Suspension System Tutorial

(AutoDesign)
Copyright © 2017 FunctionBay, Inc. All rights reserved.
User and training documentation from FunctionBay, Inc. is subjected to the copyright laws of
the Republic of Korea and other countries and is provided under a license agreement that
restricts copying, disclosure, and use of such documentation. FunctionBay, Inc. hereby grants
to the licensed user the right to make copies in printed form of this documentation if provided
on software media, but only for internal/personal use and in accordance with the license
agreement under which the applicable software is licensed. Any copy made shall include the
FunctionBay, Inc. copyright notice and any other proprietary notice provided by FunctionBay,
Inc. This documentation may not be disclosed, transferred, modified, or reduced to any form,
including electronic media, or transmitted or made publicly available by any means without
the prior written consent of FunctionBay, Inc. and no authorization is granted to make copies
for such purpose.
Information described herein is furnished for general information only, is subjected to change
without notice, and should not be construed as a warranty or commitment by FunctionBay,
Inc. FunctionBay, Inc. assumes no responsibility or liability for any errors or inaccuracies that
may appear in this document.
The software described in this document is provided under written license agreement,
contains valuable trade secrets and proprietary information, and is protected by the copyright
laws of the Republic of Korea and other countries. UNAUTHORIZED USE OF SOFTWARE OR
ITS DOCUMENTATION CAN RESULT IN CIVIL DAMAGES AND CRIMINAL PROSECUTION.

Registered Trademarks of FunctionBay, Inc. or Subsidiary

RecurDyn is a registered trademark of FunctionBay, Inc.
RecurDyn/Professional, RecurDyn/ProcessNet, RecurDyn/AutoDesign, RecurDyn/Bearing,
RecurDyn/Belt, RecurDyn/Chain, RecurDyn/CoLink, RecurDyn/Control, RecurDyn/Crank,
RecurDyn/Durability, RecurDyn/EHD, RecurDyn/Engine, RecurDyn/eTemplate,
RecurDyn/FFlex, RecurDyn/Gear, RecurDyn/HAT, RecurDyn/Linear, RecurDyn/Mesher,
RecurDyn/MTT2D, RecurDyn/MTT3D, RecurDyn/Particleworks I/F, RecurDyn/Piston,
RecurDyn/R2R2D, RecurDyn/RFlex, RecurDyn/RFlexGen, RecurDyn/SPI, RecurDyn/Spring,
RecurDyn/TimingChain, RecurDyn/Tire, RecurDyn/Track_HM, RecurDyn/Track_LM,
RecurDyn/TSG, RecurDyn/Valve
are trademarks of FunctionBay, Inc.

Edition Note
This document describes the release information of RecurDyn V9R1.
Table of Contents
Outline of Tutorial Sample D ..................................................... 1
Suspension System Design Problem ........................................... 2
Loading the Model and Viewing the Yaw & Roll ranges .................. 4
Defining the design variables .................................................... 5
Defining the performance index ................................................. 7
Running a Design Optimization .................................................. 9
Design Optimization with Screening Variables ............................ 14
 S U S P E N S I O N S Y S T E M T U T O R I A L ( A U T O D E S I G N )

Outline of Tutorial Sample D

Model Description

Suspension System Design Problem:

This design has two design objectives. Also, it has 27 design
variables. The design goal is to minimize the Yaw-range and the
Roll-Range of tire motion. This problem is not easy because it is a
multi-objective problem and has too many design variables. In
general, other design tools employ D-Optimal design or Latin
Sample D Hypercube for constructing the meta model. Suppose that
construct a quadratic response surface model, D-optimal design
fundamentally requires 406 sampling points, which is evaluate by
using 1+2*27+27*26/2. AutoDesign uses however only 44
evaluations to solve the problem without variable screening.
Key Point: Study the concept of multi-objective optimization and
the procedure of screening design variables.

