Modeling of a formula student race car in Matlab/Simulink with

Simmechanics
Citation for published version (APA):
Jong, de, B. A. J. (2004). Modeling of a formula student race car in Matlab/Simulink with Simmechanics. (DCT
rapporten; Vol. 2004.106). Technische Universiteit Eindhoven.

Document status and date:

Published: 01/01/2004

Document Version:
Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)

Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can be
important differences between the submitted version and the official published version of record. People
interested in the research are advised to contact the author for the final version of the publication, or visit the
DOI to the publisher's website.
• The final author version and the galley proof are versions of the publication after peer review.
• The final published version features the final layout of the paper including the volume, issue and page
numbers.
Link to publication

General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners
and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
• You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please
follow below link for the End User Agreement:
www.tue.nl/taverne

Take down policy

If you believe that this document breaches copyright please contact us at:
openaccess@tue.nl
providing details and we will investigate your claim.

Download date: 25. 1월. 2024

 Modeling of a Formula Student
race car in Matlab/Simulink
with Simmechanics

B.A.J. de Jong

OCT Report no: 2004.106

October 2004

Student:
Bram de Jong 0491090
B.A.J.d.Jong@student.tue.nl

Supervisor:
Igo Besselink
I.J.M.Besselink@tue.nl

/Department of Mechanical Engineering

Table of contents

1. Introduction 3
1.1 Fomlula Student 3
1.2 Background 3
1.3 Fonnulation of the assignment 3
2. Modeling of the vehicle 4
2.1 Lntroduction 4
2.2 Assumptions 5
2.3 The MATLAB/Simulink model 5
2.3.1 The Con/rols subsystem 6
2.3.2 The Model+ Visualisation subsystem 6
2.3.3 The Analysis subsystems 9
2.3.4 The parameters structure 10
3. Simulations 11
3.1 Introduction 11
3.2 Adjusting the front- and rear roll stiffness 12
3.3 Adjusting the roll damping . 12
3.4 Adjusting the track width 15
3.5 Adjusting the static camber angles 15
3.6 Conclusion 16
4. Conclusion and recommendations 17
4.1 Simmechnics 17
4.2 The vehicle 17
4.3 Recomnlendations 18
Appendix A 19
Appendix B 20
Appendix C 21
Appendix D 22
Appendix E 23
Appendix F 25
Appendix G 26
List of symbols 27
References 28

2
1. Introduction

1.1 Formula Student

Formula Student is a project for engineering students to design and build a small single-sealer racing
car. The project forms part of their academic studies and culminates in a competition where teams
from all over the world come together 10 race and compete against each other. Restrictions are placed
on the car frame and engine so the students' knowledge, creativity and imagination are tested, Four-
cycle engines up to 610 cc can be turbocharged or supercharged 10 add a new dimension to the
challenge of engine design.
Formula Student is being organised by the English Institution of Mechanical Engineers (IMechE), as a
spin-off of the American Formula SAE organised by the Society of Automotive Engineers (SAE).
The vehicles are judged in three different categories: static inspection, engineering design, solo
periormance trials and high-periormance track endurance, combining all periormance fields.
For the purpose of the competition, the students are to assume that a manufacturing firm has engaged
them to produce a prototype car for evaluation. The intended sales market is the non-professional
weekend autocross, hill climb or sprint racer. Therefore, the car must have very high periormance in
terms of its acceleration, braking and handling qualities. The car must be low in cost, easy to maintain
and reliable. In addition, the car's marketability is enhanced by other factors such as aesthetics,
comfort and use of common parts. The manufacturing firm is planning to produce 1000 cars per year
at a cost below 21.000 euro. The challenge to the team is to design and fabricate a prototype car that
best meets these objectives. Each design is compared and judged with other competing designs to
determine the best overall car.

1.2 Background
In the development of a new car computer simulations have become an essential part of the process.
With the computer a number of different design variants can be tested and an oplimalisation of the
variables can be done. This is also the case with the first Formula Student car of the Technische
Universiteit Eindhoven that is under development at the moment. Because this car is developed by,
mostly inexperienced, students, the computer simulations are also very useful in learning the handling
of such cars and from there learn how to setup these vehicles.
To do an analysis of the vehicle dynamics and the suspension geometry a multi-body model is
needed, in witch particularly the suspension is described in great detail. A bicycle model or two-track
vehicle model as presented in the Vehicle Dynamics lectures aren't sufficiently detailed for suspension
analysis.
Adams is a commonly used multi-body software package, but recently The Malhworks launched a
multi-body toolbox for the MATLAB/Simulink environment called Simmechanics. Advantages of this
toolbox are, besides the fact that most students are already familiar with MATLAB/Simulink, that the
model can easily be expanded with controllers or other toolboxes of Simulink, so every aspect of the
car can be investigated or tested before it is used on the real car. Some disadvantages can be the
graphical animations and the fact that the toolbox is relatively new, and therefore maybe a little
unstable.

1.3 Formulation of the assignment

In this assignment a vehicle model of the current design of the Formula Student race car will be made.
Some relevant simulations can be done to show the possjbilities of the model, for example driving over
a bump and a step input on the steering wheel. Also an interesting animation will be made, which can
be used for graphical analysis and PR-goals. Thirdly a short description of the model will be made so
others can use it or expand the model.
Goal of the assignment is to make a first start for a complete vehicle model (including power train etc.)
and to check whether Simmechanics is suitable for these simulations.

3
2. Modeling of the vehicle

2.1 Introduction
A Formula Student car is a small single-seater open-wheel race car. It is fitted with a BODce engine.
Due to the restrictions on the air intake the power output of this engine will be around 90 bhp. Because
of the light circuit maximum speed during the race will be around 100 kmh-'. The chassis is build of
aluminium sandwich plates. This combines stiffness and a low weight together with high safely. The
suspension is fully independent with a travel of approximately 2,5 em. The car is fitted with aluminium
wheels and the lyres are special made racing lyres with a high adhesive compound.

Figure 2.1: The first Formula Student race car of the Technische Universiteit Eindhoven

In table 2.1 some specifications and measurements of the race car are given.

