Department of Mechanical Engineering

Dynamics & Control Group

Design of a traction
control system for the
four-wheel driven race car
IM01
Bachelor Final Project

Steef de Bakker
(0841765)
DC 2017.027

Supervisor:
Dr. Ir. T.P.J. van der Sande

Eindhoven, January 2017

 Abstract

Student team Automotive Technology InMotion is working on a high-powered electric four-wheel

driven race car, the IM01. A common problem for race cars is that the torque they can deliver is
higher than the torque that the tyres can transfer to the road surface. This results in problems
when accelerating, for example, excessive wheel slip and a reduction of longitudinal acceleration.
Therefore, in this report, a traction control system for the IM01 is designed.

A lot of research has already been done in the field of traction control and different concepts
can be found in literature, this report looks into multiple traction control concepts. The advant-
ages and disadvantages of these different concepts are reviewed and one concept is used to design
the traction control system for the IM01. Also a way to determine the longitudinal vehicle velocity
is found, this is done with a Correvit optical sensor, made by Kistler.

A quarter-car model is made to do simulations and to test the traction control system. Next
to this quarter-car model, a tyre model is included, based on the Magic Formula. This tyre model
is used to simulate tyre behaviour and to calculate tyre forces. The parameters for the Magic For-
mula model are based on the parameters of a Michelin LMP1 tyre, which are used in the concept
design of the IM01. With this model, simulations have been executed to determine the maximum
longitudinal tyre force, the maximum transmissible drive torque and the theoretical maximum
acceleration.

A traction control system is designed that consists of a feed-back part, a feed-forward part and
a peak friction estimator. In the feed-back part, the measured angular wheel speed is compared
to an optimal angular wheel speed. This optimal angular wheel speed is derived from the optimal
slip ratio which can be determined with the help of the Magic Formula tyre model. The feed-
forward part uses the estimated longitudinal friction coefficient from the peak friction estimator,
the gravitational force and the wheel radius to calculate a feed-forward torque. The peak friction
estimator uses the vehicle mass, the gravitational force and the time derivative of the vehicle speed
to estimate the longitudinal friction coefficient, this is then fed into the feed-forward controller.

Simulations have been executed for different conditions: without the traction control system,
with only the feed-back part, with only the feed-forward part and with the full traction control
system. The applied drive torques, the estimates longitudinal friction coefficients, the wheel and
vehicle velocities and accelerations are then plotted for all of these situations. Also, the results for
the peak friction estimator are plotted for a sudden decrease in the longitudinal friction coefficient.

A traction control system that does not use the wheel speed of a non-driven wheel as an input
and is applicable to a four-wheel driven car has been designed. This traction control system is
designed and validated through simulations and it is shown that it results in better performances
than when no traction control system is used. A decrease of 40% in the 0 − 100km/h time is real-
ised with the help of the traction control system. The peak friction estimator that is developed is
able to estimate the longitudinal friction coefficient within approximately 1 second.
Contents

Contents ii

1 Introduction 1
1.1 Problem definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Method of investigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2 Literature review 4
2.1 Sign convention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 Magic Formula tyre model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.3 Traction control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3.1 Model following control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3.2 Maximum transmissible torque estimation . . . . . . . . . . . . . . . . . . . 7
2.3.3 Slip ratio control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3.4 Determining Vx . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

3 Modeling of the wheel and tyre 10

3.1 Wheel dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.2 Tyre model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.2.1 Longitudinal tyre forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.3 Theoretical calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.4 Model simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

4 Traction control design 16

4.1 Feed-back . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4.2 Feed-forward . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4.3 Peak friction estimator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

5 Simulation results 19

6 Conclusions & recommendations 27

6.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
6.2 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

Bibliography 28

ii Design of a traction control system for the four-wheel driven race car IM01
Chapter 1

Introduction

Automotive Technology InMotion is a student team of Eindhoven University of Technology. In-

Motion’s goal is to participate in the 24h of Le Mans, the biggest motorsport event in the world.
The goal is to do this the so called Garage 56 class, this is an innovation class that has no imposed
technical limitations except on safety. The car that is going to be built to compete in this race is
the IM01, an electric driven four-wheel drive race car. In order to be able to achieve this ambitious
goal, first a two-wheel driven electrical race car is built, the IM/e. This two-wheel driven race car
will serve multiple purposes, such as testing new technologies, learning how to build a race car as
a team and creating leverage to find new sponsors.

Figure 1.1: Render of the conceptual design of the IM01

Design of a traction control system for the four-wheel driven race car IM01 1
CHAPTER 1. INTRODUCTION

1.1 Problem definition

Lately, a lot of research and development into electric vehicles (EV’s) has been done. This is
mainly to find a more environmentally friendly alternative for internal combustion engine vehicles
(ICV’s). Apart from the environmental advantages of EV’s, the electric motors also bring another
big advantage over internal combustion engines, the torque that is generated by the electric mo-
tors can be controlled much quicker and more precisely than that of internal combustion engines.
Because this torque can be controlled this quickly and precisely, the stability of a vehicle can be
improved greatly [2]. This is because a common problem with race cars is that the torque they can
deliver is higher than the amount of torque the tyres can transfer to the road. When the applied
torque to a tyre is higher than the maximum torque it can transfer, the wheel will start slipping,
which will reduce in a loss of grip and as a result will lead to slower accelerations. In order to
limit this wheel slip and to maximize the acceleration, a traction control system is often used, this
is a controller which anticipates on how much torque can be applied and cuts down the applied
torque when wheel slip occurs. Because both the IM/e as well as the IM01 are race cars and are
built with the aim to be as fast as possible, a traction control system is needed. The objective of
this report is to design such system.

