Tire Modeling and Friction Estimation

Tire Modeling and Friction Estimation
Tire Modeling and Friction

Estimation
Jacob Svendenius
Department of Automatic Control

Lund University
Lund, April 2007
Department of Automatic Control
Lund University
Box 118
SE-221 00 LUND
Sweden
ISSN 0280–5316
ISRN LUTFD2/TFRT--1077--SE
c 2007 by Jacob Svendenius. All rights reserved.

Printed in Sweden.
Lund 2007
Abstract
New system technologies are continuously improving the performance of

vehicles regarding comfort, stability, environmental stresses, and safety.
Novel sensors are developed and used together with advanced control al-
gorithms and faster and more accurate actuators to help the driver to
maneuver the vehicle in a safer way. The physical traction limit set by the
friction between the tires and the road can, however, not be overridden.
The behavior of tires is, therefore, an important issue since it determine
the possibilities to control the vehicle. One essential task for the vehicle
control system is to be able to fully utilize these limits. Another is to as-
sist the driver choose an adequate driving style, adapted to the actual
conditions.
A method to derive the tire forces for simultaneous braking, cornering
and camber, by combining empirical models for the pure behavior of each
of respective action is presented in the thesis. The method is based on the
physical foundation brought by the simple, but well-suited brush model
theory. The pure-slip tire models can be given as empirical models or as
raw tabular data. The implementation is verified to be well-working and
computationally sound and good results are obtained in validation with
the available empirical data.
A new type of on board brush-model based friction-estimator using the
local measurements on the vehicle is also proposed. A major invention is
the way of collecting the measurements into bins, such that the available
data used for optimization is evenly weighted along the force and slip
axis. Experimental data has been collected and evaluated to ensure the
validity of the brush model during certain conditions. The estimator has
been implemented and validated on a personal car.
The wheel speed signal is an essential signal in many system appli-
cations. Based on problems observed during performed measurements, a
few ways to reduce the noise on the signal, without adding any phase shift
or time delay, are discussed in the thesis.
5
Abstract
6
Acknowledgments
First of all I would like to express my gratitude to the persons that made
the decisions, a huge amount of years ago, necessary to establish my sit-
uation as industrial PhD student, the former technical manager, Kent
Jörgensen and the former personal manager, Lena-Maria Lundberg at
Haldex Brake Products AB. I will also thank my supervisor, Björn Wit-
tenmark, at the Department of Automatic Control for the administrative
work he made to establish this collaboration. His advice, encouraging sup-
port and guiding corrections has all the time lead the projects and the
thesis writing in a rewarding direction. Thanks to my assistant supervi-
sor Per Hagander for proof reading and valuable comments on the thesis.
I would also like thank my supervisor at Haldex, Per-Axel Roth, for good
listening and finding nice parallels from his earlier experiences.
I’m very grateful to my group manager at Haldex, Anders Lindqvist,
for giving great support in all kind of matters and being the only one
at Haldex to always participate on the reference group meetings. It is a
pleasure to be a part of a great working group, having Anders Nilsson,
Ola Nockhammar, Lisette Pedersen, Sabine Marksell and Ola Ahlström
as colleagues. Many thanks to Peter Nilsson for his wise opinions and
skilled advices. I am particularly thankful Sverker Stomrud for his great
commitments in all technical and patent related issues and strong belief
in new ideas.
The work in the Road Friction Estimation project has been very inter-
esting and rewarding and I will gratefully acknowledge Johan Hultén at
Volvo Cars for good project leadership and Fredrik Bruzelius for fruitful
and inspiring cooperation and nice fellowship and proof reading of the
thesis. Both Johan and Fredrik are deeply involved the chapter concern-
ing the road friction estimator. Special thanks to Magnus Gäfvert, now at
Modelon AB, but former at both the Department of Automatic Control and
at Haldex, for his enormous working spirit. I am glad for our rewarding
discussions and nice collaboration. Magnus has contributed in both the
7
Acknowledgments
development of the semi-empirical tire model and of the friction estima-

tor and most of my publications have been written together with him. He
is to a large content involved in most sections of the thesis and I am glad
for his careful proof reading and valuable comments on the dissertation.
An advantage when having two work places is that one get many col-
leagues and I am gifted with fantastic coworkers at two different work-
places. I am very pleased with the creative spirit and open atmosphere
at the Department of Automatic Control in Lund, which has been a great
source of inspiration.
It has everyday been a pleasure to enjoy the lunch at Sparta with
"Spartagänget" that now has been reduced to Rolf Braun (I hope that you
will not need to eat alone from now on). It is also a delight to quarrel
with the always benevolent and helpful secretaries, Eva, Britt-Marie and
Agneta. Thanks to Leif for all kind of help in report-writing issues and
thanks to Brad for linguistic corrections on some of the important sections
in the thesis.
I am really grateful for the encouragement and support from my mother,
father and sister. It is good to know that they always care. Thanks to In-
ger and Christer for always stand by and being helpful. A lot of thanks
go to all my friends, many of them have somehow helped me to come to
this point.
Finally, I will thank my wife, Malin for her care and endless love and
my son, Jonah that really makes life exciting and meaningful. I love you
both very much!
Jacob
8
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.1 Background and Motivation . . . . . . . . . . . . . . . . 11
1.2 Main Results . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.3 Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.4 Related Publications . . . . . . . . . . . . . . . . . . . . 16
2. Tire Fundamentals . . . . . . . . . . . . . . . . . . . . . . . . 19
2.1 History . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.2 Design and Construction . . . . . . . . . . . . . . . . . . 20
2.3 Standards . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.4 Kinematics . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.5 Tire Mechanics . . . . . . . . . . . . . . . . . . . . . . . . 23
2.6 Tire Deformation . . . . . . . . . . . . . . . . . . . . . . 26
3. Review of Existing Tire-Models . . . . . . . . . . . . . . . . 28
3.1 Steady-State Tire Models . . . . . . . . . . . . . . . . . . 28
3.2 Combined-Slip Semi-Empirical Tire-Models . . . . . . . 32
3.3 Transient Tire-Models . . . . . . . . . . . . . . . . . . . 35
3.4 Finite Element Models . . . . . . . . . . . . . . . . . . . 37
3.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4. Tire Modeling According to the Brush Model Theory . 39
4.1 Basics of the Brush Model . . . . . . . . . . . . . . . . . 39
4.2 The Effect of Camber . . . . . . . . . . . . . . . . . . . . 55
4.3 Effects of a Flexible Carcass . . . . . . . . . . . . . . . . 59
4.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5. A Dynamic Semi-Empirical Tire-Model for Combined
Slips and Camber . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 68
5.2 Scale Factors . . . . . . . . . . . . . . . . . . . . . . . . . 70
5.3 Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 73
9
Contents
5.4 Validation and Results of the Semi-Empirical Model . . 74

5.5 Implementation and Application in Multi-Body Simula-
tion of Vehicle Handling . . . . . . . . . . . . . . . . . . 95
5.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . 97
6. Measurement and Processing of the Wheel Speed Signal 100
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 100
6.2 Effect of Tire Force Transients on the Wheel Speed Mea-
surements . . . . . . . . . . . . . . . . . . . . . . . . . . 103
6.3 Method for Elimination of Rotation Dependent Noise . 111
6.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . 119
7. Road Friction Estimation . . . . . . . . . . . . . . . . . . . . 120
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 120
7.2 Experimental Validation of the Brush Model . . . . . . 124
7.3 Tire Parameter Estimation . . . . . . . . . . . . . . . . . 143
7.4 Algorithm Implementation . . . . . . . . . . . . . . . . . 155
7.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
7.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . 166
8. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
8.1 Ideas of Further Research . . . . . . . . . . . . . . . . . 172
9. Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
A. Nomenclature . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
B. Scale Factors and Slip Conversion . . . . . . . . . . . . . . 189
C. Extension of Optimization Algorithm to Include Hori-
zontal Shift . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
10
1
Introduction
1.1 Background and Motivation

This thesis is a result of cooperation between the R&D department at
Haldex Brake Products in Landskrona and the Department of Automatic
Control at Lund University. The work is closely related to the develop-
ment of brake-by-wire systems using electrically actuated brakes for heavy
trucks and buses performed at Haldex. One objective of the collaborative
work has been to further investigate the new field for improvements of
vehicle stabilizing and safety functions, opened by the enhanced control-
lability of an electrically actuated brake compared to the conventional
pneumatic brake. For a well-working brake system, the physical proper-
ties of the brakes are not limiting, their capability to produce braking
torque is very rarely used fully. Instead the conditions of the tire and the
road surface set the functional limits for safe retardation of the vehicle.
When the Anti-lock Braking System (ABS) was introduced in the luxury
vehicles it greatly improved the surface friction utilization of the brake
system in critical situations. Nowadays, it is standard equipment in most
cars. The motivation for developing ABS relies on the observation that
a locked wheel only produces a force in the sliding direction of the tire.
Furthermore, a higher tire force can be obtained from a rolling wheel.
Preventing the wheel from locking while still generating a high braking
force is shown to reduce the braking distance. But, maybe more impor-
tant is that it enables the possibility to the driver to control the vehicle
direction by steering maneuvers.
There is an optimal wheel rotation velocity, depending on the tire prop-
erties, surface condition, and vehicle motion, for which the tire develops its
maximal friction force. Finding and keeping this value under the current
condition in critical maneuvers, while providing stable and safe motion
11
Chapter 1. Introduction
of the vehicle, is a great issue for improvement of today’s systems. The

new quicker brake actuators increase the possibilities for stable control
of the wheel velocity in such situations. When increasing the bandwidth
of the braking control loop the dynamic effects of the mechanical systems
supporting the brake and participating in the transmission of the force
to the road become crucial. Deflections in suspension, bushings, and the
tire delay the generated tire force acting on the vehicle and may cause
displacement of the wheel speed sensor. This disturbs the wheel speed
measurements. The noise filters have to be faster to catch rapid signal
variations, resulting in less attenuation of the disturbances. The inertia
in the wheel rim might cause unwanted oscillations. Hence, utilization
of the advantages in control of faster brakes requires the algorithms to
further predict and estimate the dynamics of the surrounding system.
Enhanced sensor quality and larger computational capacity increases the
possibilities to achieve such estimations.
The new and improved technology emphasizes attempts to also detect
the conditions surrounding the entire vehicle using existing sensors. The
idea is attractive since the functionality of many other vehicle systems
can be improved by such knowledge. Automatic cruise control (ACC), col-
lision avoidance warning and, electronic stability program (ESP) are a
few examples. Another important feature is to help the driver to adapt
the driving style towards the current conditions. Within the EU-project
Intelligent Roads [Vägverket, 2006a] research is performed about how the
road surface limitations should be presented to the driver. It might be as
estimated braking distance, or as a recommended maximal speed. For an
on board road-condition estimator it seems quite impossible to determine
the circumstances in front of the vehicle. It could however, be possible
by using a preview sensor or having an infrastructure that receives and
distributes frictional information about the roads. The distribution can be
both to a common central or to share between the vehicles in a close neigh-
borhood. Research regarding infrastructures that receives and distributes
estimates of the surface friction along the road is performed within the
national program Safety Road Information Structure [Vägverket, 2006c].
A major aim of this work is to explain and model the general behavior
of tires and to find properties that are common for the majority of tire
types. Simplicity in modeling is a keyword throughout the thesis. Often,
the final solutions still tend to be sufficiently comprehensive. Starting
with complex models might limit the usability in the end. Therefore, the
basic form of the brush-model theory to describe the tire behavior has
been fundamental in this work.
The developed theories are based on physical assumptions and most
of them are validated from available data or by tests performed partic-
ularly for this purpose. Due to good experimental equipment at Haldex
12
1.1 Background and Motivation
and the close collaborative work with both Volvo Cars and VTI (Swedish
National Road and transport Research Institute) a multitude of different
test scenarios have been performed. There have been many possibilities
to participate in the planning of the tests and most of them have been
directly specified to serve the method development described in the the-
sis. There is a great benefit in participating during the testing, since it
gives a better feeling for how the surrounding conditions and other un-
expected factors might influence the result. All tests have be performed
on real tires in real conditions. This has made it impossible to completely
separate the measurement of, for example, the tire behavior from distur-
bances from the road and the wheel suspension, etc. On the other hand
it gives the most realistic picture of the problems and difficulties that
have to be dealt with. Although good equipment has been available, some
of the issues addressed in the thesis, relating to very quick events, have
not been validated completely. The reason is that the sensors and mea-
surement systems that have been used have not been capable of such fast
acquisition. Although there are weeks of testing behind the results in the
thesis, further testing is still one of the most urgent issues for the future
work.
Three different, but closely related topics, are treated in the thesis.
Semi-empirical tire-modeling
Simultaneous cornering affects the behavior of the tire during a braking
phase. For some vehicle stability systems it is more important how the
vehicle cornering performance is affected when applying a brake torque
while turning. If a vehicle exhibits a large lateral acceleration such that
a roll-over may be inevitable a concurrent controlled braking can reduce
the lateral force stabilizing the vehicle, see for example [Schofield et al.,
2006]. Further, a system may not demand a higher brake force in a corner
than that the vehicle can keep its course not running off the road.
The thesis includes a description of a newly developed semi-empirical
tire-model that describes this interaction. A main focus in the model
derivation is the possibility to extract as much information as possible
from the pure slip data. This may simplify and reduce the number of cal-
ibration procedures needed to map the specific tire behavior. The longitu-
dinal and lateral forces are described by scalings of the pure-slip forces.
The used scale factors are calculated from physical relations between com-
bined and pure slip.
Transient maneuvers in one direction have a dynamic effect on the tire
behavior in both directions. The thesis derives the cross-coupling terms
and studies the transient properties of a tire for low and high slip situa-
tions.
13
Improvement of wheel speed measurement

The wheel-speed signal is one of the most essential signals for the brake
system control. Errors and disturbances on the measurements often re-
quire hard filtering that slows down the brake performance. The wheel
speed is measured as the relative rotational speed between the rim and
the axle. Transient force applications may disturb the sensor position and
cause disturbances on the measurements. The effect became particularly
evident during a test sequence where the ABS-function of the vehicle was
triggered due to a response on the wheel speed from the quick brake
torque application. The wheel was not even close to locking at that time.
Also tolerances and unevenesses in sensor arrangement may cause extra
jitter on the wheel speed signal. The thesis discusses methods to reduce
such disturbances.
Road friction estimation

Friction is a prerequisite for the modern life of human beings. Technical
solutions for transport and manufacturing would be completely different,
if ever possible, if materials did not bond to achieve frictional forces. In
many normal duties the anticipation of the friction has become automatic
and is nothing that we really reflect over. Everything from small incidents
to fatal accidents may occur if the friction differs from the expectations
or changes suddenly. A pedestrian that slips on a slippery spot or a roller
skater whose wheel bearing abruptly cuts are two examples.
A motivation for trying to estimate the friction is that the number of
accidents related to slippery road conditions might be reduced if the driver
is correctly informed about the limitations of the vehicle in the current
condition. Safety can also be increased, since the performance of many ap-
plications in the vehicle can be enhanced by knowing the current frictional
condition. The brake control can be made more efficient by knowing the
available friction, for example the gain-scheduled slip controller proposed
in [Solyom et al., 2004] bases it switching on whether the actual slip is
lower or higher than a the slip corresponding to the friction peak. This
slip value is an important parameter for obtaining the optimal tire force
and can be derived from an assumption on the estimated actual friction
and the tire characteristics.
The last part of the thesis is results from the work within a cooperative
project between Haldex, the Department of Automatic Control, and Volvo
Cars within the Intelligent Vehicle Safety Systems (IVSS) programme. This
part deals with on board model-based friction estimation based on the
local measurements on the vehicle.
The proposed algorithm includes a memory-efficient way to store fil-
tered values of the sampled data into storage bins, distributed over the
14
1.2 Main Results
force and slip axis. This will maximize the persistence of excitation of the
data feeding the estimator and it is a compact form to store and use old
data, important for characterizing the tire behavior. The tire friction and
tire stiffness parameters are derived by minimizing the error between the
adopted tire model and the stored data. The model has a practical focus
intended for implementation in production cars.
1.2 Main Results

The thesis mainly contributes in a few related areas. The first is the
derivation of a new physically based semi-empirical tire model that pre-
dicts the tire behavior for simultaneous braking, cornering, and cambering
given the tire characteristics for pure braking, cornering, and cambering.
The method incorporates the following features:
• The tire forces at combined slip are derived using any longitudinal,
lateral, and camber pure-slip model or data set.
• The scale factors for the combined tire forces are computed in a
unique way from the relation between the combined tire force and
the pure tire force. Any theoretical model describing the tire behavior
at combined slip can therefore be used to form the scale factors.
• The model is, in its presented form, based on the most basic brush-
model theory. The structure makes it straight forward to incorporate
effects from more variables and more sophisticated modeling.
• Velocity dependence is included in the model, even though this is
not explicitly present in the pure-slip models.
• The model has the possibility to convert between driving and braking
data.
• All included parameters can automatically be computed from the
shape of the pure-slip models.
The wheel speed is an important signal for the applications covered by the
thesis. A few ways to improve the quality of the wheel speed measurements
are discussed. Particularly, the contribution here is:
• The construction of a feedforward filter on the brake torque that pre-
dicts the the disturbances on the wheel speed due to the application
of the brakes.
A method for friction estimation is another contribution of this thesis. Its
major characteristics are:
15
• A filter that stores representatives of the sampled data in a compact

form to maximize the possibilities to extract information from the
force-slip relation.
• Estimations of the braking stiffness of the tire and the road-tire
friction using non-linear optimization techniques.
• The model-based estimation relies on brush-model mechanics
• Promising results are shown in validation tests at different driving
conditions.
1.3 Outline
The topics covered by the thesis are related to the wheel-tire-road in-
teraction. Chapter2 explains the fundamental properties and definitions
related to the characteristics of tires. In Chapter 3 there is a brief review
of existing tire models. A basic concept for tire modeling is the brush model
developed in the early age of the tire-modeling era. The theory behind the
brush model and some extensions are described in Chapter 4. This model
is fundamental, both in the derivation of the semi-empirical tire-model
described in Chapter 5 and for the friction estimator developed in Chap-
ter 7. Chapter 6 contains material describing a few ways to improve the
quality of the wheel speed signal, which may enhance the performance of
vehicle systems relying on this signal, e.g. the proposed friction estimator.
Proposals for further work and some concluding remarks are mentioned
in the final chapter of the thesis.
1.4 Related Publications

The thesis is based on research documented in several publications where
the author of the thesis is a coauthor or the sole author.
The following publications and reports contain theory, implementation,
and validating results of the semi-empirical tire model:
Gäfvert, M. and J. Svendenius (2003): “Construction of semi-empirical tire
models for combined slip.” Technical Report ISRN LUTFD2/TFRT--
7606--SE. Department of Automatic Control, Lund University, Sweden.
Gäfvert, M. and J. Svendenius (2004): “A semi-empirical tire-
model including the effects of camber.” Technical Report ISRN
LUTFD2/TFRT—7611–SE. Department of Automatic Control, Lund
University, Sweden.
16
1.4 Related Publications
Gäfvert, M. and J. Svendenius (2005): “A novel semi-empirical tire model

for combined slip.” Vehicle System Dynamics, 43:5.
Gäfvert, M., J. Svendenius, and J. Andreasson (2006): “Implementation
and application of a semi-empirical tire-model in multi-body simula-
tion of vehicle handling.” In Proceedings of the 8th International Sym-
posium on Advanced Vehicle Control. Taipei, Taiwan.
Svendenius, J. (2003): “Tire models for use in braking applications.”
Licentiate thesis ISRN LUTFD2/TFRT--3232--SE. Department of
Automatic Control, Lund University, Sweden.
Svendenius, J. (2006): “Examples of scale factors for a semi-empirical
tire-model.” Technical Report ISRN LUTFD2/TFRT--7614--SE. De-
partment of Automatic Control, Lund University, Sweden.
Svendenius, J. and M. Gäfvert (2004a): “A brush-model based semi-
empirical tire-model for combined slips.” SAE Technical Paper 2004-
01-1064.
Svendenius, J. and M. Gäfvert (2004b): “A semi-empirical tire-model for
transient combined-slip forces.” In AVEC ’04.
Svendenius, J. and M. Gäfvert (2005): “A semi-empirical tire model for
combined slips including the effects of cambering.” In Lugner and
M.Plöchl, Eds., Tyre Models For Vehicle Dynamics Analysis, vol. 43,
pp. 317–328. Taylor & Francis Group, Glasgow. Supplement to Vehicle
System Dynamics Volume 43.
Svendenius, J. and M. Gäfvert (2006): “A semi-empirical dynamic tire
model for combined-slip forces.” Vehicle System Dynamics, 44:2,
pp. 189 – 208. Special Issue: AVEC ’04: 7th International Symposium
on Advanced VEhicle Control, 23-27 August 2004 HAN University,
Arnhem, The Netherlands.
The conference paper [Svendenius and Gäfvert, 2004b] was chosen to-
gether with another six submissions from all contributions published at
the AVEC-04 conference to be included in the special issue of the Journal
of Vehicle System Dynamics, reflecting the highlights of the conference,
see [Svendenius and Gäfvert, 2006].
The following publications and reports are related to friction estima-
tion project:
Svendenius, J. (2007a): “Validation of the brush model towards VTI-
measurement data recorded at Hällered 2005.” Technical Report ISRN
University, Sweden. RFE-project.
17
Svendenius, J. (2007b): “Validation of the brush model towards VTI-

measurement data recorded in Arjeplog 2006.” Technical Report ISRN
To be able to realize the described method for friction estimation into
a future commercial product is has been necessary to try to protect the
ideas on an early stage. The work on the friction estimator has therefore
resulted in the following patent applications:
Svendenius, J. and M. Gäfvert (2005): “System and method for tire/road
friction estimation.” Patent application. Assignee: Haldex.
Svendenius, J., M. Gäfvert, J. Hultén, and F. Bruzelius (2007): “Systems
and methods for determining a parameter relating to a tire-to-road
contact and/or a relation between a wheel and a vehicle motion.”
Patent application. Assignee: Haldex & Ford.
18
2
Tire Fundamentals
2.1 History
The history of the tire started in 1839 when Charles Goodyear discov-
ered the rubber vulcanization process [Continental, 2003], where sulphur
is added to the natural rubber during heating. Varying the amount of
sulphur the properties of the material was shown to better resist wear,
ageing and water, and decrease the sensitivity to temperature changes.
Solid rings of vulcanized rubber later was used around the wheel rims
to reduce the vibration problem and to improve the traction properties
for the early vehicles. A couple of years later, 1845, Robert W. Thom-
son patented the idea of using air-filled rubber tubes as tires. Due to
the lower durability the invention fell into disuse [Thomson, 2003] and
in 1888 John Boyd Dunlope reinvented the pneumatic tire claiming of no
knowledge about the prior patent. The new patent was mainly directed for
bicycles, but the advantages of using pneumatic tires also for cars were
successfully examined by, among others, the brothers André and Édouard
Michelin. They competed with a car equipped with pneumatic tires in the
1895 Paris-Bordeaux road race. Although they did not win, they gained a
lot of interest due to the new type of tires. A few years later air filled tires
became an obligation for driving on the highways. Today, Michelin, which
besides Goodyear and Bridgestone, is the largest tire manufacture in the
world with approximately 20% of the market. In 2004 the company had
126 000 employees and reported net sales of 15.700 million EUR [Michelin,
2004].
19
Chapter 2. Tire Fundamentals
2.2 Design and Construction

