Documente Academic
Documente Profesional
Documente Cultură
= = A
N F F F F
I II I x
1000 0 = + = =
\
|
+
+
=
l A x A A
A A x lA
A A E
Fl
x u
1 1 2
1 2 1
1 2
) (
ln
) (
) (
Page 18 | Prof. Dr.Ing. Th. Binder | Status March 2014
Analytical solution with number values
A
1
= 40 mm
2
; A
2
= 100 mm
2
; l = 1 m;
F = 10 kN; E = 200000 MPa;
Page 19 | Prof. Dr.Ing. Th. Binder | Status March 2014
FEM: Solution with linear
approach
deformations between the nodes are approximated by means of a displacement approach, the
simplest is a linear approach:
Analytical solution: FEM approximation solution:
Note:
tensions are wrongly
calculated in the linear
approach!
Page 20 | Prof. Dr.Ing. Th. Binder | Status March 2014
Discretisation with an element
F
1000
A1 A2
Stiffness matrix and forces equilibrium of a rod element:
i.e. one must work here with the mean cross-section area
here:
For comparison, the analytical solution:
Page 21 | Prof. Dr.Ing. Th. Binder | Status March 2014
Discretisation with two elements
F
500
A1 A2
500
Element 1 Element 2
For comparison, the analytical solution:
Page 22 | Prof. Dr.Ing. Th. Binder | Status March 2014
Linear approach with two elements
linear displacement, tension
constant within an element!
==> fine discretization necessary
Page 23 | Prof. Dr.Ing. Th. Binder | Status March 2014
Gaussian idea: quadratic displacement approach
Express displacements by means of the three nodal displacements u
1
, u
2
, u
3
and determine the
parameters a
0
, a
1
, a
2
there from:
Calculate the tensions using the material law!
Quadratic approach:
u(x) = a
0
+ a
1
x + a
2
x
2
1 2 3
Mean nodes
F
III II I
III II I
III II I III II I
I
u x
l l
u x
l l
u x
l l
x u
dx
d
x
u x
l
x
l
u x
l
x
l
u x
l
x
l
x u
l
u
l
u
l
u
a
l
u
l
u
l
u
a u a
|
.
|
\
|
+
|
.
|
\
|
+
|
.
|
\
|
+ = =
|
.
|
\
|
+
|
.
|
\
|
+
|
.
|
\
|
+ =
= =
2 2 2
2
2
2
2
2
2
2 2 2 2 1 0
4 1 8 4 4 3
) ( ) (
;
2 1 4 4 2 3
1 ) (
;
2 4 2
;
4 3
;
c
Page 24 | Prof. Dr.Ing. Th. Binder | Status March 2014
Quadratic displacement approach
Stiffness matrix in quadratic displacement approach:
Quadratic approach:
u(x) = a
0
+ a
1
x + a
2
x
2
1 2 3
Mean nodes
2
1 2
:
6
11
3
7
2
3
8
6 3
1
2
3
8
3
8
3
16
3
2
3
8
6 3
1
3
2
3
8
2 3
7
A
A A
mit
=
(
(
(
(
(
(
+
o
o o
o
o o o
o
o
o
F
Page 25 | Prof. Dr.Ing. Th. Binder | Status March 2014
Quadratic displacement approach
2
1 2
3
2
1
1
3
2
1
:
6
11
3
7
2
3
8
6 3
1
2
3
8
3
8
3
16
3
2
3
8
6 3
1
3
2
3
8
2 3
7
A
A A
mit
u
u
u
l
A E
F
F
F
=
|
|
|
.
|
\
|
(
(
(
(
(
(
=
|
|
|
.
|
\
|
o
o o
o
o o o
o
o
o
Solving the FEM basic equation:
results in:
For comparison, the analytical solution:
mm u 459 , 0
2
=
mm u 756 , 0
3
=
MPa 216
1
= o
Page 26 | Prof. Dr.Ing. Th. Binder | Status March 2014
Quadratic displacement approach with
two elements
Trial function Elements
Tension [MPa]
Position
0 1/4 1/2 3/4 l l
analytical - 250 182 143 118 100
linear 1 143 143 143 143 143
2 182 182 182/118 118 118
quadratic 1 217 185 152 120 87.0
2 237 186 135/140 119 97.9
Page 27 | Prof. Dr.Ing. Th. Binder | Status March 2014
FEM: quadratic displacement
approach
Page 28 | Prof. Dr.Ing. Th. Binder | Status March 2014
Resume
Deformation is more exact than tensions
Nodal displacements are too small ==> FE model stiffer than in reality
Tensions in element middle more exact than at the boundary
Approach always from the bottom
Deformations always an order higher than tensions = ( = du/dx)
More elements increase accuracy
Displacement approach of higher order improves the result
Page 29 | Prof. Dr.Ing. Th. Binder | Status March 2014
Finite-elements work procedure
FEM procedure:
Page 30 | Prof. Dr.Ing. Th. Binder | Status March 2014
FEM applications:
Finite-elements work procedure
Page 31 | Prof. Dr.Ing. Th. Binder | Status March 2014
FEM programme:
Trend for the FE programme with special application modules and integrated pre- and
postprocessor
Finite-elements work procedure
Page 32 | Prof. Dr.Ing. Th. Binder | Status March 2014
Fig. Conversion of the problem definition in the FEM analysis
Spec. CAE software:
Ansa
Hypermesh
Patran
Medina
Finite-elements work procedure
Page 33 | Prof. Dr.Ing. Th. Binder | Status March 2014
Geometry definition
Fig. Model simplification on the basis of a perforated disk
Objective:
Reduction of computer time
Simplification of network fineness
Reduction file size
Page 34 | Prof. Dr.Ing. Th. Binder | Status March 2014
Boundary conditions
Fig. boundary conditions on a screw wrench
3 Types of boundary conditions:
Page 35 | Prof. Dr.Ing. Th. Binder | Status March 2014
Loads
Fig. Determining the equivalent nodal forces for a constant compressive load for linear
displacement approaches and different networks
Page 36 | Prof. Dr.Ing. Th. Binder | Status March 2014
Fig: stepwise refinement of the real load application over a distributed surface load
Loads
Page 37 | Prof. Dr.Ing. Th. Binder | Status March 2014
Model refinement:
Increase of the quality of results by increasing the number of elements or degrees of trial
function
Stress singularities
at the load introduction point
Networking
Page 38 | Prof. Dr.Ing. Th. Binder | Status March 2014
Fig. Partial modelling technology (submodelling), global modelling and partial modelling
Using sub models:
Networking
Page 39 | Prof. Dr.Ing. Th. Binder | Status March 2014
5.2 Finite elements application examples
Examples:
Page 40 | Prof. Dr.Ing. Th. Binder | Status March 2014
Examples: Topology optimisation for appropriate casting construction
5.2 Finite elements application examples
Page 41 | Prof. Dr.Ing. Th. Binder | Status March 2014
Problematic nature of result analysis
Lacking physical basis skill of the calculation engineer
Lacking design experience of the calculation engineer
Lacking basic knowledge of the calculation engineer in the FEM theory
Missing or incorrect characteristic values and data
Deadline pressure, hasty interpretation
Lacking practice in handling the FEM results/software
Incorrect estimation (overestimation) of the efficiency of the FEM software
Rough calculation, measurement adjustment
Error causes in the FEM analysis:
Page 42 | Prof. Dr.Ing. Th. Binder | Status March 2014
Fig. Occurring errors in the FEM
Application error of the FE method:
Model quality in the FE
system
Problematic nature of result analysis