Since the IM01 is driven a high powered race car, it can be assumed that traction control
is required. A system is needed which can reduce the applied torque to the wheel when it slips
too much. Also tyre data is used to make a first estimate of the amount of torque that can be
transferred to the road surface. Because the IM01 is a four-wheel driven car, the traction control
system has to be designed in a way it works for all four of those wheels. A problem that occurs is
that the absolute vehicle speed has to be known in order for the traction control system to work.
A way has to be found to determine this absolute vehicle speed without using the measurement
of the wheel speed of the non-driven wheels. Another problem that occurs is that no road surface
is the same, so the traction control system should be able to estimate the amount of grip that is
available at any moment, this will be done by peak friction estimator.

The purpose of this report is to design a traction control system that does not use the wheel
speed of a non-driven wheel as an input and is applicable to a four-wheel driven car.

1.2 Method of investigation

The traction control system is designed and tested with the help of a simplified model of a tyre
and a quarter of a vehicle. In this report, only the traction control for one of the four wheels of
a four-wheel driven car is considered. This is done because a race car usually has an ideal weight
distribution between the front an the rear (50% front - 50% rear). Also, because generally, a race
car has a very low centre of gravity, there is very little weight shift both longitudinal as lateral.

A few assumptions are made in order to get a simplified model of the reality, these are:

• A constant vertical force.

• Aerodynamic and rolling resistances are neglected.

• Tyre compression is neglected, so rw is constant.

• Only pure longitudinal slip is considered.

1.3 Outline
This report is built up out of six chapters. In chapter 2 a review of the existing literature about
the subject will be given. It will explain the sign convention that is used, give an introduction

2 Design of a traction control system for the four-wheel driven race car IM01
 CHAPTER 1. INTRODUCTION

into the Magic Formula tyre model and will explain different traction control methods found in
literature.

In chapter 3 the all the parts of the model that is used to design the traction control system
is explained. First, the basic, quarter-car model is explained, this model describes the dynamics
of the car and wheel. Second, the tyre model is explained, this is a model that is based on the
Magic Formula tyre model and describes the characteristics of the tyre. Third, basic theoretical
calculations are given about performances that should be possible and fourth, simulation results
of the whole model are shown.

In chapter 4, the design of the traction controller is described and explained. All parts of the
controller, namely: feed-back, feed-forward and the peak friction estimator are explained. Also, a
block diagram of the whole traction control system is shown.

After that, chapter 5 shows the simulations of the traction control system. It shows the per-
formance of the separate parts of the traction control system, it also shows what performance
gains are achieved by the traction control system.

Finally, in chapter 6 conclusions will be drawn based on the results of the simulations. It will
also give some recommendations for the implementation of the system and for future research.

Design of a traction control system for the four-wheel driven race car IM01 3
Chapter 2

Literature review

The tyres of a car are the parts that transmit all the traction forces from the road surface to the
car. Because of this, understanding tyre behaviour is necessary to be able to create a traction
control system. A lot of research has been done in the field of tyre behaviour and a lot of different
tyre modeling strategies are used, such as the brush tyre model, the Magic Formula tyre model and
Van Rijks’ tyre model. To be able to work with these tyre modeling strategies, a sign convention
has to be introduced.

2.1 Sign convention

The sign convention that is used in this report is the sign convention by Besselink [1]. Figure 2.1
shows the sign convention with all the forces and moments acting on the contact point.

Figure 2.1: Tyre sign convention [1]

As can be seen in figure 2.1, the tyre axis-system is defined as follows:

• z: normal to road surface and upwards

• x: pointing forward, in the driving direction, through the wheel plane of symmetry

• plane x, y: parallel to road surface

4 Design of a traction control system for the four-wheel driven race car IM01
 CHAPTER 2. LITERATURE REVIEW

The different forces acting on the tyre are defined as follows:

• Fx : longitudinal force (e.g. driving and braking)

• Fy : lateral force (e.g. steering), not considered here
• Fz : vertical or normal force (e.g. tyre compression)

The moments acting on the tyre are defined as follows:

• Mz : self aligning moment

Two different kinds of slip are defined, namely: lateral tyre slip α and longitudinal tyre slip κ.
Lateral tyre slip α, called side slip, is the ratio between the forward velocity, Vx , and the lateral
velocity, Vy , at the contact point [7].
 
−Vsy
α = arctan (2.1)
Vx

The longitudinal wheel slip (κ) is the ratio between the speed of the contact point and the forward
speed and is given by the following equation [7].

Vx − rw ω
κ=− (2.2)
Vx
In which rw is the effective wheel radius and ω is the angular wheel speed. From this equation it
can be derived that κ equals -1 when the wheels lock. When there is no wheel slip at all, so the
wheel rolls freely, κ is 0.

As stated in section 1.2, only pure longitudinal tyre slip is considered, so only κ is used in this
report.

