a completely different stress plot due to the
ovalisation effect.
Table 1 shows that both the two beam models with reduced stiffness give a close result to the
reaction forces of the shell model, 9.39% and 8.3% differences for reduced cross section and
reduced young’s modulus models, respectively. The reaction force for the normal beam model
is approximately twice that of the others.
It should be noted that a displacement controlled loading rather than a load controlled loading
is applied at the pipe ends. The stresses are therefore secondary (as per Pressure Vessel
Code definitions) and self-limiting, thus the beam model with reduced young’s modulus
produces the lowest stresses, and the beam model with reduced cross section gives the
highest stress values, as shown in Figure 9 and Figure 10.
If you are interested in analysing this type of structure and component, it is recommended that
you repeat this exercise with your own FE system and elements therein.
WORKED EXAMPLE
DEFINITION Page 1 of 3
Statement of Purpose:
The main purpose of this example is to demonstrate the use of 2D plane elements to calculate
the stress concentration factor for an elliptical hole in a pressurized thin cylindrical vessel.
Geometry:
WORKED EXAMPLE
DEFINITION Page 2 of 3
Idealisations:
Since the radius to thickness ratio is 100 and membrane stresses dominates in the cylinder,
the problem may be analysed as a flat plate. The schematic representation of the model with
2D plane stress elements is shown below:
Further Considerations:
(1) Make sure results are independent of “plate” width L.
(2) Model the actual cylindrical vessel with 3D shell elements or 3D solids rather than a “2D
solid” idealisation, compare results.
WORKED EXAMPLE
DEFINITION Page 3 of 3
(3) Compare results with tutorial BMT4 for the stress concentration factor for a circular hole
in an infinite plate.
(4) At what R/t ratio (for a fixed a/b ratio) would such an approach become inaccurate
within 5%?
(5) Is this approximation, which has its roots in hand calculations and early FEA, now worth
doing?
(6) Is there a better shape of hole in such a cylinder?
(7) What is the best shape for a pressurized sphere?
Useful references:
1. R.E. Peterson, Stress Concentration Factors, John Wiley, 1974.
SOLUTION
Page 1 of 5
Idealisation:
Since the radius to thickness ratio is 100 and membrane stresses dominate in the cylinder, the
problem may be analysed as a flat plate i.e. the effects of curvature will be negligible (try
modelling in 3D to check). The schematic representation of the model with 2D plane stress
elements is shown below (plane stress is the appropriate assumption given the thickness of
the vessel):
Mesh:
The model was created and solved using ANSYS v11. The element used was an 8-noded 2D
plane stress element, PLANE82 with the plane stress option.
Fig 7. 1st principal stress plot around elliptical hole, edge AE.
SOLUTION
Page 5 of 5
N/A
Conclusion(s):
For this model the SCF is defined as the ratio of maximum stress to hoop stress (PR/t) which
for this model is equal to 1x108 N/m2. This provides a converged stress concentration factor of
1.54. This compares with a value of 1.5 in ref.[1].
The convergence results in table 1 shows that capturing the elliptical profile accurately is
important for satisfactory results. The coarsest mesh examined had 8 elements along AE and
this provided a 7.22% difference in maximum von Mises stress.
If you are interested in analysing this type of structure and component, it is recommended that
you repeat this excerise with your own FE system and elements therein.
WORKED EXAMPLE
DEFINITION Page 1 of 2
Statement of Purpose:
The main purpose of this example is to carry out an elastic analysis of a flush cylindrical nozzle
in a spherical vessel, which is subjected to internal pressure only and to determine the “Limit of
proportionality” for this configuration.
Geometry:
Analysis Type(s): Material:
Linear material, static, small displacement. Nozzle, with Young’s Modulus = 200GN/m2;
Poisson’s Ratio = 0.3, yield stress=302.7
MPa.
WORKED EXAMPLE
DEFINITION Page 2 of 2
Given the geometry and loading shown, the problem is idealised as a 2D axisymmetric model.
By calculating the decay lengths of a cylindrical nozzle and a spherical vessel subject to
internal pressure, the size of the model is determined, i.e. L and φ. Constant hydrostatic end
pressure imposed along EF to simulate end cap effect. Radio edge CD is constrained so that
no movement takes place in the hoop direction.
Further Considerations:
1. DINNO K.S, GILL S.S., “An Experimental Investigation into the Plastic Behaviour of
Flush Nozzles in Spherical Pressure Vessels”. International Journal of Mechanical
Sciences, Vol. 7, pp. 817-839, 1965.
SOLUTION
Page 1 of 6
Given the geometry and loading shown, the problem is idealised as a 2D axisymmetric model.
By calculating the decay lengths of a cylindrical nozzle and a spherical vessel subject to
internal pressure, the size of the model may be determined, i.e. L and φ. For simplicity, in the
first instance, a 90 degree sector is modelled. Uniform hydrostatic end pressure imposed along
EF to simulate end cap effect. Edge CD is assumed to be a symmetry boundary.
rO
rI
Fig. 1. Idealization.
SOLUTION
Page 2 of 6
Mesh:
It should be noted that FE systems may have different rules regarding the modelling of
axisymmetric problems. In this particular case, the axis of symmetry has to be global Y and the
structure must be in the positive X-Y quadrant.
The model was created in ANSYS v11 and meshed with PLANE82, an 8-noded quadratic solid
of revolution element. A comparison with Mechanica adaptive ‘p’ elements (Wildfire 3) is also
shown.
Fig. 3. Von Mises stress plot at the intersection of nozzle and sphere.
3.0E + 08
H oop s tres s dis tribution along the ex ternal
∞ ∞
A ns ys
2.5E + 08
W eld edge
s urfac e of ves s el (1 N/m )
2
E x perimental
Theoretic al Hoop S tres s 2.0E + 08
1.5E + 08
1.0E + 08
5.0E + 07
Cylinder S phere
0.0E + 00
‐0.4 ‐0.3 ‐0.2 ‐0.1 0 0.1 0.2 0.3
D is tanc e from the middle of the weld, point A (m)
Fig. 6. Hoop stress distribution along the external surface of the vessel.
SOLUTION
Page 5 of 6
Fig. 7. Mechanica hoop stress distribution along external surface of the vessel. Note inclusion
of end cap in this analysis to show distribution of stress in this region.
Relevant Codes of Practice, Industry Standard and/or Statement of Assessment Criteria:
Conclusion(s):
Figure’s 6 and 7 show the hoop stress plots from Ansys and Mechanica at the intersection of
nozzle and sphere, the maximum stress occurs at the weld toe as expected. For the Ansys
model, theoretically elastically, this stress should be infinite and the finite element result will
tend to infinity with mesh refinement. It should also be noted that the Mechanica results have
instead a 1mm radii at the toes of the welds. This approach is sometimes used to obtain “hot-
spot” stresses for fatigue (see FEA module unit).