1
 S U S P E N S I O N S Y S T E M T U T O R I A L ( A U T O D E S I G N )

D
Sample

Suspension System Design

Problem
Let’s consider the simple model of car suspension system.
The system has 5 components such as arm, tie rod, knuckle,
shock absorber damper and tire. When the tire moves along
vertical direction, the simulation shows the damping process
and kinematical motion.
All the design variables are the geometric coordinates of
joints. The design objective is to minimize the first rotational
Yaw-Roll of tire.
This problem has 27 design variables for 9 joints.
Nevertheless, we will try to minimize the Yaw-range and the
Roll- range directly. In other words, design variable
screening is not used. Next, we will re-try to solve the same problem after screening the
design variables. Then, the optimization results will be compared.

Open files related in Sample-E

<InstallDir>
Sample
\Help\Tutorial\AutoDesign\SuspensionSystem\Examples\SAMPLE_D0.rdyn

<InstallDir>
1
\Help\Tutorial\AutoDesign\SuspensionSystem\Solutions\SAMPLE_D0.rdyn

<InstallDir>
Solution 2
\Help\Tutorial\AutoDesign\SuspensionSystem\Solutions\SAMPLE_D1.rdyn

<InstallDir>
3
\Help\Tutorial\AutoDesign\SuspensionSystem\Solutions\SAMPLE_D2.rdyn

2
 S U S P E N S I O N S Y S T E M T U T O R I A L ( A U T O D E S I G N )

Note: If you change the file path at discretion, it can be located in any folder that you specify.

3
 S U S P E N S I O N S Y S T E M T U T O R I A L ( A U T O D E S I G N )

1
Chapter

Loading the Model and Viewing

the Yaw & Roll ranges
To load the base model and view the animation:

1. On your Desktop, double-click the

RecurDyn tool.
RecurDyn starts and the Start RecurDyn
dialog box appears.
2. Close Start RecurDyn dialog box. You will
use an existing model.
1. In the toolbar, click the Open tool and select
‘Sample_D0.rdyn’ from the same directory
where this tutorial is located. The suspension
system appears in the modeling window.
(<InstallDir>\Help\Tutorial\
AutoDesign\SuspensionSystem\
Examples\ )
2. Click the Dynamic/Kinematic button.

3. Click the Plot Result button. Then, window is switched. In right side, you can see plot
database. If it is not shown, check the database window in View menu. Expression 2 &
3 are Yaw and Roll angles.

4
 S U S P E N S I O N S Y S T E M T U T O R I A L ( A U T O D E S I G N )

2
Chapter

Defining the design variables

All geometric coordinates of joints are
defined by using Parametric Points
shown at the below figure. In Figure, they
are represented as A_x, A_y, …, R_y
and R_z.

Next, the parametric values are defined in

the Parametric Value menu in
Subentity

5
 S U S P E N S I O N S Y S T E M T U T O R I A L ( A U T O D E S I G N )

All the above relations have defined in the model ‘SAMPLE_D.rdyn’.

Next, select the Design Parameter menu.
Then, you can see that 27 parametric values are linked by design parameters as shown
below.

6
 S U S P E N S I O N S Y S T E M T U T O R I A L ( A U T O D E S I G N )

3
Chapter

Defining the performance index

Let’s consider the performance indexes. We will minimize the Yaw and Roll ranges during tire
moves vertical direction. As RD does not provide those values directly, we should evaluate
them by using Expression and Variable Equation.

In order to evaluate them, we evaluate the minimum and the maximum values for Yaw and
Roll angles. Then, the deviations between Max and Min are the ranges for them.

7
 S U S P E N S I O N S Y S T E M T U T O R I A L ( A U T O D E S I G N )

 Step 1: Create the Expression for Yaw and Roll angles. The following figures show the
expressions to save Yaw and Roll angles.

 Step 2: By using Expression, ‘YAW_DEVIATION’ and ‘ROLL_DEVIATION’ are defined.

Then, you can define them as analysis responses at ‘Analysis Response’ in AutoDesign.
The following figure shows this process.

