FEA Exercise 2 – Mesh Convergence

Jousef Murad
[email protected]

1
Content
1 Introduction 3

2 Part 1: Convergence in Presence of Sharp Corners Using SimScale 4

2.1 Performance Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.2 Design Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

3 Part 2: Convergence When Bending of Thin-Structures 7

3.1 Performance Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.2 Design Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2
1 Introduction
In this exercise, you will apply some ideas of FEA and solid mechanics to do a series of sim-
ulations on the SimScale Platform and check convergence properties in various circumstances.

Assume the following situation: You are designing a truck underrun protection (Fig. 01) and
are at a certain iteration in your product development process. Now you want to validate
your design with SimScale and make decisions on how to improve your current design.

Figure 1: Underrun protection device on a semi-truck

There are several aspects to consider here. As evident in Fig. 01, there are several sharp
corners and hence in the first part, we will consider the convergence issues in the presence of
sharp corners.

Additionally, the components are thin and can be treated like beam structures. In the second
part, we will consider locking effects in such structures.

3
2 Part 1: Convergence in Presence of Sharp Corners
Using SimScale
– Copy the project

– Once the geometry has been imported, mesh the geometry (named NotchedPlate) using
an automatic coarse mesh

– Set the analysis type to Static and perform a linear static analysis simulation. The
bottom surface is fixed while a force is applied on the top surface as shown in Fig. 2

– Use steel for the material with the properties given in design details in section 2.2
Design details. This material can be directly imported from the material library.
This assumes linear elastic behaviour

Figure 2: Loading of a notched specimen for convergence analysis

– The lower surface is fully constrained using a fixed displacement boundary condition.
This is as shown in Fig. 3

– Apply a force over the top face of 1000*t. This is shown in Fig. 4 and 5

– For the same material, boundary and loading conditions, try the simulation for different
meshes. In the mesh creator, adjust the mesh fineness from 1 – 5 and run the simulations

– Optionally, one can also change the element order from first to second or alternatively
with mesh refinements at the notch

4
Figure 3: Fixed displacement boundary condition applied to the bottom surface

Figure 4: Load being applied on the top surface of the notched specimen

5
 Figure 5: Load being applied on the top surface of the notched specimen

2.1 Performance Details

– Analyze the simulation results using the online post-processor and save custom screen-
shots of each load cases showing the resulting Cauchy normal stress, along the direction
of loading

– For more details on Post-Processing, see this link

– Tabulate the maximum Cauchy stress (normal stress along the direction of loading)

– In the meshing operation event log, find the number of nodes. Each node has three
degrees of freedom and hence the total number of degrees-of-freedom is three times the
number of nodes

– Plot the total degrees of freedom vs. the tabulated maximum Cauchy stress

– Do you observe that as the number of degrees-of-freedom increases the maximum

Cauchy stress also increases?

– What are the conclusions?

6
2.2 Design Details
– Dimensions: 0.1 Ö 0.1 Ö 0.1 m3
– Young’s Modulus (E) = 205 GP a

– Poisson ratio (ν) = 0.28

– Density (ρ) = 7870 kg/m3

3 Part 2: Convergence When Bending of Thin-Structures

– We have already imported all necessary geometries in Part-I. Once the geometry has
been imported, mesh the geometry (named BeamBending) using an automatic coarse
mesh

– Set the analysis type to Static and perform a linear static analysis simulation of the
cantilever beam subjected to a uniformly distributed load as shown in the figure below

Figure 6: Cantilever beam subjected to uniform distributed loading

– The boundary conditions and loads are as shown in Fig. 7 & 8

– Use steel for the material with the properties given in design details in 3.2 Design
details. This material can be directly imported from the material library. This assumes
linear elastic behaviour.

7
 Figure 7: Fixed-end boundary condition for the cantilever beam

Figure 8: Uniform load applied on the cantilever beam

– For the same material, boundary and loading conditions, try the simulation for dif-
ferent meshes. In the mesh creator, adjust the mesh fineness from 1 – 5 and run the
simulations.

– You can also experiment with changing the element order from first to second

8
 – In total, we have about 10 simulations. With the power of SimScale, all of them can
be set to run in parallel

3.1 Performance Details

– Analyze the simulation results using the online post-processor and save custom screen-
shots of each load cases showing the resulting tip displacement
– Tabulate the tip displacement for the maximum load for each simulation
– For each maximum load, calculate the analytical result using Euler-Bernoulli beam
theory. The tip deflection can be written as

F L4
wtip =
8EI
where F is the applied load in N/m, wtip is the resulting tip displacement in, L is the
length of the beam, E is the young’s modulus and I the area moment of inertia. Here
we have a square cross-section and the area moment of inertia can be given as

a4
I=
12
where a is the length of the side. Tabulate the resulting analytical solution for each
load
– In the meshing operation event log, find the number of nodes. Each node has three
degrees of freedom and hence the total number of degrees-of-freedom is three times the
number of nodes
– Plot the total degrees of freedom vs. the tabulated tip displacement for first-order
and second-order elements. In the same plot, also plot the analytical solution. The
analytical solution will remain same irrespective of the number of degrees-of-freedom
– Do you observe the first-order elements initially behave badly while they get better
with finer mesh? Also it should be possible to observe that the second-order elements
are much better and do not show any locking
– What are the conclusions?

3.2 Design Details

– Cross-sectional dimension of the beam: 0.05 Ö 0.05 m2
– Length of the beam: 0.5 m
– Young’s Modulus (E) = 205 GP a
– Poisson ratio (ν) = 0.28
– Density (ρ) = 7870 kg/m3

9
Additional Information
– Convergence in FEA

– Tips for a high quality Mesh

10