Tabl 2.1: Main specifICations of the vehicle

Wheelbase 1600 mm
Track width - front 1250 mm
Track width - rear 1200 mm
Mass including driver 300 kO
Suspension - front UneQuallenQth A-arm with Dushrod
Susoension - rear Uneaual lenath A-arm with oushrod
Wheel size - front 13x6 inch
Wheel size - rear 13x8 inch
Tvres - front 6,2/20,0 x 13
Tvres - rear 8,2/22,0 x 13

The high power-to-mass ratia and race lyres make the longiludinal- and corner accelerations exceed
19, which make Ihe car a thrill to ride. This performance equals for example that of Porsche 911 Turbo
and is even faster then the new Ferrari 612 Scaglietti.

4
2.2 Assumptions
For the modelling of the vehicle a number of assumptions are taken into account to simplify the model
and to decrease simulation lime.

• All joints and bodies, except from the tyres and springs, are rigid. Deformation under the
applied loads is not taken into account. Because of the stiff design of the chassis and
suspension parts this assumption can be validated. The contact patch of the tyres is modeled
as a spring-damper.
• Friction in bearings and joints isn't taken into account. The aerodynamic forces on the vehicle,
like drag due to the air resistance, are also not included. With the type of aerodynamic
package on this vehicle they have little effect on the dynamic behaviour of the vehicle.
• The suspension springs and dampers used are supposed linear. In reality bump stops are
used at the end of the maximum travel of the springs. These bump stops can be seen as very
stiff springs that make sure that the suspension travel is limited.
• The tyre model used is the Magic Formula (MF-tyre) developed by TNO Automotive. The
Magic Formula aims at an accurate description of the steady-state behaviour of a tyre by
providing a set of mathematical formulae, which are partly based on a physical background.
The MF calculates the forces (F x and Fy) and moments (M x, My and Mz) acting on a tyre under
pure and combined slip conditions, using longitudinal and lateral slip (K and a), wheel camber
(y) and the vertical force (F z ) as input quantities. In addition to the Magic Formula description,
a set of differential equations is defined, representing the transient behaviour of the tyre with
respect to handling at frequencies up to 6 Hz. The most important physical parameters of the
Avon tyres are known or estimated using comparable tyres and can be seen in table 2.2. The
characteristics approximated by the model can be seen in Appendix B.

Tabl" 2.2: Parametars of the lyres used

front rear
.
vertical stiffness 16 kNm 21 kNm
.
vertical dam Dina 50 Nsm 50 Nsm
nominal vertical force 665 N 810 N
lonaitudinal SliD coefficient 1,3 . 1,4 -
lateral SliD coefficient 1,3 '-1 1,4 -
rollina resistance coefficient 0,015 I-I 0,015 1-

• The position of the centre of gravity, the mass and the inertia tensor of the chassis, induding
power train and driver, are approximated using a finite element analysis. To modellhe power
train, the engine and drive shafts, and the driver basic shapes with a uniform mass distribution
are used. For the driver parameters a 95-percentile person estimation is used. The
parameters of the wheels are also calculated with a finite element analysis. The parameters of
the suspension parts are all approximated by basic shapes. Because of the low mass of these
parts compared with the chassis and the wheels the accuracy of the approximation doesn't
influence the performance of the model much.

2,3 The MATLAB/Simulink model

As stated before the model is build in the Simulink environment of Mallab with use of the
Simmechanics toolbox. In the mechanical environment a right turning Cartesian coordinate system is
defined. The origin lies on the ground under the centre of the rear axle. The gravitation constant is
9,61 ms·2 and points into the negative z-direction. In the initial position and orientation of the vehicle
the x-axis points into the drive direction and the y-axis to the left side of the car. Numbering of joints
and parts evolves from left/under/front 10 righUup/rear.

Figure 2.2: The coordinate system defined for the model

5
The model contains a hierarchy of sUbsystems. There are 3 main subsystems, which can be seen in
figure 2.3.

-._.
.-+-19
,.~

••

--,
.... I-
-
., ~
.
., H

Figure 2.3: The first level of the model with the 3 main subsystems
- ...."
~.

2.3.1 The Controls subsystem

In the Controls subsystem, the orange subsystem in figure 2.3, the vehicle is controlled. With the first
input of this subsystem the car can be steered. The given input is the steering wheel angle in degrees.
The slider gain can be used to give step inputs with angles between a minimum and maximum.
Obviously this input, and the others too, can be replaced by every arbitrary input. Because of some
numerical issues the steering wheel input is translated with a PO-controller to an input torque. This
operation takes place in the Steeringrack subsystem, which will be described later on in the
Model+Visualisation subsystem paragraph. The second input controls the brakes of the vehicle. The
input signal can vary between 0 and 1. where 0 is no braking and 1 is full braking. The brake torque on
the front and rear axle is calculated by multiplying this signal by a brake moment gain and a brake
moment distribution between front and rear. The brake torque per wheel is then calculated by dividing
the brake torque of the relevant axle by two. The third input is the throttle. This input signal can vary
between -0.2 and 1 and controls the torque of the engine delivered to the rear wheels. No throttle is
-0.2 and engine braking takes place. The input 1 is full throttle. The throttle can be controlled
manually, but by setting the switch CruiseControl-onioffthe cruise control can be engaged and the
throttle is controlled automatically to keep the vehicle at a constant speed. This cruise speed can be
set, in ms", as the fourth input of the Controls sUbsystem. In the display speed the vehicle velocity can
be seen so it can be determined whether it is within certain boundaries near the desired velocity.

2.3.2 The Modef+Visuafisation subsystem

The vehicle model itself is placed in the Mode/+Visualisation sUbsystem. It also contains a hierarchy of
subsystems, which is diverted back from the build up of the car. In figure 2.4 on the next page the
hierarchy can be seen. Every part of the car is represented by a body block from the Simmechanics
toolbox. They can be recognised in figure 2.4 as the blocks with the picture of the part they represent
on it. In these blocks the mass and inertia of the body have to be inserted as well as the connection
points with other bodies. The degrees of freedom between the bodies are restricted by the joint blocks.
They can be recognised as the cyan coloured blocks. The red blocks are sensor· or actuator blocks.
They can measure or control translation and rotation of a body or joint and measure or apply a force

6
and torque on a body or joint The green coloured blocks are used to collect the data for the
visualisation and analysis of the performance of the vehicle.

