Isight Tutorial Number 2.

DOE and Monte Carlo

simulation of a beam stress
analysis.
Stefano Morlacchi
February 2015

yFL
σ =
4I

Simuleon B.V.
Sint Antoniestraat 7 5314 LG Bruchem
T. +31(0)418-644699 F. +31(0)418-644690 E. [email protected] W. www.simuleon.com
1. Introduction
This tutorial aims at showing an example of DOE and Monte Carlo simulation within
a stress analysis of a loaded beam. A Design of experiment (DOE) component is
used to investigate the influence of some geometric parameters on the maximum
stress found in the beam. Subsequently, a Monte Carlo simulation is defined to
evaluate the reliability of the beam’s production process by examining the influence
of production tolerances on outputs.

When you complete this tutorial, you will be able to:

- Define a DOE analysis in Isight.
- Create a simple workflow.
- Define a Monte Carlo simulation.

Preliminaries

- The simple problem studied in this tutorial is a stress analysis of a beam

loaded at the centre and supported at both ends:

yFL FL2
σ=  S max = σ ( y = L / 2) =
4I 8I
where y is the distance from the centre, F is the central load, I is the moment of
inertia and L is the length of the beam.

- The beam has a circular cross section and its moment of inertia is the following:

r4
I =π
4
- This tutorial is divided in part A for the DOE analysis and part B for the Monte
Carlo simulation.

- Open a new standalone Isight Design gateway by using the Start Menu.

Simuleon B.V.
Sint Antoniestraat 7 5314 LG Bruchem
T. +31(0)418-644699 F. +31(0)418-644690 [email protected] W. www.simuleon.com
2
PART A. DESIGN OF EXPERIMENT
2. Setting up the Isight workflow
1. Enter the Sim-flow tab and drag a Calculator component from the Application
Components Tab onto the lower black arrow. (Fig. 1)

Figure 1. Definition of the Isight workflow. Part 1.

2. Change the Task1 process component to a DOE component by right-clicking

the Task1 icon and selecting Change To  New in the menu and DOE in the
‘Select New component’ dialog box (Fig. 2).

Figure 2. Definition of the Isight workflow. Part 2.

Simuleon B.V.
Sint Antoniestraat 7 5314 LG Bruchem
T. +31(0)418-644699 F. +31(0)418-644690 [email protected] W. www.simuleon.com
3
3. Now, enter the Parameters tab and click on the Add New Parameter button on
the right to create new parameters that will be used in our workflow (Fig. 3).

Figure 3. Parameters tab.

4. In the Add parameter dialog box, enter Length as Name, keep Scalar and Input
as Structure and Mode respectively, and enter 10 as Real value. A new line will be
created in the parameters list. Notice that all the options previously defined can be
directly modified by editing the values in the table.

5. Repeat the sequence described previously to define the following parameters,

no need to define any units:

- Radius: 0.5 m
- Force: 50,000 N

Simuleon B.V.
Sint Antoniestraat 7 5314 LG Bruchem
T. +31(0)418-644699 F. +31(0)418-644690 [email protected] W. www.simuleon.com
4
3. Edit the Calculator Component
1. Double-click the Calculator Component (Fig. 4). The dialog box is composed
of a blank area on top where you can created on or more assignment statements,
a parameters list (below on the left), a calculator and a list of useful functions (right).

2. In the blank area, write the following statement to calculate the Moment of
Inertia of the column. To include the parameters in the formulas (i.e., RADIUS),
double-click the name of the parameter in the parameters list.
M_INERTIA=pi( )*RADIUS^4/4

3. Click Calculate to include the new parameter in the list on the left.

4. Enter a new statement to calculate the maximum stress found in the beam
called S_MAX and click OK (Fig. 4).
S_MAX=FORCE*LENGTH^2 / (8*M_INERTIA)

Figure 4. Edit the Calculator component.

4. Edit the DOE Component

Simuleon B.V.
Sint Antoniestraat 7 5314 LG Bruchem
T. +31(0)418-644699 F. +31(0)418-644690 [email protected] W. www.simuleon.com
5
Double-click on the DOE component.

1. In the general tab, select Optimal Latin Hypercube as DOE technique from the
drop down box. Enter 50 as number of points (Fig. 5).

Figure 5. DOE component. General Tab.

2. Enter the Factors tab and select the parameters LENGTH and RADIUS as input
for the optimization. Enter the lower and upper bounds as shown in Fig.6 using
the percentage option (Lower bound, -30%; Upper bound, +30%). Move to the
Design Matrix tab and click Generate to show the evaluation points in the design
space. If wanted, add new evaluation points in the last line of the table.

3. Enter the Postprocessing tab and select the output variable parameter S_MAX
as objective and choose to minimize it (Fig. 7). Click Ok.

4. Save the workflow by clicking on FileSave as from the main menu. Rename
the file as Isight_Beam_DOE.zmf

Simuleon B.V.
Sint Antoniestraat 7 5314 LG Bruchem
T. +31(0)418-644699 F. +31(0)418-644690 [email protected] W. www.simuleon.com
6
 Figure 6. Factors tab and Design matrix tab.

Figure 7. Post-processing tab.

5. Pre-define post-processing outputs

Simuleon B.V.
Sint Antoniestraat 7 5314 LG Bruchem
T. +31(0)418-644699 F. +31(0)418-644690 [email protected] W. www.simuleon.com
7
1. Before running the optimization, enter the Graph Templates tab to create
templates of post-processing outputs that will be populated while the model is
running (Fig. 8).

2. Click on the Create a graph for this component icon and select to
show both the Pareto graph and the Main Effects plot for the S_MAX output.