The pneumatic tire is a flexible structure that together with the rim holds
the pressure of the inflated air. The most important functions of the tire
are to reduce vibrations from unevennesses in the road and to achieve
a high friction coefficient in the interaction with the road surface. Its
main structural component is the carcass, which consists of layers of stiff
cords to hold the shape of the tire and the tension from the compressed
air. High tensile steel wires, called beads, keep the carcass to the rim
and when a load is applied on the wheel, the rim primarily hangs on
the sidewall cords and the beads. The entire structure is covered with a
wear resistant rubber compound, often styrene-butadiene, to protect the
carcass and to build up the friction to the road. There are mainly two ways
to design the carcass, radial-ply and bias-ply. Radial-ply implies that the
sidewall cords are oriented radially and the wear surface cords are laid in
layers tangentially with small angles between the cords in each layer. The
bias-ply tires have the cord layers diagonally positioned over the entire tire
surface. For a rolling bias-ply tire the deformation of the cords gives rise to
a wiping motion of the rubber tread, which causes higher wear and power
dissipation. Therefore, radial tires are nowadays mostly used for cars and
trucks, even though their manufacturing process is more complex and the
expense is about 50% higher than for a bias-ply tire [Thomson, 2003].
2.3 Standards
Many issues related to the tire are controlled by regulations and stan-
dards. One of those that might be interesting for the consumer is the
standard for the tire marking. In Europe the ECE (Economic Commis-
sion for Europe) Regulation and EEC (European Economic Community)
Directives set standards for the tire marking, which differ from the Amer-
ican way of marking the tires. However, the size of the tire is always
printed on the sidewall as a combination of four designations, for instance
P 205/75R15. The first letter (sometimes left out) signifies the vehicle for
which the tire is intended. The following number denotes the tire width
in millimeter. A slash separate the width from the height of the tire side,
which is expressed as the fraction in percent of the width. After the letter
R, the rim radius is given in inches. The size marking is often followed
by markings denoting load and speed classes.
There are also standards for the nomenclature regarding the kinemat-
ics and the mechanics of the tire. The movements and forces developed by
the tire are important signals for the control or simulation of the dynam-
ics of the vehicle. Many errors can therefore be avoided by standardizing
20
2.4 Kinematics
the coupling between the tire model and the multi-body vehicle system.
However, the multitude of different standards in this area can be confus-
ing. In 1974 SAE published a vehicle dynamics standard specifying the
choice of coordinate system and notations [SAE Recommended Practice
J670e, 1976]. Fifteen years later the European version ISO 8855 was re-
leased [ISO 8855, 1991]. One of the greatest pioneers in tire modelling, H.
B. Pacejka used an adapted version of the SAE standard [Pacejka, 1988]
in his works. In 2002, TNO together with TU Delft promoted a version
of the ISO 8855 in a standard specialized for acquisition of measurement
data for calibration of tire model parameters called TYDEX (Tyre Data
Exchange Format) [TYDEX-Working group, 1997]. In close connection to
TYDEX is the Standard Tire Interface (STI) that prescribes the interface
for the interaction between the tire model and the Multi Body Simulation-
system. All of these four standards deviate regarding sign conventions of
slips and forces, which makes it important to state which standard that
is followed or, alternatively, clearly define the signification of the used
notations.
The choice of reference system in this work follows the SAE stan-
dard [SAE Recommended Practice J670e, 1976], with the longitudinal x-
axis aligned with the wheel heading, the lateral y-axis perpendicular to
the wheel, and the vertical z-axis pointing downwards. The ISO-standard,
instead prescribes the direction of the z-axis to be upwards. Hence, the
direction of both the y- and z-axis differ between the two standards, since
their reference systems are orthogonal. In Sections 2.4–2.5 the used no-
tations and sign conventions are explained, as from [Gäfvert, 2003]. The
table in Appendix A lists the nomenclature and its deviation from ISO
8855:1991.
2.4 Kinematics
This section describes the relevant tire kinematics and introduces defi-
nitions which are used in the thesis. The entities are illustrated in Fig-
ure 2.1. Vectors have two components and are denoted by a bar as in
v̄. The corresponding
q components and magnitude are denoted by vx , vy,
and v = v2x + v2y . The x-axis is often refereed to as the longitudinal di-
rection and the y-axis as the lateral direction. The wheel-travel velocity
v̄ = ( vx , vy ) deviates from the wheel orientation by the slip angle α
vy
tan(α ) = (2.1)
vx
21
v̄
vx α
v̄s
vsx
vy vx
− Fy vsy
Mz ω
R
F̄ − Fx
x x
− Fx
y z
Figure 2.1 Kinematics of an isotropic tire during braking and cornering. Force
vectors are also included. (Left: top view; Right: side view)
The circumferential velocity of the wheel is
vc = ω R e (2.2)
where ω is the wheel angular velocity, and R e the effective rolling-radius

of the tire, defined as the ratio vx /ω 0 , where ω 0 is the wheel speed for
a free rolling wheel. The slip velocity, or the relative motion of the tire
in the contact patch to ground, that arises when a horizontal force is
transmitted, is
v̄s = ( vx − vc , vy ) (2.3)
The direction of the slip velocity is denoted by β where
vsy
tan(β ) = (2.4)
vsx
The tire slip is defined by normalizing the slip-velocity with a reference

velocity. Three slip definitions are commonly used
v̄s v̄s v̄s

σ̄ = (σ x , σ y) = ; κ̄ = (κ x , κ y) = ; s̄ = (s x , s y) = (2.5)
vc vx v
Note that the slips are collinear with the slip velocity v̄s . It is the custom
to describe tire-forces as functions of the slip rather than the slip velocity.
22
2.5 Tire Mechanics
This convention is followed also in this work. Implicitly, this assumes that
the forces do not depend on the magnitude of the slip velocity, vs . In
general, at least the sliding friction is velocity dependent. The ISO and
SAE standards [ISO 8855, 1991; SAE Recommended Practice J670e, 1976]
use −100κ x [%] to represent longitudinal slip, and α [deg] for lateral slip.
Here, the slips are defined such that signs are consistent for the different
slip definitions, and such that a generated tire force has opposite sign to
the slip. This means that braking or left cornering will result in positive
slip and negative force. For convenience the slip ratio, λ , will be used
to denote longitudinal slip as: λ = κ x . It is straightforward to translate
between the different slip representations
( λ , tan(α ) ) κ̄ s̄
σ̄ = = =q (2.6a)
(1 − λ ) 1 − κx 1 − s2y − s x
σ̄ s̄
κ̄ = ( λ , tan(α ) ) = =q (2.6b)
1 +σx 1 − s2y
σ̄ κ̄
s̄ = ( λ cos(α ), sin(α ) ) = q = q (2.6c)
(1 + σ x )2 + σ 2y 1 + κ y2
2.5 Tire Mechanics
Forces
Forces and torques working on a tire are shown in Figure 2.2, according
to the SAE-definition.
The forces of interest for vehicle handling and control purposes are the
planar lateral and longitudinal forces, Fx and Fy, and the self-aligning
torque, Mz . The longitudinal tire force Fx is generated when braking or
driving. In the following, when the word “braking” is used in the context
of longitudinal tire force generation, this will actually mean “braking or
driving” unless stated otherwise. The lateral force and the torque are
generated when cornering. The self-aligning torque results from the fact
that the planar forces have a point of action which is not positioned exactly
under the wheel center. The rolling-resistance is related to the energy
dissipation for a rolling tire and the overturning torque is the torque
necessary to camber the wheel.
Tire slip
Pure slip There is a relation between the horizontal tire force and
23
Aligning Torque Mz
Camber γ
Wheel Torque
Direction of Wheel Heading
Rolling Resistance Torque My
Tractive Force Fx
x
Slip Angle α
Direction of Wheel Travel v̄
Lateral Force Fy
y
Overturning Torque Mx
Normal Force Fz
Figure 2.2 Forces and moments acting on a tire [SAE Recommended Practice
J670e, 1976].
the slip of the tire. The velocity difference between the carcass and the
road is a result from continuous deformation of the rubber treads and
sliding between the tire and the road surfaces. At low slip the relation
is approximately linear and the forces can be described as Fx = − Cxσ x ,
longitudinally and Fy = − Cyσ y, laterally. The braking stiffness, Cx and the
cornering stiffness, Cy are correspondingly defined as the linearization of
the force-slip relation at σ x = 0 and σ y = 0 [Wong, 2001]

dFx
Cx = −
dσ x σ x =0,σ y=0
(2.7)
dFy
Cy = −
dσ y σ x =0,σ y=0
At higher slip the relation is strongly nonlinear and a more complex func-
tion is necessary to express the relation, see Sections 3.1 and 4.1. The
normal behavior for a tire on asphalt is that the force increases with slip
24
2.6 Tire Deformation
up to a certain slip value (σ ∗x ) and then diminishes slightly when σ x goes

to infinity or λ reaches unity, see the measurements in Figure 5.1.
Combined slip Combined slip signifies simultaneous longitudinal and

lateral slip, i.e. braking and cornering actions performed at the same time.
A steering maneuver during braking, generally, decreases the braking
stiffness, the longitudinal peak force, and its corresponding slip value. A
model for the interaction between the slip in both directions is therefore
inevitable for more advanced vehicle simulations. In Section 3.2 and in
Chapter 5 this topic will be discussed in detail.
Camber
The camber angle, γ , denotes the tilting angle of the tire about the x-
direction, see Figure 2.3. Normally, the sign convention is related to the
inclination relative the vehicle body [ISO 8855, 1991]. The effects from
cambering are particularly important when deriving models for motorcy-
cles that produces a large part of the cornering force by tilting. For cars
and, in particular, trucks the achieved camber angles are much smaller
and in many applications their effect can be neglected. However, many
suspension designs make the wheels to camber when the axle load varies.
Elasticity in bushings, beams, bolts and axles also allow cambering during
cornering. Cambering of a tire creates a lateral force, even though there
is no lateral slip.
Rolling resistance
The deformation of the tire due to the vertical load is often clearly vis-
ible. The wheel rotation for a moving vehicle continuously changes this
deformation and power dissipates from the system due to the visco-elastic
properties of the carcass, see Section 2.6. The effect is called rolling resis-
tance and is, in general, assumed to depend linearly on the tire load [Wong,
2001] and radius, as
My = q0 Rp Fz p (2.8)
For more accurate results the dependency on the velocity has to be in-
cluded in the tire specific factor, q0 . In [Pacejka, 2002] an addition to the
resistance torque due to the extra energy loss when a longitudinal force
is developed, is proposed, as My,add = ( R e − R) Fx . The energy loss comes
from the difference between the effective rolling radius and the tire ra-
dius that works as a lever for the torque on the rim to the transmitted
tire force to the ground.
25
y x
y
z
Figure 2.3 Cambered wheel. Left: Rear view; Right: Top view with contact patch
(dashed rectangle). Note that the contact patch is greatly exaggerated in size.

One of the main tasks for the tire is to reduce the vibrations caused by
the unevennesses in the road. The elastic and damping properties are
therefore of great importance and large vertical shape deformations are
required for efficient damping. The price is slower handling and reac-
tion on maneuvers from the driver, since a softer tire, generally, allows
larger horizontal deformation. The deformation can schematically be di-
vided between the carcass and the tread. The carcass flexibility is the
major source for the dynamic behavior and the rolling resistance of the
tire. In the longitudinal direction the minor deflection of the rubber treads
determines the force-slip characteristics, while both the tread and carcass
deflections affects it in the lateral direction. A good way to illustrate the
carcass behavior is to use the stretched string carcass model that origi-
nates from the work of von Schlippe in 1941, described in [Pacejka, 2002],
which is still used for tire modeling, see for instance [Thorvald, 1998]. The
carcass is then, as the name intends, approximated as a stretched string
attached to the rim by visco-elastic springs, see Figure 2.4. Finite element
methods, allowing more sophisticated models, are usually used to derive
more accurate results for the tire deformation, since the shape is complex
and the deflections are large [Pauwelussen et al., 1999; Gipser, 2005]. The
26
Figure 2.4 Illustration of the schematic tire structure used in this thesis. The
carcass is visualized as a stiff string attached to the rim by visco-elastic springs.
The springs are only showed in the radial direction but work also tangentially and
laterally (out of the wheel plane). The same holds for the elastic tread springs.
stretched string method gives, however, a good visualization of the physics

behind the tire behavior and in many cases the resulting equations can
be solved analytically. Restrictions can be introduced to obtain results in
fairly simple and sufficiently accurate expressions that easily can be used
for estimation and control issues.
27
3
Review of Existing
Tire-Models
An extensive amount of work has been done in the field of tire modeling.
The research covers everything from derivation of simplistic expressions
describing the tire behavior to advanced finite-element methods that pre-
dicts the important states at any point of the tire. The tire modeling was
initiated by the vehicle and air craft industry back in the 1940’s. The first
descriptions of the tire characteristics were derived from physical model-
ing. Later on the interest for finding empirical equation structures, that
easily could be adjusted to fit the measurements from the tire tests in-
creased due to the complex nature of the tire. The early tire models only
covered the static force-slip relation, but the development of fast control
systems has now increased the focus on the dynamical aspects of the tire
behavior. The larger amount of computer power has enabled the possibility
to solve the physical differential equations related to the tire deformation.
Therefore, physical modeling of both dynamical and frictional aspects has
become popular again. This chapter gives a short review of the tire models
that are relevant for the continuation of the thesis.
3.1 Steady-State Tire Models
The aim of a tire model is often to obtain a structure that can fit mea-
surement data well by optimal choice of included parameters. A detailed
description of the early tire models can be found in [Nguyen and Case,
1975], from where also the following examples in this section are gath-
ered, if not otherwise is stated.
The first tire models did only concern the steady-state relation between
the slip and the developed force. They were physically derived from vari-
ants of the brush model. The most common form of the brush model is
28
extensively presented in Section 4.1 and therefore left out here, but the
different assumptions on the brush model often results in a polynomial
including at least two parameters. One example is the proposal for the
lateral force by Smiley and Horne in 1958
2

3µ Fz

C α 1 − α

α ≤ α○ =
α 2
F0y = 3(α )
○ 2Cα (3.1)
otherwise

 µ Fz
where Cα is the cornering stiffness and µ the friction coefficient. The
definition of the slip angle α is given by (2.1). The friction is assumed to
be constant in the model, which disables its ability to also characterize the
longitudinal force accurately for longitudinal slips, since there, generally,
is a pronounced peak of the force in this direction. Interpolation between
the data points was then mostly used to model the longitudinal tire force.
By use of velocity dependent friction, a relation for combined slip that
better expressed the behavior of the brake force for large slip was derived
in [Dugoff et al., 1969], given on the form
Cx λ
Fx = f (θ ) (3.2)
1−λ
Cy tan(α )
Fy = f (θ ) (3.3)
1−λ
(
θ (2 − θ ) θ ≤1
f (θ ) = (3.4)
1 θ >1
where q
µ 0 Fz((1 − ǫv λ 2 + tan2 (α ))(1 − λ ))
θ= q (3.5)
2Cx2 λ 2 + Cy2 tan2 (α )
and µ 0 is the nominal friction coefficient and ǫ is the velocity dependency
factor. This model is one of the three models usually referred to as the
HSRI-models developed at the Highway Safety Research Institute. The
other two models are described in [Fancher et al., 1972] and [Tielking and
Mital, 1974] and are based on different assumptions on the brush model
regarding the vertical pressure distribution and sliding properties of the
rubber.
An empirical way to describe the lateral force and account for com-
bined braking and cornering by using the friction circle criterion, has
been proposed by Chiesa (1965) as
s n
Fx
Fy = 1 − Fz (a1 + a2 Fz)α + (a3 + a4 Fz )α 2 + . . . (3.6)

2µ N
29
Chapter 3. Review of Existing Tire-Models
The method requires that the longitudinal force is known instead of the
longitudinal slip, compared to the previous methods. To further increase
the flexibility in the modeling Holmes (1969) proposed an empirical struc-
ture, quite different to the others, to use for curve fitting
F0y = a0 + a1 vx + a2 v2x + a3α + a4α 2 + a5α 3 + a6 R + a7 P (3.7)
where P is a tire-pattern constant and R is a tire-tread constant. The

coefficients a1...7 have no physical interpretation. The idea to treat the
dependence of the vehicle velocity as an additional contribution was not
really accepted by the other researchers in the field. Further references
regarding these early tire models can also be found in [Bernard et al.,
1975].
In [Kiencke and Nielsen, 2000] another empirical model, developed by
Burckhardt is used, which models the real tire behavior more accurately.
The input sRes denotes the resultant slip (s2L + s2S )1/2 and an additional slip
definition (compare to Equation (2.5)) is used in s L = (vc cos(α ) − vx )/vx ,
which is the slip in the travel direction of the wheel, v̄, and perpendicular
to this, sS = vc sin(α )/vx . The model can then be used for combined slip
situations and the direction of the resulting force is collinear to the slip
vector s̄Res . Nonlinear effects of the wheel load can be accounted for in
this model which is expressed as
F = (a1 (1 − e−a2 sRes ) − a3 sRes ) e−c4 sRes v (1 − c5 Fz2 ) (3.8)
A disadvantage in the approach is that the tire characteristic is assumed

to be equal in both directions. Due to the influence of the carcass flexi-
bility, which is further discussed in Section 4.3, this is generally not the
case.
A way to use similarities between longitudinal and lateral tire behav-
ior is used in [Lugner and Mittermayr, 1991], where a slip curve for a
particular tire is separated in four regions depending on the slip. Each
region has its function structure for the approximation. For low slips the
relation is linear. At slightly higher slips up to the peak a polynomial
function describes the behavior. The negative slope of the tire model is
described by another polynomial and for very high slips the tire force is
assumed to be constant. The combined-slip forces are then derived as

σx
f x (σ x , Fz) = ȳ fst1 µ x,max ( Fz ) (3.9)
fst2 s xm ( Fz )

σy
f y (σ y, Fz) = ȳ fst1 µ x,max ( Fz) (3.10)
fst2 s ym ( Fz)
30
where ȳ(⋅) is the region-separated function-approximation. The adjusted

combined slips σ x , σ y are shaped from combinations of the pure slips
λ , α and γ and do not correspond to the physical slips from (2.5), see
the original paper for a detailed description. The frictional values at the
peak, µ x,max and µ y,max and corresponding slips s xm ( Fz ) and s ym ( Fz) are
then used to set the difference in the corresponding direction. The factors
fst1 and fst2 are used to adjust between different surfaces. Similar ideas
are proposed for deriving the self-aligning torque at combined slip. The
method is claimed to have a short on-line calculation time, and to rep-
resent the tire behavior in areas without measurements in a reasonable
way.
The paper [Bakker et al., 1987] presents the “Magic Formula”, which
quickly became the most predominating model. The model expresses the
lateral and longitudinal tire forces, as well as aligning torque, on the form
h i
y( x) = D sin C arctan (1 − E) x + ( E/ B ) arctan( B x) (3.11)

where ( x, y) is (λ , F0x ), (α , F0y), or (α , M0z). The four coefficients have

interpretations as stiffness factor ( B ), shape factor ( C ), peak factor ( D ),
and curvature factor ( E), and are unique for each of Fx , Fy, and Mz .
Approximation of the normal load dependence may be introduced as
C = a0 (3.12)
D= a1 Fz2 + a2 Fz (3.13)
B = (a3 Fz2 + a4 Fz)/( C D ea5 Fz ) (3.14)
E= a6 Fz2 + a7 Fz + a8 (3.15)
After its first presentation the Magic Formula has been extended to em-
pirically express most of the interesting properties of the tire and also
the interaction of many simultaneous actions, such as combined slip. The
model has, during cooperation between TNO and TU-Delft, become a com-
mercial product under the name MF-tyre. The entire model is presented
in [Pacejka, 2002] and the commercial interest is mainly in delivering
parameter sets for certain types a tire. The complete model includes ap-
proximately 85 parameters that has to be calibrated from measurement
data. Another direction in the progress of the Magic Formula is proposed
in [Sharp and Bettella, 2003], where one of the aims is to reduce the need
of parameters in the model. The idea of using one reference model and
having two parameters prescribing the peak force as in [Lugner and Mit-
termayr, 1991] is used, but the reference curve is described by the “Magic
Formula”-structure using “normalized” parameters. The method shows an
appreciable reduction of parameters, without losing much information.
31
3.2 Combined-Slip Semi-Empirical Tire-Models

The term semi-empirical is used for models that relies on physical model-
ing to transform empirical data collected at a certain condition to be valid
for another situation at different conditions. There is always a trade-off
since the semi-empirical model is more or less accurate. The measure-
ment data on the other hand includes errors and disturbances. The testing
circumstances also mostly differ from the driving or simulation circum-
stances. In some cases collection of data can be difficult, unreliable, or
very expensive and a theoretical model is inevitable.
Semi-empirical methods are often used to derive the tire forces at com-
bined slip situations, when the pure-condition tire-forces often are avail-
able from test-bench experiments.
The most simplistic model of combined forces is based on the friction
ellipse concept, see for instance [Wong, 2001; Ellis, 1994; Nielsen and
Eriksson, 1999]. While the friction ellipse is the envelope of the maximum
achievable forces, the ellipse is here used also for modeling intermediate
forces. It is used to compute a combined lateral force Fy(α , λ ) at a given
longitudinal force Fx , and is based on the assumption
2 2
Fy(α , λ )

Fx
∗
+ =1 (3.16)
F0x F0y(α )
where F0x∗ is the maximum achievable longitudinal force, and F (α ) de-

0y
scribe the lateral force at pure slip. The combined lateral force becomes
s 2
Fx
Fy(α , λ ) = F0y(α ) 1− ∗
(3.17)
F0x
An objection to this model is the assumption (3.16), which is not true,

since adhesion limits are not necessarily fully reached for combined forces
in the interior of the friction ellipse.
Another simple model is the Kamm Circle [Kiencke and Nielsen, 2000],
where the resultant force magnitude is described as a function of the total
slip magnitude. The force and slip vectors are then assumed to be collinear,
possibly with a corrective factor ks :
sx sy
Fx = F (s) and Fy = ks F (s) (3.18)
s s
A drawback with this model is that longitudinal and lateral characteristics

are assumed to be the same, modulo the corrective factor.
32
3.2 Combined-Slip Semi-Empirical Tire-Models
Some early efforts to model tire forces under combined-slip conditions

are described and compared in [Nguyen and Case, 1975]. One of the most
well-known is presented in [Nicholas and Comstock, 1972]:
Fx (λ ) Fy(α )λ
Fx (λ , α ) = q
λ 2 Fy2 (α ) + tan2 (α ) Fx2 (λ )
(3.19)
Fx (λ ) Fy(α ) tan(α )
Fy(λ , α ) = q
λ 2 Fy2 (α ) + tan2 (α ) Fx2 (λ )
In [Brach and Brach, 2000] this model is shown to give incorrect result
for small slips and a modified version is presented.
In [Bakker et al., 1987], a procedure for computing combined forces
for the Magic Formula is presented. It is essentially a refinement of the
Kamm Circle for non-isotropic tire characteristics and a normalization of
the slips to guarantee simultaneous sliding. The normalized slip
s 2 2
σx σy
σN = + (3.20)
σ ∗x σ ∗y
is an entity that is less than one for non-sliding conditions. It is based on

an elliptic assumption where σ ∗x and σ ∗y are the longitudinal and lateral
slips that corresponds to full sliding for pure slips, normally taken as
the slips at the peak values F0x ∗ , and F ∗ . Now the combined forces are
0y
computed as
Fx = − cos(β ○ ) F0x (σ ∗xσ N ) and Fy = −ǫd (σ N ) sin(β ○ ) F0y(σ ∗yσ N )

(3.21)
J
with tan(β ○ ) = (σ y /σ ∗y )/(σ x /σ ∗x ). For large slip conditions the factor ǫd (σ N )
must be included to give correct direction of the resulting forces. The rea-
son is the fact that for small slips real tire-forces are essentially produced
by elastic deformation, and for large slips by sliding friction. Therefore
slip vectors of the same orientation but different magnitudes may result
in forces with different orientation. It is not clear how to determine ǫd (σ N ),
and in [Bakker et al., 1989] a modified procedure was presented:
Fx = cos ((1 − ϑ )β ○ + ϑ β ) F0x

′
and Fy = sin ((1 − ϑ )β ○ + ϑ β ) F0y
′
(3.22)
J 2
with ϑ = π arctan( q1σ 2N ) and
J
′
F0x = F0x (σ N ) − sat(σ N ) ( F0x (σ N ) − F0y(σ N )) sin2 (β ○ )
(3.23)
J
′
F0y = F0y(σ N ) + sat(σ N ) ( F0x (σ N ) − F0y(σ N )) cos2 (β ○ )
33
The variables ϑ , F0x′

, and F0y
′
describe the gradual change of orientation
of the resulting force from adhesion to sliding. At large slip-magnitudes
the force is collinear with the slip vector.
In [Bayle et al., 1993] a model for combined braking and cornering
is presented, which is based on functional representation. The model is
much inspired by the Magic Formula and uses functions based on arc
tangents to describe forces under combined-slip conditions. These ideas
were used in the development of the MF-tyre model, see Section 3.1, and
is one of the differences between the descriptions in [Bakker et al., 1989]
and [Pacejka, 2002].
A method that shares ideas proposed in [Lugner and Mittermayr,
1991], where a common function describes the tire behavior in each di-
rection is described in [Radt, 1995]. The method also uses so called non-
dimensionalising of the pure-slip tire behavior. The non-dimensionalised
entities are here marked by bars. The normalised combined slip is defined
as p λ Cx α Cy
k = λ̄ 2 + ᾱ 2 ; λ̄ = ᾱ = (3.24)
µ x Fz µ y Fz
and a normalised “resultant” force as
Fx Fy
q
R( k) = F̄x2 + F̄y2 ; F̄x = F̄y = (3.25)
µ x Fz µ y Fz
The tire-force function R( k) is chosen to give most accurate fit for both
the longitudinal and lateral directions. The combined slip tire forces are
then proposed to be computed as
η ( k) R( k) tan (α ) R( k)λ
F̄y = q ; F̄x = q ; (3.26)
λ 2 + η 2 ( k) tan2 (α ) λ 2 + η 2 ( k) tan2 (α )
where
(
0.5(1 + η 0 ) − 0.5(1 − η 0 ) cos(0.5k) p kp ≤ 2π
η ( k) = (3.27)
1 p kp > 2π
This method has also been used and further developed in [Sharp, 2004].
The COMBINATOR model [Schuring et al., 1996; Pottinger et al., 1998]
is another variation on the Kamm Circle. Here the tire force magnitude
is described by
F = F0x (s) cos2 (β ) + F0y(s) sin2 (β ) (3.28a)
and the combined forces as
Fx = F cos(β ) and Fy = F sin(β ) (3.28b)
34
3.3 Transient Tire-Models
The model assumes collinearity between resulting force and the slip vector.
Further references regarding Semi-empirical tire-modeling may, for
example be found in [Böhm and Willumeit, 1996].
3.3 Transient Tire-Models

The dynamic behavior of the tire has lately gained a large interest. In
applications such as state-of-the-art active safety systems that intervene
in limited conditions by controlled braking actions the transient proper-
ties of the tire affects the vehicle dynamics. Tire forces, in general, do
not develop instantaneously at maneuvering actions, but require a cer-
tain rolling distance of the tire to build up, due to the flexible structure.
This is of practical significance in, for example, systems for control of lon-
gitudinal tire slip. The motion sensors are, here, attached to the wheel
rim, but it is the tire motion relative road contact that is of interest for
the force generation. A simple way to model this difference, was described
and validated in [Clark et al., 1971], but proposed already by von Schlippe
around thirty years earlier. The method lets a first-order transfer func-
tion describe the relation between the measured slip, λ ′ , and the slip that
generates the tire forces, λ , such that
σ ax dλ
= λ − λ′ (3.29)
vx dt
The time constant of the system σ ax /vx is then speed dependent and equal
to the time for the wheel to roll a specific distance, called relaxation length,
see Section 4.3 and Equations (4.80)–(4.82). Same approach can be ap-
plied in the lateral direction.
A different concept is employed in the LuGre model. The LuGre friction
model is known for describing special cases of friction situations and was
developed as a joint cooperation between the Department of Automatic
Control at Lund University (Sweden) and Laboratoire d’Automatique de
Grenoble (France) [Olsson, 1996]. The model describes a dynamic force
phenomenon that arises when frictional surfaces are sliding on each other.
In the formulation for tires the LuGre model assumes the frictional sur-
face to consist of bristles, with movements described by differential equa-
tions. An assumption of the vertical pressure distribution between the tire
and road is necessary to state an expression for the steady-state force-slip
relation. This is used to calibrate the included parameters to fit measure-
ment data. The deflections of bristles are then lumped together in one
variable and in [Velenis et al., 2002] the LuGre model is written, for two
dimensions, on the following form
35
σ 0i λ (vr )
z̄˙i = vri − z̄i − κ i (t)pω rp z̄ (3.30)
µ 2ki
Fi = − Fn (σ 0i zi (t) + σ 1i żi (t) + σ 2i vri )), i = x, y (3.31)
where
pp Mk2 vs pp
λ (vs ) = (3.32)
(vs )
and the friction function is redefined as
pp Mk2 vs pp
2
p Mk vs p p Mk2 vs p pvs p γ