When the vessel is under an internal pressure of 6.06 MPa, the hoop stress at point A from
numerical model is 214.67 N/m2 comparing with the experimental stress, 232.97 N/m2 – a
-7.9% difference. This error remains fairly consistent throughout the range of experimental
values and the trends in both the experimental and numerical results appear to be similar.
However, the theoretical hoop stress remote from the weld generally compares quite well with
both the Ansys and Mechanica results.
ANSYS EXPERIMENT MECHANICA
HOOP STRESS at MIDDLE
OF WELD FACE 214.64 232.97 218.13
(N/sq.mm)
MAXIMUM PRINCIPAL
STRESS at CROTCH 207.61 - 209.11
(N/sq.mm)
Table 1
SOLUTION
Page 6 of 6
The interesting forms of the stress distribution in the regions of the flat head and the weld
should also be noted. Furthermore, table 1 shows the stress values at both point A and the
crotch corner as determined by Ansys and Mechanica. It can be seen that the stress at the
crotch is lower than that at point A which may be unexpected, however the agreement
between the Ansys and Mechanica values is reassuring on the issue.
If you are interested in analysing this type of structure and component, it is recommended that
you repeat this exercise with your own FE system and elements therein.
WORKED EXAMPLE
DEFINITION Page 1 of 3
Statement of Purpose:
The main purpose of this example is to identify the limitations of modelling practices currently in
use, using plate/shell elements, for adequate representation of the stiffness and stresses in
large fabrications containing welded intersections that exhibit a slope discontinuity in shell/plate
midsurfaces.
The stresses and deflections in the fabricated detail shown are to be determined using
common industrial modelling practices. Target solution quantities required for deflection and
stresses have been specified.
Geometry:
Idealisations:
Although the problem can be analysed as 2D, the intention is that it should be representative of
large general plate/shell fabrications. With this in mind, idealisations using general 3D
plate/shell elements are required.
WORKED EXAMPLE
DEFINITION Page 3 of 3
Further Considerations:
(1) Determine the coarsest mesh that would provide you with an acceptable variation from
the following highly refined meshes.
(2) If you have the resources try a 3D solid representation (for a small sector)?
Useful references:
3. Peckover RS et al, United Kindom Offshore Steels Research Project- Phase 1 Final
Report OTH 88 282; UK Department of Energy, 1985.
SOLUTION
Page 1 of 22
The main purpose of this example is to identify the limitations of modelling practices currently
in use, using plate/shell elements, for adequate representation of the stiffness and stresses in
large fabrications containing welded intersections that exhibit a slope discontinuity in the
shell/plate midsurface.
Although the problem can be analysed as 2D, the intention is that it should be representative
of large general plate/shell fabrications. With this in mind, idealisations using general 3D
plate/shell elements and 3D solids are required.
Mesh:
Model 1_1 Solid of Revolution
The highly refined Mechanica adaptive P mesh is shown in Figure 1, with the p-levels (levels of
polynomial refinement) shown in Figure 2. Levels run from 9 (red) to 1 (blue).
Figure 3. Ansys h-element solid of revolution models. The coarse mesh has an element size of
10mm at the weld and the fine model has an element size of 4mm at the weld.
Figure 5. Ansys h-element shell models. The coarse meshes for all shell models have an
element size of 10x10mm at the weld and the fine models have an element size of 4x4mm at
the weld.
Figure 6. Shell model with weld included as a sloping band of elements (Mechanica)
SOLUTION
Page 4 of 22
Figure 8. Shell model with weld included as thicker bands of elements (Mechanica)
SOLUTION
Page 5 of 22
Various h-element models were created and solved using ANSYS v11. Model 1_1 was
meshed with element type PLANE83 which is an 8-noded axisymmetric structural solid
element. Element type SHELL93 was used for models 1_2, 1_6 and 1_7 which is an 8-noded
structural shell element.
The adaptive p-element models were solved using Mechanica 2D solid of revolution elements
and general 3D shell elements.
SOLUTION
Page 6 of 22
Table 1. (M) refers to the Mechanica p-element models and (A) refers to the ANSYS h-element
models. (F) refers to a fine mesh and (C) refers to a coarse mesh.
Deformation in “mm” and stresses in “N/mm2”. Percentage errors are given relative to the
results from the Mechanica model 1_1.
From this table, it may be concluded that all of the idealisations reported are in reasonable
agreement for the result quantities tabulated. The 36% and 26.4% differences for 1_2 should
be considered in terms of the overall magnitude of the quantities themselves. Model 1_2 is the
most flexible of all the models, as expected. The fact that it is also the simplest and most
convenient should also be borne in mind.
Section 1
250
Meridional Stress (N/sq.mm)
Model 1_7
50
0
-50 -7.5 7.5
-100
-150
-200
-250
Figure 11. Meridional stress distribution for Section 1, ANSYS coarse mesh
SOLUTION
Page 8 of 22
Figure 12. Meridional stress distributions for Section 1, ANSYS fine mesh.
Section 1
80
60 Model 1_1
Hoop Stress (N/sq.mm)
Model 1_6
20 Model 1_7
0
-20 -7.5 7.5
-40
-60
-80
-100
Figure 14. Hoop stress distributions for Section 1, ANSYS coarse mesh.
Figure 15. Hoop stress distribution for Section 1, ANSYS fine mesh.
SOLUTION
Page 10 of 22
Section 2
40
Meridional Stress (N/sq.mm)
30 Model 1_1
Model 1_6
10 Model 1_7
0
-10 -10 10
-20
-30
-40
Figure 17. Meridional stress distributions for Section 2, ANSYS coarse mesh.
SOLUTION
Page 11 of 22
Figure 18. Meridional stress distributions for Section 2, ANSYS fine mesh.
Section 2
5
4 Model 1_1
Hoop Stress (N/sq.mm)
2 Model 1_6
Model 1_7
1
0
-1 -10 10
-2
-3
-4
Figure 20. Hoop stress distributions for Section 2. ANSYS coarse mesh.
Figure 21. Hoop stress distributions for Section 2, ANSYS fine mesh.
SOLUTION
Page 13 of 22
A final point worth noting, are the differences that can arise due to the linearization procedure
itself. Figure 22 shows the non-linearised through-thickness distributions for section 1, for
Mechanica Model 1_1. As may be observed, the effect of the weld toe singularity is confined to
the quarter thickness closest to the singularity itself. For this particular problem, the first three-
quarters of the thickness exhibits a perfectly linear distribution. An engineer’s manual solution
to the linearization process would be to simply extend this linear distribution, rather than
employ a mathematical ‘best-fit straight line’ algorithm. In the latter case, the peak component
will influence the bending stress component and will in effect alter the slope of the distribution,
resulting in slightly higher stress values on the surfaces (in this case -79 cf -68 and -139 cf -
118 for the hoop and meridional stresses respectively on the singularity surface).