--
.__.Q :~
.-

-
>~ --
0
•• <-
-
-
.' " " .' -
Figure 2.4: The Mode/+Visualisalion subsystem

The central body of the model is the chassis. The world reference, the purple block, is connected to
this body and the initial position and velocity of the vehicle can be changed in the blue block. The
chassis can move totally free of the world so a Six-OoF joint is used to connect these two together.
Mounted with revolute joints to the chassis are the rockers. They transform the suspension travel
through the pushrod to compression or rebound of the spring and damper. The springs and dampers
are in the Spring/damper sUbsystem. There are 2 of these subsystems, front and rear, and in every
subsystem the suspension mechanism is placed. The design and working of the mechanism can be
seen in Appendix C. Every damper consists of 2 bodies, the casing and the piston rod. The difference
in position and speed of these parts is measured and used to calculate the orientation of the assembly
and the force it delivers on the rockers or connection points on the chassis. The parts that form the
external suspension of the car, the upper and lower A-arm, the pushrod, the tie rod and the upright,
are in the Suspension subsystems of which one can be seen in figure 2.5.

~
---
<........
---
.........
~'
i!:
.....
<.....:.. _
--,
I,
"I,
.
Ii

_...... .....-.
..... - f

\ ---
-
Ii

_..._. ..
~

..._. ..-
- ._.
Figure 2.5: The Suspension subsystem
-_.
-
..
.-
"'"'
• ._,
-, ..._, ._.
....."

7
In reality both A-arm's are connected with the chassis by so-called rod ends, which have 3 rotational
degrees of freedom in a single pivot point like a spherical joint. In the model however a telescoping
joint replaces 1 of the 2 joints of an A·arm with the chassis. A telescoping joint is a spherical joint, but
with 1 translational degree of freedom. This translation degree of freedom is in the x-direction,
because in this direction both joints of an A-arm with the chassis restrict the same movement. This
gives an overdetermined system and it therefore can happen that no solutions are found. The A-arm's
are mounted on the upright with spherical joints. The pushrod is connected with the rocker, as
described before, and with the lower A-arm. In reality the pushrod has a spherical joint on both sides.
In the model however the spherical joint on 1 side is replaced by a universal joint in which only 2
rotational degrees of freedom are defined. This is done because else the system would be singular,
since the rotation of the puSh rod round it's own axis isn't restricted or controlled then. The same holds
for the tie rod. The tie rod is the linkage between the upright and the steering rack. The steering rack is
modelled as two gears with a gear constraint between them. This ratio between both gears is very
small, which means that the radius of the second gear, the one on which the tie rods are mounted, is
much larger then that of the first gear, which is connected to the steering wheel rod. Because the
second gear is thai large it can be seen as the toothed bar in the steering rack. The first gear is torque
actuated. This is done because of some numerical problems, which occurred with the constraint
equations of the joints when using an angle or displacement as an input. Since a steering wheel angle
is used as input there is a feedback loop in this subsystem to translate the steering wheel angle to a
= =
torque input. Because of the high gain (K 200) and fast damping (0 5) of the feedback controller
this has very low effect on the input given to the vehicle due to a fast settling time and low steady state
error.

•
•

. _...-.....-
-
Figure 2.6: The Steeringrack subsystem

When the steering wheel is turned the steering rack makes the tie rods on both sides move sideways
and that makes the upright rotate about the axis through the 2 joints with the upper and lower A-arm.
In the upright lies the wheel bearing, modelled as a revolute joint. On this joint a joint actuator is
mounted. This actuator can apply a torque on the joint and is used here to apply the brake torque to
the wheel axis, and in case of the rear wheels also to apply the drive torque of the engine. Physically
this isn't totally correct, because now the engine- and brake torque are supported in the upright.
Therefore also a model with drive shafts is developed. More about this can be read in Appendix O.
Connected to the revolute joint is the wheel. The wheel is placed in the wheel subsystem. In this
subsystem the wheel body is placed, but also the interface that calculates the forces and torques on
the wheel by measuring the translation and rolation of the wheel body and the time·derivatives of this
movements. The rpm of the rear wheels is also used as an input of the engine model.
Coupled with the model is a virtual reality visualisation of the vehicle. As can be seen in figures of the
subsystems of the model the position and rotation matrix of every body is measured. This data is
needed to describe the 6 degrees of freedom of every part in the virtual world. With the green blocks
this data is send to the VR-sink block. This block is the interface between Simulink and the VRML-
viewer, the program in which the virtual reality scene runs. Note that the rolation matrix has to be

8
converted first to an equivalent representation in one axis and the rotation angle about this axis before
it is coupled with the VR-sink block, since this is the form used in Virtual Reality Modelling Language
(VRML) and the Virtual Reality Toolbox for orienting bodies. The virtual reality scene itself is made in
VRealm-builder and is a coupling of imported CAD-drawings of every part. In this VRealm-builder also
the points of view in the virtual world have to be constructed. These viewpoints can be static, but it is
also possible to import coordinates and rotations from Simulink, so for example a point of view can
drive along with the car. In the next figure the VR-sink block and a view of the visualisation can be
seen. The chassis is kept away so there is a better view over the suspension. The plotted data are the
driven path, the lateral acceleration and the yaw rate. The scene can be started by clicking on the VR-
sink block and choosing view.

• • •

FiqVf13 2.7: The VR-sink block and a view of the visualisation with some data plotted

In the VRML-viewer the viewpoints can be changed by clicking the right mouse button and then
choosing viewpoints. Viewpoints can also be changed manually. The controls for this can be seen in
appendix E. It is also possible to record the simulation as an offline movie in Matlab 7 R14. This also
can be found in Appendix E.

2.3.3 The Analvsis subsystems

During a run of the model all measured data is collected and sorted in the 2 Analysis subsystems. All
data can be seen real time in the scopes connected to the subsystems. The upper system analysis the
chassis, and the lower one the tyres. In table 2.3 can be seen which data is given as output.