(vs ) = + − exp − (3.33)
pp Mk vs pp p Mk vs p p Mk vs p vst
" # " #
µ kx 0 µ sx 0
Mk = Ms = (3.34)
0 µ ky 0 µ sy
and the parameters from the distributed steady-state description can then
be used in the lumped formulation. The derivation of κ , from (3.30) has to
be done to cover for the difference between the distributed and the lumped
formulation. More details on the model are given in [Velenis et al., 2002].
A drawback with the method is that the description of the bristle dy-
namics is complicated, resulting in comprehensive equations, while other
phenomenon are roughly simplified. The major source of the tire dynamics
still comes from the carcass behavior.
The Short Wavelength Intermediate Frequency Tyre (SWIFT) model
described, for instance, in [Pacejka, 2002] developed at TU-Delft and TNO-
Helmond is a collection of methods employed to extend the MF-tire concept
with dynamic properties and ability to handle uneven roads. The model
can handle frequencies up to 60 Hz and wavelengths larger than 0.2 m
and acts on longitudinal and lateral slip, camber, and turn slip. The tire
belt is modelled as a rigid ring with inertia. The ring has a flexible at-
tachment to the rim, with stiffnesses in all directions, compare to the
stretched string model in Section 2.6. To improve the transient properties
further, the stiffness in the contact patch rubber is modelled as a first or-
der differential equation with a time constant corresponding to the size of
the adhesion area in the patch (see Section 4.1 for explanation of adhesive
area). The pneumatic trail, used for the self-aligning torque is modelled
as a second order differential function from the lateral slip, which means
that self-aligning torque builds up slower than the lateral force. The turn-
slip torque-contribution to Mz on the rigid ring is modelled as a fourth
order polynomial function relative the turn slip in the contact patch. To
calculate the position for the contact patch when running on an uneven
36
3.4 Finite Element Models
road, the patch is described by cams, see Figure 3.1. The contact between
the road and the cam decides the vertical position of the contact patch.
To be able to manage inclination of the surface, two cams (tandem) are
necessary. Four cams (two rows of tandem cams) are needed to derive
camber angles due to road inclination in the lateral direction.
Figure 3.1 Illustration of the method of the SWIFT-model to derive the vertical
position for the contact point between the tire and the road. To the left single cam
and to the right tandem cams allowing determination of patch inclination.
3.4 Finite Element Models

The increase of available computation power has provided a new platform
for use of finite element methods to solve the complex partial differen-
tial equations describing the behavior of the tire. Briefly, this means that
the tire structure is divided into small elements. Each element has a set
of differential equations describing its physical behavior. The elements
are connected to each other through boundary surfaces. All elements con-
tains points or nodes, with constraints on the externally applied force, or
deflection. The constitutive conditions and laws of equilibrium of forces,
torques, and energy flow are used to derive matrix equations from which
the states in each node are solved. The finer meshing (small elements)
the larger matrices and memory consumption. The F-tire model [Gipser,
2005] uses between 80 and 200 sector elements to describe the carcass
deformation. Each element has five degrees of freedom denoting, longi-
tudinal, lateral and vertical displacement, rotation, and bending. Above
that, between 5 and 10 tread elements on each carcass element are used to
model the rubber treads. Each tread element contains information about
wear, temperature, and deflection. Since the tread elements outside the
37
contact area are unloaded they will not be included in the calculations
of the deflections. For a normal PC (2005) the calculations takes about
2 to 10 times real time. The model properties are determined by sets of
parameters, which can be calibrated from measurement data. F-tire is the
middle-complexity part in a family of three models. The simpler R-tire is
restricted by having a rigid carcass which is similar to the SWIFT-model
in Section 3.3. The most advanced model FE-tire is a coarse mesh finite
element tire model allowing multiple elements in the lateral direction.
The Rmod-K [Oertel and Fandre, 1999] is another tire model that
similarly to the FE-tire is based on coarse-mesh finite-elements.
3.5 Conclusions
Much research is devoted to tire-modeling related issues and the brief

survey above does by no means cover the area. The aim is more to show
on the variety and width of the expressions and approaches that have
been proposed in the literature than to try to give a complete view of the
results presented in this area.
When choosing among the tire models it is important to consider the
purpose of the particular application. If the time constraints are suffi-
ciently long such that steady-state model can be used or which fidelity
requirement should be set on the description of the dynamics. For vibra-
tional and deformation analysis at high frequencies the choice of a finite
element method is often inevitable, while some vehicle handling analysis
situations may be well described without any dynamic effects. Using a too
comprehensive model is inefficient use of time and computation resources.
If only general behavior of a tire is needed in a certain application, a very
simple physical model might be used. For a requiring algorithm calibra-
tion a more accurate model may be needed and an empirical method might
be good choice. It is then necessary to be convinced that the actual param-
eter sets are derived from representable test conditions and that changed
conditions have a large effect on the tire characteristics.
38
4
Tire Modeling According to
the Brush Model Theory
The objective of this chapter is to explain the underlying physics for the
derivation of the semi-empirical tire-model presented in Chapter 5. The
main prerequisite for the generation of the forces in the contact patch is
the presumption of the brush-model theory assuming a stiff carcass, which
is described in the first section. The next section explains the effect of
camber on the forces, and torque. Finally, modeling of some consequences
due to the elasticity of the tire carcass, that can enhance the performance
of the model, are discussed. In particular, the dynamic properties of the
development of the tire forces are examined. The main part of the contents
in this chapter is gathered from previous publications by other authors and
rewritten with a slightly new touch.
4.1 Basics of the Brush Model
The brush model is a well-known approach to model tire forces, see e.g.
[Dugoff et al., 1969], [Pacejka, 1988], or [Wong, 2001]. The model was very
popular in the 1960’s and 1970’s before the empirical approaches became
dominating and describes the physics behind the tire behavior in an educa-
tional way. In this section the brush-model concept is applied to combined
slips, much like the approach of [Gim and Nikravesh, 1991]. The brush
model describes the generation of tire forces based on partitioning of the
contact patch into an adhesion and a sliding region. Forces in the adhe-
sive region are assumed to be caused by elastic deformations in the rubber
volume that is between the tire carcass and the ground. The carcass is
assumed to be stiff, which means that effects of carcass deformation are
neglected. In the sliding region, forces are caused by sliding friction.
The model is obtained by dividing the rubber volume in the contact
39
Chapter 4. Tire Modeling According to the Brush Model Theory
Fz
Ca
rca
ss vsx
Ru
bb
er
Road Slide Adhesion
x
−a xs 0 a
vsy
δ xb ( x)
y δ yb ( x)
Figure 4.1 The deformation of the rubber layer between the tire carcass and the
road according to the brush model. The carcass moves with the velocity (vsx , vsy)
relative the road. The contact zone moves with the vehicle velocity (vx , vy = vsy).
The break-away point, xs , is the coordinate that separates the sliding and adhesive
areas. (Top: side view; Bottom: top view)
region into small brush elements. Each element stretches laterally ( y-

direction) over the entire contact region, but their length is infinitesimal
in the longitudinal, x, direction. The elements are regarded as elastic rect-
angular blades, or bristles, see Figure 4.1. Even though rubber, in general,
is not linearly elastic, this assumption is made in the brush model. Posi-
tions in the contact region are expressed in a reference system attached
to the carcass, with the origin located in the center of the contact region.
The length of the contact region is 2a. Each bristle is assumed to deform
independently in the longitudinal and lateral directions. In the adhesive
region the bristles adhere to the road surface and the deformation force is
carried by static friction. In the sliding region the bristles slide on the road
surface under influence of sliding friction. Hence, in the sliding region the
resulting force is independent of the bristle deformations.
40
Adhesive bristle forces Regard the specific infinitesimal bristle which

is attached to the carcass at position x relative the origin in the center
of the contact patch. Assume that this bristle belongs to the adhesive
region. The bristle is in contact with the road surface at position xr ( x),
yr ( x), see Figure 4.2. Since there is no sliding in the adhesive region the
contact-point position may be described by
Z t c ( x)
xr ( x ) = a − vx dt
0
Z t c ( x)
(4.1)
yr ( x) = − vy dt
0
where tc ( x) is the time elapsed since the bristle entered the contact re-
gion. The velocities vc , vx and vy are assumed to be constant as a bristle
travels through the adhesive region of the contact patch, i.e. during the
integration interval [0, tc ( x)]. Hence, the bristle position is x = a − vc tc ( x),
and tc ( x) = (a − x)/vc . The deformation becomes
δ xs ( x) = xr ( x) − x
(4.2)
δ ys ( x) = yr ( x)
Insertion of (4.1) and the expression for tc ( x) yields
vx − vc
δ xs ( x) = − (a − x) = −σ x (a − x)
vc
vy (4.3)
δ ys ( x) = − (a − x) = −σ y (a − x)
vc
where the slip definition from (2.5) is used in the last equality. As the
carcass is assumed to be stiff the the bristle deformations are accordingly,
δ xb = δ xs and δ yb = δ ys . With the assumption of linear elasticity, the
deformation force working on the bristles becomes
dFax ( x) = cpx dx δ xb ( x)
(4.4)
dFay( x) = cpy dx δ yb ( x)
where cpx and cpy are the longitudinal and lateral bristle stiffnesses per
unit length. The assumption of constant vc , vx , vy in the interval [0, tc ( x)]
is relaxed to the assumption of slow variations in σ x and σ y with respect
to the duration 2a/vc , which is the maximum time for a bristle to travel
through the adhesion region. The total adhesive tire force is computed by
41
yr ( x) x
xr ( x)
x
Figure 4.2 The deformation of a bristle element in the contact patch. Compare
with Figure 4.1.
integration of (4.4) over the adhesive region. With (4.3) this gives
Z a Z a
Fax = dFax ( x) = −cpxσ x (a − x) dx
x x
Z sa Z as (4.5)
Fay = dFay ( x) = −cpyσ y (a − x) dx
xs xs
where xs is the position in the contact patch which divides the adhesive
and sliding regions. To compute the total adhesive force it is necessary to
know xs .
The size of the adhesion region The size of the adhesive region is
determined by the available static friction. The deformation will be limited
by the largest force that can be carried by the static friction between the
tire and the road. The static friction is assumed to be anisotropic (i.e.
depending on the sliding direction) with the friction coefficients µ sx and
µ sy, respectively. With a normal force dFz ( x) acting on the infinitesimal
bristle at position x, the available static friction force is described by the
elliptic constraint
2 2
dFax ( x) dFay( x)
+ ≤1 (4.6)
dFz ( x)µ sx dFz ( x)µ sy
As a result, the magnitude of the available static friction force is depen-

dent of the direction of the deformation force dF̄a ( x), defined by (4.4).
The static friction constraint is illustrated in Figure 4.3. When dF̄a ( x)
exceeds the static friction constraint the bristle will leave the adhesive
42
µ sy dFz( x)
dFy( x)
dF
dFx ( x)
µ sx dFz( x)
Figure 4.3 Illustration of the elliptic static friction constraint at anisotropic fric-
tion and rubber characteristics. Note that the direction of dF̄ ( x) and σ̄ is equal only
if c px /c py = µ sx /µ sy.
region and start to slide. Introduce the pressure distribution qz( x), with
dFz ( x) = qz( x) dx. By combining (4.3) and (4.4) with (4.6) the static fric-
tion constraint may be written as
s 2 2
cpxσ x cpyσ y
+ (a − x) ≤ qz ( x) (4.7)
µ sx µ sy
The position xs in the contact area is the break-away point where the static
friction limit is reached and the bristles starts to slide. If the pressure
distribution qz( x) is known then xs can be calculated by setting equality
in (4.7) with x = xs .
A common assumption is to describe the pressure distribution in the
contact patch as a symmetric parabolic function:
3Fz x 2

qz ( x) = 1− (4.8)
4a a
This is proposed, for example, in [Tielking and Mital, 1974] and has shown
to give a good agreement with experimental longitudinal force-slip curves
for real tires. In [Svendenius, 2003] the influence of the pressure distribu-
tion is further examined. A more realistic pressure distribution is obtained
by moving the pressure-peak slightly forward in the contact patch. The
gain in accuracy is, however, small compared to increase in complexity of
the resulting formulas and it can be concluded that the proposed parabolic
assumption is sufficient for the applications related to the brush model in
this work. Inserting (4.8) in (4.7) with equality gives
s 2 2
cpxσ x cpyσ y 3Fz
+ ( a − xs ) = ( a − xs ) ( a + xs ) (4.9)
µ sx µ sy 4a3
43
The solution for the break-away point xs is then

s 2 2
3
4a

cpxσ x cpyσ y
xs (σ x , σ y) = + −a (4.10)
3Fz µ sx µ sy
Since xs is a point in the contact patch it must belong to the interval

[−a, a]. If xs = a the entire contact patch is sliding. In the case of pure
slip, i.e. either σ x or σ y is zero, this will occur at the limit slips σ x = σ ○x
or σ y = σ ○y with σ ○x and σ ○y given by the following
J 3Fz µ sx J 3Fz µ sy
σ ○x = ; σ ○y = (4.11)
2a2 cpx 2a2 cpy
Introduction of normalized slips with respect to the limit slips will simplify
the notation in the following. The normalized slip is defined as
s 2 2
J σx σy
ψ (σ x , σ y) = + (4.12)
σ ○x σ ○y
Equation (4.10) may now be rewritten as
xs (σ x , σ y) = (2ψ (σ x , σ y) − 1)a (4.13)
It is clear that partial sliding occurs when ψ (σ x , σ y) < 1. At full sliding

then ψ (σ x , σ y) ≥ 1 and Fax (σ x , σ y) = Fay(σ x , σ y) = 0. In the following
the construction of adhesive and sliding forces at partial sliding will be
determined.
Adhesion force Knowing the size of the adhesive region, xs (σ x , σ y),

given by (4.13), the adhesive forces are obtained by solving the integrals
(4.5) yielding
2
Fax (σ x , σ y) = −2a2 cpxσ x (1 − ψ (σ x , σ y))
(4.14)
Fay(σ x , σ y) = −2a2 cpyσ y (1 − ψ (σ x , σ y))2
for ψ (σ x , σ y) < 1. At full sliding, ψ (σ x , σ y) ≥ 1, the adhesion forces be-

comes Fax (σ x , σ y) = Fay(σ x , σ y) = 0. Special notations for the forces at
pure slip are introduced as
J
F0ax (σ x ) = Fax (σ x , 0)
(4.15)
J
F0ay(σ y ) = Fay(0, σ y)
44
µ sx qz( x)
σ x c px σ x○ c px
Fax (σ x , 0)
−a xs (σ x , 0) 0 a x
Figure 4.4 Illustration of the adhesive tire-force for pure longitudinal slip. The
elastic deformation force for an element at x in the adhesive region depends linearly
on x as c pxσ x ( x − a), where the slope is proportional to the slip σ x . The transition
from adhesion to slide occurs at the intersection of the lines at the break-away point
xs . For slips σ x > σ x○ full sliding occur in the contact area since there is then no
intersection.
Note that it follows from (4.4) and (4.3) that the produced adhesive force
per unit length in the adhesion region is not affected by combined slips:
dFax (σ x , x)
= −cpxσ x (a − x)
dx
(4.16)
dFay(σ y , x)
= −cpyσ y (a − x)
dx
The adhesive forces thus grow linearly with slopes cpxσ x and cpyσ y as the
contact element moves into the adhesion region. To illustrate the gener-
ation of the adhesive force the case of pure longitudinal slip is regarded,
i.e. σ y = 0. From (4.7) the size of the contact region is determined by
the point where cpxσ x (a − x) = µ sx qz( x). That is, where the straight line
describing the produced force per unit length intersects µ sx qz( x), as is
shown in Figure 4.4. The striped area under the line corresponds to the
total adhesion force. The slope corresponding to full sliding, i.e. σ x = σ ○x ,
is also shown. The case of pure lateral slip is analogous.
45
Combined-slip slide forces The normal force acting on the sliding

region at partial sliding may be computed from (4.8) and (4.13) as
Z xs (σ x ,σ y )
Fsz(σ x , σ y) = qz ( x) dx = Fzψ 2 (σ x , σ y) (3 − 2ψ (σ x , σ y)) (4.17)
−a
In case of isotropic sliding friction with the friction coefficient µ k ,

the friction force is collinear with the slip velocity with the magnitude
Fsz (σ x , σ y)µ k . Its components are given by
Fsx (σ x , σ y) = − cos (β )µ k Fsz(σ x , σ y)

(4.18)
Fsy(σ x , σ y) = − sin (β ) µ k Fsz (σ x , σ y)
where β is defined by (2.4). Assumptions on isotropic sliding friction are

common in tire modeling, see for instance [Schuring et al., 1996].
If the sliding-friction is anisotropic with the different friction coeffi-
cients µ kx and µ ky, there are several ways to calculate the magnitude and
the direction of the resulting force. Three different methods are presented
in the following and which one to choose depends on the assumptions
made on the friction behavior for the actual case.
Collinear slide forces This method should be used if the friction be-
tween two surfaces is supposed to be isotropic, but the values of µ kx and
µ ky are unequal. A reason for that could, for instance, be errors in the
measurements of the pure-slip forces. The friction forces are given by
Col
Fsx (σ x , σ y) = − cos (β ′ )µ kx Fsx (σ x , σ y)
Col
(4.19)
Fsy (σ x , σ y) = − sin (β ′ )µ ky Fsy(σ x , σ y)
where β ′ is defined as
−1
J µ ky vsy
tan(β ′ ) = (4.20)
µ kx vsx
The choice of β ′ ensures that F̄s acts in the opposite direction to the
sliding motion, with a friction coefficient that is somewhere in the interval
[µ kx , µ ky] depending on the sliding angle β .
Maximum dissipation rate The correct way to treat anisotropic fric-

tion according to the literature is to apply the Maximum Dissipation Rate
(MDR) principle. The theory which is further presented in [Goyal, 1989]
claims that the resulting sliding-friction force F̄sMDR is generated in such
46
a way that the mechanical work W = −v̄s ⋅ F̄sMDR is maximized under the
constraint !2
MDR 2 MDR
Fsx Fsy
+ ≤1 (4.21)
Fsz µ kx Fsz µ ky
This results in the sliding forces
MDR µ 2kx vsx

Fsx (σ x , σ y) = − p Fsz (σ x , σ y) = −µ kx cos (β SPM ) Fsz
(µ kx vsx )2 + (µ kyvsy )2
MDR
µ 2kyvsy
Fsy (σ x , σ y) = − p Fsz (σ x , σ y) = −µ ky sin (β SPM ) Fsz
(µ kx vsx )2 + (µ kyvsy )2
(4.22)
where β SPM is defined as
J µ ky vsy
tan(β SPM ) = (4.23)
µ kx vsx
The angle of the resulting force F̄sMDR is denoted by β MDR and is given by
2
µ ky vsy
tan(β MDR ) = (4.24)
µ kx vsx
Slip-projection method An intermediate approach to model aniso-

tropic sliding friction is to simply replace µ k in (4.18) with µ kx and µ ky in
the corresponding directions:
SPM
Fsx (σ x , σ y) = − cos (β ) µ kx Fsz(σ x , σ y)
SPM
(4.25)
Fsy (σ x , σ y) = − sin (β )µ ky Fsz(σ x , σ y)
This means a projection of the pure-slip sliding-forces on the slip vector.

The angle of the resulting force is then equal to β SPM . From the definitions
of β (2.4), β SPM (4.23), and β MDR (4.24), it is clear that the direction of
F̄sSPM is between the directions of F̄sCol and F̄sMDR , see Figure 4.5.
To summarize: The sliding forces are described by
Fsx (σ x , σ y) = − cos(β f )µ kx Fsz (σ x , σ y)

(4.26)
Fsy(σ x , σ y) = − sin(β f )µ ky Fsz (σ x , σ y)
with
Fsz(σ x , σ y) = Fzψ 2 (σ x , σ y) (3 − 2ψ (σ x , σ y)) (4.27)
47
FsSPM
FsCol FsMDR
β SPM
β MDR
β
µ ky Fsz
µ kx Fsz
σ̄
Figure 4.5 Illustration of methods to describe kinetic friction in case of different

longitudinal and lateral friction coefficients.
and β f is any of β ′ (collinear), β (slip-projection) or β SPM (MDR) depend-

ing on choice of friction model:
!−1
J µ ky vsy J vsy J µ ky vsy
tan(β ′ ) = ; tan(β ) = ; tan(β SPM ) = (4.28)
µ kx vsx vsx µ kx vsx
In the special case of pure-slip the sliding-forces are
F0sx (σ x ) = −µ kx Fsz(σ x , 0) sgn(σ x )

(4.29)
F0sy(σ y) = −µ ky Fsz(0, σ y) sgn(σ y)
In Figure 4.6 the case of pure longitudinal slip is again regarded, now with
also the sliding force introduced. Since qz( x) is the normal force per unit
length, the sliding force per unit length is simply µ kx qz ( x), as marked in
the figure. The horizontally striped area corresponds to the total sliding
force.
Effects of combined slips The total tire force is given by adding the
adhesive forces of (4.14) and the sliding forces of (4.26):
Fx (σ x , σ y) = Fax (σ x , σ y) + Fsx (σ x , σ y)
(4.30)
Fy(σ x , σ y) = Fay(σ x , σ y) + Fsy (σ x , σ y)
48
µ sx qz( x)
µ kx qz( x)
σ x c px
Fax (σ x , 0)
Fsx (σ x , 0)
−a xs (σ x , 0) 0 a x
Figure 4.6 Illustration of partition of the contact area into a sliding and an ad-
hesive region for the case of pure longitudinal slip. The slide force for an element at
x is determined by the pressure distribution µ kx qz( x) dx. The horizontally striped
area is the total slide force.
To illustrate the effect of combined slips Figure 4.7 shows the produc-
tion of longitudinal force in the case of combined longitudinal and lateral
slip (σ x , σ y) with σ x ,= 0, σ y ,= 0. Equation (4.13) shows that the ad-
hering region shrinks compared to the case with pure slip (σ x , 0). The
sliding region grows accordingly. From (4.16) it is clear that the adhesive
force per unit length is the same for the combined slip (σ x , σ y) as for the
pure-slip (σ x , 0). Hence, the slope is the same, but the area corresponding
to the force is smaller since the adhering region is smaller. The corre-
sponding adhesive-force slope derived from (4.16) is cpxσ ○xψ (σ x , σ y). The
corresponding expression applies for the lateral force. It is therefore clear
that sliding will occur simultaneously in both directions as ψ (σ x , σ y) ap-
proaches unity. It is important to note that the indicated area under the
pressure distribution no longer corresponds to the resulting sliding force.
Instead it describes µ kx Fsz (σ x , σ y), which is the force that would result
for pure longitudinal sliding with the sliding region xs (σ x , σ y). This force
must be limited by a friction constraint according to Section 4.1.
The braking and cornering stiffnesses are the linearizations of the
pure-slip friction curves at small slips and may be computed by derivation
49
µ sx qz( x)
µ kx qz( x)
σ x c px σ x○ψ (σ x , σ y)c px σ x○ c px
µ kx Fsz(σ x , σ y) Fax (σ x , σ y)
−a xs (σ x , 0) xs (σ x , σ y) 0 a x
Figure 4.7 Illustration of the effect of combined slip. The combined-slip has the
effect of decreasing the size of the adhesive region, compare with Figure 4.6.
of (4.30):

Fx (σ x , 0)
Cx = − = 2cpx a2
σ x

σ =0
x (4.31)
Fy(0, σ y)
Cy = − = 2cpy a2
σ y

σ y =0
Self-aligning torque The self-aligning torque consists of two parts.