Before addressing the issue of assessment, it would be useful to consider the general issue of
‘hot-spot’ extrapolation. In this case it is argued that such extrapolation is unnecessary for the
shell idealisations as no singularity exists in these models. Surface extrapolation as
recommended by the International Institute of Welding (see module) will be confined to the
principal stress distributions for Model 1_1. Furthermore, it is clear from the linearised results
that the maximum stresses occur on section 1. Surface extrapolation will be confined to the
vertical shell in this case.
SOLUTION
Page 14 of 22
Surface distributions of meridional and hoop stresses leading up to section 1 are shown in
Figures 23 - 34 for both the inner and outer surfaces. Results for the simple shell model 1_2
are shown for comparison. Vertical lines are shown at locations corresponding to the wall
centreline for the lower plate, the upper surface of the lower plate, the weld toe and 1,2,3
upper shell thicknesses from the weld toe. The vertical lines on the graph enable the form of
the stress distributions to be better appreciated. The two distributions would be in better
agreement if the thin shell distribution were to be displaced by half a lower plate thickness to
the right. While this fact is interesting, it is unnecessary for the purposes of surface
extrapolation of the shell of revolution results. The UKOSRP project (see module) in the study
of joints for offshore structures noted that the distance that such thin shell graphical
distributions had to ‘displaced’ was also a function of the intersection angle as well as the shell
thicknesses.
Figure 24. Outer surface meridional stress distributions, ANSYS h-element, coarse mesh.
Figure 25. Outer surface meridional stress distributions, ANSYS h-element, fine mesh.
SOLUTION
Page 16 of 22
Figure 27. Outer surface hoop stress distributions, ANSYS h-elements, coarse mesh.
SOLUTION
Page 17 of 22
Figure 28. Outer surface hoop stress distributions, ANSYS h-element, fine mesh.
Figure 30. Inner surface meridional stress distributions. ANSYS h-elements, coarse mesh.
Figure 31. Inner surface meridional stress distributions, ANSYS h-elements, fine mesh.
SOLUTION
Page 19 of 22
Figure 33. Inner surface hoop stress distributions. ANSYS h-elements, coarse mesh.
SOLUTION
Page 20 of 22
Figure 34. Inner surface hoop stress distributions, ANSYS h-elements, fine mesh.
From these distributions, various extrapolated hot-spot stresses have been derived using the
linear and quadratic recommendations discussed in the module, as shown in Table 2. It is
realised that in fact such extrapolation is not required for the inner surface, as the fatigue
assessment of the weld root requires use of a nominal stress rather than a ‘hot-spot’ value, as
recommended by the IIW and various Codes of Practice. These issues are addressed in the
module. A comparison is made for this surface non-the-less. Similarly no regard is given to
guidance relating to Type ‘a’ and ‘b’ hot-spots or coarse/fine meshes at this stage.
NB Figures 24, 25, 27, 29, 30, 31, 33 and 34 show the danger of using averaged nodal
stresses at intersections. This is the cause of the discontinuity in the distributions. This error in
the last point of the graph (ie at the intersection) may also affect the extrapolation procedures
in this case. This can be a common problem with graph plotting procedures in FEA systems.
Unaveraged stresses should be plotted for the last point in the distribution. In this regard the
Ansys results should be used with caution, while the Mechanica results have been corrected
for this problem.
SOLUTION
Page 21 of 22
For the outer surface, extrapolation is to the weld toe and for the inner surface it is to the re-
entrant corner corresponding to the full penetration weld root.
SOLUTION
Page 22 of 22
N/A
Conclusion(s):
Statement of Purpose:
The main purpose of this example is to demonstrate the use of axi-symmetric shell elements to
model a cylindrical vessel with a skirt support and study the stresses at the shell intersection.
Geometry:
Linear material, static, small displacement. Steel, with Young’s Modulus = 210 GPa;
Poisson’s Ratio = 0.3.
Idealisations:
Since the geometry, loading and material do not vary with θ, an axisymmetric idealisation is
appropriate. The radius to thickness ratio is 100, indicating that the thin shell representation
would be appropriate.
Further Considerations:
(1) Identify other likely axisymmetric loadings.
(2) Study convergence.
(3) Plot graph of meridional and hoop stresses along edge BD and AD and identify location
of maximum bending. Comment on the forms of the distributions and the nature of the
results at the intersection. Compare the decay lengths with the standard formulae for
edge loaded cylinders and spheres in notes. Try imposing a boundary condition at D to
see if the significant results change.
(4) Where would you check for possible buckling? Would an axisymmetric (non axi-Fourier)
WORKED EXAMPLE
DEFINITION Page 3 of 3
Useful references:
1. D., Hitchings, “Linear Statics Benchmarks”, NAFEMS Report LSB2, Nov, 1987.
SOLUTION
Page 1 of 10
Since the geometry, loading and material do not vary with θ, an axisymmetric idealisation is
appropriate. The radius to thickness ratio is 100, indicating that the thin shell representation
would be appropriate.
The lack of some form of constraint at D for this loading is not entirely practical. However
target results are available for this scenario. Results have also been provided for a pressure
end-cap effect i.e. with meridional stresses in the cylinder. This would normally be simulated
by applying a force (F) equal to the internal pressure multiplied by the internal cross-sectional
area (P x Ai).
SOLUTION
Page 2 of 10
Mesh:
The mesh contains 50 elements along vessel head as well as the upper and lower cylinders. A
mesh spacing ratio of 4 was used for each section with finer elements towards point C.
It should be noted that FE systems may have different rules regarding the modelling of
axisymmetric problems. In this particular case, the axis of symmetry has to be global Y and the
structure must be in the positive X-Y quadrant.
The ANSYS 3-D shell model used the same mesh which was rotated 90 degrees.
SOLUTION
Page 3 of 10
Fig 6. Mechanica p-element 2-D solid of revolution model, showing automatic refinement in
vicinity of re-entrant corners.
SOLUTION
Page 5 of 10
The h-element models were created and solved using ANSYS v11. The elements used were
SHELL209, which is a 3-node quadratic finite strain axisymmetric shell element and
SHELL93 which is an 8-noded quadratic structural shell element. The solid of
revolution model was meshed with PLANE82, an 8-noded quadratic plane element.
The p-element models were created and solved in Mechanica, using adaptive p
technology. Such elements can utilize up to a 9th order polynomial where necessary.
For the axisymmetric shell and solid of revolution models a coarse mesh and a fine
mesh were used.
It should be noted that all these results are for no-end-cap pressure case.
Maximum stresses at such re-entrant corners (without a fillet) should not be used directly as
the value obtained from any FEA is a function of mesh refinement.
Fig 10. Mechanica 3-D shell stress results. Displacements are exaggerated.