To ble 2.': Tha outout data of the model

Chassis Tvres
x-nosition CG chassis m slio anole rad
v-nosition CG chassis m lonaitudinal slin coefficient (-)
z-position CG chassis m camber an Ie (rad)
lonaitudinal sneed (ms' lonaitudinal force (N)
lonaitudinal acceleration (ms' lateral force N)
vaw rate (rads") vertical force (N)
roll an Ie (de self-alianina moment Nm)
oitch anale deal
lateral acceleration (ms'

9
The data from the tyres is given by the MF-tyre interface. Most of the data of the chassis follows
directly out of the sensor blocks of Simmechanics. Only the roll angle, pitch angle and lateral
acceleration are calculated out of measured data. The roll angle and pilch angle are calculated using
the m-fiIe pr_angle.m. The rotation-matrix of the chassis with the world coordinate system as base is
measured. By integrating the yaw rate the rotation angle round the yaw axis of the vehicle also can be
obtained. With this angle a local coordinate system is defined which turns with the car when making a
corner. Now it is possible to obtain the rotations round the x-axis (roll angle) and the y-axis (pitch
angle) of this local coordinate system by using the measured rotation-matrix. The m-fiIe pr_angle.m
can be found in Appendix F. The lateral acceleration is calculated with formula 2.3, which follows from
formula 2.1 and 2.2. The speed V and yaw rate r of the vehicle are measured values. The corner
radius R has to be eliminated.

a
v'
~- (2.1 )
, R

R ~!:... (2.2)
r
Q y = Vr (23)

While running a simulation all the analysed data is also written to mal-files. After the simulation run the
m-fiIe telemetrie.m can be run. This m-file will collect all the data and then plots it in sorted figures.
First it will show the data from the chassis. By pressing any key it will show all lyre data. Then by
pressing any key it will ask to save the data. If chosen so, it will save all data in one .mal-file named as
specified at the start. The arrangement of the data in the .mat-file can be seen in Appendix G.

2.3.4 The parameters structure

All parameters of the model, for example the body masses and inertias, but also the steering ratio and
spring rates, are stored in a Mallab-structure. This has a couple of advantages. Firstly it is better
organised. It is also better attainable and therefore easy to make some changes. Other setups can be
simulated very rapidly. Secondly, because the parameters are not in the model itself, the model can be
used for every other vehicle as long it has the same suspension layout. In table 2.4 the first part of the
schematic representation of the structure is given. It can be seen that the lowest level in the first part
of the schematic representation are the italic printed parts. This is the main level in the sequel part of
the representation in table 2.5.

Table 2.4s: Schematic reoresentation of the Darameters structure

model
info chassis susDension wheel
units numberin world constants steerin rack fTontleft fTontri hI rearlett rearri ht fTonlleft frontri ht rearleft rearri ht
mass
inerlia
;.
, gear1
~ gear2
~
rocker rocker rocker
A-arm1 A-arm1 A-ann1
rocker
A-ann1
~ ~ ~ ~
length b A-arm2 A·ann2 A·ann2 A-arm2 ~ ~ ~ ~
S
angle
'.
brake gain ~
~
tie rod tie rod tie rod
pushrod pushrod pushrod
tie rod
pushrod
S
~
~
S
~
~
S
~
~
S
~
~
brake distr. upright upright upright upright

damper damper

,, ,
T,bl24bSeoue/Da
J rl 0 fth
'"hemer IC reoresen fafIOn

part
mass inertia CG CS1 CS ...
Th, Th, base1 axis base2 rotation coordinates orientation connection base1 axis base2 rotation coordinates orientation
mass a inertia Th, Th, Th<> Th, Th, Th, Th,
th, tensor of origin axis origin rotation translation rotation description
body the body of the of of the 0","' over the on the ofrhe
coordi the rotati defined defined connection
nate '00 on coordinate rotation
syste rdin syste system system
m
'f'
syst
m
,m

10
 3. Simulations
3.1 Introduction
Because of the great physical detail of the model, an enormous amount of things can be tested. The
model includes: spring and damper rates, anti-roll rates, bump steer, steer camber, steer forces, scrub
radius and pneumatic trial of lyres. roll and swing centre effects (anti-roll, anti-pitch), reaction forces in
suspension elements and compliance camber due to lateral force, tractive/braking force, aligning
torque and overturning moment of lyres. Static camber, castor, Ackermann, loe inlout and the track
width are several possible setup changes. To give a brief notice of the possibilities a few experiments
are done. Goal of these experiments is to improve the performance of the vehicle by making some
simple setup changes. First it is tried to improve the performance by adjusting the roll stiffness on the
front and rear of the car. Then there will be looked at the influence of the roll dampers. Also a different
setup for the track width and static camber of the front wheels will be tested. For the tests the model
without drive shafts is used, because of the shorter simulation time. From earlier measurements it
occurred that it doesn't make any visible difference in the behaviour and performance of the vehicle.
1
The test will be done at a speed of 72 kmh· with a step·input on the steering wheel of 65 degrees,
which can be seen in figure 3.1a. The cruise control makes sure the car will keep the desired speed,
with a little steady state error of 0,2%. The throttle output of the controller can be seen in figure 3.1 b.
The start·up reaction, which can be seen in every output of the vehicle, is a result of the car finding its
equilibrium at the start of the simulation.

- ",,----------'-"'-----------,

"

I"

Ei£l~a:
'. •
The input on the steering wheel cifJure 3~
•
The throttle input
•

In figure 3.2a and b the behaviour of the vehicle can be seen with the standard setup. II can be seen
in figure 3.2b that the car is understeered. The slip angles of the front wheels are larger than that of
the rear wheels.