The main part is Mz′ , which is the torque developed by the non symmetric
distribution of the lateral force Fy. The additional part Mz′′ is due to the
deformation of the tire.
The torque dMz′ developed at position x in the contact region is
dMz′ ( x) = dFy ( x) x (4.32)
In the adhesive part of the contact region the expression for dFy ( x) is
given by (4.4) together with (4.3). In the sliding zone it is given by differ-
entiating (4.26) using dFz ( x) = qz ( x) dx from (4.8). Integration over the
50
adhesive and sliding area separately gives
Z a
′
Maz (σ x , σ y) = −cpyσ y x (a − x) dx
xs (σ x ,σ y )
2
= −cpy a3σ y (1 − ψ (σ x , σ y))2 (4ψ (σ x , σ y) − 1) (4.33)
3
Z xs (σ x ,σ y )
′
Msz (σ x , σ y) = −µ ky sin (β ) x qz( x) dx
−a
= −3µ kx sin (β )aFzψ 2 (σ x , σ y)(1 − ψ (σ x , σ y))2 (4.34)
Mz′ (σ x , σ y) = Maz
′
(σ x , σ y) + Msz
′
(σ x , σ y) (4.35)
When there is a lateral slip the tire deflects laterally and the point
of action for the longitudinal force will have an offset from the central
plane of the wheel. This produces an additional deformation torque in the
z-direction. A longitudinal deflection together with a lateral force has the
same effect. Since it is assumed that the carcass is stiff the deformation is
here described by bristle deflections, see also Section 4.3. The deformation
torque developed at position x in the contact region is described by
dMz′′ ( x) = dFy ( x)δ xb ( x) − dFx ( x)δ yb ( x) (4.36)
In the same way as above, integration over the adhesive and the sliding
regions is performed separately. The deformation δ xb ( x) is computed from
(4.3) in the adhesive region and from (4.4) using the infinitesimal sliding
force in the sliding region. Hence
Z a
′′
Maz (σ x ,σ y) = cpyσ y(a − x)σ x (a − x) dx
xs (σ x ,σ y)
Z a
− cpxσ x (a − x)σ y(a − x) dx
xs (σ x ,σ y )
(4.37)
4
= ( Cy − Cx )aσ xσ y(1 − ψ (σ x , σ y))3
3
4 1 1

a
= − Fax (σ x , σ y) Fay(σ x , σ y)
3 Cx Cy (1 − ψ (σ x , σ y))
51
Z xs (σ x ,σ y)
1
′′
Msz (σ x , σ y) = µ ky sin (β f ) qz ( x)µ 2kx cos (β f ) qz( x) dx
−a c px
Z xs (σ x ,σ y)
1
− µ kx cos (β f ) qz( x)µ ky sin (β f ) qz ( x) dx
−a cpy
6 1 1

= − µ kx µ ky a sin (β f ) cos (β f ) Fz2 (4.38)
5 Cx Cy
⋅ ψ 3 (σ x , σ y)(10 − 15ψ (σ x , σ y) + 6ψ 2 (σ x , σ y))
6 1 1 a(10 − 15ψ (σ x , σ y) + 6ψ 2 (σ x , σ y))

= −
5 Cx Cy ψ (σ x , σ y)(3 − 2ψ (σ x , σ y))2
⋅ Fsx (σ x , σ y) Fsy (σ x , σ y)
where (4.14) and (4.26) have been used in the last step. The total addi-
tional torque is
Mz′′ (σ x , σ y) = Maz
′′
(σ x , σ y) + Msz
′′
(σ x , σ y) (4.39)
and finally the total self-aligning torque is
Mz (σ x , σ y) = Mz′ (σ x , σ y) + Mz′′ (σ x , σ y) (4.40)
A commonly used parameter is the pneumatic trail, which denotes the

distance between the center of the tire and point of action for the lateral
force. It is defined as t(σ x , σ y) = Mz(σ x , σ y)/ Fy(σ x , σ y). The coordinate
for the point of action for the adhesive force is denoted by ta (σ x , σ y) and
for the sliding force by ts (σ x , σ y). By using (4.33) and (4.14) respectively
(4.34) and (4.26) the contributions from Mz′ (σ x , σ y) to the pneumatic trail,
t′a (σ x , σ y) and t′s (σ x , σ y), are given by
′
Maz (σ x , σ y) a
t′a (σ x , σ y) = = (4ψ (σ x , σ y) − 1)
Fay(σ x , σ y) 3
(4.41)
M ′ (σ x , σ y) (1 − ψ (σ x , σ y))2
t′s (σ x , σ y) = sz = −3a
Fsy (σ x , σ y) (3 − 2ψ (σ x , σ y))
The contributions from Mz′′ (σ x , σ y) can be read directly from (4.37) and
(4.38).
In the same way as for the braking and cornering stiffness, the aligning
stiffness is defined as

Mz 2 a
Cz = = cpy a3 = Cy (4.42)
σ y 3 3

σ x ,σ y=0
52
40
−F [kN]
20
x 0
0 20 40 60 80 100
40
−F [kN]
20
y
0
0 20 40 60 80 100
1
M [Nm]
0.5
z
−0.5
0 20 40 60 80 100
λ [%]
Figure 4.8 Tire forces as function of λ with α = [0, 5, 10, 20] deg. The dotted line
shows the force from the adhesive region, the dashed line shows it from the sliding
region. The solid line is the total force and the dashed-dotted line is the reference
curve generated from a Magic Formula approximation of real tire data [Gäfvert
and Svendenius, 2003]. For the self aligning torque the dotted line denotes the
deformation torque Mz′′ and the dashed line M ′ . The slip definition λ is used since it
is most common when data is visualized. Refer to Equation (2.6) for transformation
between λ and σ x .
Properties of the brush-model

Figures 4.8 and 4.9 show the behavior of the brush model at pure and
combined slip. The pure slip curves are compared to a Magic-Formula
parametrisation of a classified truck tire. The brush-model parameters are
chosen so that the pure-slip curves have the same braking and cornering
stiffnesses and the same peak force as of the Magic-Formula reference-
curve. The length of the contact-patch is chosen such that the aligning-
stiffness agrees with the self-aligning torque reference-curve. For small
pure longitudinal slip the coherence between the brush model and the
reference curve is good. For slips larger than the value corresponding to
the peak force, the curves do not agree since no velocity-dependency is
modelled for the friction. This have the effect that the modelled tire force
is constant for higher slip, while here and most commonly the tire force
decreases in the region. Different ways to introduce velocity dependence
in the friction is, for example, discussed in [Svendenius, 2003]. At partial
sliding (ψ < 1) this approximation normally has small effects.
53
40
−F [kN]
20
x
0
0 10 20 30 40 50
40
−F [kN]
20
y
0
0 10 20 30 40 50
1
M [Nm]
0.5
z
−0.5
0 10 20 30 40 50
α [deg]
Figure 4.9 Tire forces as function of α with λ = [0, 5, 10, 20] deg. The dotted line
shows the force from the adhesive region, the dashed line shows it from the sliding
region. The solid line is the total force and the dashed dotted line is the Magic-
Formula reference curve. For the self aligning torque the dotted line denotes the
deformation torque Mz′′ and the dashed line M ′ . The slip definition α is used since it
is most common when data is visualized. Refer to Equation (2.6) for transformation
between α and σ y.
For pure lateral slip there are disparities in the lateral force and in
the self-aligning torque compared to the Magic-Formula reference, even
at lower slips. Probably, the main reason for this is the assumption of a
stiff carcass, which is reasonably accurate in the longitudinal direction.
Laterally, where the carcass is weaker, the effects of this simplification is
noticeable.
There are several ways to include carcass flexibility based on assump-
tions on stretched string or beam behavior in the brush model. Some of
them are further discussed in Section 4.3. Also the self-aligning torque
depends on the flexibility of the carcass which might explain some of its
disagreement to the reference. A factor that can explain the deviation of
the self-aligning torque, particularly at high lateral slip is the deviation
of the actual pressure distribution from the parabolic assumption in (4.8).
If the position of the center of the vertical load differs from the hub cen-
ter, the pneumatic trail will not become zero at higher slips. Hence, the
lateral force then gives a torque contribution. This have not been further
examined in this thesis, but the effects on the longitudinal tire force, due
to non-uniform pressure distribution is penetrated in [Svendenius and
Wittenmark, 2003].
54
4.2 The Effect of Camber
xs
i
δ xs
y δ y,cam δ ys
−a 0 a x
Figure 4.10 Schematic illustration of the contact patch showing the bristle defor-
mation due to longitudinal and lateral slip and cambering.
Due to lack of data there were no possibilities to verify the deformation

torque, Mz′′ , which probably is underestimated since only the rubber de-
formations are considered here. Torque addition due to carcass deflection
is discussed in Section 4.3.
Bristle deflection
The developed tire force due to tilting of the tire, can with some approxi-
mations, be explained by the brush model. In Figure 2.3 a cambered tire
is shown together with the orbit, an ellipse, that describes the projection
on the road surface of a point on the carcass during rolling motion. The
deviation from the straight contact patch from a non-cambered tire is
p √
y = − sin(γ ) R2 − x2 − R2 − a2 (4.43)
where R is the average wheel radius and a is half the contact length.
Figure 4.10 also shows the deformation of bristle element due to cam-
bering and lateral slip according to the brush model. Relation (4.43) will
result in difficult expressions when the standard parabolic pressure distri-
bution is employed. In, for example, [Gim and Nikravesh, 1991] and [Pace-
jka, 2002] the deviation due to cambering is approximated as a parabolic
function similar to the assumed pressure distribution as
δ y,cam ( x) = −γ k a2 − x2 (4.44)

55
where k is chosen such that the average camber deflection is correct in

the contact patch
√
a a
3 R − R2 − a2
Z Z
δ y,cam dx = ydx [ k ( (4.45)
−a −a 4 a2
The approximation simplifies the calculations considerably. The deforma-

tion of a bristle that adheres to the road in the contact patch is then
−δ y,cam and the total bristle deformation including the effects caused by
slip, under the assumption of a stiff carcass, can be written as
δ xb ( x) = −σ x (a − x) (4.46)
2 2
δ yb( x) = −σ y(a − x) + γ k(a − x ) (4.47)
Size of adhesive region

The calculations in Section 4.1 can be used to determine the size of the
adhesion region. By using (4.5) in (4.6) with the addition from camber on
δ yb from (4.44), the expression for deriving the break-away point is given
as
s 2 2
cpxσ x cpy (σ y + γ k(a + xs ))
+ (a − xs ) ≤ qz ( xs ) (4.48)
µ sx µ sy
with the assumption of the pressure distribution from (4.8) and the defi-
nition of the normalized slip, ψ , compared to (4.12), is extended to
 s 
○2
2 2 2
J γ  σ yγ + σx σy σ xγ
ψ (σ x , σ y, γ ) = + −
γ ○2 − γ 2 σ ○yγ ○

σ ○x σ ○y σ ○xγ ○
(4.49)
where the limit slips, σ ○x , σ ○y are defined in (4.11), and the camber limit
angle γ ○ are defined as
J 3Fz µ sy
γ○ = (4.50)
2Cy ka
The break-away point can be written as in (4.13)
xs (σ x , σ y, γ ) = (2ψ (σ x , σ y, γ ) − 1) a (4.51)
The partitioning of the contact patch and the significance of ψ and xs is

visualized in Figure 4.11. It is clear that values of ψ larger than unity lack
physical interpretation. Since the bristle deformation due to cambering
56
1
0
1
0
1
0
replacements 1
0
1
0
0
1
1
0
1 µ dFz( x)
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
Bristle Force
1
0
1
0
1111111111
0000000000 1
0
1
0
0000000000
1111111111
0000000000000000000000000
1111111111111111111111111
c pyδ yb 1
0000000000
1111111111
0
1
0
0000000000000000000000000
1111111111111111111111111
1
0
0000000000000000000000000
1111111111111111111111111
0000000000
1111111111
1
0
0000000000000000000000000
1111111111111111111111111
1
0
0000000000000000000000000
1111111111111111111111111
0000000000
1111111111
1
0
0000000000000000000000000
1111111111111111111111111
1
0
0000000000000000000000000
1111111111111111111111111
1
0
0000000000000000000000000
1111111111111111111111111
1
0
0000000000000000000000000
1111111111111111111111111
1
0
0000000000000000000000000
1111111111111111111111111
1
0
0000000000000000000000000
1111111111111111111111111
c pyδ ys 1
0
0000000000000000000000000
1111111111111111111111111
1
0
0000000000000000000000000
1111111111111111111111111
1
0
0000000000000000000000000
1111111111111111111111111
1
0
0000000000000000000000000
1111111111111111111111111
1
0
0000000000000000000000000
1111111111111111111111111
1
0
0000000000000000000000000
1111111111111111111111111
1
0
0000000000000000000000000
1111111111111111111111111
1
0
0000000000000000000000000
1111111111111111111111111
1
0
2aψ 0000000000000000000000000
1111111111111111111111111
1
0
0000000000000000000000000
1111111111111111111111111
1
0
0000000000000000000000000
1111111111111111111111111
1
0
0000000000000000000000000
1111111111111111111111111
c pyδ y,cam 1
0
0000000000000000000000000
1111111111111111111111111
1
0
0000000000000000000000000
1111111111111111111111111
1
0
0000000000000000000000000
1111111111111111111111111
1
0
0000000000000000000000000
1111111111111111111111111
1
0
0000000000000000000000000a
1111111111111111111111111
1
0
−a 0000000000000000000000000
1111111111111111111111111
xs
1
0
1
0
1
0
0
Contact Patch
Figure 4.11 Force generation in the contact patch due to slip and camber. The
shaded area is the force generated in the sliding region. The lower area with diagonal
lines denotes the adhesion force generated by slip and the upper the contribution
due to camber. The dashed straight line indicates the sliding region in case of zero
camber.
has the same shape as the pressure distribution along the contact patch,
ψ will be zero as long as σ x = σ y = 0. At the camber angle γ ○ the whole
contact patch starts to slide against the road and ψ can not be computed
since the denominator reaches zero. In this case the model is not valid
and if accurate results are to be obtained a different tire model should be
used. Such large camber angles are, however, not relevant for studies on
trucks and cars.
Forces and torque

The forces generated in the contact patch are in Section 4.1 derived sepa-
rately for the adhesive and sliding regions. As mentioned, camber affects
the natural path of the carcass in the contact region, which affects the
force contribution from the adhesive region. In the sliding area the de-
formation, δ x , δ y, changes, since the vertical force on the bristle varies
according to the parabolic pressure distribution. The rate of change of the
deformation is small and the effect on the sliding velocity is neglected. The
curvature of the bristle path resulting from cambering is small for small
camber angles and the effect on the sliding friction forces is neglected. In
other words, the only effect cambering has on the sliding force is that it
changes the size of the sliding region, which already is included by use of
ψ from (4.49)
57
Adhesion The forces from the adhesive region derived in (4.14) with use
of ψ from (4.49) are still valid. In the lateral direction the force addition
due to the camber is the sum of the force contribution from each bristle
given by (4.44) with (4.4) as
Z a
Fay,cam (σ x , σ y, γ ) = cpyγ k(a2 − x2 )dx
xs (σ x ,σ y ,γ )
2
= γ kaCy 2ψ 3 (σ x , σ y, γ ) − 3ψ 2 (σ x , σ y, γ ) + 1 (4.52)

3
The camber stiffness, Cγ is defined as
Fy(0, 0, γ ) 2kaCy

Fz µ sy
Cγ = = = (4.53)
γ
γ =0 3 γ○
In case of pure cambering for γ < γ ○ the lateral force is
F0,cam(γ ) = Cγ γ (4.54)
The total adhesive lateral force is
Fay(σ x , σ y, γ ) = − Cyσ y (1 − ψ (σ x , σ y, γ ))2 + Fay,cam(σ x , σ y, γ ) (4.55)
Self-aligning torque
According to the conditions assumed here, pure cambering will not give
rise to any self-aligning torque. As long as there is no sliding region in
the contact patch the bristle deformation is symmetric. A simultaneous
slip causes a sliding region and the addition to the torque due to camber
in the adhesive region can be written as
Z a
2
Mz,cam = cpy xδ y,cam dx = 2γ ka2 Cyψ 2 (σ x , σ y, γ ) (1 − ψ (σ x , σ y, γ ))
xs (σ x ,σ y ,γ )
(4.56)
The total self-aligning torque can be expressed as
Mz (σ x , σ y, γ ) = Maz (σ x , σ y, γ ) + Msz (σ x , σ y, γ )
+ Mz,cam (σ x , σ y, γ ) + Mz,add (σ x , σ y, γ ) (4.57)
In Section 5.4 results on how the combined slip forces and torque are
affected by camber are shown.
58
4.3 Effects of a Flexible Carcass
Lateral deformation of the carcass

As mentioned previously the brush model described in Section 4.1 is based
on the assumption of a stiff carcass. In reality the carcass is flexible, as
mentioned in Section 2.6 and exhibits significant deformation. Figure 4.12
illustrates how the carcass deformation affects the brush model laterally.
Mainly, the break-away point will move backwards and the deflection of
the bristles increases differently in the contact patch. The bristles and
the carcass can be seen as two spring elements connected serially, with
the distinction that each bristle deflects individually, when the carcass is
a coherent unit. The total carcass deflection is denoted by δ yctot , and δ yc
is the relative deflection, compared to the deflection in the leading point
of the contact patch, ( x = a), as δ yc ( x) = δ yctot ( x) − δ yctot (a). The total
deformation relative the carcass deformation in same point is the sum of
the bristle and carcass deflections δ yb ( x) and δ yc ( x)
δ ys ( x) = δ yb( x) + δ yc ( x) (4.58)
and
dFyb ( x) = dFyc ( x) = dFy( x) (4.59)
Let Fy′ (σ y) denote the lateral tire-force for a tire with flexible carcass.
Regard again how the deformation at position x in the adhesive region,
δ y( x), is described by (4.3). This is a purely kinematic relation which holds
also in the case of flexible carcass. The force acting on the bristle element
at x will then be
dFy′ ( x) = cpyδ yb ( x) dx = −cpy(σ x (a − x) + δ yc ( x)) (4.60)
δ yb
δ yb
α
−a a x −a δ yc a α x
xs y x′s y
δ yctot
Figure 4.12 Lateral tire deformation in the contact patch according to the brush-
model. Left: Stiff carcass. Right: Flexible carcass.
59
δ yb
−a a α x
x′s δ yc y
Figure 4.13 Lateral tire deformation in the contact patch with carcass deforma-
tion according to the assumption in Section 4.3.
when the bristle starts to slide the force is

dFy′ ( x) = µ dFz ( x) (4.61)
The relation between the carcass deflection and the lateral force distribu-
tion has to be established and the literature propose several ways. Com-
mon proposals are to treat the carcass as a string or as a beam. For the
string model the force-deflection relation has the following differential
form
d2δ yc ( x)
S = qyc ( x) (4.62)
dx2
where S is the tension in the thread and qyc = qy is the lateral force per
length unit equal to dFy/ dx. A system of equations consisting of an expres-
sion for the break-away point given by the equality of (4.60) and (4.61)
and one differential expression for lateral force per unit for xs ≤ x ≤ a
from (4.62) and (4.60) and for −a ≤ x ≤ xs given by (4.61), can be es-
tablished. The approach requires extensive calculations and will not be
further treated here. It has, however, been solved and a few results is
presented in [Pacejka, 1988]. A simpler approach is to assume a certain
shape of the carcass deformation with an amplitude depending on the to-
tal lateral force. Here the simplest one is used, where it is assumed that
the carcass is straight in the adhesion area, but have a deviation towards
the rolling direction. The incline is proportional to the lateral force, see
Figure 4.13. The deflection is given by
Fy′ (σ y)
δ yc ( x) = − (a − x) (4.63)
Cc
where Cc relates to the lateral the carcass stiffness and Fy′ the lateral
tire force under the assumption of a flexible carcass. The proposal is in
analogy with an assumption by von Schlippe discussed in [Pacejka, 2002].
From (4.58), (4.63) and (4.3) it holds that
Fy′ (σ y)

′
dFy ( x) = cpyδ yb ( x) dx = −cpy σ y − (a − x) dx (4.64)
Cc
60
Using (4.6) at pure lateral slip together with a parabolic pressure distri-
bution (4.8) the position for the break-away point can be solved from
Fy′ (σ y) 3µ ay Fz

cpy σy − = ( a + xs ) (4.65)
Cc 4a3
The total lateral force can be derived from the following equation
Z xs ( Fy′ (σ y ))
a Fy′ (σ y )
Z
Fy′ (σ y) = cpy σy − (a − x)dx + µ qz dx (4.66)
xs ( Fy′ (σ y )) Cc −a
It is realized that the analytical solution to this strongly simplified ap-

proach will not become a smooth expression. Numerical solution can eas-
ily be found and a comparison between this compensation and the ordi-
nary brush model can be seen in Figure 4.14. Important information can,
however, be derived from this model considering the effect of the flexible
carcass on σ ○y , see Equation (4.11). Regard the case when the entire con-
tact patch slides, i.e xs = a, σ y = σ ○y ′ , and Fy′ (σ ○y ′ ) = µ ky Fz . Solving (4.65)
for σ ○y ′ under these conditions gives
3Fz µ sy 3µ sy

µ ky Fz µ ky
σ ○y ′ = 2
+ = Fz 2
+ (4.67)
2a cpy Cc 2a cpy Cc
The next step is to derive an expression for the relation between Cc and the
cornering stiffness Cy′ . The apostrophe marks that the cornering stiffness
is derived under the assumption of a flexible carcass. At very small slips
σ y ( 0 there is no sliding in the contact patch and the tire force only
consists of adhesive force. Therefore,
dFy′ (σ y) a
Z
d
Cy′ =

=− cpyδ yb ( x) dx
dσ y σ y =0 dσ y
−a σ y =0
!
′
1 dFy(σ y ) Cy′

= 2a2 cpy 1− 2
= 2a cpy 1 − (4.68)

Cc dσ y σ y =0 Cc
Hence, the cornering stiffness including the flexible carcass is given by
Cc 2a2 cpy
Cy′ = (4.69)
Cc + 2a2 cpy
Further, the rubber will be assumed to behave isotropically, cpy = cpx ,

which is a realistic assumption. In Section 4.1 the carcass deformation
was included in the lateral rubber stiffness and in that case cpy ,= cpx .
61
35
30
25
Lateral tire force [kNm]
20
15
10
0
0 5 10 15 20 25
α [deg]
Figure 4.14 Comparison of the brush-model with (dashed) and without (dashed
dotted) compensation for a flexible carcass. Asterisks denotes points for total sliding
α ○′ and α ○ . Solid line is a Magic-Formula parametrisation of empirical data further
presented in [Gäfvert and Svendenius, 2003].
The divergence between the longitudinal and lateral stiffnesses may now
be explained by the carcass stiffness, which can be calculated from (4.69)
as
Cx Cy′
Cc = (4.70)
Cx − Cy′
where 2a2 cpx = Cx according to (4.31). Using (4.70) then the limit-slip
adjusted for carcass deformation of (4.67), σ ○y ′ , can be written as
2 1

σ ○y ′ = Fz µ y + ′ (4.71)
Cx Cy
Discussion Introducing flexibility in the carcass as described above

improves the accuracy of the brush model in the lateral direction signif-
icantly, as can be seen in Figure 4.14. The approach can be used even
for combined slip. The reliability of the brush model is still better in the
longitudinal direction, but assuming more realistic lateral deflections will
complicate the expressions considerably. The major aim of this section is
62
the derivation of an enhanced prediction of the slip corresponding to the

lateral peak-force value in (4.71).
Effects on the self-aligning torque

In Section 4.1 a second-order effect of combined slip on the self-aligning
torque, due to the bristle deformation, was derived in (4.36). Probably
this effect is small compared to the torque generated by the displacement
of the points of action for the tire forces due to the carcass deformation.
This torque can be described as
1 1

Mcz(σ x , σ y, γ ) = Fx (σ x , σ y, γ ) Fy(σ x , σ y, γ )
′′
− (4.72)
Ccx Ccy
where Ccx and Ccy are the carcass-stiffness coefficients, which may be
calculated from (4.82). The self-aligning torque contribution Mcz
′′
should
be added to the result in Equation (4.40).
Tire dynamics
In a tire with rigid carcass the motion in tire-road interface may be directly
described by the motion of the wheel rim. With a flexible carcass there is
a dynamic relation between these motions. When deriving the relation a
few entities have to be redefined, compare to Figure 2.1 and Section 2.4.
The slip speed in the contact patch of a tire with rigid carcass is given by
v′sx = vx − ω R e ; v′sy = vy (4.73)
where the motion of the rim is described by the velocities v′x and v′y, the
wheel rotational velocity Ω , and the effective rolling radius R e . To be able
to distinguish between motions related to the carcass and the motions
related to the rim, the latter entities are marked by an apostrophe. The
corresponding slip velocities with a flexible carcass are
vsx = v′sx + δ˙x ; vsy = v′sy + δ˙y (4.74)
where δ x and δ y are the deformations of the carcass as illustrated in

Figure 4.15. The slip velocities may be normalized to form the tire slips. In
practical models they are best normalized with the longitudinal velocity
vx . The slips corresponding to tires with rigid (primed variables) and
flexible carcasses can be written as
v′sx δ˙x
λ′ = ; λ = λ′ +
vx vx
(4.75)
v′sy δ˙y
tan (α ′ ) = ; tan (α ) = tan (α ) +′
vx vx
63
The tire forces that are generated in the contact patch, Fx (λ , α , γ ) and
Fy(λ , α , γ ), are transmitted to the wheel rim by the flexible carcass. The
contact patch forces depend on the motion in the friction interface, which
may be described by the tire slips, and the camber angle γ . In this sec-
tion (λ , α ) are chosen to represent the slip instead of (σ x , σ y) as used
previously. The reason is that (λ , α ) is most commonly used when visu-
alizing data and results and the conversion between the slips are often
straight forward by using (2.6). Here, those slips are used to show the
extra step that is necessary for the differentiation of tan(α ), see (4.79).
Use of (σ x , σ y) simplifies the calculations in that λ , tan(α ) and vx are
replaced by σ x , σ y and vc in (4.75).
y x
z z
−δ y −δ x
− Fy vsy − Fx vsx
v′sy v′sx
Figure 4.15 Carcass deformation of a tire. To the left: Lateral deformation. To

the right: Longitudinal deformation for a braked wheel.
The tire carcass is assumed to behave like a linear spring and damper
such that
Fx = Ccxδ x + D xδ˙x ; Fy = Ccyδ y + D yδ˙y (4.76)
where Ccx and Ccy are the carcass stiffnesses in respective directions.
Using (4.75) together with the time derivative of (4.76) gives
Ḟx D xδ¨x
λ − λ′ = −
Ccx vx Ccx vx
(4.77)
Ḟy D yδ¨y
tan(α ) − tan(α ′ ) = −
Ccyvx Ccyvx
64
Then, by expansion of the time derivative of the forces as
Fx Fx Fx
Ḟx = λ̇ + α̇ + γ˙
λ α γ
(4.78)
Fy Fy Fy
Ḟy = λ̇ + α̇ + γ˙
λ α γ
and by differentiation of the right column of (4.75) under the assumption

that the vehicle velocity varies slowly, the carcass dynamics are formulated
as
1

Dx ˙′ Fx Fx Fx
λ −λ =′
λ + − D x vx λ̇ + α̇ + γ˙ (4.79a)
Ccx Ccx vx λ α γ
Dy 1
tan(α ) − tan(α ′ ) = (1 + tan2 (α ′ ))α˙ ′ +
Ccy Ccyvx