4.00E+08
Outside
3.00E+08 Inside
Outside (with endcap)
Axial stress (N/sq.m)
-3.00E+08
-4.00E+08
Distancefrom
Distance frompoint
point
DD
Fig 11. Un-averaged stress plot for ANSYS 2-D axisymmetric shell model.
SOLUTION
Page 8 of 10
Fig 13. Mechanica plot of axial stress for fine 2-D axisymmetric shell model.
SOLUTION
Page 9 of 10
Both Figure12 and 13 show the danger of using averaged nodal stresses at intersections. This
is the cause of the discontinuity in the distributions.
N/A
In Figure 7, axisymmetric elements have been expanded to show stress contours. In ANSYS,
this is done by issuing a post-processing command “/expand” which allows the creation of a
larger graphic display than represented by the actual finite element analysis model. In this
worked-example, A 3D fringe image is produced for what is in essence a 2D axisymmetric
problem.
Conclusion(s):
This target stress value would not necessarily be the focus in practice, as it is compressive and
there are higher tensile stresses on the inside of the vessel as seen from the 2-D solid of
revolution plot in figure 8 and also the graph of stresses in figure 11.
Figure 11 also shows that with the addition of an end-cap effect, the compressive stress on the
outside is reduced and the tensile stress on the inside is increased. Neglecting the end-cap
effect is un-conservative.
The graphs in figures 12 and 13 show a common problem with results from a shell model at
SOLUTION
Page 10 of 10
intersections. For the common node at the intersection most systems will incorrectly use the
averaged stress when graphing results using such common nodes. This results in an incorrect
evaluation (invariably an underestimate) of the maximum thin shell intersection stress. It is
important therefore to use the un-averaged stress as has been done for the graph in figure 11.
The Mechanica plots in figures 9 and 10 show the stresses throughout the model. As would be
expected, the main region of distortion is at the head-shell intersection due to bending. The 3D
shell model also shows that there is localised bending occurring at the constrained bottom
edge. Away from these areas there are no bending stresses and only membrane stresses
exist.
The results from the thin shell models agree well with the target result of -319.9 MPa, in
general, for all idealisations. As would be expected, displacements are also well represented.
For example, the radial displacement at point C from the ANSYS 2D axisymmetric thin shell
was 0.27641x10-3 m which is close to that obtained from the reference which was 0.2797x10-3
m and a theoretical displacement of 0.2847x10-3 m.
The stresses from the axisymmetric solid of revolution models are in fact more realistic and do
not suffer from the approximations inherent in thin shell idealisations. The exception to this is
at the re-entrant corners on the 2D geometry. At these locations the stresses are theoretically
infinite. Unlike the shell intersection results, which are finite, 2D solid of revolution and 3D solid
idealisations produce un-converged finite results. Such values should not be used directly in
assessment. The FEA and DBA modules examine ways of producing realistic hot-spot
stresses for such re-entrant corners.
If you are interested in analysing this type of structure and component, it is recommended that
you repeat this exercise with your own FE system and elements therein.
WORKED EXAMPLE
DEFINITION Page 1 of 2
The purpose of this example is to perform stress categorisation on a thin un-welded flat end.
This example is taken from the CEN DBA manual example 1.2.
Geometry:
WORKED EXAMPLE
DEFINITION Page 2 of 2
Idealisations:
Axisymmetric model.
Further Considerations:
Students may consider using different mesh densities and higher order elements to check the
effect on the results.
Strength of materials thin cylinder equations, Lame’s equations and circular disk equations can
be used to check the software results at certain classification lines such as at A and E.
Comparison with the “Direct Method” as detailed in EN13445 would provide an interesting an
perhaps simpler approach.
Useful references:
Due to the symmetry of the example, the geometry can be represented by an axisymmetric
model, and using 4-noded quadrilateral elements.
Mesh:
Enlarged view
Elements
N/A
ASME Boiler and Pressure Vessel Code, Section VIII, Rules for Construction of Pressure
Vessels, Division 2 – Alternative Rules; American Society of Mechanical Engineers, 2007
Analysis data
Loading
Material parameters
The following material parameters are used for analysis (given in the example description).
Analysis steps
- Using the FEA software post processor (Ansys software was used in this case), the
linearized stresses along the defined classification lines are extracted.
The Tresca equivalent stress is used. This is given directly by the software so it is not
required to do the calculations manually.
- The linearized stresses are checked against the allowable stress limits. In this example
the allowable stress limits and terminology used are those given in EN13445-3
AnnexC.
Description of Results
The figure shown below shows the elastic stress distribution for the applied internal pressure of
4.2Nmm-2. The maximum stress intensity is at the inside corner with a value of 290.93Nmm-2.
SOLUTION
Page 4 of 9
For each classification line, the stresses are linearized by the FEA software (Ansys). The
graphs shown below show the linearization results for each classification line. The graphs are
plotting the Tresca equivalent stress (Stress intensity, SINT) across the section thickness.
It may be noted that the membrane plus bending plot is not linear across the section thickness.
At the stress component level, the bending stress across the thickness is in fact linear.
However the graphs shown are for the Tresca equivalent stress which due to the nature of its
calculation will result in the contours shown.
The table lists the linearization results for classification line B. The results are grouped by
type, namely; membrane, bending, membrane plus bending, peak and total. The FEA
software lists both the component linearized stresses and the calculated Tresca’s and von
Mises’ equivalent stress. In this example the Tresca’s equivalent stress is used.
The software first linearizes the stresses at a component level and then calculates the
equivalent stress on the results.
SOLUTION
Page 5 of 9
The maximum membrane, membrane plus bending, peak and total stresses for all five
classification lines are listed in the next table. For each classification line, the table also shows
the assigned stress categories, allowable and calculated stresses.
Note:
In this example there is no local stress concentration effects or thermal loads applied.
Therefore no peak stress can exist.
The calculated “peak” stress given by Ansys is a feature of the mathematical
linearization procedure. In this case, the peak stress is simply the difference between
the linearised membrane plus bending stress and the actual membrane plus bending
distribution. In EN13345-3 Annex C, this is referred to as the non-linear part.
As here there is no peak stress, the membrane plus non-linear bending stress
distribution is equivalent to the “total” stress distribution. Therefore, for this case the
calculated total equivalent stress is used in the assessment rather than the linearised
membrane plus bending equivalent stress.
SOLUTION
Page 8 of 9
Note
For classification line D, the bending stress could either be classified as primary or as
secondary. The choice in the classification depends on whether the plate edge bending
reduces the bending stress at the plate centre. Both the ASME and EN13445 codes make
reference to this situation. The ASME code basically says that if the bending moment at the
plate edge is required to maintain the bending stress in the centre region within acceptable
limits, the edge bending is classified as primary (Pb) otherwise it is classified as secondary (Q).