I
- 0>

..
- Ie-§:
0>
II
•
"
.. I"
." ,
"
.,.
.• of--
.•,k-------:c-----~ ..; _----_!,.,
Il')

Figure 3.2: The lateral acceleration with the standard setup

•", ,
Figura 3.20: The slip angles
'. " "

11
3.2 Adjusting the front- and rear roll stiffnp.ss
By adjusting the roll stiffness the distribution of the vertical forces between the inside- and outside
wheels while driving through a corner can be changed. A car has the most grip when the difference of
the vertical loads between the inside- and outside tyres is as small as possible. Because the car is
understeered, the front wheels need more grip, so the front roll stiffness has to be lowered and more
weight will be transferred on the outside rear wheel. In the standard setup the front roll stiffness is 20
1
kNm· .ln this test this is lowered to 15 kNm-'. The lateral acceleration increases from 9,4 ms-2 to 9,6
2
ms· , Because of the lower roll stiffness and the higher lateral acceleration the roll angle of the vehicle
also increases from 1.3 degrees to 1,7 degrees. The slip angles increase, but the difference stays the
same, so the car is less understeered. It can be seen in figure 3.3 that the difference in vertical loads
between the frontleft and fronlright tyre has decreased and between the rearlefl and rearright tyre has
increased.
~,----~--='----~---;==;==;=~ ..'
I §:I
,~

j
" ,~

~ ,-

~.•
-

••
1(0) " "
FiQure 3.3: The vertical loads on the lYres with selup1

The same effect of decreasing the front roll stiffness can be obtained by increasing the rear roll
1
stiffness, In the standard setup the stiffness is 20 kNm- . In this setup it is increased to 30 kNm-\ the
1
front stiffness is kept at 15 kNm- , In figure 3.4a and b the results can be seen. Figure 3.4a shows that
2
the lateral acceleration has increased to 9,9 ms- , The slip angles also have increased, but the
difference has decreased. This means that the car is even less understeered.

-
---
"I------~-=-----c=~_~

"
• ."
I
I ..
o•

••f-----t-----~.----_!"
... .••f-----t-----~.----_!"
'"
Figure 3.4a: The lateral acceleration with selup2 Figure 3.4b: The slip angles with setup2

3.3 Adjusting the roll damping

Because of the special suspension system the roll damping can be adjusted very easily. The damping
character of a vehicle is especially very important in transient behaviour, so when driving through fast
on each other following corners or a slalom. In a slep steer experiment the effect and importance of
the damping behaviour can't be observed very properly. but it can be seen in figure 3.5 that the
vehicle reacts faster by increasing the front- and rear roll damping. The settling time of the vehicle is

12
smaller, which implies are better response of the vehicle. Damper constants of 1500, 3000 and 5000
1
Nsm- are tested.

,
-- b.... '''lCINom·'
b.... lOCll .....·'
b.... 50DIl NIInr'

I
~, .,
• t" ~

.,•

1101
.. "
Figure 3.5: The lateral acceleration for different roll damper settings

To make a better assessment of the damper-setup sensitivity, a test is done with a lane change
manoeuvre. In figure 3.6 the steering input can be seen .

..
..

~.~. --,-,--,-,--,---:-----i ."

Figure 3.6: The steering wheel input for /he lane change manoeuvre

In figure 3.7 on the next page the driven lines with the different damper-setups are plotted. It can be
seen that with higher damper constants the lines are tighter. This is because of the better response of
the car. Figure 3.8 shows that with lower damper constants the reaction of the rear tyres gets a bigger
delay in respect with the front tyres and therefore the car drives a wider line and is more oversteered
during change of direction. In this figure also can be seen that the tighter lines are driven with the
higher damping constants, because the slip angles are smaller.

13
 • -..... alpha", broil" 1500 Nsm'l
aI~r' b,oI" 1500 Nsm,l

I
•
1
a1phatt- broil" 1500 Nsm-
a1pha,,_ brot '" 1500 Nsm· 1
a1pha'l' brol: 3000 Nsm"
,. a1phafr' brol : 3000 N5m,l
>
1 alpha,., b,oI" 3000 Nsm'l
• alpha
fr'
b
roll
.. 3000 Nsm- 1
alpha!l' brol " 5000 Nsm,l
1
a1pha,,_ broil- 5000 Nsm-
, alpha,., brol " 5000 Nsm- 1
alpharr' b,ot .. 5000 NSm"

'i-.--,.;;---,~:c-- ••:c-----..:c----.:,• •---.:,•.------;,~

.~,

Fiqurv 3.7: The driven lines for different damper setups Fiqyre 3.81: The legend for figure 3.8b

lane change
0.25.

o2l -
I
0.151- J
0.11- I
0.05'
~
I
j o.
=-~ J I

I(
I
! .a,OS /
/

<>.1

·(1.15
\\, i

/~
J
-0.25
0 2 3 4 5 , 7
1(5)

rltW!ll 3. 8b: The slip angles of the different damper setups

14
3.4 Adjusting the track width
By adjusting the track width it is possible, like with the roll stiffness, to influence the distribution of the
vertical loads between inside- and outside wheels. When making the track width wider less weight will
be transferred to the outside wheels. The car is still understeered so first it is tried to widen the front
track width. The standard front track width of 1250mm is increased to 1300mm. As can be seen in
figure 3.6, the distribution of the vertical loads has changed. Compared to the previous setup the
lateral acceleration increases to 10,0 ms·2 .

-
-,--------="---------,=="" .
I §:I

!. ,~~ ,
~-~,-
••I
/'---- --- -
----"-L-=;--'-----~ .., ----___i..,
''''
Figure 3.6: The vertical loads on the tyres with the increased front track width

The balance of the car can also be improved by making the rear track width smaller. Decreasing the
rear track width from 1200 mm to 1150mm however didn't sort any effect.

3.5 Adjusting the static camber angles

While cornering a car also has a little roll over the tyres itself. Therefore the contact patch of these
tyres doesn't make full contact anymore with the road surface. This can be improved by giving the
wheels a static positive camber angle, so the outside tyres make full contact with the road while
making a turn and are under a small angle when driving on a straight line. Normally only the front
wheels are given a camber angle, because the rear wheels also have to apply the drive torque to the
road surface, so they better can have full contact with the road on a straight line.
In figure 3.7 the camber angles of the wheels during the test can be seen.

••
•••
- I ----
••

,~
L,-.
• • ',~
i
••
,
•••
••• " • . "
'"
Figure 3.7: The camber angles of the wheels during the test

15
The figure shows that the camber angles of the front wheels are positive in the corner. This is due to
steer camber, which is a result of the castor angle of the kingpin. So a static camber angle is not
needed in this case. Maybe it should be considered to decrease the castor angle.