2.2 Magic Formula tyre model

An often used way to model tyre behaviour is the Magic Formula tyre model, this is a semi-
empirical tyre behaviour model which is used to model tyre force and moment characteristics in
a steady-state situation. Since the first development of the model, multiple versions of the model
have been used. The version used here and is given below is that by Besselink [1]. The Magic
Formula tyre model is probably the most popular tyre model for vehicle handling simulations
worldwide. The basic form of the Magic Formula tyre model is given by

F = D sin(C arctan((1 − E)Bx + E arctan(Bx))). (2.3)

The Magic Formula tyre model for lateral forces is a function of both the vertical force Fz as the
lateral slip α as given below.
Fy = M F (α, Fz ) (2.4)
The Magic Formula tyre model for longitudinal forces is a function of both the vertical force Fz
as the longitudinal slip κ as given below.

Fx = M F (κ, Fz ) (2.5)

As stated in section 1.2, only pure longitudinal tyre slip is considered, so only equation 2.5 is
used in this report.

The model used in this report is fully based on the basic form of the Magic Formula tyre model
as given in equation 2.3. A further explanation of the model will be given in section 3.2.

Design of a traction control system for the four-wheel driven race car IM01 5
CHAPTER 2. LITERATURE REVIEW

2.3 Traction control

Traction control makes sure that the tyres of a car do not spin when accelerating, which is done by
controlling the torque applied by the motor of the vehicle. This is necessary when the torque that
is delivered to the wheel is higher than the maximum torque it can transfer to the road. Because
then, the wheels will start slipping and the slip ratio gets too high, the transmissible tyre force will
decline and so will the maximum transmissible drive torque. Because electric motors can deliver
high torques at low speed and the IM01 is a four-wheel driven car that uses four electric motors,
one for each wheel, the need for traction control in the IM01 is very high, this will be further
discussed in section 3.4.

Figure 2.2 shows the longitudinal tyre force Fx as a function of the longitudinal slip ratio κ for
a typical tyre. The friction coefficient µ is defined as
Fi
µi = (2.6)
Fz
with i equal to x in the case of longitudinal slip and i equal to y in the case of lateral slip. As can be
seen in figure 2.2, there is a slip ratio at which longitudinal tyre force, Fx , is at it’s maximum, and
so is the longitudinal friction coefficient, µx , after that maximum, the friction coefficient decreases,
and so does the tyre force. The traction control system attempts to make sure the tyre always
operates as close to this maximum as possible to maximize the longitudinal acceleration, ax .

Longitudinal friction F as function of slip κ

x
x
Longitudinal tyre force F

0
Longitudinal slip ratio κ

Figure 2.2: Longitudinal tyre force Fx function of longitudinal slip ratio κ

A lot of research has already been done in different approaches of traction control systems.
Three of them will be further explained in the following sections, namely: model following control,
maximum transmissible torque estimation and slip ratio control. The advantages and disadvant-
ages of all these approaches will be explained and an approach that is most suitable for the purpose
of this report will be chosen.

2.3.1 Model following control

Model following control is an approach to control the slip ratio, it’s a type of traction control
that doesn’t use any tyre data. Because of this, it is a relative simple type of traction control.
Model following control will control the torque such that there will be no slip at all, κ = 0. Model
following control uses parameters, most of which are constant. It uses the inertia of the wheel, the
vehicle mass and the tyre radius to calculate a reference inertia, Jmodel [2]. Jmodel is then given
by
Jmodel = Jwheel + m · re2 (2.7)

6 Design of a traction control system for the four-wheel driven race car IM01
 CHAPTER 2. LITERATURE REVIEW

What model following control then does is comparing the reference inertia Jmodel to the vehicles
inertia which is defined as:
J = Jwheel + m · re2 · (1 − κ) (2.8)
In equation 2.8, κ is calculated by
rw ω − Vx
κ= (2.9)
rw ω
The difference between Jmodel and J is then fed back into a controller as an error which is
given in equation 2.10.
e = Jmodel − J (2.10)
The controller then uses this error to regulate the drive torque that is delivered to the wheel. The
result is that this controller reduces the slip ratio to zero. As can be seen in figure 2.2, this is not
the optimal slip ratio because the longitudinal friction is not at its maximum at that point. This
is why model following control does not result in maximum longitudinal acceleration, ax . Also
model following control is a very rough approach, if we want to regulate the slip ratio better, a
more precise approach is needed, that is why model following control is not used. A block diagram
of a model following controller is given in figure 2.3.

Figure 2.3: Block diagram of a model following controller [6]

2.3.2 Maximum transmissible torque estimation

Maximum Transmissible Torque Estimation (MTTE) [5] [8], is a traction control approach that
does not use the actual vehicle speed Vx as an input. This is ideal in the case of an four-wheel
driven car, as is the IM01, because no additional sensors are needed to determine Vx . On top of
that, MTTE doesn’t require any tyre data. MTTE uses the applied torque to a wheel and the
angular acceleration of that wheel to make an estimation of the maximum amount of torque that
can be transmitted to the road.

Design of a traction control system for the four-wheel driven race car IM01 7
CHAPTER 2. LITERATURE REVIEW

In MTTE, the friction force of a tyre is given by

T Jwheel · V̇wheel
Fd = − 2
(2.11)
rw rw
The maximum transmissible torque is then given by
 
Jwheel
Tmax = 2
· rw · Fd (2.12)
α · m · rw

In which α is the relaxation factor of the tyre, which approximates the ratio of the acceleration of
the wheel and the chassis, given by

V̇ (Fd − Fdrag ) · Jwheel

α= = (2.13)
V̇wheel (Tmax − rw · Fd ) · rw · m

The most important parameter to regulate the estimation of the maximum transmissible torque
is α. The value of α has to be chosen such that the performance is optimal. Parameter α can
be used to balance the anti-slip and acceleration performances. When slip occurs, the wheel will
accelerate too fast, Fd will reduce and then Tmax will reduce as well.