Fy Fy Fy
⋅ λ̇ + − D yvx (1 + tan2 (α )) α̇ + γ˙ (4.79b)
λ α γ
which relates the contact patch slips to the wheel-rim slips. Note that the
dynamics include cross-couplings.
Without damping the system becomes unstable in slip regions where
the slip-force characteristics have positive slope. To avoid this, the damp-
ing coefficients should be chosen so that D x vx,min > sup( Fx /λ ) and
D yvx,min > sup( Fy/α ) for an arbitrary choice of slowest vx,min > 0. The
system is then well posed for all vx ≥ vx,min . In [Pacejka, 2002] it is sug-
gested to solve this problem by preventing the diagonal terms, Fx /λ ,
Fy/α from being positive by limitation. The introduction of damping
may be a more physically motivated modification.
The carcass stiffnesses are properties that might not be available for
the specific tire, but quite often the relaxation length, σ a of a tire is mea-
sured. The relaxation length is usually defined for small slips [Pacejka,
2002] and relate the dynamics between the rim and the carcass as
σ ax σ ay
λ̇ + λ = λ ′ ; α̇ + α = α ′ (4.80)
vx vx
Note that vx is time varying, and that (4.80) therefore is not a linear
time-invariant system. Instead, the differential equations are linear space-
invariant, since the independent time variable t can be exchanged for the
space variable s (rolling distance) as λ̇ = dλ / ds ⋅ ds/ dt and ds/ dt = vx ,
65
and correspondingly for the lateral slip. In the case of small pure slip, zero
camber, and neglecting the damping effects (4.79) might be simplified to
1 Fx 1 Fy
λ − λ′ = λ̇ ; α −α′ = α̇ (4.81)
Ccx vx λ Ccyvx α
It is then clear that the relaxation lengths are related to the stiffnesses
as
Cx Cy
σ ax = ; σ ay = (4.82)
Ccx Ccy
4.4 Summary
In this chapter the well-known brush model has been presented in detail.
The brush model in its ordinary form renders simplistic expressions for
the tire-force and slip relations. There are, however, many shortcomings,
which can be overcome by more extensive modeling. The trade-off prob-
lem between simplicity and accuracy is obvious and it is important to be
aware about the demands of the application when adopting a model for a
particular purpose. Each extra feature increases the complexity consider-
ably. An example shown here is the introduction of camber, which still is
possible with reasonable complexity.
In the presented form, the brush model gives a good description of
the longitudinal tire characteristic at low slip up to the peak force. If
good accuracy for higher slip is necessary the velocity dependence has
to be included in friction coefficient, which is discussed in Section 5.2
and more generally in e. g. [Svendenius and Wittenmark, 2003; Pacejka,
1988]. An extensive evaluation of the accuracy of the longitudinal proper-
ties of the brush model can be found in Section 7.2. In the lateral direction
also the behavior of the carcass effects the tire characteristics. An accu-
rate modeling of the carcass require more sophisticated software, such as
Finite-Element tools, see Section 3.4.
The most simple corrections available in the literature have been dis-
cussed here. A new touch is given by the practical way of calculating the
lateral slip corresponding to the peak force in Equation (4.71). Most of
the contents in this section are gathered from earlier presented material
by other authors.
66
5
A Dynamic Semi-Empirical
Tire-Model for Combined
Slips and Camber
This chapter presents a method to derive the tire forces during com-
bined slip and camber, given the empirical tire-force models at pure slip.
The method has previously been presented in a number of journal arti-
cles [Gäfvert and Svendenius, 2005; Svendenius and Gäfvert, 2005; Sven-
denius and Gäfvert, 2006] and conference proceedings [Svendenius and
Gäfvert, 2004a; Svendenius and Gäfvert, 2004b; Gäfvert et al., 2006]. A de-
tailed presentation of the theory behind the model can be found in [Gäfvert
and Svendenius, 2003], while the first publication was in [Gäfvert, 2003].
The steady-state part of the model, described in Sections 5.1–5.3, is
based on the observation that the pure-slip characteristics contains all
necessary information to reproduce combined-slip behavior with reason-
able accuracy. The separation of the empirical pure-slip forces into com-
ponents of adhesion and sliding and the different treatment of respective
contribution are fundamental parts of the method. The model is “semi-
empirical”, since the pure-slip models may be entirely empirical, while
the method to construct the combined-slip characteristics strictly follows
from theory on brush-model mechanics, explained in Section 4.1. This
distinguishes the presented model from similar previous methods that
are purely empirical, for example [Bakker et al., 1989; Schuring et al.,
1996; Lugner and Mittermayr, 1991; Sharp, 2004]. Due to the physical
assumptions adopted, it is possible to include velocity dependency in the
model, even if this is not explicitly present in the pure-slip models.
A validation and a presentation of the results of the model are given
in Section 5.4 using measurement data from a Jeep Cherokee. The Magic
Formula is used to approximate the pure-slip data. An implementation
67
Chapter 5. A Dynamic Semi-Empirical Tire-Model . . .
and a simulation of the dynamic model derived in Section 4.3 are de-
scribed, here. The dynamics behavior is explained by the deflection of the
tire carcass, that causes differences between the forces and movements
of the rim and the forces and movements in the contact patch. This is of
practical significance in, for example, systems for control of the longitudi-
nal tire slip where motion sensors are, in general, attached to the wheel
rim and the force are generated by the motions in the contact patch.
5.1 Introduction
Semi-empirical methods are often used to derive the tire force at combined
slip situations, since pure-condition tire-forces may be described well by
rather compact and simple empirical models. Pure-slip tire data from test-
bench experiments are also often available for calibration. The situation
for mixed conditions is somewhat different. The transition from one to two
dimensions makes it more difficult to apply functional approximations.
Empirical models tend to be either rough approximations or quite com-
plex, difficult to understand, and rely on parameters that need to be cali-
brated with mixed-condition experimental data. This is a drawback since
such data are expensive and time consuming to collect and therefore rarely
available for a specific tire. Semi-empirical models for combined-slip uses
the information from the pure-slip models to generate the tire forces when
the vehicle brakes and turns simultaneously. Based on the mechanics of
the tire and the available empirical data, a number of criteria for com-
bined models may be stated (in the spirit of [Brach and Brach, 2000]):
1. The combined force F̄ (λ , α ) should preferably be constructed from
pure slip models F0x (λ ) and F0y(α ), with few additional parameters.
2. The computations involved in the model must be numerically feasible

and efficient.
3. The formulas should preferably be physically motivated.
4. The combined force F̄ (λ , α ) should reduce to F0x (λ ) and F0y(α ) at

pure braking or cornering:
F̄ (λ , 0) = [ F0x (λ ), 0 ]
F̄ (0, α ) = [ 0, F0y(α ) ]
5. Sliding must occur simultaneously in longitudinal and lateral direc-

tions.
68
5.1 Introduction
6. The resulting force magnitudes should stay within the friction con-
straints.
7. The combined force should be F̄ = − Fz µ v̄s /vs at full sliding for tires
with isotropic friction characteristics, i.e. equal friction properties in
all directions.
These criteria are used when developing the semi-empirical tire-model
for combined slip presented in the following.
Main approach
The general idea of this method to derive the forces at a combined slip
(σ x , σ y, γ ) is to scale the forces given by the empirical pure-slip model at
certain pure slips σ 0x (σ x , σ y, γ ) and σ 0y(σ x , σ y, γ ). The pure slips can be
chosen in various ways, but their relations to σ x , σ y and γ have to be well
motivated and different proposals are discussed below. For convenience,
the arguments (σ x , σ y, γ ) for σ 0x and σ 0y are left out in the following.
The scale factors depend on the combined longitudinal and lateral slip,
the camber angle, and the relation between the used pure slip and the
cambered combined-slip. The arguments (σ x , σ y, γ ) are also left out for the
scale factors, G i j , in the following. Since the generation of forces from the
adhesive and the sliding regions are built on different physical phenomena
they are treated separately. The following equation shows the form
F̂x (σ x , σ y, γ ) = G ax F̂0x (σ 0xa ) + G sx F̂0x (σ 0xs ) (5.1)

F̂y(σ x , σ y, γ ) = G ay F̂0y(σ 0ya ) + G sy F̂0y(σ 0ys ) + Gcamy F̂0cam (γ ) (5.2)
where F̂0x and F̂0y are the empirical pure slip models and F̂0cam is the
empirical pure cambering model. The pure-slip tire behavior can be given
in any form, either as an empirical model or as tabular data, as long
as the vertical and horizontal shift are zero, i.e. that the curves cross
the origin. If not known the pure camber model may be as simple as the
cambering stiffness, Ĉγ , times the camber angle as assumed in (4.54). If
no information about the camber properties is available, then Cγ can be
calculated from the cornering stiffness using (4.53). The half patch-length
a is either assumed or derived from (4.42) using the aligning stiffness.
The tire radius R and the contact patch length is used to calculate k by
Equation (4.45).
Analogously, the combined self-aligning torque can be derived as
M̂z (σ x , σ y, γ ) = G f z F̂0y(σ 0z) + G mz M̂0z(σ 0z) + Gcamz F̂0cam (γ ) (5.3)
The scale factors, G i j , are derived from the analytical expressions of the
69
brush model, as
Fax (σ x , σ y, γ )
Fax (σ x , σ y, γ ) = F0x (σ 0xa ) = G ax F0x (σ 0xa ) ( G ax F̂0x (σ 0xa )
F0x (σ 0xa )
(5.4)
Fay(σ x , σ y, γ ) Fay,cam(σ x , σ y, γ )
Fay(σ x , σ y, γ ) = F0y(σ 0ya ) + F0cam (γ )
F0y(σ 0ya ) F0cam (γ )
= G ay F0y(σ 0ya ) + Gcamy F0cam (γ ) ( G ay F̂0y(σ 0ya ) + Gcamy F̂0cam (γ ) (5.5)
Fsx (σ x , σ y, γ )
Fsx (σ x , σ y, γ ) = F0x (σ 0xs ) = G sx F0x (σ 0xs ) ( G sx F̂0x (σ 0xs )
F0x (σ 0xs )
(5.6)
and
Fsy(σ x , σ y, γ )
Fsy(σ x , σ y, γ ) = F0y(σ 0ys ) = G sy F0y(σ 0ys ) ( G sy F̂0y(σ 0ys )
F0y(σ 0ys
(5.7)
For the self-aligning torque the structure is slightly different, and the
reader is referred to the computations in Appendix B.
5.2 Scale Factors
The structure for deriving the scale factors require the assumption of a
model that describes the relation between the combined-slip and the pure-
slip forces. It also gives a freedom in the choice of the relation between the
pure slips (σ 0x , σ 0y) and the combined slip situation (σ x , σ y, γ ). This sec-
tion proposes the choices for the pure slip relations and derives the scale
factor using the basic brush-model assumptions proposed in Section 4.1.
Adhesive region
The bristle deformations are the source of the adhesion forces. Therefore,
for adhesion forces it makes sense to regard pure slips that result in the
same deformation as the combined slip. The deformation state depends on
the slip, σ̄ . The pure slip is therefore constructed to maintain σ̄ constant.
Hence
σ 0xa = σ x ; σ 0ya = σ y (5.8)
The scale factors G ax and G ay can now be calculated as
3 (1 − ψ (σ x , σ y, γ ))2 3 (1 − ψ (σ x , σ y, γ ))2
G ax = ; G ay = (5.9)
ϒ(σ x , 0, 0) ϒ(0, σ y, 0)
70
5.2 Scale Factors
when ψ < 1, otherwise G ax = G ay = 0, with

J
ϒ( x, y, z) = ψ 2 ( x, y, z) − 3ψ ( x, y, z) + 3 (5.10)
Here Equations (5.4) and (5.5) are used together with (4.14), (4.29), and
(4.30). Details on the computation of the scale factors can be found in
Appendix B. The scale factor of the camber force can be calculated from
(5.5), (4.52), and (4.54), as
Gcamy = 2ψ 3 (σ x , σ y, γ ) − 3ψ 2 (σ x , σ y, γ ) + 1 (5.11)
when ψ < 1, otherwise Gcamy = 0.
Sliding region
In the literature, slip-velocity is mentioned as a significant factor that
influences the friction coefficient for a specific tire on a certain road foun-
dation [Wong, 2001]. Therefore, it is reasonable to define the pure slip
used for the sliding forces such that the slip velocity is invariant. The
sliding-velocity invariant pure-slips are defined as
q
v σ 2x + σ 2y sgn(σ x )
vel
σ 0x = q q
v0 (1 + σ x )2 + σ 2y − v σ 2x + σ 2y sgn (σ x )
q (5.12)
v σ 2x + σ 2y sgn(σ y)
vel
σ 0y = q
v20 (1 + σ x )2 + σ 2y − v2 (σ 2x + σ 2y )

vel vel
Hence (σ 0x , 0, 0) and (0, σ 0y , 0) at the wheel-travel velocity v0 result in the
same slip velocity, vs , as the combined slip (σ x , σ y, γ ) at the wheel-travel
velocity v. Note that v is the actual wheel travel velocity and v0 the velocity
at which the pure slip model is valid. For details on the computation the
reader is referred to Appendix B. The sliding pure-slip are chosen as,
vel vel
σ 0xs = σ 0x , and, σ 0ys = σ 0y , and the scaling factors G sx and G sy can be
computed from (4.14), (4.15), (4.29) and (4.26), using β f = β ′ as
G sx = pcos (β ′ )p ⋅ Γ x ; G sy = psin (β ′ )p ⋅ Γ y (5.13)

with
 q q
v (1 + σ )2 + σ 2 − v σ 2 + σ 2 sgn (σ ) Λσ ○

0
 x x



y x y x


ψ (σ x , σ y, γ )(3 − 2ψ (σ x , σ y, γ ))

J
Γx = ⋅ vel
vel
if ψ (σ 0x , 0, 0) < 1 (5.14a)


 v ϒ( σ 0x , 0, 0)


 2 vel
ψ (σ x , σ y, γ )(3 − 2ψ (σ x , σ y, γ )) if ψ (σ 0x , 0, 0) ≥ 1

71
and

2 2 2 2 2 2
21
 v0 (1 + σ x ) + σ y − v (σ x + σ y ) Λσ y
○






ψ (σ x , σ y, γ )(3 − 2ψ (σ x , σ y, γ ))

J
Γy = vel (5.14b)
 ⋅ vel
if ψ (0, σ 0y , 0) < 1
v ϒ( 0, σ , 0)



 0y

 2 vel
ψ (σ x , σ y, γ )(3 − 2ψ (σ x , σ y, γ )) if ψ (0, σ 0y , 0) ≥ 1

and
 v 
γ ○2  sin(β ) γ u cos(β ) 2 2 !
sin
2
u
J γ ( β )
Λ = ○2 +t 1− +
γ −γ 2

σ ○y γ ○ σ ○x γ○ σ ○y
(5.15)
if ψ (σ x , σ y, γ ) < 1, otherwise
( −1 vel
J
ψ (σ 0x , 0, 0)ϒ −1 (σ 0x
vel
, 0, 0) vel
if ψ (σ 0x , 0, 0) < 1
Γx = (5.16a)
vel
1 if ψ (σ 0x , 0, 0) ≥ 1
and
( −1 vel
J
ψ (0, σ 0y , 0)ϒ −1 (0, σ 0y
vel
, 0) vel
if ψ (0, σ 0y , 0) < 1
Γy = (5.16b)
vel
1 if ψ (0, σ 0y , 0) ≥ 1
The friction-constraint angle β ′ can be derived from

vel
σ y Γ x F̂0x (σ 0x )
tan(β ′ ) = vel
(5.17)
σ x Γ y F̂0y(σ 0y )
see (4.20).
Self-aligning torque
For simplicity, another pure-slip definition is used for the self-aligning
reg reg
torque. The pure slips (σ 0x , 0, 0) and (0, σ 0y , 0), with
reg reg
σ 0x = σ ○xψ (σ x , σ y, γ ) sgn(σ x ) ; σ 0y = σ ○yψ (σ x , σ y, γ ) sgn(σ y ) (5.18)
result in adhesion and sliding regions of the same size as the combined
reg
slip (σ x , σ y, γ ). It is therefore obvious to choose σ 0z = σ 0y and the self-
aligning torque is derived as
Mz (σ x , σ y, γ ) = G f z F0y(σ reg reg

y ) + G mz M0z (σ y ) + G camz F0cam (γ ) (5.19)
72
5.3 Parameters
where
G mz = p sin(β ′ )p (5.20)
(4ψ − 1)(1 − ψ )2 p sin(β )p

Gfz = a − p sin ( β ′
)p (5.21)
ϒ σ ○y Λ z
and
Gcamz = 3aψ 2 (σ x , σ y, γ ) (1 − ψ (σ x , σ y, γ ))2 (5.22)
with
J γ ○2
Λ z = ○2
γ −γ 2
 s 
2 2 2
sin(β ) γ cos(β ) sin(β ) cos(β )γ
⋅ + + − sgn(βσ y)
σ ○y γ ○ σ ○x σ ○y σ ○xγ ○
(5.23)
Note that the brush model states that the contact patch length can be de-
rived by the relation between the cornering stiffness and the self-aligning
stiffness, see (4.42), as a = 3Cz/ Cy. When using the region-invariant slips
the friction angle is computed from
σ y F̂0x (σ reg
x )
tan(β ′ ) = (5.24)
σ x F̂0y(σ reg
y )
5.3 Parameters
Four parameters are needed in the steady-state part of the model and two
if the transient properties are to be included. The parameters
• σ ○x , longitudinal limit slip
• σ ○y , lateral limit slip
• γ ○ , adhesive camber limit
• v0 , velocity for the pure-slip model
• Ccx , longitudinal carcass stiffness
• Ccy, lateral carcass stiffness
73
all have clear physical interpretations. The parameters σ ○x and σ ○y , de-

scribe the pure slips where transition from partial to full sliding occur.
They are needed to compute the normalized slip, ψ (σ x , σ y, γ ). A common
assumption is that these transitions occur when the tire forces obtain
their maximum. Some road foundations, such as gravel and snow might
not provide a clear maximum point. In any case, the parameters may sim-
ply be set to the slip values corresponding to the maximum of the tire
forces, F0x
∗ and F ∗ . From the brush model it can be shown that
0y
3 F̂0x
∗ 2 F̂0y
∗ ∗
F̂0y
σ ○x ( ; σ ○y ( + (5.25)
Ĉx Ĉx Ĉy
which is a result from (4.11) and (4.71).

The parameter v0 denotes the wheel-travel velocity at which the empir-
ical pure-slip model is valid. The actual wheel-travel velocity v is assumed
to be a model input. If v0 is not known then v/v0 = 1 may be used, which
will neglect any velocity dependence. This is the common assumption in
most other models.
The camber parameter, γ ○ , is outside the cambering range that is nor-
mally measured on truck and car wheels. Therefore, it is not possible to ob-
tain it directly from measurement data. Instead it can be used that (4.53)
is well approximated by
γ ○ ( F̂0y
∗
/ Ĉγ (5.26)
The camber stiffness can either be given from the empirical camber model
or be computed from tire parameters according to (4.53) using (4.45)
and (4.42).
The carcass stiffnesses can be derived from (4.82) as
Ĉx Ĉy
Ccx = ; Ccy = (5.27)
σ ax σ ay
if the relaxation lengths are known. This is further discussed in the last
part of Subsection 5.4
5.4 Validation and Results of the Semi-Empirical Model
Empirical data, provided by NHTSA, US, for P225/75R15 Goodyear Wran-

gler RT/S, OWL, all terrain steel belted radials [Salaani et al., 1999], are
used for validation. The reference speed used was 48 km/h and inflation
pressure 2.4 bar. The data consist of pure-slip forces and aligning mo-
ments in the slip ranges λ = [−50, 50]% and α = [−15, 15] deg. Forces
74
Measurement data, for the vertical loads 1.1, 2.2, 3.3, 4.4, 5.3/5.5 [kN] Measurement data, for the vertical loads 2.4, 4.2, 6.0, 7.8, 9.6 [kN]
6 10
8
4
6
2 4
−F [kN]
2
F [kN]
y
x
−2 −2
−4
−4
−6
−6 −8
150 200 250 300 350 400 450 500 550 −15 −10 −5 0 5 10 15
Wheel rotational speed [RPM] Slip Angle [deg]
Figure 5.1 Available measurement data. Left: Longitudinal force for different
wheel rotational speeds; Right: Lateral force for sweeps of α .
at combined slip at α = [±2, ±4] deg for sweeps of λ and forces at some
camber angles were also included. The different vertical loads for the mea-
surements of the pure longitudinal behavior are 1.1, 2.2, 3.3, 4.4, 5.3 or
5.5 kN. It is unclear from the data whether the highest load for sweeps
of λ and the only load for the combined slip is 5.3 kN or 5.5 kN. For the
pure lateral and cambering behavior 2.4, 4.2, 6.0, 7.8, 9.6 kN. The avail-
able steady-state measurement data are shown in Figures 5.1, 5.2 and 5.3.
Measurements of the transient properties of the tire are available as si-
nusoidal excitations of the slip angle at nine different frequencies. The
data are shown in Figure 5.17 and further discussed in Section 5.4.
Validation of tire-forces at combined slip

The measurements of the pure lateral properties were performed at differ-
ent conditions than the longitudinal and combined properties. The vertical
loads do not correspond between the different setups. The longitudinal
pure slip model and the combined validation data are taken at highest
available load and the lateral pure slip model at 6.0 kN. A linear correc-
tion is applied to adjust for the different loads. The pure slip models are
Magic Formula parametrisation of the raw data, but for the torque model
interpolation of the tabular data was used. The empirical pure slip mod-
els are shown in Figure 5.4. In Figure 5.5 the forces and torque from the
model are compared with the measured values for sweeps of λ . In Figure
5.6 the pure slip measurement for sweep of α is shown together with the
pure slip Magic Formula parametrisation and the computed characteris-
tics for combined slip. Note that (λ , α ) is used to signify the slip instead
of (σ x , σ y), since it is the most common slip definition when dealing with
75
Measurement data, for Combined Slips α=2 −2 4 −4 [deg]

10
F [kN] 0
x
−5
−10
200 250 300 350 400 450 500 550
5
Fy [kN]
−5
200 250 300 350 400 450 500 550
Wheel rotational speed [RPM]
Figure 5.2 Available measurement data for combined slip. Longitudinal and lat-
eral force at Fz=5.3 or 5.5 kN, see comment in text.
Camber Force for F =2.4, 4.2, 6.0, 7.8, 9.6 [kN]

z
0.4
0.3
0.2
0.1
Lateral Force [kN]
−0.1
−0.2
−0.3
−0.4
−0.5
−6 −4 −2 0 2 4 6
Camber Angle [deg]
Figure 5.3 Available camber data. Lateral force as a function of the camber angle
(γ ) for different vertical loads.
76
10
−F [kN]
5
x 0
−5
0 5 10 15 20 25 30 35 40 45 50
Longitudinal Slip [%]
10
−F [kN]
0
y
−10
−20 −15 −10 −5 0 5 10 15 20
Lateral Slip [deg]
0.2
M [kNm]
0
z
−0.2
−20 −15 −10 −5 0 5 10 15 20
Lateral Slip [deg]
Figure 5.4 Pure-slip measurement data (dots) with Magic Formula parametrisa-
tion (solid lines) for the forces and interpolation of raw data for the aligning moment.
The limit slips calculated from (5.25) are marked by crosses in the two upper plots.
measurement data and may look more familiar to the reader. Conversions
between the two slips can be found in Appendix B. The overall behavior
of the model is very similar to the real measurements. Particularly good
agreement is shown at small and large slips. The largest deviations can
be found at slips around the force peak value. Since the tire behavior is
sensitive to many factors it is very important that the pure slip empir-
ical models are generated at the same conditions as the combined slip
measurements. The change of test equipment and the different vertical
loads between the collection of longitudinal and lateral data sets affects
the reliability of the data and the validation.
Comparison with other models In Figure 5.7, comparisons are shown

between the proposed model, the BPL model [Bakker et al., 1989], and
the COMBINATOR model [Schuring et al., 1996]. It is clear that the pro-
posed model performs better than the COMBINATOR model at small slip-
angles. The COMBINATOR model may be written as
Fx = F cos(β ) ; Fy = F sin(β ) (5.28)
where
F = F0x (s) cos2 (β ) + F0y(s) sin2 (β ) (5.29)
77
Combined Slip Forces at α=−4 ,−2, 0, 2, 4 [deg]

6
−F [kN]
x
4
0
0 5 10 15 20 25 30 35
5
−F [kN]
0
y
−5
0 5 10 15 20 25 30 35
0.2
−M [kNm]
0
z
−0.2
0 5 10 15 20 25 30 35
Figure 5.5 Combined measurement data compared to the presented model (solid
line) for sweeps of λ at different slip angles.
Combined Slip Forces at λ=0, 2, 4, 10 [%]

6
−F [kN]
4
x
0
−30 −20 −10 0 10 20 30
5
−F [kN]
0
y
−5
−30 −20 −10 0 10 20 30
0.2
−M [kN]
0
z
−0.2
−30 −20 −10 0 10 20 30
Lateral Slip α [deg]
Figure 5.6 Lateral behavior for combined slip of the model. The measured data
at pure slip is also showed.
78
Combined Slip Forces at α=−4 ,−2, 2, 4 [deg]

5
1
−F [kN]
0
y
−1
−2
−3
−4
−5
−1 0 1 2 3 4 5 6
−F [kN]
x
Figure 5.7 Combined-slip forces with the proposed model (solid line) compared
to the COMBINATOR model (dashed-dotted line) and the BPL model (dashed-line),
for fixed slip angles α and varying λ ranging from 0 to 35%. for comparison.
The resulting force is always collinear with the slip vector, which is an
assumption with weak physical motivation. In the COMBINATOR model
the lateral force initially increases, as a longitudinal slip is applied. This is
a result of the assumption of a collinear combined-slip tire-force in the full
slip range, in combination with the use of the combined-slip magnitude
in the empirical pure-slip models.
The proposed method performs similar to the BPL model. In fact the
essence of the two methods are much like each other, even though the
structures may appear different. The BPL model, see (3.2), uses convex
combination of angles corresponding to adhesion and sliding, to deter-
mine the orientation of the resulting force. In the proposed method the
magnitude of respective force component also relies on convex combina-
tions of adhesion and sliding force contributions. The pure-slip used for
the adhesion force, most prominent at low slip, behaves differently than
the region-invariant slips defined in (5.18) that are used in the BPL-
vel vel
method. However, the sliding-velocity invariant pure-slip (σ 0x , σ 0y ) de-
79
. Velocity dependency for v=[0.5 1 1.5 2] ⋅ v0