3.6 Conclusion
It can be seen in the last paragraphs that by simple changing some setup parameters the performance
of the vehicle can be changed. The steady state lateral acceleration for example is improved from 9,3
ms·2 10 10,0 ms'2 . This can be even better if a good optimalisation of these parameters is made. Note
that this isn't the maximum performance of the vehicle. With larger steering wheel inputs the car will
achieve larger lateral accelerations, but for this test 65 degrees is chosen.
These steps of improving the setup is just a simple example of what is possible with the vehicle model.

16
4. Conclusion and recommendations

4.1 Simmechnics
The first things that can be concluded is that Simmechanics is a good tool to make multi-body models
like vehicle models. It is quite easy to learn and understand, because it uses the basic multi-body
dynamics equipment with some useful features added to that. Working with this in the Simulink
environment makes it very clarifying. The downside on this is that the block structure of Simulink
makes it a very lime-consuming job creating complex vehicle models, also because Simmechanics
doesn't contain special automotive features like other vehicle modeling software. For example Adams
contains a standard steering rack model as well as standard gear for the visualisation.
But the advantages of using Simulink compensate for this. Such an advantage of Simulink is the wide
range of the toolboxes. The model can be easily extended with tools to test all sorts of things. For
example the power train is completely designed separately from this vehicle model. Integration can be
done afterwards. With the extension of a power train the complete management of the vehicle can be
tested on a model level. With the possibility of Matlab to program Electronic Control Units (ECU) this
can be very useful. The wide range of the toolboxes also contributes to the fact that Matlab/Simulink is
widely used. Therefore many people are able to work with this software, which makes the model more
accessible.
The numerical performance of Simmechanics is quite good, although it depends on the quality of the
model. Yet with some multi-body dynamics knowledge this shouldn't be a problem. Only in certain
occasions on certain moments the model appears to be unstable and no solution is found. A clear
relation in these moments couldn't be found, but by changing some controller gains in the model this
could be overcome. The running time of the simulation is also quite good, but that also depends on the
performance of the computer used. The test procedure as well has its influence. In steady-stale
situations, like driving in a straight line with constant speed or cornering with a constant steering wheel
angle, the simulation runs faster then when changing from direction or speed.
The standard vehicle model software packages also have nice graphical visualisations of the vehicle.
With the Virtual Reality toolbox in Simulink this is also possible. This toolbox includes the VRealm-
builder with which virtual worlds can be created. Simulink can communicate with this program.

Tliblfl 4.1: The advantaoes and disadvantaaes of $immechanics

advantages disadvantages
easy to learn and understand time consuming interface
wide range of toolboxes no special automotive features
model based design
accessible, familiar
numerical stable
good simulations limes
possibility for nice visualisation

4.2 The vehicle

The model already appeared to be very useful in detecting initial design errors in the suspension
configuration. From the results of the simulations in chapter 3 can be seen that is also very useful in
improving the performance of the vehicle just by adjusting the setup. This is very instructive for setting
up the real vehicle. Off course it is also possible to look at the design parameters and change the
design with obtained information from the simulations.
The results of the simulations however can be doubted, because of the lack of a good model of the
tyres used. As described in paragraph 2 of chapter 2 only a few of the parameters in the MF-tyre
model are adapted to meet the race tyres. So for a common view of the handling of the vehicle this
model is very useful, but for a very accurate representation the tyre model has to be improved.

17
4.3 Recommendations
Firstly, for a smooth modeling process multi-body dynamics knowledge is needed. It is also advisable
to take some notice of the used iterations methods. This can be very useful optimising the simulation
time and when numerical problems are encountered. Next to this off course a basic understanding of
Matlab/Simulink is needed. For the visualisation it is recommended to model the parts of the vehicle in
programs that are solely made for that purpose like Unigraphics or AutoCAD. These objects can then
be imported into the VRealm-builder to form the total vehicle. In the VRealm·builder itself the parts
also can be modeled, but it is not some more difficult and gives less better results. The VRealm·
builder asks for some practice to work with, but is a good tool.
Improvements can be made on the lyre model. Because of the big influence of the tyres on the total
vehicle handling a more accurate model is needed to describe the behaviour of the real vehicle more
precisely. The model used now is a model of another tyre, with a few adaptations made for some
parameters. To obtain a better tyre model the used tyres can be tested on a flatbank lyre tester.
Other improvements on the tyre model could be the expansion of the model with a variable tyre
pressure, a temperature distribution and the wear of the tyre. These are three major issues of racing
tyres and have a big influence on the performance of a race car. Note that there is big coherence in
these three variables. An expansion of the model with a model for the power train would be another
improvement. With a model for the power train differential settings or controllers can be tested. If this
model also includes a good model of the engine and the clutch traction- and launch control or the
earlier mentioned ECU also could be developed.
Finally, a good investigation on the influence of the assumptions can be done. Firstly, to look at the
existing assumptions and expand the model to exclude one of these assumptions, for example to look
at the influence of flexibility of the bodies. On the other side some extra assumptions can be added to
decrease the complexity of the model and thereby shorten the simulation times.

18
Appendix A
All files belonging 10 the model are on the enclosed cd in the directory model, The files can be seen in
the table. To run the model alleasl Matlab 6.5 R13 with Simulink 5.0.2 and Simmechanics 2.0 is
required. The Virtual Reality toolbox has also to be installed. For some exira options of this toolbox
Matlab 7.0 R14 is recommended. It is also recommended to use a computer with enough CPU power
for a smooth running of the model. When the model is opened all data will be loaded automatically.
The license of the lyre model is valid until January 2005. If you are required, a new license can be
given by Igo Bessefink (LJ.M.Besselink@tue.nl).
In the directory report a digital version of the report and some animation files can be found.

File Function
FSmodel.mdl Standard Simulink file of vehicle model
Fsmodel2.mdl Simulink file of vehicle model with drive shafts
data FSmodel.mal File contain ina vehicle parameters structure
~tvre.tpf Files containina tvre Maaic Formula oarameters
mfread.dl1 File to read "tvre.lof for MF-tvre model
s mftvre sti.dl1 File for MF-tvre model
tu nov.lic License for MF-tvre model
or anloe.m File for calculatina roll- and oitch anale of vehicle
".wrl Files containin arts for the virtual world
·.JPG Files conlainin 1pictures of parts
lelemelrie.m File to collect and olot vehicle data
J turn.mat Steerina wheel inDut for drivino J-turn
lane change. mal Steering wheel input for drivin lane change

19
 Appendix B
The next figures show the basic characteristics of the frool- and rear lyres. In the first 2 figures the
lateral force is plotted against the slip angle for different vertical loads. In the lower 2 figures the
longitudinal force is plotted against the longitudinal slip coefficient.