2.3.3 Slip ratio control

Slip ratio control is a traction control method which regulates the angular wheel speed such that
the actual slip ratio is as close to the optimal slip ratio as possible. It does this by comparing
the actual measured wheel speed to an optimal angular wheel speed, ωopt [2][6][3]. This optimal
angular wheel speed is related to the optimal slip ratio, κopt as given in equation: 2.14. This
optimal slip ratio has to be found by using the Magic Formula tyre model as described in section
3.3.
Vx − rw ωopt
κopt = − (2.14)
Vx
From this optimal slip ratio, the optimal angular wheel speed can be derived with equation 2.15.

Vx (κopt + 1)
ωopt = (2.15)
rw

In equation 2.15, Vx is the longitudinal velocity of the vehicle.

Figure 2.4: Block diagram of a slip ratio controller [6]

This slip ratio control method leads to the best results, as the wheel speed can be controlled
best [6] [3]. That is why this method is used to design the traction control system. The design of
the traction control system will be further explained in chapter 4.

8 Design of a traction control system for the four-wheel driven race car IM01
 CHAPTER 2. LITERATURE REVIEW

2.3.4 Determining Vx
As the longitudinal speed of the vehicle, Vx needs to be determined to determine ωopt , is has to
measured in a good way. Usually, on two-wheel driven cars, Vx is calculated from the angular
wheel speed of the non-driven wheels, e.g. the front wheels in the case of a rear wheel-driven
car. Because the IM01 is a four-wheel driven car, this is not possible because there simply are no
non-driven wheels. That’s why another way has to be found to determine Vx . Below, suggestions
are given for determining Vx .

GPS
A way to determine Vx is by using GPS. GPS can determine the location of a vehicle and by that,
the velocity in any direction of a vehicle. So determining Vx through GPS is possible, however,
GPS has some downsides. Because we want to determine Vx highly accurate with a high sampling
frequency, GPS is not the ideal way. Also, GPS requires an unobstructed line of sight to four or
more GPS satellites, which can’t guaranteed at all time.

Pitot tube
A Pitot tube is a pressure measurement instrument, used to measure fluid flow velocity. It is
widely used to determine the airspeed of an aircraft or the water speed of a boat, but is also used
on high-end race cars. A downside of a Pitot tube is that it measures the airspeed of a car relative
to the wind speed, so not its actual speed.

Correvit sensor
A Correvit optical sensor, by Kistler [4], is a sensor that can measure longitudinal and lateral
vehicle dynamics in a slip-free way. Although this sensor is quite costly, it has a high level of
accuracy, that’s why it is used in for instance, F1, LMP1 and LMP2 race cars.

Conclusion
Because of the disadvantages of GPS and Pitot tubes, because the Correvit optical sensor has
already proven itself in several professional racing series, and because it is a slip-free way of meas-
uring Vx , it is recommended to use a Correvit sensor to measure and determine Vx in the IM01.

Design of a traction control system for the four-wheel driven race car IM01 9
Chapter 3

Modeling of the wheel and tyre

As explained in section 1.2, only a single wheel of the vehicle is evaluated in this model. As stated
in chapter 2, all the traction forces from the road surface to the car are transmitted through the
tyres, therefor tyre characteristics play a major role in the modeling of a traction control system.
In this chapter, the model that is used to simulate the car and tyre behaviour will be explained.
This model consists of two different parts, namely: the wheel dynamics model and the tyre forces
model. First, in section 3.1, the dynamics of one wheel and a quarter of the car will be explained.
Second, in section 3.2, the tyre model that is used to simulate the tyre forces is explained. Third,
in section 3.3, some basic, theoretical calculations based on both the wheel dynamics model as on
the tyre model are given. Last, in section 3.4, simulation results of the model are given.

3.1 Wheel dynamics

The dynamics of one wheel of the car will be modelled for the application of a drive torque applied
by the motor to the wheel. Because only a quarter of the car is considered, a quarter of the total
mass of the car is connected to this wheel. Figure 3.1 shows the sign convention that will be used
in this report.

In figure, 3.1 ω is the angular wheel speed of the considered wheel in [rad/s]. The longitudinal
velocity of the center of the wheel is represented as Vx in [m/s] in here. Td represents the drive
torque in [N m] applied to the wheel. Fx is the longitudinal tyre force in [N ] acting on the tyre.
Fz is the vertical tyre force in [N ] acting on the tyre. Iw is the inertia of the wheel in [kgm2 ].
The wheel radius is given by rw in [m], and m represents the mass of a quarter of the vehicle in [kg].

The dynamics of the angular velocity of the wheel, ω, and the mass of a quarter of the vehicle,
m, can be described with the following equations of motion

Iw ω̇ = Td − Fx rw (3.1)

Fx = max (3.2)
Table 3.1 shows the values of the parameters used to describe the wheel dynamics in equations
3.1 and 3.2 which represent the parameters as they are set for the concept design of the IM01.

m 300kg
rw 0.356m
Iw 2.7kgm2

Table 3.1: Parameters for the wheel dynamics used in this model

10 Design of a traction control system for the four-wheel driven race car IM01
 CHAPTER 3. MODELING OF THE WHEEL AND TYRE

Figure 3.1: Dynamic wheel model

3.2 Tyre model

As stated in section 2.2, the tyre model that is used in this report is based on the Magic Formula
tyre model, the version that is used here is the one given by Besselink [1].

