Chapter 7:
Chapter 7 Practical
Considerations in Modeling:
with examples from other
chapters
ME 501 chapter 7 1
Key considerations when
generating a FE model
Number of elements being used in the mesh
Referred to as mesh density
The type of element
1-D or 2-D (with and w/o bending), 3-D
Plane assumption vs. shell assumption (chapt 12)
Linear or some higher order displacement function
What is the quality of the mesh
This has a significant effect on the solution (crashing)
Classic question:
Use many simple elements or few complex element
ME 501 chapter 7 2
© 2002 Brooks/Cole Publishing / Thomson Learning™
Figure 7-1a (a) Beam with
loading: effects of the aspect
ratio (AR) illustrated by the
five cases with different
aspect ratios
May not have effect in a
Constant strain field, such
as axial load
ME 501 chapter 7 3
© 2002 Brooks/Cole Publishing / Thomson Learning™
--Using Element from 10.2 in Logan,
a LINEAR strain rectangular
element
--Other formulations are
probably LESS sensitive
Figure 7-1b (b) Inaccuracy of solution as a function of the aspect
ratio (numbers in parentheses correspond to the cases listed in
Table 7-1)
ME 501 chapter 7 4
© 2002 Brooks/Cole Publishing / Thomson Learning™
Always try to avoid generating a mesh with too many elements that
have these characteristics. You can get away with a few.
Figure 7-2 Elements with poor shapes
ME 501 chapter 7 5
Transition triangle
used to go from
fine to coarse
© 2002 Brooks/Cole Publishing / Thomson Learning™
Figure 7-4 Use of symmetry applied to a uniaxially loaded
member with a fillet
ME 501 chapter 7 6
Rather than using “transition triangles”, you can also use quads
when you transition from a fine mesh to a coarse mesh.
You can also
use a 3:1 ratio
ME 501 chapter 7 7
© 2002 Brooks/Cole Publishing / Thomson Learning™
Figure 7-5 Problem reduction
using axes of symmetry applied
to a plate with a hole subjected
to tensile force
If you choose to solve the full
problem, make sure you apply
appropriate BCs to eliminate
rigid body motion.
ME 501 chapter 7 8
Sharp Corners or
Discontinuities
If they exist, local yielding will occur
Linear FEM will not calculate infinite stress
Use realistic radius in design and FEM
model
Or account for yielding (plasticity model)
ME 501 chapter 7 9
Concentrated forces and solid
elements
Per linear theory of elasticity, Point load
gives infinite stress & displacement
Actual material would yield, redistribute
force
In finite elements, displacement function
does not allow infinite displacement or
stress (approaches with very fine mesh)
ME 501 chapter 7 10
© 2002 Brooks/Cole Publishing / Thomson Learning™
If you actually decide to model
a point force, you need to resolve
the mesh in that region in order
to capture the stress and strain
gradient.
Figure 7-6 Natural subdivisions
at discontinuities, i.e., where to
put element edges.
Resolve the mesh
near corners.
Use elements to
distribute load.
ME 501 chapter 7 11
Concentrated Forces in Solids
Either ignore locally high stresses
i.e., don’t use stress right at force location
Or, input a more realistic loading
distribution (non-uniform pressure)
Hertzian contact stress distribution
ME 501 chapter 7 12
© 2002 Brooks/Cole Publishing / Thomson Learning™
Figure 7-17 (a) Connecting rod subjected to tensile loading and (b)
principal stress throughout the rod
ME 501 chapter 7 13
Mesh Refinement
h-Method
Keep element formulation the same
Refine by creating more elements
Evaluate error
p-method
Keep mesh the same
Use increasingly higher polynomials for shape
function
Evaluate error
ME 501 chapter 7 14
This gives more
refined solution
inside the element
ME 501 chapter 7 15
Element Selection
Personal Preference
Sometimes depends on industry
Type of System
What are the physics being modeled
Solids, fluids, heat transfer
FE program Emphasis
The program Continuity (cardiac modeling)
only uses cubic hermite elements
ME 501 chapter 7 16
CST
Generally too stiff in bending
Requires too many elements to resolve
Higher-order element is needed
Approach 1 – More external nodes
higher order means more DOF
Approach 2 – Use 4 node quad elements
These can be derived in several ways.