~
-~
',-
',lIlXl - '.-
',_ICICI
',""50
',"'0000
~ .-
- ',""'ll!
',""00

~
~ - - - - -. --
- - ----

I • •
"
\ ---.
\- - -~

., • .02 0
-(..oj
02
•• .. .. , .. .. .. " • .. . .. . ,

~
-~

'.- ',-

--
',ollolXl F,_""
--
f-
',."50 -
F
,_,..xl
,
~-
-
-- -
'."l150
',"'0100
--
-
r
- - --

j •
•" •
"
- -; J
f-- --j
-,~ -,~

-, •• •• .. " • . . .. . , ., .. .. .. " • .. . . . ,

20
Appendix C
During the design of the race car il is tried to separate the bump· and roll movement of the vehicle in
the suspension. The advantage of this is thai no compromises in this area have 10 be made. A normal
race car is fitted with four separate spring/dampers and a front- and rear anti-roll bar. In this car the
main spring/damper is connected between both rockers. The roll movement of the car is supported by
the rolation mechanism (third rocker) and the spring and damper connected to that.

21
AppendixD

As stated in paragraph 2,3.2 the assumption to apply the drive and brake torque to the revolute joint.
which in reality can be seen as the wheel bearing, isn't totally correct. In this case the engine torque is
supported in the centre of the wheel bearing, while in reality this is on the chassis where the engine is
mounted. The same holds for the brake torque. To meet this problem, the drive shafts are also
modeled. From measurements it occurred that it didn't make observable difference for the results from
the vehicle. A disadvantage of adding the drive shafts to Ihe model is that it significantly slows the
simulation. But when a power train is added to the model and will be tested Ihe drive shafts wilt be
very useful. In the next figure the Simulink model of the driveshaft can be seen.

The JointActuator applies the drive torque to the driveshaft. Axis1a is connected with the gearbox. The
brake discs are also mounted on this axis to lower the unsprung mass and the brake torque is applied
by the BodyActuator. The first part of the axis is separated in two and divided by a prismatic joint. In
reality this is the spline in the axis which compensates the axial movement of the axis when a wheel
travels. Both the u-joints are represented by the universal joints. These joints have the two axis
perpendicular to the axis of the driveshaft as degrees of freedom. They are connected by the second
part of the drive shaft. Axis3 is the axis connected to the wheel with the weld joint and connected to
the upright with the revolute joint.
Disadvantage of a driveshaft with u-joints or flex plates is that the input torque through the axis will get
an oscillation. This is due to the changing ratio of the axis when it is under an angle. When the wheel
travel isn't very large this oscillation will not be seen back in the results form the vehicle. With larger
angles in the driveshaft the oscillation also can be seen in the output of the vehicle.

22
AppendixE

In the next table the keyboard control buttons for the VRML·viewer can be found.

Keyboard Navigation
Keyboard Navigation Function
Command

0 Resetlhe camera so that it is pointed straight ahead

8 Toggle the alpha blending on/off. If alpha blending is on, objects that have a
sel alpha value are displayed as transparent.

FIG Zoom in/out

H Toggle the headlight on/off

I Show/hide information on viewpoint name and camera position

L Toggle the lighting on/off

T Toggle the object textures on/off

.V Toggle the navigation zones on/off

W Toggle the wireframe option on/off

E" Return to default viewpoint

R,Home Relurn to current viewpoint

Page Up, Page Move between preset viewpoin1s

Down
S,X, <, > Ralale the viewpoint aboullhe origin of the scene
left Arrow, Right Move the camara left and right
Arrow

Up Arraw, Dawn Move the camera forward and backward

Arrow

N,M Tilt the camera right and left

\ Camera is bound/unbound from the viewpoint

Shift-W Set the navigation method 10 Walk

By righi-clicking the mouse on the view-screen a pop-up menu appears in which some of above
described things can be done.

It is also possible to record offline movie files from the virtual reality scene. Note that this is only
possible in Matlab 7.0 R14. How to do this;s described in the topic on the next page.

23
Manual 2-D .avi or 3-D VRML Animation Recording

This topic describes how to manually record a 2-D .avi or 3-D animation using the MATLAB interface
for a virtual world that is associated with a Simulink model. The timing of the animation file derives
from the simulation time. One second of the recorded animation time corresponds to one second of
Simulink time. You create and record the animation file by interactively starting and stopping the
recording from the MATLAB Command Window. All the steps described below can also be set in the
menus of the VRML viewer.

1. Run the Simulink model.

2. Open the Virtual Reality Toolbox viewer for that model is loaded and becomes active.
3. To work with the virtual world associated with the model from the MATLAB interface, retrieve
the virtual world handle. Use the vrwhos command. Type vrwhos.
4. If the result shows that only one vrworld object is in the workspace, assign its handle directly
=
to a variable. Type myworld vrwho.
5. If multiple virtual worlds are listed, you must select which of these virtual worlds you want to
manipulate. To select the virtual world, you can use indexing or a selection method using a
=
string comparison of virtual world descriptions. For the indexing method, type worlds vrwho
=
and after that myworJd worJds(1)
6. To have the Virtual Reality Toolbox manually record the animation, set the RecordMode
property to manual. Type set(myworJd, 'RecordMode', 'manuan
7. To retrieve the handle to the currently displayed the Virtual Reality Toolbox viewer figure, type

f=get(myworld, 'Figures')

8. Direct the Virtual Reality Toolbox to record the animation to .avi or VRML format file. Type
set(f, 'Record2D', 'on') or set(myworld, 'Record3D', 'on')
9. It is possible to turn off the navigation panel of the viewer for a clearer view. Type
set({. 'Pane/Mode', 'none') or set(myworfd, 'PanelMode', 'none')
10. Run the Simulink model. From the Simulation menu, select Normal, then click Start.
Alternatively, if you are using the Virtual Reality Toolbox default viewer, you can run the
Simulink model with one of the following from the viewer.

o From the menu bar, select the Simulation menu Start option to start or stop the
simulation.
o From the tool bar, click Start/pause/continue simulation to start the simulation.
o From the keyboard, press Ctrl+T to start the simulation.