3.2.1 Longitudinal tyre forces

For convenience reasons, only longitudinal tyre forces are taken into account in this report, as
stated in section 1.2. The Magic Formula calculates the longitudinal tyre force Fx as a func-
tion of Fz and κ, the vertical force and the longitudinal slip respectively. The value for Fz
is based on a quarter of the mass of the vehicle as stated in section 3.1, it is then given by:
Fz = 300 · 9.81 = 2943N .

Fx = M F (κ, Fz ) (3.3)
The Magic Formula is then given by:

Fx = Dx sin[Cx arctan{(1 − Ex )Bx κ + Ex arctan(Bx κ)}] (3.4)

in which
Dx = Fz µx (3.5)
µx = pDx1 + PDx2 dfz (3.6)
Cx = pCx1 (3.7)
Ex = pEx1 + pEx2 dfz (3.8)

Design of a traction control system for the four-wheel driven race car IM01 11
CHAPTER 3. MODELING OF THE WHEEL AND TYRE

CF κ = Kx = Bx Cx Dx = Fz (pKx1 + pKx2 dfz ) (3.9)

Fz − Fz,nom
dfz = (3.10)
Fz,nom
Bx can then be calculated from Cx , Dx and Kx by:
Kx
Bx = (3.11)
Cx Dx
κ as described by [7] is given in equation 2.2.

Vx − rw ω
κ=− (3.12)
max(|Vx |, Vmin )

In equation 3.12 a minimum velocity Vmin is introduced in the definition of κ to avoid a division
by zero when the vehicle is standing still, so when Vx = 0 [1].

The model coefficients are usually determined by numerical optimisation, they are compared
to measurements and are then tuned to minimise the difference between model and measurements.
Table 3.2 shows the fitting parameters as used in this model which are based on the parameters
of a type of Michelin LMP1 tyres which is used in the concept design of the IM01.

Bx 9.5528N
Cx 1.6642
Dx 5.7637 · 103 N
Ex 0.7409
Fz,nom 4414N
µx 1.9584

Table 3.2: Fitting parameters of the Magic Formula tyre model

3.3 Theoretical calculations

In this section, the optimal slip ratio, theoretical calculations of the maximum transmissible tyre
force, Fx,max , and drive torque, Td,max , of the tyre and the maximum theoretical acceleration,
amax , will be determined.

The optimal slip ratio κopt is found by differentiating equation 3.4 as given in equation 3.13
∂Fx
and solving it for = 0. In figure 3.2, κ is plotted against Fx . The value of κ corresponding
∂κ
to the maximum value of Fx is then found, which is κopt .
 
Bx Ex
Cx Dx cos(Cx arctan(Ex arctan(Bx κ) − Bx κ(Ex − 1))) · Bx (Ex − 1) − 2 2
∂Fx Bx κ + 1
=−
∂κ (Ex arctan(Bx κ) − Bx κ(Ex − 1))2 + 1
(3.13)
The value for κopt in the case of pure longitudinal slip is then κopt = 0.2269.

The maximum transmissible tyre force, Fx,max , can then be found by simply filling in κopt in
equation 3.4, the maximum transmissible tyre force that is then found is: Fx,max = 5763.7N .

The maximum transmissible drive torque is then found by multiplying the maximum trans-
missible tyre force, Fx,max , with the wheel radius, rw .

Td,max = Fx,max · rw = 5763.7 · 0.356 = 2051.9N m (3.14)

12 Design of a traction control system for the four-wheel driven race car IM01
 CHAPTER 3. MODELING OF THE WHEEL AND TYRE

Longitudinal friction Fx as function of slip κ

6000

4000

2000
Fx [N]

-2000

-4000

-6000
-0.6 -0.4 -0.2 0 0.2 0.4 0.6
κ

Figure 3.2: Longitudinal tyre force Fx as function of the longitudinal slip ratio κ

By using equations 3.1, 3.2 and the value for Fx,max that is found before, the theoretical
maximum acceleration, amax , can be calculated with equation 3.15.
2
Fx,max rw
amax = 2
= 17.9385m/s2 (3.15)
mrw + Iw

3.4 Model simulations

Simulations have been done to show the need of a traction control system in a high-powered race
car like the IM01. Three simulations have been done for different applied drive torques. The first
one is with a drive torque of 1000N m, so a lot lower than Td,max as determined in section 3.3, the
second is with a drive torque of Td,max = 2051.9N m, the third is with a drive torque of 3000N m,
which is a lot higher than Td,max . These simulations have been done to show that with a drive
torque below Td,max , performance is not at its maximum and with a drive torque above Td,max ,
the wheels will start slipping because then ω · rw rises much faster than Vx . The results of these
simulations can be seen in figure 3.3.

Second, the longitudinal accelerations for the same drive torques as before are plotted in figure
3.4. It shows that if the applied drive torque gets too high, the longitudinal acceleration ax is
lower than when the applied drive torque is set at Td,max .

Third, the angular wheel accelerations, ω̇, corresponding to the same drive torques as before
are plotted in figure 3.4. As can be derived from figure 3.3, the angular velocity for a drive torque
of 3000N m is much higher than that for a drive torque of Td,max = 2051.9N m.