EN13445 says that the classification of bending stress into the primary (Pb) category ensures
that no plastic deformation can occur in the region under consideration during normal service.
So to be conservative it is best to classify the bending stress as primary bending.
Conclusion(s):
From this simple example it is evident that the process of stress classification can sometimes
be unclear, and further calculations (when possible) may be necessary to correctly determine
the appropriate category. The use of conservative assumptions can sometimes be used to
speed up assessment at little or no penalty.
The CEN ‘Design by Analysis Manual’ gives guidelines on how to do stress linearization. Other
guidelines that are based on research work done by the US Pressure Vessel Research
Council project (PVRC) ‘Three dimensional stress criteria’ are given in the ASME code.
The student is encouraged to review these guidelines as a means of learning more on stress
categorization. Obviously the guidelines to be followed need to be the ones given in the
pressure vessel code being followed.
It would be wise to compare the FEA results with flat plate results at the centre of the head at
E and with thick cylinder results remote from end (at A) as a check on the accuracy of the field
stresses. These checks provide necessary validation but not sufficient however.
WORKED EXAMPLE
DEFINITION Page 1 of 3
The main purpose of this example is to determine the plastic load of the given thick
hemisphere when subjected to an internal pressure.
Geometry:
WORKED EXAMPLE
DEFINITION Page 2 of 3
Idealisations:
Further Considerations:
Students may repeat the example to see the effect of using different mesh densities, lower
order elements, large deformation theory, and also using a 3D model. Repeating the exercise
with both 8-noded and 4-noded quads would give a good insight into the minimum acceptable
mesh in 3D. Results from the latter would be expected to be the same as for the axisymmetric
model although computational time will increase considerably.
A single element wide sector model can be used to reduce run-time further, using symmetrical
boundary conditions in a non-global direction.
The student can also repeat the example using different bore and outside diameters or even
maybe find the plastic pressure for a thick cylinder.
Use of the twice elastic slope and/or the plastic work can also be used to calculate the plastic
load. The student is encouraged to compare results and effort required when using the
different plastic work criteria.
The student may also calculate the plastic collapse load using the methods now provided
WORKED EXAMPLE
DEFINITION Page 3 of 3
within the EN13445-3 Annex B and the new 2007 ASME Section VIII Division 2 Part 5. These
codes are covered in the notes of ‘DBA codes of practice’ unit of this module.
Useful references:
Due to the symmetry of the example, the geometry can be represented by an axisymmetric
model, using 8-noded quadrilateral elements.
As stated in the problem description a model with linear elastic-plastic material with Bilinear
hardening and non-linear geometry is used.
Mesh:
Elements
ASME Boiler and Pressure Vessel Code, Section VIII, Rules for Construction of Pressure
Vessels, Division 2 – Alternative Rules; American Society of Mechanical Engineers, 2007
Analysis data
Loading
Maximum internal pressure, P. This value is not known and needs to be calculated.
Material parameters
The following material parameters are used for analysis (given in the example description).
Analysis steps
The objective of the analysis is to obtain a load – displacement graph from which the plastic
load is found using the tangent intersection criterion.
The internal pressure is applied in gradual steps. Preferably, the load increment step size
needs to be sized such that a smooth load-displacement contour is obtained.
The deformation parameter used is the radial displacement at the bore. For this simple
geometry the radial displacement is the same for all points on the inside of the hemisphere.
The solution need not be extended until it fails to converge. Since the objective is to use the
tangent-intersection method, an arbitrary pressure value may be chosen that gives an
adequate load-displacement contour from which to draw the tangent lines.
Description of Results
The internal pressure to apply is not given in the example. From some preliminary analysis on
the FEA model a pressure of 400Nmm-2 appears to be adequate to get a suitable load-
displacement graph.
The figure shown below shows the resulting load-displacement graph. The displacement
taken is in the radial direction.
Tangents were drawn as shown. The plastic load is the pressure value at the intersection point
of the two tangents. This was determined to be 332.5Nmm-2.
For comparison purposes the limit pressure calculated in worked example WE3 is 332.7Nmm-2
which uses an elastic-perfectly plastic material model and small deformation theory.
SOLUTION
Page 4 of 4
The plot below shows the von Mises stress distribution for a pressure Pti=332.5Nmm-2. It is
noted that the inside of the hemisphere has undergone some hardening (stress value is higher
than yield). The hardening process appears to have spread to around half of the material
thickness. On the other hand, the outside of the hemisphere is still below the yield stress.
Conclusion(s):
For the example considered, the plastic load using the tangent intersection method was
determined to be 332.5Nmm-2.
It is noted that due to the effects of strain hardening, the stress distribution is different from that
of the limit analysis model obtained in example WE3. The inside of the hemisphere has
undergone some hardening, while the outside is still below yield.
To summarize:
1. Limit Load (ASME Code definition), with small displacements and elastic-perfectly-
plastic material = 332.7 N/sq.mm (last converged solution). Radial displacement at the
bore = 0.53mm.
2. Plastic Collapse Load (Code definition), with large displacements and strain hardening
= 1396 N/sq.mm (last converged solution). Radial displacement at the bore =
49.95mm. This result is not quite as per the Code in that the code requires "When
using this material model, the hardening behavior shall be included up to the true
ultimate stress and perfect plasticity behaviour (i.e. the slope of the stress-strain curves
is zero) beyond lhis limit” - ASME VIII. This result is therefore unrealistic.
3. “Plastic Load” using the Tangent Modulus method with strain-hardening and large
displacement analysis = 332.5 N/sq.mm.
WORKED EXAMPLE
DEFINITION Page 1 of 2
The main purpose of this example is to perform a buckling check on a torishperical head under
internal pressure according to the requirements given in ASME VIII Div2 part 5.
The check is to be carried out using the type 3 buckling assessment method;
Geometry:
WORKED EXAMPLE
DEFINITION Page 2 of 2
N/A
Idealisations:
Further Considerations:
1. Consider varying the mesh density and using lower order elements.
2. The example may also be attempted using an axisymmetric model in order to
understand the importance of capturing non axisymmetric buckling modes.
Useful references:
1. ASME Boiler and Pressure Vessel Code, Section VIII, Rules for Construction of
Pressure Vessels, Division 2 – Alternative Rules; American Society of Mechanical
Engineers, 2007.
SOLUTION
Page 1 of 5
The geometry can be modelled using shell elements. A full 360 degree model is used to avoid
missing any unsymmetrical buckling modes.
Mesh:
N/A
ASME Boiler and Pressure Vessel Code, Section VIII, Rules for Construction of Pressure
Vessels, Division 2 – Alternative Rules; American Society of Mechanical Engineers, 2007.