11. As the simulation runs, start recording the animation by setting the virtual world
Recording property. Type set(myworld, 'Recording', 'on')
12. When you want to stop the recording operation, type set(myworid, 'Recording', 'off).
The Virtual Reality Toolbox stops recording the animation. The Virtual Reality Toolbox creates
the file model_anim1.avi if the RecordMode is 2-D or model_anim1.wrl if the RecordMode is 3-
D. The files are saved in the current working directory. If the simulation stops before you stop
recording, the recording operation stops and creates the animation file.
13. Stop the simulation. You can use one of the following from the viewer.

o From the menu bar, select the Simulation menu Stop option to stop the simulation.
o From the toolbar. click Stop simulation to stop the simulation.
o From the keyboard. press Ctrl+T to stop the simulation.

14. You do not need to manually stop the recording before stopping the simulation. If you do not
manually stop the recording, the recording operation does not stop and create the animation
file when the simulation stops.
15. Close and delete the objects if you do not want to continue using them.

24
AppendixF

To calculate the ro11- and pitch angle of the vehicle a Matlab-function is used in the model.
This function is written in the m-file pr_angle.m.

%Roll angle and pitch angle

function Qutput=pr_angle(R)

%calculale yaw angle with rotalionmalnx

R11=R(1);
R21=R(2);
R31 =R(3);
R12=R(4);
R22=R(5);
R32=R(6);
R13=R(7);
R23=RIB);
R33=R(9);
13w=R(10):

RW=[R 11,R12,R13:R21 ,R22,R23:R31 ,R32,R33]:

Aw=inv(Rw);

%t3w is yaw angle

t2w=asin(Aw(3,1 »:
%t3w=asi n(·Aw{2.1 )fcos(12w»;
11w=asin(.Aw(3,2)fcos(t2w»;

%e1 is world axis in drive direction, e2 is chassis axis

e 1=(cos(13w),sin(13w),O:-sin(13w),cos(13w),0;0,0,1];
e2=Aw;

%determine A-matrix and angles

A=e2"inv(e1 );
ra=asin(A(2,3));
pa=asin(A(3,1 »:
output= [( lBO/pi)' ra, (lBO/pi )·pa];

25
Appendix G
When a .mat-file is saved with telemerie.m the data is stored in 3 matrices. The vehicle-input is stored
in the matrix inp, the chassis variables are stored in the matrix chs and the lyre data is stored in the
matrix tys. In the hereunder-plotted tables can be seen which row in every matrix contains which data.

ino tvs
1 time is\ 1 time s)
2 steeri~wheel an;:;le j~ut d~ 2 slio anale frontleft rad
3 real steerino wheel an Ie deal 3 slin anole frontrioht (rad)
4 steerinn wheel annie error den' 4 sljn annie rearleft rad
5 steerinnwheeJ !ornue fNm 5 slio an Ie rearriaht (rad)
6 brake moment frontleftTNm 6 slio coefficient frontleft -
7 brake moment frontrinhl Nm\ 7 slin coefficient frontrinht '- \
B brake moment rearlett tNm\ B slin coefficient rearlefl -
9 brake moment rearrinhl Nril'- 9 slio coefficient rearri ht -I
10 throttle inout (-) 10 camber anole frontleft rad
11 camber annie frontrinht rad
chs 12 camber annIe rearlefl rad\
1 time is\ 13 camber anale rearri ht rad
2 x-:;::;Osition of CG chassis m 14 Ion itudinal force fronlleft N
3 I v-nosition of CG chassis m 15 lonnitudinal force frontrinht N\
4 z-nosition of CG-chassis m 16 Ion itudinal force rearleft (Nl
17 Ion itudinal force rearri ht N
5 lonnjtudinal sneed ms· 1 \
6 lonaitudinal acceleration(m$2f 18 lateral force fronUefl N
7 I vaw rate {rads' 19 lateral force frontrinht N
B roll annie den' 20 lateral force rearleft (N
9 ~tcha~e d~
21 lateral force rearriaht N
10 lateral acceleration (ms· 22 vertical force frontleft N
23 vertical force frontriaht (N)
24 vertical force rearlefl N
25 vertical force rearrioht N
26 overlurninn moment front/eft Nm
27 overlurnin moment frontri ht Nm)
28 overturnino moment rear/eft fNm
29 overturninn moment rearrinht Nm
30 M Nm front/eft INm
31 M Nm frontriaht fNm
32 M Nm rear/eft Nm)
33 M Nm rearrinht INm
34 self-alianina moment fronlleft (Nm)
35 self-alionina moment fronlriaht (Nm
36 self-alinninn moment rearlefl (Nm
37 self-alinninn moment rearrinht {Nm

26
List of symbols

I time
V lonaitudinal saeed
a Ion itudinal acceleration
r yaw rate
a lateral acceleration
a slio annIe
K Jon itudinal slip coefficient
v camber anale
F lonaitudinal force
F lateral force
F vertical force
M, overturnina moment
M moment round v-axis
M self-ali nin momenl

27
References

[1] www.mathworks.nVsupport

[2] Igo Besselink, Vehicle Dynamics lecture sheets, Eindhoven, 2003

[3] The Malhworks, SimMechanics User's Guide, Natick, 2002

[4] TNO Automotive, Tyre models Users manual, Delft, 2002

{5] Nathan van de Wouw, Multibody Dynamics Lecture Notes, Eindhoven, 2003

[6] Bernard Kolman, Introductory Linear Algebra with application, Prentice hall, Upper Saddle
River, New Jersey, 1997

{7] Hans B. Pacejka, Tyre and vehicle dynamics, Butterworth·Heinemann, Oxford, 2002

[8] Wouler Berkhoul, Graduation assignment, Eindhoven, 2004

[9I William F. Milliken - Douglas L. Milliken, - L. Daniel Metz, Race car vehicle dynamics,
Warrendale: SAE International, 1995-1996

28