Design of a traction control system for the four-wheel driven race car IM01 13
CHAPTER 3. MODELING OF THE WHEEL AND TYRE

Wheel and vehicle velocities at different drive torques

80
V 1000 Nm
x
ω*rw 1000 Nm
70 V 2051.9 Nm
x
ω*rw 2051.9 Nm
V 3000 Nm
x
60 ω*r 3000 Nm
w

50
Vx [m/s]

40

30

20

10

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time [s]

Figure 3.3: Vehicle and tyre velocities at different drive torques

Looking at these figures 3.3 and 3.4, the need for a traction control system in the case of the
IM01 becomes very clear. When the applied drive torque is too high the wheels will start slipping
as can be seen in figure 3.3, the longitudinal acceleration will drop as can be seen in figure 3.4.
Figure 3.4 shows that the angular acceleration at an applied drive torque of 3000N m way too high
what also shows that excessive wheel slip occurs.

14 Design of a traction control system for the four-wheel driven race car IM01
 CHAPTER 3. MODELING OF THE WHEEL AND TYRE

Longitudinal accelerations
20
1000 Nm
18 2051.9 Nm
3000 Nm
16

14

12
a x [m/s 2]

10

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time [s]

Angular accelerations
1200
1000 Nm
2051.9 Nm
1000 3000 Nm

800
ω̇[rad/s2 ]

600

400

200

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time [s]

Figure 3.4: Longitudinal and angular accelerations at different drive torques

Design of a traction control system for the four-wheel driven race car IM01 15
Chapter 4

Traction control design

The traction control system is designed to regulate the torque delivered to the wheels in a way
such that traction is maximized. This controller consists of three parts, namely a feed-back loop,
a feed-forward loop and the peak friction estimator. These will be discussed separately in the
following sections. The concept of the controller system is illustrated in figure 4.1.

The block diagram as shown in figure 4.1 consists of several block who all represent a part of
the traction control system. The peak friction estimator uses the wheel speed signal, u, to estimate
the value of µx , this value of µx is then fed into the feed-forward controller together with vertical
tyre force Fz . The tyre model uses both the wheel speed signal, u, as the vertical tyre force Fz to
determine an optimal angular wheel speed, ωopt , from this optimal angular wheel speed, an error
signal is generated according to equation 4.1.

ωerror = ωopt − ωmeasured (4.1)

This error signal is fed into the feed-back controller. The two signals from both the feed-forward
as the feed-back controllers are then added into a controller torque. Because the driver should
always have full control over the car, only the minimum of the controller torque and the requested
torque by the driver is used to make up the torque request that is put into the motor. This
torque request is then applied to the wheel which generates a wheel speed signal and a measured
angular wheel speed, ωmeasured out of it, these are fed back into the peak friction estimator and
into equation 4.1 respectively.

Results of all the separate parts of the traction control system and of the complete traction
control system will be shown in chapter 5.

16 Design of a traction control system for the four-wheel driven race car IM01
 CHAPTER 4. TRACTION CONTROL DESIGN

Figure 4.1: Block diagram of the traction control system based on [6].

4.1 Feed-back
The feed-back loop is constructed as explained in section 2.3.3. The optimal angular wheel speed,
ωopt , serves as a set-point signal for the feed-back loop. That optimal angular wheel speed is
compared with the measured angular wheel speed ωmeasured and this is fed back into the feed-
back controller as an error signal, ωerror , see equation 4.1. This error signal is then controlled
by a PI-controller to minimize it as quickly as possible in order to get the best performance possible.

4.2 Feed-forward
The feed-forward loop is constructed as can be seen in figure 4.1. This part of the controller
uses the estimated longitudinal friction coefficient µx of the tyre estimated by the peak friction
estimator, as explained in section 4.3. The vertical tyre force is assumed to be constant as given
in equation 4.2:
Fz = 300 · 9.81 = 2943N (4.2)
The feed-forward torque is then calculated by:

TF eed−f orward = µx · Fz · rw (4.3)

The feed-forward torque is used to give a torque which corresponds to the expected maximum
value of the longitudinal tyre force, that is the force at which the tyre is expected to start slipping.

Design of a traction control system for the four-wheel driven race car IM01 17
CHAPTER 4. TRACTION CONTROL DESIGN

4.3 Peak friction estimator

The peak friction estimator is based on the method introduced by Y. Hori et al. [2] that is used to
quickly estimate the value of µx . This estimator is necessary because track conditions are never
the same, so the amount of grip that is available won’t always be the same. Because the vehicle
speed Vx can be measured directly, µx can be estimated by equation 4.4 [2]:

M dVx
µx = (4.4)
Fz dt
In which M is the vehicle mass and Fz is given by equation 4.2. Results of the peak friction
estimator will be shown in chapter 5.

18 Design of a traction control system for the four-wheel driven race car IM01
Chapter 5

Simulation results

In this chapter, the simulations will be shown. The model as given in chapter 3 is used to simulate
a situation in which the driver gives full throttle after 1 second. The drive torque that is requested
is equal to 3000N m, this drive torque represents the expected maximum drive torque for the IM01.
The drive torque requested by the driver is the same for all the simulations and is shown in figure
5.1.

Requested drive torque

3500

3000
Requested drive torque [Nm]

2500

2000

1500

1000

500

0
0 1 2 3 4 5 6 7 8 9 10
Time [s]

Figure 5.1: The requested drive torque for all simulations

First, the model without any form of traction control is simulated and the torque as plotted
in figure 5.1 is requested. This is the situation in which the driver would requested the maximum
amount of torque possible from the motors without letting go of the throttle. The results of this
simulation are shown in figure 5.2.

