Ansys Work Book-1
Ansys Work Book-1
Goals
This exercise will teach you how to perform a simple, yet complete, finite element
analysis (FEA) consisting of pre-processing, solving the FE model and subsequent
post-processing using the static structural module of ANSYS Workbench.
Beam under 3-Pt Bending
We want to simulate a beam under three point bending with a force F applied at the
center as shown in Figure 1.
F t
h
w
The following geometry and material data are required to model our problem:
σyiel
d
Questions
With respect to this classic two-dimensional mechanical problem, we can state two
questions:
2 / 23
Taking advantage of symmetries
Can we take advantage of symmetries? Please, draw a simplified beam model,
which takes advantage of potential symmetries (Figure 2).
Choose appropriate boundary conditions for the simplified beam such that
you would get the same displacement results than for the 3-pt bending
all rigid body movements are fixed.
Figure 2: Space for drawing a simplified beam model taking advantage of symmetries.
3 / 23
Starting and Configuring ANSYS Workbench
First, log in using your credentials. From the Windows start menu select and run
ANSYS Work- bench (Figure 3), opening up ANSYS Workbench’s project view
(Figure 4).
launched Work- bench go to Tools → License Preferences and make sure that
Important: Please make sure to choose the right license type. After you have
ANSYS Academic Teach Ad- vanced is the default (i.e.: top-most) license
option (Figure 5). Otherwise, use the Move up and Move down buttons to
correct and finally Apply the settings.
4 / 23
The project schematic
visual- izes the work- and
data-flow between the
different project
components and modules.
5 / 23
A. Pre-Processing: Setting up the Model
Before building the actual model, you need to create a new static-structural FE
analysis by dragging and dropping the Static Structural analysis system onto
the empty project schematic (Figure 6).
Create the solid beam by choosing (from the main menu) Create → Primitives →
Box. Use the
Details pane to specify the desired dimensions of the new primitive.
Please ensure that the origin of the coordinate system is located on the plane and
at the center of the cross section of the beam. The beam’s long axis must be
oriented along the global x- axis (Figure 7).
6 / 23
Figure 7: The beam in DesignModeler
In your static structural analysis system in the Workbench project view, double
click the Engi- neering Data cell. This opens up a window titled Outline of
Schematic A2: Engineering Data. “Structural steel” is the default material and
is always predefined. Click the row beneath (where it says “Click here to add a
new material”) and enter any name for your new material.
In the Toolbox to the left, expand the Linear Elastic node and drag and drop
Isotropic Elas- ticity onto the Material column of your material (Figure 8).
Enter the appropriate values into the Properties window (Young’s modulus,
Poisson’s ratio), before clicking Return to Project.
7 / 23
Figure 8: Defining a new linear-elastic material
Make sure that the correct material model is assigned: In the Mechanical
module’s Outline pane (to the left) select the solid body representing the beam
(under the Geometry node). Then select the material in the Details pane
(Details → Material → Assignment).
A.4 Meshing
The next pre-processing step is concerned with discretizing the continuous solid
body geom- etry, also known as meshing.
The outline tree view also contains a node called Mesh with a little yellow flash
symbol. Right click on Mesh and select Insert → Mapped Face Meshing from
the context menu (Figure 9). Select all 6 faces of the beam and click Scope →
Geometry → Apply in the details pane of the Mapped Face Meshing node.
In the same way, add a Sizing sub-node to the Mesh node. This time, select the
whole body
and again apply your selection. In the Details pane of the (Body) Sizing node, select
Definition
→ Type → Element Size and set the element size to 15 mm. In the Outline, right
click on Mesh
→ Generate Mesh. The result should resemble Figure 10.
8 / 23
Figure 9: Adding a meshing method
9 / 23
A.5 Applying Loads and Boundary Conditions
We now have to apply the loads and boundary conditions in such a way that the
FE model represents our ideas from Figure 2.
We therefore fix all degrees of freedom of one end. In the Outline right click on
the Static Structural (A5) node and select Insert → Fixed Support from the
context menu. Select an appropriate face of your solid body to be fixed.
Next, we need to apply the force to the other end of the beam. Again, right click
on the Static Structural (A5) node, but this time Insert → Force. Select the
correct face and apply a force of the appropriate magnitude and direction. The
result should resemble Figure 11.
Figure 11: The model after applying loads and boundary conditions
10 /
23
B. Solving
Because this is a simple linear problem, we do not need to modify the solver
options manually (Analysis Settings in the Outline). Instead, simply right click
on the Static Structural (A5) node and select Solve. This will bring up a status
window, which should disappear again after a few seconds of computing.
- Total Deformation
- Normal Elastic Strain (in the beams axial direction)
- Normal Stress (in the beams axial direction)
- Equivalent (von Mises) Stress
11 /
23
C.2 Total Deformation
When performing FE analyses, it is always wise to first perform some plausibility
checking. Cre- ate a contour plot of the total deformation (Figure 13).
The predicted displacements seem to be totally fine at first glance. On closer look
the maximum total displacement is more than 1000 mm, according to the scale
to the left!
The issue with this plot is that by default ANSYS automatically scales the
displayed defor- mations so that they are “easily visible.” For very small
displacements this behavior is totally fine, as they wouldn’t be visible at all
otherwise. In our case, however, this setting is deceptive. Changing the scaling
factor to 1.0 (Results toolbar) yields a completely different picture, mak- ing it
crystal clear that something has gone wrong – awfully wrong:
12 /
23
assumed a much too high force. If we correct the force to be 𝐹 = 5000 N instead,
The reason for this huge displacement is that in Table 1, we (deliberately)
Figure 15: (Unscaled) contour plot of the total deformations after applying the
correct load
13 /
23
Figure 18: Von Mises stress
Answering the
Questions:
1. Will the beam break? If so, where would it fail?
maximal predicted von Mises stress reaches values of 𝜎pred ≈ 220 N/mm2,
With the corrected force (F = 5,000 N) the beam will not break. The
and thus less than the ultimate yield stress of 𝜎yield = 235 N/mm2. That
means the failure criterion 𝜎pred > 𝜎yield is not fulfilled. The difference
between the two values however is small (𝜎predreaches 94 % of
𝜎yield). Many technical applications require a safety factor (SF) of 2.0 or
higher. In our example, the safety factor SF = 𝜎yield/𝜎pred is much smaller.