3. Click on the Constraint Overlay 3D graph icon to create a 3D surface

graph based on 3 parameters. In this case, select Length parameter as X axis,
Radius as Y axis and S_MAX as Z axis.

Figure 8. Graph templates tab.

4. Before running the optimization, check that the parameters are correctly mapped
by showing the dataflow of the analysis. Enter the Dataflow tab and then the
Overview tab. Three parameters (LENGTH, RADIUS and FORCE) should be
entering the calculator component from the DOE process while the parameter
S_MAX should be mapped in the opposite direction.

6. Run the Model

Simuleon B.V.
Sint Antoniestraat 7 5314 LG Bruchem
T. +31(0)418-644699 F. +31(0)418-644690 [email protected] W. www.simuleon.com
8
1. Save the Isight workflow. From the main drop down menu, click on FileSave.

2. Run the model by clicking on the blue arrow in the main toolbar. Once the Isight
Run Gateway opens, go to the history tab and check the analysis results.

7. Post-Processing and results visualization

1. Scroll the values of the History tab until you find the Optimal solution (green
line). (Fig. 9)

2. Enter the Graphs tab of the Isight Runtime Gateway and make sure the
Optimization component is selected in the left model tree. Look at the graphs
previously defined, now populated with the outcomes of this run (Fig. 10). What
can you see from these results? What parameter has the main effect?

3. If wanted, spend some time to discover and create new post-processing options
such as Scatter Plot 2D graphs, Constraint 2D Overlay Graphs etc.

Figure 9. History tab. Results.

Simuleon B.V.
Sint Antoniestraat 7 5314 LG Bruchem
T. +31(0)418-644699 F. +31(0)418-644690 [email protected] W. www.simuleon.com
9
 Figure 10. Graph tab. Results.

PART B. MONTE CARLO SIMULATION

After our DOE analysis, the optimal design found which minimizes the stress in
the beam is characterized by a radius equal to 0.6194 and a beam length equal to
7.367. Stress is 2.93 MPa.
However, we are aware that the production process of the beam has an internal
variability on the radius and the length variables due to manufacturing
tolerances. Mont Carlo component can be used to address this uncertainty and
the reliability of the solution.
A maximum stress limit of 3.1 MPa can still be accepted and Monte Carlo
simulation will provide an estimate on the probability that the product will fail this
constrain,

8. Configuration of the Monte Carlo component

1. Go back to the Isight Design Gateway and right-click the DOE component. Select
Change To  New in the menu and Monte Carlo in the ‘Select New component’
dialog box.

Simuleon B.V.
Sint Antoniestraat 7 5314 LG Bruchem
T. +31(0)418-644699 F. +31(0)418-644690 [email protected] W. www.simuleon.com
10
2. Double-click the Monte Carlo component and enter the General Tab. Choose
Simple Random Technique as Sampling technique and enter 1000 as number of
points.

3. Enter the Random Variables Tab. Tick the parameter Length and define its
distribution information by selecting Normal as distribution, entering 7.367 as
mean value and 0.01 as Coefficient of Variation. Click Fix close to the Coefficient
of Variation cell.

4. Tick the parameter Radius and define its distribution information by selecting
Normal as distribution, entering 0.6194 as mean value and 0.01 as Coefficient
of Variation. Click Fix close to the Coefficient of Variation cell. (Fig. 11)

5. Enter the Responses tab and enter 3,100,000 Pa as upper limit for the S_MAX
parameter. This limit is the maximum acceptable S_MAX value (Fig. 12).

Figure 11. Monte Carlo component. Random Variables Tab.

Simuleon B.V.
Sint Antoniestraat 7 5314 LG Bruchem
T. +31(0)418-644699 F. +31(0)418-644690 [email protected] W. www.simuleon.com
11
 Figure 12. Monte Carlo component. Responses Tab.

9. Pre-define post-processing outputs

1. Before running the optimization, enter the Graph Templates tab to create
templates of post-processing outputs that will be populated while the model is
running (Fig. 13).

2. Click on the Create a graph for this component icon and select to
show both the Cumulative distribution and the Probability distribution for the
S_MAX output.

3. Click again on the Create a graph for this component icon and
select the Scatter Plot 2D graph. Select both Length and Radius as X axis and
S_MAX as Y axis.

Simuleon B.V.
Sint Antoniestraat 7 5314 LG Bruchem
T. +31(0)418-644699 F. +31(0)418-644690 [email protected] W. www.simuleon.com
12
10. Run the Model
1. Save the Isight workflow. From the main drop down menu, click on FileSave
as and rename the file as Isight_Beam_MS.zmf.

2. Run the model by clicking on the blue arrow in the main toolbar. Once the Isight
Run Gateway opens, go to the history tab and check the analysis results.

Figure 13. Monte Carlo component. Post-processing outputs.

Simuleon B.V.
Sint Antoniestraat 7 5314 LG Bruchem
T. +31(0)418-644699 F. +31(0)418-644690 [email protected] W. www.simuleon.com
13
11. Monte Carlo simulation results
1. Once the Isight Runtime Gateway opens, enter the History Tab which shows
the values of input and output design variables for each run. Red lines correspond
to the runs which fail to respect the upper limit constraint on the S_MAX parameter.

2. Move to the Graphs tab and look at the results shown. Due to the manufacturing
process uncertainty, what is the probability that the response is over the specified
limit? (Fig. 14)

Figure 14. Monte Carlo component. Post-processing results.

Simuleon B.V.
Sint Antoniestraat 7 5314 LG Bruchem
T. +31(0)418-644699 F. +31(0)418-644690 [email protected] W. www.simuleon.com
14