Analysis data
Loading
In this example the applied internal pressure is not given. The analysis will be carried out to
determine the maximum allowable internal pressure. Buckles in the toroid region can occur
under the action of internal pressure because of compressive hoop stresses that occur due to
the geometry of the head.
Material parameters
The following material parameters are used for analysis (given in the example description).
Analysis steps
- Linear solution
The linear solution corresponds to the classical / bifurcation solution in order to determine the
first deformation shapes. It is convenient that the results from the linear solution are such that
the maximum deviation from the perfect shape is unity.
SOLUTION
Page 3 of 5
- Non-Linear solution
The non-linear solution corresponds to taking the deviations obtained from the linear solution
and applying them as pre-deformations on the design model. In this case the pre-
deformations are scaled (scaling is done on the deviations from the perfect shape) to
correspond to the allowed tolerances for formed shell heads (Part 4 of ASME VIII div 2). Some
commercial software provide a means to update the geometry with deformations taken from a
previous analysis. This simplifies the process considerably.
The loading is then applied to the model in gradual increments until solution convergence is no
longer possible. The applied load at the last converged solution is then noted.
The maximum allowable value of the internal pressure is then given by applying a load and
resistance factor design (LRFD) to the applied pressure at the last converged solution.
For elastic plastic analysis and internal pressure (global criteria) the factor is 2.4
Description of Results
The analysis was done for the first mode. The displacement plot shown below shows the 1st
mode obtained from the linear solution.
- Non-Linear solution
In this procedure, the nodal displacements are extracted from the linear eigenvalue solution
and superimposed on the original shape (as mentioned previously, finite element systems
usually have a facility for adding a scaled version of the eigenvector onto the original
undeformed shape, for the subsequent large deformation analysis). In this case, the pre-
deformed geometry was adjusted such that the inner surface of the shell deviated from the
specified shape by 1.25% of the inner diameter D (refer to Part 4 of ASME VIII div 2).
Therefore the pre-deformed geometry was adjusted such that the maximum inner diameter
difference from the mean value was 24.75mm.
In the non-linear solution the pressure is applied in gradual steps until the solution failed to
converge.
Conclusion(s):
From the buckling analysis carried out the maximum internal pressure that can be applied is
1.25Nmm-2. This value compares with ???? for the linear buckling analysis.
It is interesting to note that the linear eigenvalue buckling analysis resulted in a buckling load
of 32 N/sq.mm. Following Type 1 buckling analysis (ASME VIII Div2 Part5) this requires a
design factor of 16.2 (2/βcr = 2/0.124) that should then be applied to the Euler buckling load to
obtain the design pressure. Therefore Type 1 buckling analysis results in an allowable
pressure of 1.98 N/sq.mm. It is interesting to note that in this case the Type 1 (simpler)
buckling analysis results in a larger allowable external pressure than that calculated using
Type 3 (non-linear) – 1.25 N/sq.mm.
5. Pressure Vessel Related Images.
If anyone would like to add to this collection of images, please send any images,
with information relating to what it is and the image source to:
jim@wood-home.myzen.co.uk.
Enjoy!
Cockenzie Power Station, Scotland, drum failure. Right-hand end of steam drum.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Cockenzie Power Station, Scotland, drum failure. Fractures at nozzles.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Cockenzie Power Station, Scotland, drum failure. Crack on nozzle.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Dounreay Nuclear Power Station – the sphere under construction.
Image source: unknown.
Dounreay Nuclear Power Station – the finished sphere.
Image source: unknown.
Dounreay Nuclear Power Station – the sphere under construction.
Image source: unknown.
Babcock and Wilcox, Scotland. Injector vessel for proton synchrotron at UKAEA Harwell.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox on site construction of Hinkley power station. B&W 400 ton Goliath crane in action.
Image source: Babcock and Wilcox, Renfrew, Scotland.
A model of Hinkley power station – not a CAD system in sight!
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox fabrication shop. Construction of Hinkley power station diagrid.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox fabrication shop. Construction of blower casing for Hinkley power station.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox fabrication shop. Construction of cascade corners for Hinkley power station. I spent too many years of
my life studying mitred pipe bends!
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox fabrication shop. Construction of course 2 for the reactor vessel for Hinkley power station.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox fabrication shop. Construction of course 5 for the reactor vessel for Hinkley power station.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox fabrication shop. Construction of internal skirt for Hinkley power station.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox fabrication shop. Welding a section of a steam raising unit head for Hinkley power station.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox Drum Shop, Renfrew, Scotland. Hinkley power station steam raising units under construction.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox Fabrication Shop, Renfrew, Scotland. Electro-slag welding of Ferrybridge power station course 1.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox fabrication shop. Ammonia converter vessel for ICI plant at Severnside.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox tube shop. Reheater panels for Kincardine coal-fired power station.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox tube shop. Front wall panels for Kincardine coal-fired power station.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox tube shop. Membrane wall welding machine in action.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox tube shop. Reheater panels for Kincardine coal-fired power station, in shipping frames.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox tube shop. Side wall panels for Kincardine coal-fired power station.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox fabrication shop. Accumulator vessel for RTB Newport.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox fabrication shop. Stainless steel vessels for Lummus.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox fabrication shop. Stainless steel gas drier for Lummus.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox fabrication shop. Gas inlet nozzle for Sizewell nuclear power station.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox fabrication shop. Gas inlet nozzle for Sizewell nuclear power station.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox fabrication shop. Reactor vessel course assembly for Sizewell nuclear power station.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox fabrication shop. Steam raising unit, course 1, for Sizewell nuclear power station.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox fabrication shop. Steam raising unit heads, for Sizewell nuclear power station.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Sizewell nuclear power station - heat exchanger failure.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Repair in process at Sizewell “A” nuclear power station. What do you think this guy is doing?
Image source: unknown.
Repair in process at Sizewell “A” nuclear power station. What is happening above and below the weld?
Image source: unknown.
Babcock and Wilcox drum shop. Steam drum for Thorpe Marsh power station.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox fabrication shop. Reactor vessel – assembly of course 4, for Trawsfynydd nuclear power station.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox fabrication shop. Diagrid for Trawsfynydd nuclear power station.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox fabrication shop. Unusual spherical corners used in the ducting for Trawsfynydd nuclear power
station. No - he isn’t trying to create an initiation site for a fatigue crack with his centre punch … its simply part of the
method used to remove a large hole for a horizontal duct nozzle to be welded on.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox on site construction of Trawsfynydd nuclear power station. B&W 400 ton Goliath crane in action.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox tube shop.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox – transport by rail of the steam drum for West Burton power station. This number of nozzles would
have kept the welders busy on site for a while!
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox tube shop. Burner walls for West Thurrock power station.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Babcock and Wilcox – transport by rail of the steam drum for West Thurrock power station.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Magnetic particle dry powder inspection of a weld.