6 6
5 5
4 4
−Fx [kN]
−Fy [kN]
3 3
2 2
1 1
0 0
0 5 10 15 20 25 30 35 40 0 5 10 15
λ [%] Lateral Slip α [deg]
Figure 5.8 Velocity dependence at pure slips using the models shown in Fig-
ure 5.4. Velocities: v = 0.5v0 , (dotted) v0 (solid), 1.5v0 (dashed), 2v0 (dash-dotted).
fined in (5.12) used for the sliding contribution behaves more similar to
the region-invariant slips, when v = v0 . The resultant magnitude for the
BPL model at full sliding is
F = F0x cos2 (β ○ ) + F0y sin2 (β ○ ) (5.30)
Note that β ○ defined as (tan(β ○ ) = (σ yσ ∗x )/(σ xσ ∗y )) depends on the tire

stiffnesses and the static friction coefficient, which do not have any phys-
ical significance at full sliding. For the proposed model the corresponding
magnitude is q
F0x F0y/ 2
F0x sin2 (β ) + F0y
2
cos2 (β ) (5.31)
Compare with (5.29) for the COMBINATOR model. The resulting force is
collinear to the slip vector in all three methods at high slip.
Velocity dependence The dependence on wheel-travel velocity is il-

lustrated in Figures 5.8 and 5.9, by varying v. The reference speed v0
is approximately 48 km/h. The pure-slip Magic Formula model is cal-
ibrated with data from tire measurements at the wheel-travel velocity
v0 . Tire forces are shown for velocities that are 0.5–2 times the wheel-
travel velocity of the pure-slip model. The longitudinal direction behaves
qualitatively well with what is reported in e.g. [Pacejka, 2002], while the
behavior in the lateral direction is unexpected with an increase in tire
force for higher velocities. The semi-empirical tire-model generates the
80
Velocity dependency at combined slip for v=[0.5 1 1.5 2] ⋅ v

0
6
4
Fy [kN]
0
−6 −4 −2 0 2 4 6
Fx [kN]
Figure 5.9 Velocity dependence at combined slips using the pure-slip models
shown in Figure 5.4. Sweeps of λ for α = −2, −4 [deg] and sweeps of α for λ = ±2, ±4
[%]. Velocities: v = 0.5v0 , (dotted) v0 (solid), 1.5v0 (dashed), 2v0 (dash-dotted).
forces from the sliding area by extracting the friction coefficients in each
direction for the actual slip velocity. The slip velocity at a combined slip λ ,
1/2
α , is vs = v (λ cos (α ))2 + sin2 (α ) . The (α , λ )-definition is used here,
since it is the common slip definition for calibration of Magic Formula,
see Appendix B for conversion between slip definitions. Corresponding
pure slips for the empirical pure-slip model are given by vs = λ vel 0 v0 and
vs = sin(α 0vel )v0 . Figure 5.10 shows the extracted friction coefficients, cal-
vel
culated from Γ x,y F̂0x,y(σ 0x ) / Fsz(σ x , σ y, γ ), see (5.13) and (4.27), as func-
tions of the slip velocity, which explains the reason behind the strange
result of the model. It clearly shows that the extracted lateral friction
first increases with the speed up to a certain value and then decreases.
It is realistic to assume that the friction, ideally, should be identical
in the both directions. In reality, test disturbances, uncertainties in mea-
surements and modeling approximations shows something else. However,
the choice of limit slip is very important for the friction extraction and
the rule of thumb for the choice of α ○ seems not to be sufficiently good.
A higher limit slip than proposed in (5.25) gives the lateral friction more
similar appearance to the longitudinal direction. Note that this behav-
ior may change from tire to tire and it is not possible to state general
81
°
Velocity dependency of extracted friction coefficient λ =[8.1 10.6 12.9] %
1.4
1.2
1
Long. Friction
0.8
0.6
0.4
0.2
0
0 2 4 6 8 10 12 14 16 18 20
Sliding velocity [km/h]
Velocity dependency of extracted friction coefficient α°=[6.69 8.38 10.1] deg
1.2
1
Lat. Friction
0.8
0.6
0.4
0.2
0
0 2 4 6 8 10 12 14 16 18 20
Sliding velocity [km/h]
Figure 5.10 Calculated friction coefficient for different choices of limit slip. Re-
spective limit slip, in increasing order, are denoted by dashed, solid and dash-dotted
line.
recommendations for the choice of limit slip in this aspect.

If it is necessary to elaborate with the lateral limit slip to get proper
velocity dependence it is also important to study the effect on the combined
slip behavior. Figure 5.11 shows the parameter sensitivity of the force
envelopes at combined slip for variation of α ○ . The figure shows that the
resulting tire force gets larger for higher values on α ○ . A higher limit slip
increases the proportion of the adhesion contribution for lower slip than
this limit. The force limitation due to the friction constraints, therefore,
starts to act later. This will give higher forces before the limit slip is
reached.
The expressions for pure slips can be solved for λ vel vel
0 < 100% (σ 0x < 1 )
and α 0 < 90 deg (σ 0y < 1), which is when (v/v0 ) λ cos (α ) + sin2 (α )
vel vel 2 2 2
≤ 1 is fulfilled, see (5.12) or (B.20). This restricts the interval of v, λ , and

α . If λ vel vel
0 or α 0 are outside the valid range of the pure-slip model, then
extrapolation is necessary. This is actually when the model tries to extract
the friction properties from speeds not available in the empirical model. A
straightforward method is to use the end points of the models. The Magic
82
Parameter dependency at combined slip for α°=[6.69 8.38 10.1] deg

6
4
Fy [kN]
0
−6 −4 −2 0 2 4 6
Fx [kN]
Figure 5.11 Illustration of the parameter sensitivity of the force envelopes at

combined slip with sweeps of λ for α =-2, -4 [deg] and sweeps of α for λ = ±2, ±4.
The different limit slip, in increasing order, are denoted by dashed, solid and dash-
dotted lines, respectively.
Formula with default shape factors has decent extrapolation properties

and may be used with slip ratios λ vel vel
0 > 100%, but α 0 is limited to 90
deg. In general, extrapolation of empirical data must be carried out with
great care.
At actual speeds, v, lower than v0 the friction properties are extracted
from lower slip at v0 to the ensure that the sliding velocity is invariant
(obvious from (5.12)). At low slip the sliding contributions are small and
they have to be magnified to match the proportion of sliding friction at v.
Normally neglectable uncertainties of the model are then also magnified
which might cause strange and non-accurate results.
Utilizing braking data to generate driving data In the proposed

model it is assumed that the given longitudinal pure-slip model, F̂0x (λ ),
is valid both at braking (λ > 0) and driving (λ < 0). The Magic Formula
is an odd function and therefore F̂0x (λ ) = − F̂0x (−λ ). Generally, this is
not in accordance with empirical observations. If the pure-slip model is
assumed to be valid for braking, then the following procedure suggests
how to modify the argument to the pure-slip model at driving, so that a
more accurate force is obtained.
The brush model states that the adhesion force developed at braking
with a slip σ x > 0 will have the same size and the opposite sign, as the
83
Driving vs Braking
6
2
F [kN]
0
x
−2
−4
−6
−15 −10 −5 0 5 10 15
Figure 5.12 Illustration of the difference between driving and braking. Solid line
is the proposed method to convert between braking and driving, with sliding contri-
bution as the dashed dotted line and the adhesive part as the dotted line. Dashed
line shows the Magic Formula tire-model, F0x (λ ) = − F0x (−λ ).
force generated at driving with slip −σ x :
F0ax (σ x ) = − F0ax (−σ x ) (5.32)
With σ x = −λ driving /(1 − λ driving ) and σ x = λ braking /(1 − λ braking) this means
that
λ driving
λ braking = − (5.33)
1 − 2λ driving
Hence, when computing adhesion forces using an empirical model for brak-
ing, the pure-slip forces
−λ

− F̂
0x λ<0
1 − 2λ (5.34)
F̂0x (λ ) λ≥0

are best used for driving and braking, respectively. For the sliding case it
is more natural to let the force depend on the relative velocity vs = λ vx .
Then, −vs will simply correspond to −λ .
84
Tire Forces at γ=0, −5, −10, −15 [deg]
−F [kN] 0
x
−5
−15 −10 −5 0 5 10 15
2
−Fy [kN]
0
−15 −10 −5 0 5 10 15
0.08
0.06
Mz [kNm]
0.04
0.02
0
−0.02
−15 −10 −5 0 5 10 15
Figure 5.13 The effect of camber on the tire forces. The dashed dotted lines are
the adhesion force and the dashed line the sliding forces. The camber effect when
included as a shift on the lateral slip is shown by the dotted line for γ = 0, −5, −10,
and −15 deg.
Therefore, when computing sliding forces using an empirical model for

braking, the pure-slip forces
λ <0
(
− F̂0x (−λ )
(5.35)
F̂0x (λ ) λ ≥0
are best used for driving and braking respectively.

The effect of the described procedure is shown in Figure 5.12. It is
clear that the influences are small and neglectable compared to the un-
certainties in measurements. Interesting is, however, to see the differ-
ence between the proportion adhesion and sliding force for driving and
braking respectively. The magnitude of the longitudinal limit-slip is, the-
oretically, equal if expressed as σ ○x . If using λ ○ it is important to notice
that pλ ○driving p > pλ ○braking p as a consequence of the definition, see Equation
(B.16).
Results of cambering
The information about the camber properties for the tire in the avail-
able data was restricted to a few measurement points at different loads,
85
Tire Forces at γ=0, −5, −10 [deg]
5
−F [kN]
0
y
−5
−15 −10 −5 0 5 10 15
0.15
0.1
0.05
Mz [kNm]
−0.05
−0.1
−15 −10 −5 0 5 10 15
Lateral Slip [deg]
Figure 5.14 Illustration of the effects of camber on the tire forces. Solid lines
are the total force and aligning torque. The dashed dotted lines are the camber
forces. For the torque the dashed dotted lines denote the additional torque from the
cambering forces. The camber effect when included as a shift on the lateral slip is
shown by the dotted lines for γ = 0, −5, and −10 deg.
see Figure 5.3. A validation of the effect of camber on combined slip was
therefore, not possible. Instead the properties and results are illustrated
in the following. The parameters of interest for cambering are Cy = 1.4
kN/deg, Cz = 54 Nm/deg, and Cγ = 26 N/deg. It can be noticed that
the influence of camber on the lateral tire force is very low for the tire,
i.e. Cγ is small, and can be neglected for any practical case. Interestingly,
the camber stiffness calculated from the aligning stiffness, as described
in Section 4.2 by combining (4.53), (4.45) and (4.42), Ĉγ′ = 2kĈz , is sig-
nificantly larger, Ĉγ′ = 120 N/deg. To be able to visualize the effects of
camber, the computed larger stiffness was chosen. In Figure 5.13 the ef-
fect of camber is shown for sweeps of λ and in Figure 5.14 for sweeps of
α.
It can be seen from the result that camber hardly affects the longitu-
dinal force. It affects the partitioning between sliding and adhesion, but
the lower adhesion force is compensated by a higher sliding force. In other
tire models, for example presented in [Hirschberg et al., 2002] the camber
effect is included as a shift on the lateral slip, i.e. the lateral slip gets
an additional term of Cγ γ / Cy . This much simpler approach to handle the
86
Tire Forces at γ=0, −5, −10 [deg] and α=0, 3, 6 [deg]

6
4
−F [kN]
3
y
0
−6 −4 −2 0 2 4 6
−F [kN]
x
Figure 5.15 Illustration of the effects of camber on the lateral force as a function
of the longitudinal tire force at combined slip. The solid lines shows zero camber,
dashed lines γ = −5 deg and dashed-dotted lines γ = −10 deg. The transitions to
full sliding are marked with asterisks.
camber is plotted in the figures and the results for pure slip cases are
similar. For the combined-slip cases the results differ significantly, since
the offset has an undesired influence on the sliding velocity and also on
the sliding-force component. The self-aligning torque also differs between
the two methods. Figure 5.15 shows the camber effect on the lateral force
as a function of the longitudinal tire force at combined slip. Note that
camber has a strong influence on the point of transition to full sliding at
larger lateral slip.
Discussion The difference between the theoretical camber stiffness, Ĉγ′

and the measured camber stiffness, Ĉγ for six different tires is shown in
Table 5.1. It is noticed that the deviation between the calculated and the
measured camber stiffness is smaller for tires for personal cars and motor-
cycles than for tires for heavier vehicles as trucks or the Jeep, used in the
example above. A theory that is discussed in [Pacejka, 2002] is that the
assumed bristle deformation due to the camber is most accurate for tires
with a rounded shoulder. For wider and flatter tires the camber effect
might be better explained by deformation in the tire walls and changes in
the contact pressure distribution instead of bristle deflections and carcass
87
Table 5.1 Comparison between measured and computed camber stiffness for dif-
ferent tires and loads. Note that â is calculated as 3 Ĉy / Ĉz for all tires except 185/60
R14, where it is measured. The calculation is an approximation which gives unre-
alistic values at higher vertical loads.
P225/75R15 7.60-15 [Pacejka, 1988]
Fz [kN] 2.4 4.2 6.0 7.8 9.6 2.5 4.5 6.3 8.1
R [mm] 350 350 350 350 350 317 317 317 317
Ĉy [kN/deg] 0.75 1.2 1.4 1.4 1.4 0.7 0.8 0.9 1
Ĉz [Nm/deg] 16 33 54 75 94 20 30 50 70
â [mm] 64 83 116 161 201 86 112 167 210
Ĉγ [N/deg] 9.2 17 26 40 61 100 125 133 145

Ĉγ′ [N/deg] 40 82 135 187 235 50 75 125 175
195/65 R15 185/60 R14 315/80 R22.5 160/70ZR17

[Pacejka, 2002] [Kageyama and [Pacejka, 2002] [Pacejka, 2002]
Kuwahara, 2002]
Fz 7 1.9 2.5 2.9 41 3
R 317 289 289 289 540 330
Ĉy 1 1 1.2 1.4 4 0.5
Ĉz 60 x x x 125 15
â 180 75 87 96 94 90
Ĉγ 100 30 45 50 <100 50
Ĉγ′ 150 62 87 112 312 37
deformations, see Figure 5.16. Through the difference between Ĉγ and Ĉγ′
it may be possible to calculate the stiffness of the carcass during camber,
since γ tot = γ + Ccγ Mx , Ccγ is the carcass stiffness in the actual direction
and Mx the overturning torque. It would also hold that Ĉγ′ γ = Ĉγ γ tot for
small angles, which gives Ccγ = γ ( Ĉγ′ − Ĉγ )/ Mx . A further validation has
not been performed and the reasoning should be seen as a possible ex-
planation to the distinction between the measured and calculated camber
stiffnesses.
The dynamic properties of the tire model

The available measurements from NHTSA included data for evaluation of
the lateral dynamic properties at one velocity. The transient data is shown
in Figure 5.17 and generated by a sinusoidal excitation of the slip angle.
The amplitude of the oscillations is small and the tire can be assumed
88
γ γ tot
γ
y
z
Figure 5.16 Schematic picture of a cambered wheel, illustrating the effect of flex-
ibility in the carcass, which may explain the difference between the theoretic and
the actual camber stiffness
to behave linearly. According to the the relaxation-length concept from

(4.80) the relation between the wheel rim slip angle and the lateral force
is given by the transfer function
Cy
Fy = α′ (5.36)
σ ay
s+1
vx
The parameters in (5.36) were identified from the measurement data as

Cy = 1.4 kN/deg and σ ay = 0.57 m. In Figure 5.18 the identified transfer
function is compared with the data in a Bode-plot. The misfit of the phase
was supposed to be the effect of a time delay between the collection of the
slip-angle signal and the force signal. The fit is greatly improved with the
addition of a delay of a half sampling period. The identified stiffness also
corresponds well to the static measurements in Figure 5.4 where Cy =1.3
kN/deg and Cx = 1.4 kN/%. Since no measurements are available in the
longitudinal direction σ ax = 0.5 m is used. Relation (4.82) now gives the
carcass stiffnesses as Ccy =2.3 kN/deg⋅m and Ccx =2.8 kN/%⋅m. Carcass
damping coefficients were chosen to D x = 4 s/%⋅m and D y = 7 s/deg⋅m.
The aligning stiffness was identified from Figure 5.4 to Cz =60 Nm/deg.
89
Dynamic measurements in lateral direction, v=48 km/h, Fz=6 kN

1
Slip angle [deg]

0.5
−0.5
−1
0 10 20 30 40 50 60 70
1
Fy [kN]
−1
−2
0 10 20 30 40 50 60 70
Time [s]
Figure 5.17 Measurement of the force response for sinusoidal slip-angle inputs at
different frequencies. The sampling frequency is 50 Hz.
The test velocity was given as v0 = 48 km/h. All results below is for v = v0 .
Result for transient input The dynamic system (4.79) was simulated
on the form
Fx Fx
" #" #
λ
− D x vx “ α ” λ̇
Fy Fy
λ
cos2 (α ) cos(α ′ ) α
cos2 (α ) − D yvx cos(α ′ ) α̇
" #
−1
Cxσ ax v x (λ − λ ′ )
= −1
(5.37)
Cyσ ay vx cos(α ) sin(α − α ′ )
where the contributions from λ˙′ and α˙ ′ are neglected. Note the trigono-
metric re-writings to avoid numerical problems at slips angles of 90 deg.
Partial derivatives of tire forces with respect to slips are required for sim-
ulation of the model. In general, these are not analytically available and
in the following they are computed by finite differences in each simulation
step.
Figure 5.19 shows the response of the contact slip, λ , and longitudinal
force for rim-slip steps, λ ′ , of different magnitude at pure longitudinal
slip and constant lateral rim slip α ′ = 5 deg. The results agree qualita-
tively well with reports in the literature [Pacejka, 2002]. Note the quicker
90
Comparison between estimated system and measurements, v=48 km/h, Fz=6 kN

1
10
0
Amplitude 10
−1
10
−2
10 0 1 2 3
10 10 10 10
0
Phase (degrees)
−20
−40
−60
−80
0 1 2 3
10 10 10 10
Frequency (rad/s)
Figure 5.18 Bode plot showing the fit of a linear relaxation-length model (solid)
with the measurement data (stars). Addition of a time delay (10 ms) results in
better fit for the phase (dashed).
response for larger steps. The figure also illustrates the influence of com-
bined slip on the dynamics. Figure 5.20 illustrates the dynamic cross-
coupling between λ and α at step inputs in λ ′ and α ′ . The cross-coupling
is larger and the dynamics are faster at larger slips.
To visualize the effect of the dynamics of the tire in a more realis-
tic setup the following results consider the dynamic of a whole wheel. In
automotive applications the control of λ ′ using the brake torque is a del-
icate problem. Regard the wheel in Figure 4.15. Torque equilibrium can
be stated as
Jω̇ = T − Fx (λ ) R e (5.38)
where R e is the wheel radius and T is the applied torque. Assuming
constant vehicle velocity vx the equilibrium can be rewritten as
J λ̇ ′ vx = ( Fx (λ ) R e − T ) R e (5.39)
Note that the tire is not a rigid body and that the assumption of a fixed
inertia J is an approximation. The inertia for the tested wheel was not
known and was set to J = 5 kg⋅m2 . It is also pointed out that the effec-
tive rolling radius R e might not be identical to the moment arm by which
91
14
12
10
Slip [Percent] 8
4
α′=0 [deg]
2
α′=5 [deg]
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
4
−Fx [kN]
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Distance [m]
Figure 5.19 Illustration of the responses in λ and Fx for 10%, 20% and 30% steps
in λ ′ at α ′ = 0 deg and α ′ = 10 deg.
6 6
Long. Slip [Percent]
Lat. Slip [Deg]

4 4
2 2
0 0
0 2 4 6 8 10 12 14 16 18
5
Tire Force [kN]
0
0 2 4 6 8 10 12 14 16 18
Distance [m]
Figure 5.20 Illustration of the effect of dynamic cross-coupling on λ , α , Fx , Fy at

step inputs in λ ′ and α ′ (solid: longitudinal; dashed: lateral).
92
6 6
Long.
Lat.
4 4
−Fx [kN]
−Fy [kN]
2 2
0 0
0 0.05 0.1 0.15 0.2 0.25 0.3
8 8
Long. Slip [percent]
6 6
Lat. Slip [deg]

4 4
2 2
0 0
0 0.05 0.1 0.15 0.2 0.25 0.3
Time [s]
Figure 5.21 Simulation of step responses in the brake torque (500, 1000, 1500
Nm) while keeping the rim-slip angle constant at 2 deg. Upper plot: contact forces;
Lower plot: contact slips.
the tire force applies a moment on the wheel. However, the difference
is in practice small and they are assumed to be equal in the following.
Figure 5.21 shows the response of the wheel dynamics (5.39) in combi-
nation with the tire model (5.37) for step inputs in the brake torque at a
constant lateral rim-slip. Note that the effect of cross-coupling is greatly
reduced when the wheel dynamics are introduced. In practice, the brake
torque may appear like a spike in which case the cross-coupling effect
would again have more influence. It is clear that the responses are less
damped for smaller brake applications. This may be understood from a
closer examination of the combined dynamics.
The characteristic polynomial of the linearized constant-speed wheel
dynamics at a stationary operating point with constant brake torque is
ϕ R2e D x − Ccx vx J ϕ R2e Ccx

s2 + s+ (5.40)
J (ϕ − D x vx ) J (ϕ − D x vx )
where ϕ = Fx (λ )/λ p○ is the gain from contact-patch slip to longitudinal

contact-force at the operating slip. Note that ϕ < 0 in the small-slip re-
gion. Figure 5.22 shows the root locus of the characteristic equation with
respect to ϕ . This corresponds to the slope of the force-slip curve that
is transversed during a braking action. For large negative ϕ that corre-
93
60
−∞
40
20
φ φ
0
0
Im
−20
−40 φ
−∞
−60
−400 −350 −300 −250 −200 −150 −100 −50 0 50
Re
Figure 5.22 Root locus of the characteristic polynomial (5.40).
sponds to the braking stiffness at small slip the poles are complex and
poorly damped. When ϕ increases the poles become real. This explains
the increased damping at larger brake torques. As ϕ approaches zero one
pole goes to a stable limit and the other approaches zero. The introduc-
tion of the damping D x lets ϕ grow to the positive value specified by the
conditions in Section 4.3 while maintaining stability for the carcass dy-
namics. The wheel dynamics, however, becomes unstable as the slope of
the force-slip curve gets positive. The root locus clearly shows that wheel
and tire dynamics can not be separated and that any control application
that involves the wheel rotation benefits from taking the tire dynamics
into account. It is well known that the vehicle velocity vx greatly affects
the wheel dynamics and that slip control is harder at low velocities. In-
spection of the characteristic polynomial at limit conditions reveals the
interesting fact that the dynamics change equally for variations in the
slip operating point and the vehicle velocity. For ϕ → −∞ or vx → 0 the
characteristic polynomial goes to
R2e D x R2 Ccx
s2 + s+ e (5.41)
J J
94
5.5 Implementation and Application in MBS Systems
and for ϕ → 0 or vx → ∞ to

Ccx
s s+ (5.42)
Dx
Future work will address quantitative validation with measured tran-

sient data. A problem in this context is the possibilities to separate the
contribution of dynamic behavior from the wheel and tire from deflections,
back-lash, and hysteresis of the wheel suspension in a real application.
Section 6.2 treats these questions in a more practically way.
It is also of interest to reveal if the cross coupling between the longi-
tudinal and lateral dynamics is relevant in practical applications.
5.5 Implementation and Application in Multi-Body