8
 S U S P E N S I O N S Y S T E M T U T O R I A L ( A U T O D E S I G N )

4
Chapter

Running a Design Optimization

When you select the ‘Design Optimization’, all the design variables, check in DV, will be
included in the design. The performance index however will be empty. Then, you add ARs to
the Performance Index.
The optimization problem is to minimize the Yaw Range and the Roll Range while satisfying
those two ranges less than them of the current design. Also, we think that Yaw Range is
more important than Roll Range. Thus, its’ weight is twice greater than that of Roll.
Minimize Yaw_Deviation*2 & Roll_Deviation
subject to Yaw_Deviation =< 0.67
Roll_Deviation =< 2.12,
where the values of 0.67 and 2.12 are the range values for the current design. Those two
inequality constraints seem to be redundant because two objectives should be minimized. It
is however not guaranteed. Thus, it is difficult to solve multi-objective optimization problem.
In the Part II manual: Guideline for AutoDesign, the chapter 5 explains the multi-objective
optimization.

Let’s consider the above design problem from the view of multi-objective optimization
process. If necessary, let’s denote f1  Yaw _ Deviation and f 2  Roll _ Deviation . When
the initial samples construct the approximate functions of and , which are called as
metamodels. Then, f1 x and f 2 x  is the minimum values from the initial samples.

 f1 (x)  f1*   f 2 (x)  f 2* 

Min max  ,  
 f1*  f 2* 
Thus, the quality of multi-objectives optimization depends on the number of initial samples.
Unfortunately, this problem has 27 design variables, which requires too many sample points.
Those two inequality constraints can avoid the pre-mature convergence, even though the
number of initial samples is minimized.

9
 S U S P E N S I O N S Y S T E M T U T O R I A L ( A U T O D E S I G N )

Let’s start to solve the multi-objective optimization problem:

1. Click the Performance Index Tab. Then, define the above optimization formulation as
follows:

2. Click the Optimization Control Tab. The default values are directly used. Then, click
the Execution button.

Then, you can see the summary of design formulation. Check Design Variables,
Performance Index and Meta-Model information. If all information is correct, then
click the OK button. Then, optimization process is progressed.

10
 S U S P E N S I O N S Y S T E M T U T O R I A L ( A U T O D E S I G N )

3. When the optimization process is completed, the Result Sheet Tab is automatically
shown. The optimization process is converged in 14 iterations. Thus, AutoDesign solves
the suspension system design having 27 design variables only for 47 analyses that
includes 33 analyses for the initial sampling points. The final design gives that
AR1=0.302 and AR2=2.025, which can minimize the Yaw deviation by 58.6 % and the
Roll deviation by 4.2 %.

11
 S U S P E N S I O N S Y S T E M T U T O R I A L ( A U T O D E S I G N )

 Convergence History

 Summary Sheet

12
 S U S P E N S I O N S Y S T E M T U T O R I A L ( A U T O D E S I G N )

In the summary sheet, shown in the above, is newly provided. The optimization results
are summarized in the design variables and analysis responses lists. Also, the SAO
information is summarized, which shows that SAO requires 14 iterations. Thus, the
number of total evaluations is 47. And the analysis result of optimal design is ‘DO_014’.
4. Finally, let’s compare the yaw and roll ranges for the initial design and the final design.
DO_014 is the final design. Also, DOE33 is the initial design. The following figures show
those comparisons. When the ISCD-1 and ISCD-2 compose the initial samples, the final
one of the initial samples is the current design. In the following figures, the red color line
is an optimum result.
Yaw(Toe) Roll(Camber)

13
 S U S P E N S I O N S Y S T E M T U T O R I A L ( A U T O D E S I G N )

5
Chapter

Design Optimization with

Screening Variables
Now, let’s compare the optimization results for considering all design variables and the
screened design variables. Thus, save the model file ‘Sample_D0.rdyn’ as
‘Sample_D1.rdyn’ and delete all the results in the Simulation History.
First, let’s try to screen the design variables. As screening is based on the effect analysis
results, you select the design study in AutoDesign. Following steps explain the screening
process:
1. Enter the Design Study menu.
2. Then, you can see the list of design variables in the Design Variable tab. As the
number of design variables is 27, we select the 2-Level Orthogonal Array to reduce
the number of trials. If 3-Level Orthogonal Array is selected, the number of trials is 81.