Image source: unknown.
Babcock and Wilcox – X-ray machine.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Large Whessoe vessel. Often lamp posts had to be removed when transporting such large vessels from the work or to the
final destination.
Image source: Babcock and Wilcox, Renfrew, Scotland.
Vessel with external helical heating coils.
Image source: Apollo Engineering, Troon, Scotland.
Oil refinery plant.
Image source: unknown.
Smooth pipe bend undergoing an in-plane bending test. What happens at the centre of the bend and how does this affect
the deformation, end-reactions and the stresses?
Image source: Babcock and Wilcox, Renfrew, Scotland.
A collage with a couple of interesting images. The curved tube with nozzles is impressive, as is the number of nozzles on
the lower vessel.
Image source: unknown.
Pipe laying in the North Sea. During this process the pipe is coiled onto a reel and then straightened while being laid from
the back of the vessel.
Image source: unknown.
Motherwell Bridge vessel on a low loader.
Image source: Motherwell Bridge Engineering, Motherwell, Scotland.
Motherwell Bridge reactor vessel.
Image source: Motherwell Bridge Fabricators, Motherwell, Scotland.
Motherwell Bridge vessel being lifted. Three-point lift is good … but why no “spreader-beam”?
Image source: Motherwell Bridge Fabricators, Motherwell, Scotland.
Large diameter flanged joint.
Image source: Motherwell Bridge Fabricators, Motherwell, Scotland.
Nice horizontal vessel with only one nozzle in the knuckle region – which is not bad I suppose.
Image source: Motherwell Bridge Fabricators, Motherwell, Scotland.
Another large diameter flanged joint … just look at the thickness of these flanges!
Image source: Motherwell Bridge Fabricators, Motherwell, Scotland.
No safety harnesses here then!
Image source: unknown.
Wonder why he doesn’t grind the weld while he is at it! What difference would that make to the assessment of the nozzle?
Image source: Motherwell Bridge Fabricators, Motherwell, Scotland.
Nice vessel with impressive flanges and array of small reinforced penetrations. What is that on the knuckle I wonder?
Image source: Motherwell Bridge Fabricators, Motherwell, Scotland.
Unusual vessel.
Image source: Motherwell Bridge Fabricators, Motherwell, Scotland.
Large oblique reinforced nozzle on a cylindrical shell.
Image source: Motherwell Bridge Fabricators, Motherwell, Scotland.
Looks like another large diameter flanged joint in the making.
Image source: Motherwell Bridge Fabricators, Motherwell, Scotland.
Nice view of the inside of a cylindrical skirt.
Image source: Motherwell Bridge Fabricators, Motherwell, Scotland.
Large nozzles in a vessel – thick walled?
Image source: Motherwell Bridge Fabricators, Motherwell, Scotland.
More impressive large diameter flanges.
Image source: Motherwell Bridge Fabricators, Motherwell, Scotland.
Construction underway. Is the bracing at the end temporary … a construction loading case perhaps?
Image source: Motherwell Bridge Fabricators, Motherwell, Scotland.
Imagine having this in your back garden! Nice horizontal vessels mind you – not sure about the gasometers though.
Image source: Motherwell Bridge Fabricators, Motherwell, Scotland.
Saddles at right angles – unusual.
Image source: Motherwell Bridge Fabricators, Motherwell, Scotland.
Four saddles this time! Unusual end detail.
Image source: Motherwell Bridge Fabricators, Motherwell, Scotland.
Nice shot of a vessel being lifted. I do hope the force exerted by his legs doesn’t start the rolls moving! I presume the end
nozzles are designed to be lifted in this way.
Image source: Motherwell Bridge Fabricators, Motherwell, Scotland.
From low-loader to ship on the banks of the Clyde in Scotland. The vessel certainly looks the part.
Image source: Motherwell Bridge Fabricators, Motherwell, Scotland.
A lovely image – used by one of my colleagues as a front sheet for his pressure vessel design notes! Somehow you can
just follow the designer’s thinking for a space-saving layout.
Image source: unknown (what I mean is I don’t think he took the photograph).
Talk about diversity in design! Doesn’t this simply look well-designed? Look at the detail in the saddle support and the
junctions with the small diameter cylinder.
Image source: Motherwell Bridge Fabricators, Motherwell, Scotland.
http://www.johnstonboiler.com/images/new/1800-2500_HP_PFTS-BOILER.jpg
Image source: www.johnstonboiler.com.
Nice Dorman Long vessel on a low-loader. Note the reinforcement around nozzles and leg supports.
Image source: unknown.
Horizontal vessel with saddle supports and large oblique nozzle on torispherical head.
Image source: unknown.
A refinery at night – what a sight!
Image source: unknown.
Close-up of a flange weld.
Image source: http://www.imageafter.com/.
Do you think they had a 3D cad system to lay this out?
Image source: http://www.imageafter.com/.
US Nuclear submarine “Texas”. An externally pressurized vessel!
Image source: unknown.
Babcock and Wilcox fabrication shop. Stress relieving of nuclear submarine prototype reactor.
Image source: Babcock and Wilcox, Renfrew, Scotland.
A submersible. Interaction effects apparent?
Image source: http://www.imageafter.com/.
Nut, bolt and a washer … now how do you model pre-load again?
Image source: http://www.imageafter.com/.
Concrete storage tanks – hydrostatic loading, roof loads, snow, wind … anything else?
Image source: http://www.imageafter.com/.
A membrane – tricky analysis. What are the loads? How would you model the seam?
Image source: http://www.imageafter.com/.
Vessel, pipework and steelwork.
Image source: http://www.imageafter.com/.
Nice collection of large diameter bends, T-pieces, valves and reducers. Discontinuity stresses, ovalization, fatigue perhaps?
Image source: http://www.imageafter.com/.
More large diameter bends, valves and reducers. Interesting support on the bend – wonder if it is reinforced?
Image source: http://www.imageafter.com/.
Pumps, valves, reducers and T-pieces.
Image source: http://www.imageafter.com/.
A better view of the supports on the bend – and no they are not reinforced? No lagging, probably water at ambient temperature ..
low stresses anyway perhaps.
Image source: http://www.imageafter.com/.
Interesting pipe supports and hangers.
Image source: http://www.imageafter.com/
Horizontal rail transportation vessels – supported on longitudinal beams? Sloshing – whats that?
Image source: http://www.imageafter.com/.
Someone has to design the walk-ways as well!
Image source: http://www.imageafter.com/.
Nice reflective image. Strakes visible – therefore steel? Wonder how thick at bottom and top? What size of section around the top
do you think?
Image source: http://www.imageafter.com/.
Unusual plant – I wonder why it is all under a canopy?
Image source: http://www.imageafter.com/.
A flare stack, two spherical vessels on legs and a conventional roofed storage tank … nice! Do you think all these legs are
necessary – or even a good idea?