Isoparametric formulation (Chapt 10)
Static condensation (Chapt 7)
ME 501 chapter 7 17
Example: 4 CST 1 Quad
Figure 7-13 Quadrilateral element with an internal node
Assemble quad from 4 CST, then
“condense” out internal DOF (node 5) prior
to assembly of global stiffness matrix
ME 501 chapter 7 18
10 X 10 8X8
1 2 3 4 5
# # # # # U1xy
# U
# # # # 2 xy
The mathematical details are
# # # # # U 3 xy given in Section 7.5. We will
only focus on the isoparametric
formulation in Chapt 10.
# U
# # # # 4 xy
# # # # # U 5 xy
ME 501 chapter 7 19
Additional Elements
8-node PLANE82 in ANSYS
(quadratic)
Lin-quad
LST
Linear Strain
Triangle
cubic
ME 501 chapter 7 20
Quick look at shell elements
(chapter 12 – plate bending elements)
The key difference between plane
elements and shell elements is the ability
to model transverse loads that create
bending, and rotational DOF’s
ME 501 chapter 7 21
Shells – very useful for projects
ME 501 chapter 7 22
The same principle holds for 3-D solid elements and shell elements.
8-node hex (brick) elements only solve for displacments, whereas
4-node shell elements solve for displacement and rotation.
7-23
Stresses in Solid Elements
Depend on element shape function
Linear stress = const
Quadratic stress = linear
In general, Less accurate than
displacements
ME 501 chapter 7 24
Element vs. Nodal
CST A B
BC
2
B 2
A C
A D
D CD
Calculate stress at 2 2
center of element
A B C D
2
ME 501 chapter 7 25
More Complex Elements
f ( x, y ) (not constant)
Not continuous across element boundaries
Still need contour plots
ME 501 chapter 7 26
Typical Approach
Within each element, calculate stress at
nxn matrix of loc’ns (gauss or
integration points – chapter 10)
Curve fit to extrapolate stress to nodes
(SX, SY, SXY)
Average stress at common nodes
ME 501 chapter 7 27
Draw Contour Plots
10
50
. . . These are usually isobars,
that represent lines of constant
value. So, the values should
. . . 90
scale in between these lines
. . .
ME 501 chapter 7 28
Typically average component
Stresses (SX, SY, SXY),
then compute principal stresses
SIG1,SIG2,SIG3 1 2 3
SIGE Distortion Energy
SI 2* max stress intensity
Use for design calcs – theory of yielding
Theory of failure
ME 501 chapter 7 29
© 2002 Brooks/Cole Publishing / Thomson Learning™
Note the local
mesh refinement
near the hole,
where stresses
are expected to
be high.
Figure 7-7 Plate of steel (20 in. long, 20 in. wide,l in. thick, and
with a l-in.-radius hole) discretized using a preprocessor program
[15] with automatic mesh generation
ME 501 chapter 7 30
© 2002 Brooks/Cole Publishing / Thomson Learning™
Figure 7-9 Maximum
principal stress contour
plot for a plate with a hole
(original plot is in color).
Largest principal stresses
of 301 psi occur at the top
and the bottom of the hole,
which indicated a stress
concentration of 3. Stresses
were obtained by using an
average of the nodal values
(called smoothing)
ME 501 chapter 7 31
Verification of Elements/Results
Run a “patch test” to ensure the element
satisfies our standard requirements
Comparison to handbook solution
Create coarse model and refine mesh for
convergence check
Thisis a typical requirement when publishing
results from an FE study
ME 501 chapter 7 32
© 2002 Brooks/Cole Publishing / Thomson Learning™
Incompatible element
Figure 7-11 Convergence of a finite element solution based on the
compatible displacement formulation
ME 501 chapter 7 33
© 2002 Brooks/Cole Publishing / Thomson Learning™
Figure 7-16 Flowchart
of plane/strain finite
element process
ME 501 chapter 7 34
Subtle Boundary Condition Effects
UX1 = UX2 = 0 UY1 = UY2 = 0
3002
-5000
2
Reactions
1
-5000
Load
-3002
10000 lb
total
ME 501 chapter 7 35
Another option
UX1 = UX2 = 0 UY1 = 0
0
-5000
2
Reactions
1
-5000
Load
0
10000 lb
total
ME 501 chapter 7 36
Solid Elements (some
examples)
SOLID45
3D Solid element, linear with extra modes
8 node E,I,J,K,L,M,N,O,P
Other shape options
SOLID95
20 node element with mid side nodes, quadratic
Other shape options
Solid 92 Tetrahedron
quadratic
ME 501 chapter 7 37
Thermal Elements
Thermal55 Solid 2D
Thermal34 Radiation
Thermal70 3D Solid
KXX conductivity
C thermal capacity
DENS Density
ME 501 chapter 7 38