The critical region where we would expect the beam to start failing is
located at the left end of the half beam, at the location of maximum
stresses. For the full length beam the critical region would therefore be
located in the middle of the beam where the force was applied.
2. Assuming that it would not fail, what would be the maximum deflection w?
We predicted a maximum deflection of w = 11 mm appearing at the free
end (right side) of the simplified half model. The full length beam under
three point bending will show a maximum deflection of the same amount
in the middle.
14 /
23
D. Bonus Task: Parameterization & Optimization
Given that we trust the predictions of our model, we now know that the beam can
hardly handle the desired load without safety issues. Thus, one might be
interested in how the beam has to be changed to meet the safety factor goal of
of the beam such that we achieve a SF ≥ 2.0 while using a minimum amount of
2.0. In this optional task, we are therefore now going to optimize the thickness
material, as we don’t want to waste expensive steel. This is typical example for a
constrained optimization problem.
Open the DesignModeler and select the beam primitive from the Tree Outline.
In the Details View click the empty box next to the FD8, Diagonal Z
Component entry (or whatever com- ponent you chose to use as the thickness
of the beam); a blue D should now appear inside the box (Figure 19). Then enter
an appropriate parameter name into the dialog window and click OK.
As we want the origin to always be located at the center of the cross section, we
also need to parameterize FD5, Point 1 Z Coordinate (i.e. the coordinates of
the first corner of the box primitive). We will make this a derived parameter (i.e.
dependent on BeamThickness) in the next step. For now, perform the same steps
as before and call this new parameter something like “BeamZOffset.”
Back in the Project View, you will notice that there appeared an additional
Parameter Set block right beneath the Static Structural analysis system.
Notice how there is only one arrow going from the Parameter Set to the
analysis system; this means that we have defined input parameters so far.
15 /
23
Figure 20: Project View after adding our two input parameters
Double click on the Parameter Set block to edit the parameter properties. As
BeamThickness
BeamZOffset should always be − enter a corresponding arithmetic
2
expression in the properties
outline for parameter P2 (= BeamZOffset). Note that you have to use the internal
parameter ID P1 instead of BeamThickness (Figure 21).
While we are at it, click on the cell with the light-gray New name entry (ID: New
input param- eter) to define a new constant. Call it TensileYieldStrength and
assign it a value of 235 [MPa] (units must be given in brackets). We will use that
later on to compute the safety factor for the current design.
16 /
23
Figure 21: Defining a derived parameter
(BeamZOffset)
17 /
23
Return to the project view and double click on the Paramter Set block. The
output parameter P3 - Equivalent Stress Maximum should now be visible under
Output Paramters. As we did before with the TensileYieldStrength, define a new
derived parameter SafetyFactor as
SafetyFactor =TensileYieldStrength
(Figure 23).
EquivalentStressMaxi
mum
D.3 Optimization
Back in the Project View drag and drop the Response Surface Optimization
system from the
Toolbox → Design Exploration onto the project schematic (Figure 24).
18 /
23
Figure 24 Adding the optimization system
19 /
23
Figure 25: Selecting a central composite DoE
Next, select the input parameter P1 (BeamThickness) and set its lower and upper
bound to 20 and 60 respectively (i.e. we assume the optimum thickness is
somewhere between 20 and 60 mm; Figure 26).
20 /
23
TensileYieldStrength is a constant and should not vary. Thus, uncheck the
Enabled check box right to the P4 – TensileYieldStrength row in the Outline
of Schematic B2: Design of Ex- periments window.
Right click on the Design of Experiments row and select Update; this generates
the five design points (= five samples) and evaluates these five differently
parameterized FE models. After- wards, click Return to Project.
Right click on Response Surface and select Update to fit the response surface
model to the sample points. When ANSYS is done, select the Goodness of Fit
node under the Metrics node. The plot as well as the computed error metrics
shows that the quadratic surrogate model is able to predict the behavior of the
full FE model (Figure 27).
Figure 27 Goodness of fit plot (prediction based on the response surface model vs. FE
model)
21 /
23
We can also plot the created model (= response surface) itself by selecting the
Response node from Response Points → Response Point. Set the x-axis to
BeamThickness and the y-axis to Equivalent Stress Maximum to produce a plot
resembling Figure 28. From this plot we can already tell, that the ideal thickness
should be located somewhere around 35 mm.
Figure 28 Response surface plot (max. von Mises stress vs. beam thickness)
When you are done investigating the response surface, click Return to Project.
Next, select the Objectives and Constraints row. In the table to the right
(Table of Schematic B4: Optimization) add constraints and objectives as
shown in Figure 29.
22 /
23
Figure 29 Objectives and constraints definitions
Candidate Point 1 seems to be the best bet for an optimum thickness. To check
whether this response surface based optimum agrees with the prediction of the
FE model, right click on Candidate Point 1 and select Verify by Design Point
Update. This will initiate a full FEA using a BeamThickness of 36.55 mm. As it
turns out, the predicted safety factor is almost identical with the one determined
via the FE model; we can therefore safely accept BeamThickness = 36,55 mm as
the optimal beam thickness (Figure 31).
23 /
23