Design of a traction control system for the four-wheel driven race car IM01 19
CHAPTER 5. SIMULATION RESULTS

Applied drive torque without TC

4000
Drive torque [Nm]

3000

2000

1000

0
0 1 2 3 4 5 6 7 8 9 10
Time [s]
µx
4

3
µx [-]

0
0 1 2 3 4 5 6 7 8 9 10
Time [s]
Wheel and vehicle velocities without TC
2500
2000 ω*rw
V [m/s]

1500 Vx
1000
x

500
0
0 1 2 3 4 5 6 7 8 9 10
Time [s]
Wheel and vehicle accelerations without TC
300

dω/dt*r w
a x [m/s ]

200
2

ax

100

0
0 1 2 3 4 5 6 7 8 9 10
Time [s]

Figure 5.2: Simulation results without traction control

As can be seen in figure 5.2, the wheel velocity rises much faster than the vehicle velocity
and the wheel acceleration is much higher than the vehicle acceleration, so the wheels will start
spinning almost immediately after the torque request is drawn from the electric motors.

20 Design of a traction control system for the four-wheel driven race car IM01
 CHAPTER 5. SIMULATION RESULTS

Second, the model with only the feed-forward loop implemented is simulated with, again the
torque request as plotted in figure 5.1. So, in this simulation, no feed-back controller is used. The
drive torque is only controlled by the TF eed−f orward signal as explained and calculated in section
4.2. The results of this simulation are shown in figure 5.3.

Applied drive torque with feed-forward loop only

2500
Drive torque [Nm]

2000
1500
1000
500
0
0 1 2 3 4 5 6 7 8 9 10
Time [s]
Estimated µ x
2

1.5
µ [-]

1
x

0.5

0
0 1 2 3 4 5 6 7 8 9 10
Time [s]
Wheel and vehicle velocities with feed-forward loop only
200

150 ω*rw
Vx [m/s]

Vx
100

50

0
0 1 2 3 4 5 6 7 8 9 10
Time [s]
Wheel and vehicle accelerations with feed-forward loop only
30

dω/dt*r w
a [m/s 2 ]

20
ax
x

10

0
0 1 2 3 4 5 6 7 8 9 10
Time [s]

Figure 5.3: Simulation results with the feed-forward loop only

Design of a traction control system for the four-wheel driven race car IM01 21
CHAPTER 5. SIMULATION RESULTS

As can be seen in figure 5.3, the feed-forward controller cuts down the requested drive torque
immediately and keeps the applied drive torque to the motor constant at TF eed−f orward for the
rest of the simulation.

Third, the model with only the feed-back loop implemented is simulated with, again, the torque
request as plotted in figure 5.1. So, in this simulation, no feed-forward controller is used. The
drive torque is only controlled by the ωerror signal and the PI-controller as explained in section
4.1. The results of this simulation are shown in figure 5.4.

22 Design of a traction control system for the four-wheel driven race car IM01
 CHAPTER 5. SIMULATION RESULTS

Applied drive torque with feed-back loop only

2500
Drive torque [Nm]

2000
1500
1000
500
0
0 1 2 3 4 5 6 7 8 9 10
Time [s]
Estimated µ x
2

1.5
µ [-]

1
x

0.5

0
0 1 2 3 4 5 6 7 8 9 10
Time [s]
Wheel and vehicle velocities with feed-back loop only
150

ω*rw
100
Vx [m/s]

Vx

50

0
0 1 2 3 4 5 6 7 8 9 10
Time [s]
Wheel and vehicle accelerations with feed-back loop only
25
20 dω/dt*r w
a [m/s 2 ]

15 ax
10
x

5
0
0 1 2 3 4 5 6 7 8 9 10
Time [s]

Figure 5.4: Simulation results with the feed-back loop only

As can be seen in figure 5.4, the feed-back controller cuts down the requested torque immedi-
ately, after that, the feed-back controller lets the applied drive torque increase slowly.

Design of a traction control system for the four-wheel driven race car IM01 23
CHAPTER 5. SIMULATION RESULTS

Fourth, the model with the full traction control system is simulated with, again, the torque
request as plotted in figure 5.1. So, in this simulation, all the parts of the traction control system
are implemented. The drive torque is both controlled by the feed-forward loop and the feed-back
loop. The results of this simulation are shown in figure 5.5.

Applied drive torque with TC

2500
Drive torque [Nm]

2000
1500
1000
500
0
0 1 2 3 4 5 6 7 8 9 10
Time [s]
Estimated µ x
2

1.5
µ [-]

1
x

0.5

0
0 1 2 3 4 5 6 7 8 9 10
Time [s]
Wheel and vehicle velocities with TC
200

150 ω*rw
Vx [m/s]

Vx
100

50

0
0 1 2 3 4 5 6 7 8 9 10
Time [s]
Wheel and vehicle accelerations with TC
25
20 dω/dt*r w
a [m/s 2 ]

15 ax
10
x

5
0
0 1 2 3 4 5 6 7 8 9 10
Time [s]

Figure 5.5: Simulation results with the full traction control system

24 Design of a traction control system for the four-wheel driven race car IM01
 CHAPTER 5. SIMULATION RESULTS

Vehicle accelerations with a drive torque of 3000 Nm

20

18

Without TC
16
Theoretical acceleration
With TC

14

12
a [m/s 2 ]

10
x

0
0 1 2 3 4 5 6 7 8 9 10
Time [s]

Figure 5.6: Vehicle accelerations with and without traction control and the theoretical maximum

As can be seen in figure 5.5, the traction control system lets the applied drive torque gradually
rise to the value of TF eed−f orward . This results in a longitudinal acceleration which is very close
to the theoretical maximum acceleration, the longitudinal vehicle accelerations are shown in more
detail in figure 5.6.