14
 S U S P E N S I O N S Y S T E M T U T O R I A L ( A U T O D E S I G N )

3. Next, in the Performance Index tab, two analysis responses such as AR1 and AR2 will be
shown. If they are not, you retry to make ‘Sample_D1.rdyn’ from ‘Sample_D0.rdyn’.

4. Next, click the Simulation Control tab. If you save the analysis results each trial, check
the Save Results and enter the Folder name. Then, click the Execution button.

5. After all analyses have been completed, the design study window is activated. Then,
click the Effect Analysis tab. Then, you can see the effect analysis chart by checking
the check boxes in the Effect Values column and clicking the Draw button. The effect
analysis results are shown as follows:

15
 S U S P E N S I O N S Y S T E M T U T O R I A L ( A U T O D E S I G N )

Two figures show that the different variables are sensitive to PI_1 and PI_2. Generally,
users select the sensitive variables from the above charts. It is however not easy. Thus,
AutoDesign provides a statistical guideline for screening variables.

6. Click the Screening Variables tab. Then, select AR_1 in the PI box. Then, you can see
the scatter points. The right positioned points are more sensitive to the left ones. Move
the controller to divide design variable groups. Then, the Cutoff Value as 0.3 and click
the Screening DV button.

16
 S U S P E N S I O N S Y S T E M T U T O R I A L ( A U T O D E S I G N )

Then, the red line divides the scatter points into two groups. The right group is the
remaining variables. Then, the screened variables are marked as ‘On’. Others are done
as ‘Off’.

Next, select AR2 in the PI box and do the same process similarly except defining Cutoff
Value as 0.06. Then, the screened variables are summarized as follows:

The screening DV gives that the active design variables (DV) are DP3, DP6, DP8, DP10,
DP12, DP15, DP17, DP19, DP21 and DP27.
Now, create the new model by clicking the Save button with checking Create New
Model. The new model file name is ‘Sample_D2.rdyn’.

7. Now, you can see that the model file is changed as ‘Sample_D2.rdyn’. Click the
Design Parameter icon in AutoDesign. Unlike the model ‘Sample_D1.rdyn’, the
screened variables are only checked in DV column. This represents that DV1=DP3,
DV2=DP6, DV3=DP8, DV4=DP10, DV5=DP12, DV6=DP15, DV7=DP17, DV8=DP19,
DV9=DP21 and DV10=DP27. Most of the active design parameters are z-coordinate
values.

17
 S U S P E N S I O N S Y S T E M T U T O R I A L ( A U T O D E S I G N )

8. Next, Click the Design Optimization icon. Then, the screened variables are shown as
follows:

9. Click the Performance Index tab. The optimization formulation will be the same as
that of ‘Sample_D1.rdyn’. Let’s use the same formulation.

18
 S U S P E N S I O N S Y S T E M T U T O R I A L ( A U T O D E S I G N )

10. In the Optimization Control tab, all the convergence tolerances use the default values.
The result files are saved in the folder ‘Screening_Variables’.

11. After the optimization is completed, let’s see the result sheet. AutoDesign is converged
in 2 iterations. The optimized AR1 and AR2 are 0.515 and 2.082, respectively. AR1 and
AR2 are similar with those of the design without screening.
 Convergence History

19
 S U S P E N S I O N S Y S T E M T U T O R I A L ( A U T O D E S I G N )

Unlike the convergence history for the design problem without screening, the above
convergence history simply decreased. Let’s compare the summary of two optimization
problems.

With Screening Without Screening

Number of design variables 10 27

Initial samples 13 33

Number of SAO runs 18 14

Samples for screening 32 -

Optimum response 0.515, 2.082 0.302, 2.025

Thanks for participating in this tutorial!

20