Simulation of Vehicle Handling
There are many different software environments available for modeling
and simulation of multi-body mechanics, both for general mechanics and
applications specific for vehicle dynamics. The VehicleDynamics LibraryTM -
[Modelon, 2006] for DymolaTM [Dynasim, 2006] is chosen, due to the close
connection to the developers, to extend the usability and the possibility
of validation of the model, see [Gäfvert et al., 2006]. Dymola is a model-
ing and simulation environment for general multi-domain modeling and
simulation. It is based on the open high-level ModelicaF R language that
is maintained by the Modelica Association [Mod, 2005]. The VehicleDy-
namics Library is an add-on to Dymola that offers complete and detailed
multi-body vehicle models for vehicle dynamics analysis.
The multi-body model describes the vehicle as a set of rigid bodies con-
nected by joints (constraints) or force elements. The description is based
on orthonormal reference frames that are attached to the bodies. The ref-
erence frames follows the ISO 8855 standard with x pointing forward, y
to the left, and z upwards.
The wheels are treated as bodies with force elements that describe
the interactions between the chassis and the track surface. The rim of
each wheel is rigidly connected to the suspension wheel carrier by the
hub, and may be kinematically defined by the chassis and wheel-spin
states. The tire is a vertically deformable body that is attached to the
rim and interacts with the ground in a contact patch of varying size and
location. In principle, the contact patch is represented by a contact point
at which the forces and moments are assumed to apply, located where a
line through the wheel center, aligned with the wheel plane, intersects
the track surface. Moreover, the deformation of the tire is normally only
95
Figure 5.23 To the left: Simulated test rig for analyzing the transient and sta-
tionary behaviour of tyre models. The test rig can apply longitudinal and lateral
slips as well as camber at desired loads and velocities. To the right: Visualization
of simulated NHTSA-Fishhook handling maneuver.
explicitly described in the radial direction. In this way, many effects of

the tire deformation are lumped into the description of contact forces in
the contact point. This also makes it easy to use empirical data from
experiments to characterize the tire. The orthonormal reference frames of
the wheels are chosen to support this tire-modeling paradigm, and follow
the TYDEX conventions [TYDEX-Working group, 1997].
Simulation Results
The combined-slip model is combined with the Magic Formula pure-slip
models described in [Bakker et al., 1987] with parameters from the same
source. The combined-slip characteristics of the model were first deter-
mined by simulations in the virtual test rig shown to the left in Fig-
ure 5.23. The experiment is a sequences of sweeps in either λ or α , with
the non-swept slip held constant. The result can be visualized as in Fig-
ure 5.24 and give a good view of the characteristics and limits of the tire.
The first figure shows the resulting forces for zero inclination without ve-
locity dependence. It is seen that small slips give negligible combined-slip
influence, while at larger slips the force envelope describes the ellipse-
formed limits of the tire. The second plot shows the shift effects of adding
camber. The third and fourth plots show how lower speeds will increase
and higher speeds decrease the force envelopes as the velocity dependence
is turned on. The tire model was also used in simulation of handling
96
5.6 Conclusions
Figure 5.24 Combined-slip force characteristics at 4kN load as plots of lateral

force as function of longitudinal force for experiment in tyre test rig with slip sweeps
λ ∈ [−1, 1] and α ∈ [−89, 89] deg at constant levels λ ∈ {0, ±logspace(−3, 0, 10)}
and α ∈ {0, ±logspace(0.1, 89, 10)}. Upper right shows the effect of adding 4 deg
camber, and the lower left and right plots show velocity dependence at v = 0.5v0
and v = 2v0 , correspondingly. Note that the plots include some straight lines that
are the result of re-adjusting the rig between the slip sweeps.
maneuvers with a 40-DOF chassis model with elasto-kinematic suspen-

sion models, see the right hand side of Figure 5.23. The presented tire
model gave similar results as Magic Formula [Pacejka, 2002] and TMEasy
[Hirschberg et al., 2002] with similar simulation performance. Absence of
calibrated parameters for the three compared models for one specific tire
makes it difficult to draw detailed conclusions.
5.6 Conclusions
This chapter has presented a new method to derive the tire forces for si-
multaneous braking, cornering, and camber, by combining empirical mod-
els for pure braking, cornering, and cambering. The proposed model is
97
based on understanding of the physical sources of tire forces, as given

by the theoretical rigid-carcass brush-model. Based on the brush model,
the combined-slip forces may be described by a scaling of corresponding
empirical pure-slip forces. There is a freedom in choosing the pure-slip
forces, which may be used to emphasize different physical effects. This
made it easy to include velocity dependence in the model, in a unique
way. The generation of the camber force has been described in a simple,
but physically motivated manner. Its influence on the partitioning of the
sliding and adhesive regions in the tire contact patch, affects the scale
factors for the tire-forces. For pure slips it is common to introduce camber
by an offset on the lateral slip. However, for the self-aligning torque and
the combined slip forces the proposed model gives significantly different
result. It has also been discussed that the effects of camber, with varying
accuracy, can be approximated from the lateral properties of the tire. The
model is simple to use since it does not introduce any new parameters.
All necessary information is given by the used pure-slip model.
Even though the model in its present state is useful for a number of
applications, the addition of effects of turn-slip and flexible carcass would
further enhance the applicability. The physical foundation of the model is
expected to make this possible with reasonable effort.
Relations to similar, previously published combined-slip models are
analyzed. Previous models are partly heuristically based, while the pro-
posed model is entirely based on physical principles. An advantage with
the presented model is that it combines the data-fitting accuracy of em-
pirical pure slip models with the stringency of physical modeling. The
calibration of physically-based models to measurements is often limited
by the small set of parameters and the approximations and idealized or
neglected effects in such models. Therefore, more flexible heuristic mod-
els such as the Magic Formula [Pacejka, 2002] are more commonly used
nowadays both for pure and combined slip situations. Despite good fit at
calibration with measurement data, deviations from real behaviour will
still arise as the normal operating conditions for a tire usually differ sig-
nificantly from those at calibration-measurement experiments. Empirical
models are normally valid only within the input intervals where calibra-
tion measurements exists and extrapolation properties are often very un-
reliable. Moreover, the large amount of factors that affects the tire behav-
ior makes it practically impossible to collect and manage data for all kind
of influencing factors. These observations raise the question whether it is
better to use physically based models with less perfect fit to measurement
data, but with better extrapolation properties outside the measurements
ranges.
Good results are obtained in validation with the available empirical
data. Experimental validation indicates that the proposed model and the
98
5.6 Conclusions
BPL model [Bakker et al., 1989] performs similarly. The COMBINATOR

model [Schuring et al., 1996] behaves slightly differently at small com-
bined slips. None of these previous models include velocity dependence.
The tire model is successfully implemented in Matlab code and in the
multi body simulation environment, Dymola. It works well and is compu-
tationally sound on both platforms.
The proposed dynamic extension of the proposed combined-slip model
exhibits a qualitatively realistic tire relaxation behaviour. The model in-
cludes cross-couplings between longitudinal and lateral dynamics, that
are not commonly found in other models of similar complexity. The issue
of stability is handled by including a physically motivated damping. Ex-
amination of the combined dynamics of a wheel and the tire model gives
insights that might be relevant for brake-control applications.
99
6
Measurement and
Processing of the Wheel
Speed Signal
6.1 Introduction
The rotational velocity of the wheels are some of the most essential sig-
nals for vehicle control and stabilization. The signals contain information
about several factors, relating to the road foundation, the tires, and the
motion of the vehicle. In some road detection systems the variance of
the wheel speed signal is used to distinguish between smooth and un-
even foundations. In vehicle simulations and in many friction estimators
the normalized difference between the wheel and vehicle speed, slip, is
very important in its relation to the tire force. The proportion between
the wheel speeds on each tire can reveal information about tire pressure
or tread wear. For the stability of the vehicle it is of major concern that
the wheel do not loose the grip on to the road. Any tendency to wheel-lock
must be detected early from the speed signal and then be safely prevented.
The high degree of information makes the speed signals valuable, but on
the other hand noisy and hard to utilize, if the different effects are not
correctly separated from each other.
The common way to measure the wheel speed is to use a sensor that
registers the magnetic field between the sensor and a tooth-wheel. The
sensor is mounted at the end of the axle and the tooth wheel is attached
to the wheel rim, see Figure 6.1. The general idea is to measure and
count repeated variations in the magnetic field due to the rotation of the
tooth wheel. The measurement methods are mostly based on induction, or
100
6.1 Introduction
direct measurement of the magnitude of the magnetic field. An inductive

sensor consists of a core and a coil, which generates a voltage due to the
changes of the magnetic field, see [ProAuto Limited, 2006]. The output,
when the wheel is rolling, is a wave formed signal, centered around zero.
The frequency and amplitude depends on the number of passed teeth per
time unit, which makes the velocity signal hard to distinguish from the
noise at low speeds. This sets a lower limit of the working range for this
type of sensor.
The magnitude of the magnetic field can be measured by a Hall-
element. A Hall-element has to be used with a magnetic tooth wheel or
with a magnetic material placed behind the sensor such that the tooth
wheel can affect the magnetic field surrounding the sensor. Practically,
the Hall-element produces a voltage depending on the distance to the
closest teeth. A sensor with Hall-element is often refereed to as an ac-
tive sensor, since a circuit for signal processing often is included in the
sensor package. The processing unit is necessary to identify the passage
of the tooth flanks, which is more difficult for a Hall-element since the
output is not centered around zero and the bias, sensitive to the gap be-
tween the sensor and the tooth wheel, might vary with time and around
the peripheral. Two Hall-elements are often used together to improve the
repeatability, and to make the sensor possess the ability for status diag-
nosing and determination of the wheel-rotation direction. The processed
output use to be, each time a tooth flank passes, a short pulse contain-
ing coded information from the sensor. More detailed information can be
found in, for example, [Ana, 2001]. The speed of the wheel is derived from
the inverse of the time between the pulses from the sensor. Modern mi-
croprocessors provide flexible timers, separated from the main core, that
counts the elapsed time without affecting the execution of the real-time
kernel. An interrupt at each pulse might be necessary for storage and data
processing in more advanced routines. At high speed the pulses come of-
ten and the number of performed instructions at each pulse are limited
by the controller capability.
It is important to get an accurate disturbance free wheel-speed signal,
since the noise may be heavily amplified in certain computations. Partic-
ularly in applications depending on the slip or on the actual tire force.
The slip is the difference between two almost equal signals and a noise
level of up to one percent of the wheel speed is much compared to a slip
signal in the range of 0 to 20 percent of the wheel speed. The derivative of
the speed is used to calculate the magnitude of the wheel-inertia torque
that opposes the brake torque when deriving the actual tire force. Differ-
entiation of a noisy signal is a crucial operation requiring effective filters
which often deteriorates the timing capability.
Normal noise filters that introduce extra phase lag has to be used
101
Chapter 6. Measurement and Processing of the Wheel Speed Signal
Figure 6.1 Picture of a rear truck-axle showing axle, rim, brake unit, tooth wheel
(TW) and the placement of the speed sensor (SP) (Modified, by courtesy of Haldex
Brake Products AB).
with great care, not to delay the detection of a beginning wheel-lock situ-
ation. In some conditions, particularly at low speed, the wheel might lock
quickly and any delay on the signal might be hazardous for the stabiliza-
tion system. However, the signal processing has to be efficient enough not
to unnecessarily invoke e. g. the anti-lock system.
The sensors used nowadays often measure accurately, but it is impor-
tant to regard that they measure the magnetic field due to the passage of
a tooth in front of the sensor pick-up. This do not always coincide with the
tire speed that is useful for the vehicle control system. There are three
main noise sources, among others, that affect the measurement of the
wheel speed:
• Brake force transients There is a dynamic response from the

mechanical system around the wheel axle when a braking torque
is applied onto the wheel. Two different physical effects are then
present. First, the movement of the wheel-axle end, where the brake
and wheel speed sensor is mounted, when the axle is subjected to
a reaction torque from the brake affects the wheel speed measure-
ment. Since the relative speed between the wheel and the sensor is
102
6.2 Effect of Tire Force Transients on the Wheel Speed Measurements
measured the sensor movement has to be accounted for in such a

situation. The second effect is the dynamic system consisting of the
wheel inertia and the tire stiffness. The tire deflects and affects the
force transmission from the brake pads to the tire-road contact.
• Rotational dependent noise Irregularities of the tire shape and

asymmetries of the tooth wheel causes disturbance patterns on the
speed measurements. The disturbance patterns are repeated each
wheel revolution.
• Road irregularities Bumps and unevenesses in the road might

temporarily effect the rolling radius of the wheel. For this type of
disturbances there might be a correlation between the front wheels
and the rear wheel with a delay corresponding to the wheel base of
the car. In some axle constructions, suspension movements affects
the sensor position, which causes even further measurement errors.
A simple model of the axle behavior and a proposal on how the influence of
the transient brake force on the measured wheel speed can be accounted
for is described in the following section. The tire dynamics is previously
treated in Section 2.6 and 4.3. Section 6.3 reviews and describes some
methods to eliminate the rotation dependent noise from the speed signal.
Processing the noise from road irregularities is a large area for further
work. In, for example, [Gustafsson et al., 2006a] the dependency between
the front and the rear wheel speed covariances are used for determination
of the absolute velocity of the car. This field in not further treated in this
thesis. Additional noise sources are, e. g. mentioned and discussed in
[Schwarz, 1999]
6.2 Effect of Tire Force Transients on the Wheel Speed

Measurements
This section will mainly discuss the disturbance on the wheel speed mea-
surements caused by movements of the sensor position. The effect is par-
ticularly obvious for front axles, since these, generally, are relatively weak
and flexible mechanical constructions. The brake that creates and carries
the reacting braking torque is mounted on the axle end, which deflects
when supporting the load. Forces acting on the tire due to, for example,
the engine torque are also subjected to a dynamic behavior, but do not, in
the same way, affect the wheel speed measurement. Rear axles are often
of more robust design and deflects less, while the dynamic behavior of the
tire still is present.
103
Veh Ret. [m/s ] Br. Torque [kNm] Whl Sp. [km/h]

35
30
25
48.4 48.6 48.8 49 49.2 49.4
10
0
48.4 48.6 48.8 49 49.2 49.4
5
2
−5
48.4 48.6 48.8 49 49.2 49.4
0.2
0.1
Slip
0
−0.1
48.4 48.6 48.8 49 49.2 49.4
Time [s]
Figure 6.2 Plot showing the effect of a front-axle brake-force transient on the
wheel-speed measurement. A step command on the brake pedal causes an apparent
decrease of the wheel speed. The ABS reacts and lowers the brake force. After an
overshoot the wheel speed returns to a stable level and the brake system can come
back to normal mode. The rear axle is totally unbraked. The solid lines denotes the
front left, dashed lines the front right, dashed dotted the rear left and dotted line
describes the rear right wheel. The vehicle retardation (third plot) are derived from
accelerometer measurements.
An example of the effect can be viewed in Figure 6.2, that shows mea-
surements collected from a Scania test truck at Haldex. The transient
of the brake torque, in the interval t=48.4–48.6 s, causes a dynamic be-
havior of the wheel speed signal, which is clearly seen from the top and
bottom subplots. The overshoot in slip during the application phase largely
exceeds the settled value at the end of the data series even though the
brake torque is as largest at the end.
In [Schwarz et al., 1997b] the effect is modelled as longitudinal suspen-
sion stiffness and accounted for in the slip determination. In [Sugai et al.,
2003] a method is presented to on-line estimate a coefficient denoting the
entire influence of the longitudinal suspension motion and tire carcass
stiffness. Here, the behavior will be modelled and discussed. Based on the
available measurements, the requirements on a model that predicts the
tire-wheel-axle dynamics are studied in an empirical way.
104
ϕa
MB
Figure 6.3 Schematic figure of the axle and wheel hub. The axle rotation are
here supposed to carry the effects from all kind of deflections of the suspension and
tire. The movement of the sensor causes the disturbances on the wheel rim speed
measurements.
Dynamic model
The reaction of the wheel speed measurements, due to transients in the
brake forces, is explained by many different effects. One part is the move-
ment of the wheel hub and wheel speed sensor due to deflection of the
axle and suspension. Another part is the dynamics of the tire, caused by
the carcass flexibility and the inertia of the wheel, discussed in detail in
Sections 4.3 and 5.4.
In the following it is assumed that the movement of the sensor relative
the tooth wheel is caused by axle torsion, see Figure 6.3. In reality, the
movements of the axle are much more complex, see e. g. [Matschinsky,
1997], and the sensor together with the wheel axle will additionally exhibit
linear motions. The deflections are here lumped together in one equation,
since the major part of the respective dynamics can be described as a
mechanical system, characterized as
1
ϕa = MB (6.1)
Ja s2 + Da s + K a
It is therefore emphasised that, the inertia, Ja , the damping, Da , and,

the stiffness, K a are generic parameters gathered from the entire system.
They do not directly correspond to the physical, measurable quantities of
105
the axle, since many parallel systems are incorporated in this equation.
The deflection angle, ϕ a , denotes the movements of the wheel speed sensor
and the brake torque is denoted by MB . The actual speed of the rim ω
can be calculated from
ω = ω̂ − ϕ̇ a (6.2)
where ω̂ is the measured wheel speed. For the wheel speed measurement
there is a difference if the sensor makes a rotational movement or if the
sensor and the wheel axle together gets a linear deflection. In the first case
the speed measurement is actually wrong, since the sensor moves with
the wheel. In the other case the speed is correct, but the wheel reference
speed, ω actually deviates from the vehicle reference. It is assumed that
lumping of these motions into one rotational motion is a good engineering
approximation.
Inserting (6.1) in (6.2) gives
s
ω = ω̂ − MB (6.3)
Ja s2 + Da s + K a
Included in ω there is another dynamic response of the force transient due

to the tire deformation. The response is velocity dependent and should be
considered when the speed of the tire carcass in the contact with the road
is to be derived, which normally is the case when modeling tire forces
from the slip. The relation between the wheel rim speed and tire carcass
speed is in detail discussed in Section 5.4. In Figure 6.4 the result from
a simulation of the axle, wheel, and tire including these effects is shown.
For the axle simulation Equation (6.1) is used and the tire and wheel is
modelled according to (5.37) and (5.38). The brake torque sequence visi-
ble in Figure 6.2 is used as input signal to the system. It can be seen that
major part of the dynamics is caught by the model. Proper parameters for
the system is, however, difficult to find since the tire and axle parameters
are not entirely known. The unknown parameters used in the simulation
example are more or less manually chosen to give acceptable result and
more effort can be done to further increase the accuracy. Tire characteris-
tics given by the manufacturer is used, even though the reliability is low
since the conditions at the testing facilities certainly, are different from
those at the measurement occasion.
Identification of the dynamics

The System Identification toolbox in Matlab is employed to increase the
agreement of the model. The aim is to find an accurate relation between
the brake torque and the wheel speed during transients. Finding a sim-
ple transfer function with good accuracy is of more interest than finding
the optimal parameters to the physical model from the prior section. The
106
34
33
32
Time [s]
31
30
29
28
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
Figure 6.4 Simulated wheel speed response (solid line) compared to measure-
ments (dashed line) for the scenario shown in Figure 6.2.
identified model dynamics is assumed to behave as any first or second-

order transfer function, G (s). The signals necessary for the identification
are derived from Equation (6.3), rewritten as
ω − ω̂ = G (s) MB (6.4)
The real wheel velocity ω can not be measured. Here, it is approximated

from wheel speed measurements of the unaffected rear axle, but compen-
sated for the slip (due to the applied brake force), such that there is no
stationary gain in the identified system. The slip is caused by the tire be-
havior and should not be incorporated in the axle behavior. In that way the
transient effects of the torque application becomes clear, see Figure 6.5.
The velocity dependency of the dynamics caused by the tire carcass
is then completely neglected, since it is incorporated in the velocity inde-
pendent system for identification. In practise it is difficult to distinguish
between the different effects. Tests at different speed are required for a
further evaluation.
Using the Matlab command arx, the discrete equivalence H ( q) to G (s)
is identified on the form
b0 + b1 q−1 + b2 q−2
H ( q) = (6.5)
a0 + a1 q−1 + a2 q−2
107
35
34
33
Wheel Speed [kph]
32
31
30
29
28
27
0 0.05 0.1 0.15 0.2 0.25 0.3
Time [s]
Figure 6.5 Illustration of measured wheel velocity, R eω̂ (full line, left wheel and
dashed, right wheel) and the constructed accurate wheel velocities R eω (dashed-
dotted line for the left wheel and dashed for the right wheel).
The measured data set contains two occasions where the effects of the
torque transients are clearly visible. The first is shown in Figure 6.2 from
where the time slot [48.4, 48.65] is used for identification. The second
occasion is used only for validation. The identified parameter values are
presented in Table 6.1 and the obtained dynamics are graphically shown
in Figure 6.6. The result is almost similar for the first and second order
system and no accuracy is gained by higher order than one. The right
wheel shows better accuracy than the left-side from where there seems to
be a timing error of approximately 0.01 s between the simulation result
and the measurements. Reasons for this are discussed later on.
The result from the identification is used to reduce the disturbance of
the brake force application on the wheel speed. The corrected rim speed
is then calculated as
ω (t) = ω̂ (t) − H ( q) MB (t) (6.6)
In Figures 6.7 and 6.8 the correction is validated towards the measure-
ment data for the two available test occasions. It is clear that the dis-
turbance on the wheel speed is greatly reduced. The best result is seen
in the first transient on the right side, where the disturbance is almost
108
8 8
Input (Br Tq) Input (Br Tq)
Output Output
1st ord. sys 6 1st ord. sys
6
Wheel Speed [kph] / Br Torque [kNm]
Wheel Speed [kph] / Br Torque [kNm]

2nd ord. sys 2nd ord. sys
Full model Full model
4
4
2
2
0
0
−2
−2
−4
−4 −6
48.4 48.45 48.5 48.55 48.6 48.65 48.4 48.45 48.5 48.55 48.6 48.65
Time [s] Time [s]
Figure 6.6 Comparison between the output from the identified system and the
measurements ω −ω̂ for left (left plot) and right (right plot) front wheels. The dotted
line denotes the result from the full-model simulation described in Section 6.2
34 39
Wheel Speed Wheel Speed
1st ord. corr 1st ord. corr
38
2nd ord. corr 2nd ord. corr
33
37
32 36
Wheel Speed [kph]
Wheel Speed [kph]
35
31
34
30 33
32
29
31
28 30
48.3 48.4 48.5 48.6 48.7 48.8 137 137.1 137.2 137.3 137.4 137.5
Time [s] Time [s]
Figure 6.7 Comparison between the measured wheel velocity and the corrections
applied with the two different filters according to (6.6). To the left: The first brake
application where the disturbance is seen, which also is used for the identification.
To right: The second brake application for validation. Data from the left-front wheel.
removed. In the other plots the disturbances still are present, but reduced
by at least 50% and the first order system performs slightly better than
the second order system.
Conclusion The aim of this section has been to find a way to com-
pensate for disturbances on the wheel speed signal created by axle and
suspension deflections during the brake application. From the validation
it can be seen that the disturbance is not entirely eliminated, but its
amplitude is greatly reduced by very simple means. The correction on the
wheel speed signal can be seen as a feed-forward filter of the brake torque,
predicting the disturbance. A first order filter is sufficient and higher or-
109
35 39
Wheel Speed Wheel Speed
1st ord. corr 1st ord. corr
34 38
2nd ord. corr 2nd ord. corr
37
33
36
Wheel Speed [kph]
Wheel Speed [kph]

32
35
31
34
30
33
29
32
28 31
27 30
48.3 48.4 48.5 48.6 48.7 48.8 137 137.1 137.2 137.3 137.4 137.5
Time [s] Time [s]
Figure 6.8 Data from the right front wheel, compare to Figure 6.7.
Table 6.1 Result of the parameter identification of (6.5).
Side a0 a1 a2 b0 b1 b2
Left 1 0.283 0 -0.00249 0.00255 0

Right 1 0.411 0 -0.00301 0.00299 0
Left 1 -0.131 -0.217 -0.00293 0.00497 -0.00206
Right 1 -0.15 -0.348 -0.00319 0.00531 -0.00213
der filters do not improve the results. A remarkable result is that poles
of the identified discrete system are on the negative real axis. For the
second order system one pole stays on the negative and the other on the
positive stable axis. This makes it impossible do a physical interpretation
of the systems. Comparing the pole to the slightly unstable zero in the
first order system shows that behavior due to the pole is overridden by
the effect of the zero. The behavior can almost be seen as a pure differ-
entiation, which also will be the result if the the denominator is set a
constant while re-doing the identification.
The disturbances are reduced by at least 50% in the available valida-
tion data and this is probably sufficient to increase the performance of
vehicle systems relying on the wheel-speed signal during the brake-force
transients. A drawback, so far, is the limited amount of measurement
data. More test occasions are necessary for a more comprehensive valida-
tion. Tests on different types of axles are another necessary requirement.
A question mark is if the neglection of the tire characteristics still is ac-
ceptable in the identification of a stiffer axle or at a lower vehicle velocity.
Since the relative contribution of the tire then is larger. Also, the stiffer
110
6.3 Method for Elimination of Rotation Dependent Noise
axles have less impact on the wheel speed and the dynamics then might
be completely disregarded. In practise, it is difficult to distinguish the tire
dynamics from the axle dynamics. A result from Section 5.4 is that the im-
pact of the carcass deflections are less obvious at higher slip. Comparing
the dynamics at high and low slip might be an accessible way to separate
between wheel and axle dynamics. Performing experiment at different ve-
hicle velocities is another. The method of approximating the suspension
and tire stiffnesses in one coefficient and identify it on-line as described
in [Sugai et al., 2003] seems to be an realistic approach, since the major
part of the dynamics shows to be described by this coefficient. A problem
with on-line estimation in this case may be that quick and substantial
excitations necessary for the estimation might occur rarely during normal
run. When they suddenly occurs it might be in a limited condition and
optimal performance of the system is a demand.
On the other hand there are many ways to further improve the param-
eter identification. The system is very fast and the sample time of 0.01 s,
used in the collected measurements, is quite slow compared to the reac-
tion of the wheel speed. It is therefore assumed that shorter sampling time
would be advantageous for the estimation. For such fast system dynamics
delays and jitter on the signals caused by the CAN-bus transmission and
different sensor characteristics can be crucial. In Figure 6.6 there is a
delay of 0.01 s between the measurements and the identified systems for
the left side, while the right axle shows better agreement. The setup is
identical, but with the hard timing constraints the error may be explained
by varying latency on the signals.