In section 3.4, the theoretical maximum acceleration is calculated in equation 3.15 and is equal
to: amax = 17.9385m/s2 . This theoretical maximum acceleration is plotted in figure 5.6 together
with the acceleration with and without traction control for a requested drive torque of 3000N m.
As can be seen in figure 5.6, the vehicle acceleration that is realised by using the traction control
system is almost equal to the theoretical maximum acceleration and is way higher than the vehicle
acceleration without traction control, from this, and from comparing figure 5.2 and figure 5.5, it
can be concluded that the designed traction control system works.

Also the 0 − 100km/h times with and without traction control are determined, the results are
shown in table 5.1. The 0 − 100km/h is approximately 40 % higher with traction control than
without.

Design of a traction control system for the four-wheel driven race car IM01 25
CHAPTER 5. SIMULATION RESULTS

Without TC 2.88s
With TC 2.05s

Table 5.1: 0 − 100km/h times with and without TC

In figure 5.7, the results of the µx estimator are shown, the estimated µx is plotted together
with the estimated µx which is decreased with 1 after 2.5s. The estimated µx becomes almost
equal to the real value of µx as given in section 3.2.1. Also, µx which is decreased with 1 after
2.5s is estimated to the real value of µx minus 1. So it can be concluded that the µx estimator
approximates the real value of µx and tunes its estimated value of µx after a decrease.

µ x estimator
2

1.8

1.6
Estimated µ without decrease
x
Estimated µx with decrease
1.4
Real µ without decrease
x
Real µ with decrease
x

1.2
µ [s]

1
x

0.8

0.6

0.4

0.2

0
0 1 2 3 4 5 6 7 8 9 10
Time [s]

Figure 5.7: Results of the µx estimator

26 Design of a traction control system for the four-wheel driven race car IM01
Chapter 6

Conclusions & recommendations

6.1 Conclusions
The objective of this report is to design a traction control system that does not use the wheel
speed of a non-driven wheel as an input and is applicable to a four-wheel driven car.

In this report, a model is developed to simulate the dynamics of one wheel of the car for a drive
torque applied by the motor to the wheel. A tyre model based on the Magic Formula tyre model
is developed to simulate tyre behaviour and tyre forces. A traction control system is designed
that does not use the wheel speed of a non-driven wheel as an input and that is applicable to a
four-wheel driven race car as is the IM01. This traction control system is designed on one wheel
only and is based on the slip ratio control approach. The traction control system consists of a
feed-forward and a feed-back part and is then expanded with a peak friction estimator which estim-
ates µx . The validation of the traction control system is done with the help of the developed model.

Using the feed-forward loop only, the best performances can be achieved, however, variations in
the road-surface and disturbances can not be controlled. That’s why the feed-back loop is needed
in the traction control system. Together, the feed-forward and feed-back loop can achieve better
performances than without, but not as good as with the feed-forward loop only.

The traction control system that is designed and validated through simulations works and
results in better performances than when no traction control system is used. The peak friction
estimator that is developed works as well and is able to estimate the value of µx .

6.2 Recommendations
Because some assumptions have been made in the dynamical model, as stated in section 1.2, the
model can be improved by including the aerodynamic and rolling resistance for example. By doing
this, the simulations will come closer to the real world and so, the design of the traction control
system can be improved so it will deliver a higher level of performance. Also, tyre deformation
can be taken into account what will lead to better results as well.

The peak friction estimator as it is designed for this report is based on the method introduced
by Y. Hori et al. [2]. Other papers introduce other ways of estimating µx such as using a recursive
least squares method, which should be investigated and will possibly lead to better results and to
a faster working estimator.

Design of a traction control system for the four-wheel driven race car IM01 27
Bibliography

[1] I.J.M. Besselink. Introduction Vehicle Dynamics and Powertrains , 4AUB0. pages 1–198, 2015.
4, 5, 11, 12

[2] Y. Hori, Y. Toyoda, and Y. Tsuruoka. Traction Control of Electric Vehicle: Basic Experimental
Results Using the Test EV UOT Electric March. IEEE Transactions on Industry Applications,
34(5):1131–1138, 1998. 2, 6, 8, 18, 27
[3] N. Janssen. Designing control systems for InMotion ’ s KP & T IM / e. 2016. 8

[4] Kistler Group. Correvit

R S-350 Datasheet. 2015. 9

[5] J. Li, Z. Song, Z. Shuai, L. Xu, and M. Ouyang. Wheel Slip Control Using Sliding-Mode
Technique and Maximum Transmissible Torque Estimation. Journal of Dynamic Systems,
Measurement, and Control, 137(11):111010, 2015. 7

[6] J. W. Loof. Traction control of an electric formula student racing car , modeling and validation.
(JUNE), 2013. 7, 8, 17
[7] H. B. Pacejka. Tyre Characteristics and Vehicle Handling and Stability. 2006. 5, 12
[8] D. Yin, S. Oh, and Y. Hori. A novel traction control for EV based on maximum transmissible
torque estimation. IEEE Transactions on Industrial Electronics, 56(6):2086–2094, 2009. 7

28 Design of a traction control system for the four-wheel driven race car IM01