Image source: http://www.imageafter.com/.
Cylindrical vessel with a conical discharge at the bottom. Wonder if axial buckling of the cylinder is a possibility with this type of
content?
Image source: http://www.imageafter.com/.
Not a pressure vessel I know, but a nice photograph non-the-less! A gravity structure subjected to wind loading though. Why are
the metal bands necessary?
Image source: http://www.imageafter.com/.
A more modern reinforced concrete chimney – and no metal bands (not visible at least)?
Image source: http://www.imageafter.com/.
More spherical vessels with leg supports – fewer legs? Why is the leg junction at this height? Lobster-back or multi-mitred bends in
the fore-ground.
Image source: http://www.imageafter.com/.
Pipe-work looks nice!
Image source: http://www.imageafter.com/.
Again, not a pressure vessel, but another lovely image! Symmetrical under what loading cases? Moment or shear connections at
the joints?
Image source: http://www.imageafter.com/.
Just can’t resist a wonderful structure! This one is in Rostock, Germany outside a conference centre. First year mechanics class –
why no gross bending of the members? Identify the tension and compression members! So that’s what a pin-joint looks like!
Remember Maxwell’s Lemma regarding optimum structures?
Image source: Jim Wood
A process plant in miniature. Why green and yellow colours I wonder?
Image source: http://www.imageafter.com/.
Nice process plant image.
Image source: http://www.imageafter.com/.
There is something wrong here – can you spot what it is?
Image source: http://www.imageafter.com/.
Corrosion allowance … why?
Image source: http://www.imageafter.com/.
Is that a logarithmic spiral? Cant help thinking that modern optimisation tools could shed some weight here – then again it might not
be there for us to admire?
Image source: http://www.imageafter.com/.
Took this picture as I walked into Oliver Tambo airport in Johannesburg, South Africa – after a wonderful holiday! What shape are
these cooling towers again – and which loading would allow me to use these highly efficient axisymmetric thin shells elements?
Image source: Jim Wood
Is 3 saddle supports a good idea in general? Is there any reinforcement?
Image source: http://www.imageafter.com/.
Nice package unit on a skid base waiting to be connected up. Cuts down on-site work. Looks like whole thing is lifted by the two
lugs on the vessels (surely not). They do look rather large I suppose. What do you think? Good practice? Wonder if they assumed
lug loads the same?
Image source: http://www.imageafter.com/.
A colourful symmetrical construction!
Image source: http://www.imageafter.com/.
A collection of storage tanks of various sizes. If the roof of such a tank collapsed as the tank was being emptied, what would you
first of all suspect?
Image source: http://www.imageafter.com/
Locomotive boilers … before the days of welding!
Image source: unknown.
Steam drum being lifted into place during construction of a power station. One of the highlights of my career as a Junior Site
Engineer with Babcock Construction some 30 odd years ago was noticing from the drawings, that such a lift was 180 degrees out.
This was only apparent from slight differences in the nozzle patterns on both sides. The chief rigger never forgave me!
Image source: unknown.
Now isn’t that nice! Wonder how much FEA was required for this? So why do these chimneys have spirals on the outside?
Image source: unknown.
Inside a spherical reactor vessel in a nuclear power station? Is that yellow chap Homer Simpson? Wonder if they considered the
scenario of the crane collapsing onto the core and its effect on the diagrid?
Image source: unknown.
A nice collection of tall shiny vessels. Would one failing affect those adjacent I wonder?
Image source: unknown.
A collection of small vessels.
Image source: unknown.
Go on then … switch it on! Spot the lack of symmetry.
Image source: unknown.
Its amazing where boilers turn up.
Image source: unknown.
Milk transportation – any particular material requirement? The head looks very flat – what form does it have I wonder?
Image source: unknown.
Nice cylindrical skirt. What is the purpose of the big hole in the skirt?
Image source: unknown.
This looks unusual plant?
Image source: unknown.
Nice stainless vessel with all the “action” on the head. Is that some kind of “stirrer” on the top? Flange design rules don’t usually
cover rotating machinery being bolted on directly.
Image source: unknown.
If one of these tanks collapsed, do you think the walls would contain the spill? How might you analyse this?
Image source: unknown.
Pressure … as in pressure vessel!
Image source: unknown.
An old vessel fabrication image … do you think they have a problem … there is a man in a suit after all?
Image source: unknown.
This picture simply exudes quality design … and that’s without seeing any sums!
Image source: unknown.
Plastic storage vessels. What complexities do they bring?
Image source: unknown.
A nice autoclave – the end swings open!
Image source: unknown.
OOPS … this doesn’t look the fault of the vessel designer though!
Image source: unknown.
Two saddles - with reinforcement on saddles and end nozzle. Head and seam welds also visible.
Image source: unknown.
You see … its not just me that thinks there is something nice about old gasometers and big rusting lumps of plant!
Image source: unknown.
The end of a design life.
Image source: http://www.imageafter.com/.
Just a nice photograph!
Image source: unknown.
Another couple of old rusty boilers at the end of their working life. Tubesheet analysis – now there is an interesting problem.
Image source: unknown.
A steam roller or tractor perhaps? Well before the days of FEA!
Image source: unknown.
A vessel “graveyard” shot. What are these wired studs connected to vessel head do you think?
Image source: unknown.
I used to know someone who used to buy old vessels, clean them with a wire brush and sell them again …. before the “art” of
residual life assessment came along!
Image source: unknown.
These legs don’t quite look adequate? ASME III vessel … what do you think?
Image source: unknown.
Is this what is meant by “moth-balled”?
Image source: unknown.
These “constant strength” shells were actually built. Problem is they were difficult to fabricate and were only “constant strength”
when full.
Image source: Jim Wood (collage).
A torus of a very complex shape. The W7-X “stellarator” fusion reactor under construction at the IPP in Greifswald, Germany.
Image source: Jim Wood.
Part of the cryogenic plant at the IPP in Greifswald, Germany. Cryogenics .. what does this requirement imply?
Image source: Jim Wood.
A nice 90 degree single un-reinforced mitred pipe bend at the IPP in Greifswald, Germany.
Image source: Jim Wood.
A vacuum vessel at the IPP in Greifswald, Germany. The top half literally lifts off! The hooks hanging down hold the flanges
together.
Image source: Jim Wood.
A guide for aligning the two halves of the vacuum vessel at the IPP in Greifswald, Germany.
Image source: Jim Wood.
A finished nozzle on the vacuum vessel at the IPP in Greifswald, Germany.
Image source: Jim Wood.
A novel way of creating a pressure “wall”. Two sheets are welded together along the lines shown. The cavity between the sheets is
then plastically “inflated” to form the necessary flow cavity for the wall. Component on display at the IPP in Greifswald, Germany.
Image source: Jim Wood.
.