There is, in general, noise on the wheel speed signal that is repeated ev-
ery revolution, with a scheme that is depending on the rotation angle of
the wheel. This type of noise might be hard to discern since its pattern
is distorted after sampling and filtering. To clearly see this kind of dis-
turbance, the signal has to be unfiltered and should be sampled at each
passage of a tooth flank or faster. This is not commonly done even on test
vehicles. The sources for the disturbances come, for example, from irregu-
larities of the teeth, varying air-gap between sensor, and a distorted tooth
wheel, and uneven worn or oval tires. To be able to eliminate such a noise
source, the position of the rim has to be known during the wheel rotation.
It is not necessary to determine the absolute angle, if there are sufficient
time for calibration of the algorithm after each restart of the system. The
important thing is to be able to compare a pulse signal at a certain po-
sition to the pulses from the same position at the prior revolutions. The
111
Figure 6.9 Illustration of a part of a tooth wheel, with angle θ between the teeth,
ideally.
general idea for the noise elimination is to derive a vector containing one
element per pulse generated by the sensor during one revolution of the
tooth wheel. The vector values are used for correction of the actual wheel
speed signal and calibrated in such a way that the signal irregularities
are cancelled. Similar methods have been presented in [Schwarz et al.,
1997a] and [Persson, 2002].
The measurement system

Let N denote the number of pulses from the sensor during one revolution.
Each pulse is indexed by k. Hence, tk is the time at pulse k, which is
generated by the tooth flank ik (when i is used as index, k will be left
out). The rotation angle of the rim between each pulse is ideally θ = 2π / N,
see Figure 6.9. Due to the circumstances mentioned above the pulses are
not evenly distributed along the travelled distance for one revolution of
the wheel. The angle corresponding to the rolled distance between two
pulses at the tooth i, see Figure 6.10, is
θi = θ + δθi (6.7)
where δ θ i is a measure for the distortions in the tooth wheel and uneven
rolling radius around the peripheral of the tire and belongs to the vector
δθ,
XN
δθ = [ δ θ 1 δ θ 2 .. δ θ N ], δθi = 0 (6.8)
i=1
The measurement of the rotational velocity can be expressed as a time
112
R eθ i−1 R eθ i R eθ i+1
∆ t k−1 ∆ t k ∆ t k+1
t k−2 t k−1 t k t k+1
Figure 6.10 Illustration of the pulses from the wheel speed sensor relative the
rotation of the wheel. The space between the pulses relates to the distance travelled
by the wheel, when it is not subjected to a slip. Any irregularities in the tooth wheel
and the dynamic tire radius are accounted for in δ θ . Therefore, R e might be treated
as constant.
varying system on state-space form as
x(tk+1 ) = Φ x(tk ) + e p (6.9)

y(tk ) = C (ik , tk ) x(tk ) + em (6.10)
where the state vector x is

T
1

x( t k ) = δ θ1 δ θ2 .. δ θ N (6.11)
ω (tk )
The system matrix, Φ is the identity matrix with N + 1 rows and columns.
The mean velocity during the time between tk−1 and tk is denoted by ω (tk ).
The process error, e p relates to the assumed variation of the inverted
wheel speed and the irregularities of the teeth. The irregularities are
assumed not to change, hence
e p = [ eω 0 0 .. 0 ]T (6.12)
The observation y(tk ) = [∆ tk ] = tk − tk−1 is the time elapsed between the

two latest pulses. The observation vector becomes
1
C (i k , tk ) = θ 0 .. 0 0 .. 0 (6.13)

ω (t k )
where the factor 1/ω (tk ) is placed corresponding to the element δ θ i , i. e.

on instance ik + 1. The measurement error, em is assumed to be white noise
113
due to road irregularities and disturbances from the signal conversion in

the sensor. Hence, the observation of the inverse velocity is
θ δθi
y(tk ) = ∆ tk = + + e(tk ) (6.14)
ω (tk ) ω (tk )
From the observations of ∆ t the aim is to identify the correct velocity, or

in this case, its inverse. By estimation of the irregularities of the teeth,
their influence on the noise on the speed can be eliminated.
Wheel speed filtering by use of a minimal-variance technique

If the disturbances on the system are Gaussian and respective covariances
are known sufficiently accurately, than a Kalman filter is an optimal way
to estimate the state vector even for time-variant system [Åström, 1970].
The recursive Kalman filter and the state estimation scheme is formulated
as
x̂(tk+1 ) = x̂(tk ) + K ( y( k) − C (ik , tk ) x̂(tk )) (6.15)

T T −1
K (tk ) = P(tk ) C (ik , tk )( R2 + C (ik , tk ) P(tk ) C (ik , tk )) (6.16)
P(tk+1 ) = P(tk ) + R1
−( P(tk ) C (ik , tk )T )( R2 + C (ik , tk ) P(tk ) C (ik , tk )T )−1 ( C (ik , tk ) P(tk ))
(6.17)
P(0) = R0 (6.18)
The condition of having the sum of δθ equal to zero is not yet included in
the formulations.
P To fulfill this specification, the observer, y, is extended
to also observe (δθ). Hence,
1
 
∆ tk ) 0 .. 0 0 .. 0 et

θ
yE (tk ) = ; C E (i, k) =  ω (tk ) ; em =
0 0
0 1 .. 1 1 1 .. 1
(6.19)
which means that an error-less observation that the sum is zero is per-
formed at each instant and the estimated states will adapt to this condi-
tion. The covariance matrices can be chosen as
rω 0 ⋅ ⋅ ⋅ 0
 
 0 r ⋅⋅⋅ 0 
rω 0(1, N ) rt 0

θ
; ;
 
R0 = 
 .. .. . . . R1 = R2 =
. ..

 . .

 0( N, 1) 0( N, N ) 0 0
0 0 0 rθ
(6.20)
114
where rω is the variance relating to the change rate of the wheel speed,
or more correctly, the inverse of the wheel speed. The variance of non-
repeating measurement errors due to, i.e road irregularities are incor-
porated in r t . It should, however, be a balance between r t and rω such
that quick wheel speed changes not are misinterpreted as temporal dis-
turbances. To be on the safe side, then r t are set very low. The covariance
of the irregularities, rθ , might not be known at the beginning, but after a
successful state estimation, it can be computed as cov(δθ). The vector δθ
does not change during normal run. However, there are occasions where
the counter can miss a tooth. Therefore, both at start up and when a
missed tooth is suspected the filter has to be reinitialized. Earlier com-
putations of cov(δθ) can then be used for rω . An alternative is to set the
diagonal elements in the process covariance matrix, R1 in the same way
as those of R0 and in that way introduce a forgetting factor in the estima-
tions. A drawback is that a trade off problem is encountered in choosing
between a quick re-adaptation and minimum variance on the estimates.
Different ways to supervise the algorithm, using different update modes,
can be more effective and give and better output in this case.
As an illustration of the method, a tooth wheel and a sensor that
outputs 10 pulses per revolution are simulated at constant speed. A vector
of tooth asymmetries is created, but there are no other disturbances on the
signals, so that the function of the state estimator can be easily illustrated.
Signal disturbances would increase the convergence time. The difference
of the measured output (time between the pulses, ∆ tk ) and the corrected
output (1/ω (tk )) is shown in Figure 6.11 using the proposed method. It
can clearly be seen that the variance of the corrected signal decreases
while the values in δθ-vector adjust. The result of the δθ estimation is
shown in Figure 6.12. Note, that all values in the vector changes at each
update such that the sum of the entire vector is zero.
The presented observer is regarded as an optimal observer, predict-
ing the states with minimal variance relating the to available measure-
ments. Setting r t low, the major tuning parameter is rω . It is important to
have a sufficiently high value on rω such that the velocity estimate reacts
fast and is calculated with minimum delay. If the actual wheel velocity
changes rapidly, for example, in a wheel-lock situation and the estimated
velocity can not adapt quick enough, there will be a disturbance on the
δθ-estimates. On the other hand, if the velocity reacts to fast the δθ-
values will not be calculated correctly, since the tooth-irregularities can
not be separated from velocity changes. In the method illustrated above
the variances are set as rθ = 0.04, rω = 0.0005, and r t = 0. In a realistic
approach the variances much be set more careful, concerning the temporal
disturbances on the wheel speed signal and the desired ability of the al-
gorithm to catch velocity changes of the wheel. A separate algorithm may
115
Figure 6.11 Results from minimum variance estimation of simulated tooth irreg-
ularities. The upper plot shows the “measured” pulse time and the lower plot shows
the pulse time corrected for the estimated δθ. The algorithm starts at 0.5 s.
be employed to compute the corrected velocity, ω , directly using (6.14)

backwards and the working areas of the algorithm may be limited to the
cases when the wheel velocity is not rapidly changing.
In practise, the presented minimum-variance method is too compu-
tationally demanding and memory consuming to be implemented in a
limited real-time system for this problem. Processors used in the automo-
tive sector constantly increase in power, and a demanding solution today
might be applicable in a few years.
Wheel speed filtering for suitable for on-line implementation

There are some ways to do the wheel speed filtering more computationally
efficiently. One example can be found in [Schwarz et al., 1997a] protected
by the patent [Isermann et al., 2000], where a vector, v, with calibration
factors, one value per pulse, is created and updated on-line. The mea-
sured velocity is corrected as ω corr (ti ) = 2π vi / y(ti ), where vi is iteratively
calculated from
2π

vi ( j + 1) = vi ( j ) + γ ω̄ − vi ( j ) (6.21)
∆t
on the corresponding tooth. The revolution number is denoted by j. The

update factor γ is updated according to the Recursive Least Squares
116
Figure 6.12 Estimated normalized angle errors compared to the true error
(dashed-dotted lines)
method for effective elimination of disturbances on v. A method to cor-

rect and detect a missed tooth by comparing the sequences of the latest
vi to the older values before update is also proposed. The actual speed
ω (tk ) is reconstructed one lap afterwards according to a fuzzy fusion of
the measured speed of the teeth around the updated tooth. The algorithm
requires a good quantity of memory and computation power, but can be
implemented in a 16-bit microprocessor. The implementation is not pro-
vided in detail, but as understood from its presentation the need of storage
for y(tk ) and v, are 2N values. At each detected tooth a parameter update
for a first order system, according to (6.21), is performed together with a
second order least squares identification, to get the wheel velocity.
In [Persson and Gustafsson, 2001] a similar method is proposed with
the alternative updated scheme of the calibration factors as
2π

δ θ i ( j + 1) = δ θ i ( j ) + γ ω (tk )∆ tk − − δ θ i( j ) (6.22)
N
The reference speed is calculated in a more simple way compared to the

previous method as ω (tk ) = 2π /(tk − tk− N ). The calibration factor corre-
sponds directly to the angle error of the teeth. In [Persson, 2002] the same
author proposes another method to cope with the disturbances caused by
the tooth irregularities. In an algorithm for detection of low tire pressure
the resonance frequency of the tire is one of the major detection points
117
for indicating the tire pressure. Therefore the frequency spectra for the
wheel speed signal is derived during the travel of the vehicle. It has been
showed that the tooth irregularities creates disturbing frequency peaks in
the spectrum plot at the frequencies f K (m) = mω k /(2π ) and the proposed
method do only attempt to suppress the harmonics in the frequency win-
dow of interest. A Fourier series is employed to describe the disturbance
pattern around the tooth wheel peripheral
N /2 −1)/2
( NX
2π 2π
X
δ θ k = a0 + am cos m k + bm sin m k (6.23)
N N
m=1 m=1
and the included parameters a and b are estimated from the recursive
scheme
θ = [ am bm ]T (6.24)
1 2π 2π

ϕk = cos m k sin m k (6.25)
∆ tk N N
ε k = yk − ϕ kT θ k (6.26)
θ k+1 = θ k + µ kϕ k ε k (6.27)
where yk = 2π (1/∆ tk − 1/(tk+ N /2 − tk− N /2 )). The harmonics to be attenu-

ated is determined by the value on m. It is concluded that the complexity
of this algorithm increases for each frequency reduction. The number of
memorized variables are 2M ( M denotes the number of suppressed har-
monics), required additions 6M + 3 and multiplications 12M + 4. It is
emphasised that the method is derived for suppressing a limited number
of harmonics and it is not particularly efficient and usable for wheel speed
filtering in the time domain.
Conclusive remarks
Measurements performed on cars have shown a rotational dependent noise
level up to 1.5% percent of the wheel speed. The expectation is that the
disturbances be heavily reduced leaving noise levels of maximum 0.5%
on smooth asphalt roads during free rolling. In [Pavkovi et al., 2006] even
better results are shown. The variance of the remaining noise in a general
case is difficult to determine since it depends on many factors, for exam-
ple, the actual surface conditions and the sensitivity towards the force
transients.
118
6.4 Conclusion
6.4 Conclusion
The functionality of many vehicle system relies on the wheel speed signals.
In some important applications traditional noise filtering is difficult since
the timing constraints on the signal are high. This section has discussed
a few methods to improve the quality of the wheel speed signal, without
causing any phase shift or time delay.
On, particularly, front axle wheels force transients caused by quick
changes of the brake command can give large disturbances on the wheel
speed signal. The disturbances arise due to wheel axle deflection and a
corresponding movement of the speed sensor position, since the measured
speed is the relative speed between the wheel and the sensor. The result
shows that the effect of such a disturbance can by simple means be reduced
by over 50% by appling a first-order feed forward filter from the brake
torque signal.
Further, the wheel speed signal has a rotation-dependent noise, due to,
tolerance errors on the tooth wheel that together with a sensor generate
the pulses to the frequency counter that calculates the wheel speed. The
noise pattern can be removed by creating a vector that adaptively keeps
track on the error related to each tooth. The expectation is that noise of
1.5% on the speed signal could be reduced to around 0.5% for free rolling
on smooth asphalt roads.
119
7
Road Friction Estimation
This chapter cover work done within the sub project, Model-based road
friction estimation in the IVSS-programme, Road Friction Estimation,
RFE.
IVSS, Intelligent Vehicle Safety Systems, is a unique joint venture by
public-sector agencies, private-sector companies, and industry organiza-
tions, that was set up to stimulate research and development for the road
safety of the future and aims at moving the emphasis from passive solu-
tions to active systems. In other words, preventing problems from arising
in the first place [Vägverket, 2006b].
7.1 Introduction
The Road Friction Estimation project

Road vehicles rely strongly on friction. Their large masses that often move
at high speeds may cause fatal damage if they loose steerability. The
controlling tire forces are generated by and dependent on a sometimes
abruptly changing friction. A large safety margin in the traffic should be
compulsory, but is often not sufficiently regarded by the drivers. Modern
vehicle control systems can, to some extent, correct for incautious actions
from the driver, but a correct appraisal of the driving circumstances is
mandatory for safe driving.
Many investigations show a correlation between the road condition and
the accident risk, see e.g. [Wallman and Åström, 2001]. The output from
a road friction estimator might be used as a detecting device that warns
the driver about a bad or suddenly changed road condition. Information
about the friction can also be used to enhance the functionality of ac-
tive and adaptive control systems within the vehicle or sent to a global
infrastructure that receives and transmits information about the roads.
120
7.1 Introduction
The aim of the RFE project is to estimate the friction between tire and
road and to evaluate and optimize the reliability as well as the delay of the
estimation. Three different concepts are analysed within the RFE-project.
Sensor equipment for accurate determination of tire forces Fric-

tion estimators often relies on estimated tire forces. Hence, errors on the
derived forces might have an crucial effect on the reliability of the fric-
tion estimate. To enhance the signal quality SAAB investigates the pos-
sibilities to mount force sensors close to the wheel hub to enable direct
measurements of the lateral and longitudinal force and the self-aligning
torque.
Surface detection by preview techniques Volvo Technology and Luleå

Technical University study different concepts to estimate the surface con-
dition in front of the vehicle by optical means. A proposal under evaluation
is a sensor called RoadEye [Optical Sensors, 2006] that transmits a couple
of laser frequencies on to the road and measures the reflections. With this
technique it is possible to detect whether there is a coating layer on the
road surface, such as water, snow, or ice, see [Casselgren et al., 2007].
Model-based road friction estimation The thesis presents work per-

formed within this sub project. The main aim is to develop algorithms
for friction estimation during longitudinal tire force excitations. The re-
sults are derived in close cooperation with Johan Hultén at Volvo Cars,
Fredrik Bruzelius at Volvo Technology, and Magnus Gäfvert at Modelon
AB. Besides IVSS Haldex Brake Products AB and Volvo Cars finance the
work.
Model-based road friction estimation

A popular and in many circumstances successful approach to assess the
friction is to estimate the braking stiffness, i.e. the incline of the tire force
relative the slip at low slips and from this value distinguishing between
different surface conditions. Note that the term “braking” in this context
refers to longitudinal slips and forces and is also used to denote the force-
slip inclination at low slips during driving. A linear tire model is hereby
used with an assumption on the relation between the slope and the fric-
tion condition, see e.g. [Gustafsson, 1997]. The braking stiffness depends
on many factors and a generic relation between the slope and the exact
friction coefficient is therefore not possible to obtain. Even the surface
detection is not always reliable due to the large variation of the braking
stiffness caused by other reasons, see for instance [Pavkovi et al., 2006].
This work aims at estimating the value of the friction coefficient by
using a well-defined accurate model for the effect of the friction on the
121
Chapter 7. Road Friction Estimation
tire behavior. The model adopted for this purpose is the physically based
brush-model, that describes the effects of the rubber deformation and fric-
tional limitations in the contact patch and specifies the curvature of the
force-slip relation. The information of the friction is derived from the de-
viation from linearity in the force-slip slope, which is clearly detectable at
larger slips. The estimation of the linearity at small slips is a similarity
to the previous method, but its slope is not used as indicator of the fric-
tion. In the proposed method the detection of the incline at low slip is a
necessity for deriving the curvature.
One advantage of the method is that the algorithm is based on a clear,
verified relation between the measurements and the friction, which makes
the estimation more reliable. The drawback is it that two tire-road related
parameters have to be estimated and the method requires higher force and
slip excitations compared to stiffness-based methods.
Another problem that arises when the curvature of a slope is to be
determined is that the measurement data need to be sufficiently spread
in the force and slip plane for an accurate estimate. If the data is clus-
tered around one point the solution become ambiguous and any parameter
couple that results in a curve that crosses this point may be a possible
solution. The robustness of the estimation decreases and noise, model er-
ror, and other temporal disturbances will have a very large impact on the
result. To overcome this, a particular data storage system is developed to,
as good as possible, utilize the spread of the available measurements.
Outline
A main prerequisite for the choice of a model-based estimator for the
road friction is a clear formulation of the model, which describes how the
friction coefficient influences the behavior of the tire in the interaction
with the road. Here, it is assumed that the friction affects the relation
between the slip and the developed tire force and that these signals are
available for measurement or estimation. The model also has to reflect the
variation rate of the included parameters and should specify the expected
model and measurement noise distributions.
The function of this kind of estimator can be divided into three sub
problems:
• Choice of the tire model
• Determination of signals to the tire model. In this case the normal-
ized tire force and the slip.
• Construction of the estimator algorithm
In this work the brush model, see Section 4.1 is used to describe the
tire behavior. Section 7.2 contains an evaluation of its reliability for fric-
122
7.1 Introduction
tion estimation by a validation towards measurement data. Section 7.3

presents the idea of storing sampled data in data bins, which is a way
to lump measurements in separate regions distributed over the force and
slip axles together. This is a way to remember and filter old data and
to decrease the sensitivity to disturbances. The states in the data bin,
representing the normalized longitudinal tire force and the tire slip are
used in the cost function from which the optimal parameters denoting the
braking stiffness and friction coefficient are calculated. The least squares
method and the Gauss-Newton algorithm are proposed for this purpose.
Implementation details and the additional vehicle estimators neces-
sary for deriving the tire force and slip signal are described in Section 7.4
followed by Section 7.5 presenting the results from tests in reality. The
chapter concludes with a discussion about the estimator properties and
the requirements on further work.
Friction estimation — Prior art

Much effort is world-widely spent on developing algorithms for friction
estimation. A literature study in this area is quite difficult, since pub-
lished articles often describe academic nice solutions, but these are often
restricted to certain circumstances with few real implementations. The
companies, having powerful test equipment, mostly do not publish articles
due to the hard competition. Instead the ideas are protected by patents,
that are difficult to penetrate and understand. One exception is work done
by NIRA-dynamics, which is been presented in many articles, for exam-
ple [Gustafsson, 1997] and also developed into a product. The product
is an algorithm that uses information from the wheel speed signal, the
vehicle velocity, and the engine torque to distinguish between the road
foundations: gravel, ice, snow, and asphalt. The idea is to evaluate the
braking stiffness and correlate its value to the surface friction. Using the
braking stiffness as a friction indicator is not always reliable and the al-
gorithm is limited to certain conditions. The variance in the wheel speed
signal is used to separate between rough and smooth surfaces. Further
developments and discussion about this method can be found in [Müller
and Uchanski, 2001] and [Müller et al., 2003]. Another approach has been
published in [Ray, 1997] where the vehicle longitudinal and lateral accel-
erations, the vehicle yaw rate, and the speed of the wheels are measured.
The corresponding tire forces are estimated by an extended Kalman-Busy
filter and among several tire models the most probable is chosen through
Bayesian learning. An expectation is that such an algorithm is depending
on extensive calibration and that the choice of tire models has to take the
behavior of the actual tire into account.
One method based on the brush model is published in [Pasterkamp
and Pacejka, 1997], where the relation between the self-aligning torque
123
and the lateral tire force is used to estimate the friction while turning.
The approximations in the brush model becomes apparent in the real
application and neural networking is employed to learn the estimator to
determine the friction. A simpler method for using the relation between
the self-aligning torque and the lateral tire force can be found in [Yasui
et al., 2004].
Several different friction-estimation techniques are discussed in [Eich-
horn and Roth, 1992]. One unique approach that is presented is to mount
a microphone close to the tire-road contact and by the acoustics separate
the different road surfaces. It is concluded that the recorded sound con-
tains information about the friction, but distinguishing it from the other
possible noise sources is difficult. The paper also includes an evaluation of
a tread strain sensor, which shows promising result at least for detecting
low friction. This kind of sensor is further described in [Pohl et al., 1999].
Due to the varying pressure distribution in the contact patch there is of-
ten a region where the tread slides on the road. The friction is derived by
evaluating the vertical and horizontal tread forces in that region.
In [Umeno et al., 2002] the frequency contents of the wheel speed signal
is analysed. From the frequency and magnitude of the resonance peak,
the tire stiffness is estimated. The method can not predict the friction
coefficient, but might work as a detector of a change of the road surface.
The number of patents dealing with this kind of problem is voluminous.
Searching on friction estimation of tires give almost 500 hits at this date.
In this project, a share of the found patents were selected for further
examination, but a suitable approach for friction estimation was not found
among those. Probably, related innovations are hidden from being found by
forming the title without mentioning words as “friction” and “estimation”.
One such example is [Levy and Fangeat, 2004] which describes a device
that measures the secant G (λ ) = f x (λ )/λ in the force-slip plane. From the
presumed tire model, G (λ ) = a − bλ , predicting the secant behavior, a and
b are estimated. By knowing a and b, further conclusions about µ and λ ○ ,
substantiated by measurement data, can be drawn. Similar expectations
can be derived from the brush model where σ ○ ( b/ a, λ ○ = 1/(σ ○ + 1) and
µ ( a2 /(3b) even if these differ slightly from the conclusions in the patent.
The proposed method resembles this approach in that two parameters
are estimated from the tire characteristics. The estimation routines are,
however, completely different.
7.2 Experimental Validation of the Brush Model

This section presents an experimental validation of the brush-tire model
used in the Road Friction Estimation project. The aim is to validate the
124
agreement of the tire model towards real conditions and to investigate the
sensitivity of the included parameters toward various factors that may
change during normal run of the vehicle. It is important to validate the
model to measurement data collected at real driving. The environmental
disturbances and the limited measurement conditions in a production car
have to be regarded when evaluating the fit of the data to the brush model.
Otherwise, the results may not be valid and the estimator will not work
as expected in the real implementation. However, in the first verification
of the usability of the brush model, it is desirable to minimize the un-
certainties from the signal estimations, necessary for the limited sensor
situation in a normal car. Therefore, the measurements are performed by
a special tire test-truck from VTI, but the tires and road conditions are
normal. Additional available tests results, not collected within the project
will also be referenced in this section.
Test equipment The test vehicle, denoted BV12, is a Scania truck

LB80, see Figure 7.1, owned and run by VTI and equipped with a fifth
wheel for various measurements and slip and force excitations of tires
for personal cars. The test wheel is pressed against the road by the pres-
sure from a hydraulic cylinder, which can achieve different vertical loads
in the range 1–6 kN. A varying brake torque can be applied on to the
wheel through a disc brake controlled by a hydraulic brake system. The
angle between the vehicle travel and the wheel rotational direction can
also be changed during a test sequence. There are sensors on the wheel
suspension for measurements of the vertical and horizontal forces and the
vertical torque working on the wheel hub. There is also a sensor to mea-
sure the rotational velocity of the test wheel. The vehicle reference speed
is obtained from the left-front wheel of the truck. Further information is
available from [Nordström and Åström, 2001].
Test facilities To be able to collect information about a number of

tires on several different road foundations with rich variation, tests were
performed at two test grounds.
• Hällered is a proving ground close to Borås owned by Volvo Com-

pany. The facility has an area of 700 hectares and includes tracks
for high-speed, endurance, off-road, and comfort testing tracks [Volvo
AB, 2006]. For the testing in this project mainly the brake and hand-
ling area was used, where high and low-friction tracks are available
with the possibility of water spraying. The high friction surface is
normal asphalt, having the same properties as a well treated high-
way. The low friction area is built up by basalt bricks providing low
friction when sprayed with water.
125
Figure 7.1 The test truck BV12 in Arjeplog.
• Arjeplog is a famous place in northern Sweden for winter test-

ing of vehicles. Many companies provide test facilities for the au-
tomotive industry. The tests in the project were performed at the
proving ground run by Colmis AB, see [Colmis AB, 2006], since both
Haldex and GM have contracts with them. The available surfaces are
ice, snow and asphalt. The asphalt patch is held free from snow by
heat from elements in the ground, but when it is snowing the lanes
quickly fills with water and melting snow. The frictional properties
of both snow and ice are temperature sensitive and the temperature
can change in a large interval from day to day.
Test procedure The longitudinal tire data was generated by applying

a braking torque on the wheel as a ramp function from free rolling to
complete lock-up of the wheel. When the wheel got locked the brake pres-
sure was released and this sequence was repeated for a number of times
depending on the available road length. Between 2–5 sequences where col-
lected in one file. For each test setup 1–3 files were recorded. In Hällered
the vehicle speed was 70 km/h during the braking phases and in the Ar-
jeplog it was 50 km/h. The winter tire with 4 kN vertical load was chosen
as reference setup and was subjected to all available test conditions. The
load dependence was measured by varying the vertical load between 2, 4,
and 6 kN. Dry asphalt was tested both for summer and winter tires. Re-
sult from wet basalt giving a low-friction firm surface were only obtained
for the winter tires. Snow and ice was tested with all three tire types,
126
which were:
• Winter tire: Continental ContiWinterContact TS810 215/55R16

• Summer tire: Continental ContiSportContact 225/45R17 91W
• Studded tire: Gislaved Nordfrost 3 215/55R16
Data from lateral steering tests and braking with a few different slip
angles were also collected and results can be found in [Svendenius, 2007a]
and [Svendenius, 2007b].
Test results
The result from the measurements are shown in figures with the slip, λ ,
see definition in (2.5), on the x-axis and the normalized longitudinal brake
force − Fx / Fz on the y-axis. The different plot-styles of the data points
denote which of the repeated sequences in the test file the measurements
belong to. The solid line is the optimal-fitting brush-model adaptation to
the data. The brush model is given by
2 3
− C σ + 1 Cx σ x pσ x p − 1 ( Cxσ x )

if pσ x p < σ ○x

x x 2
Fx = 3 µ Fz 27 (µ Fz) (7.1)
−µ Fzsign(σ x ) otherwise


where σ ○x = 3µ Fz/ Cx and the optimal values on C0x = Cx / Fz and µ are

derived by minimizing the squared error along the y-axis. The optimiza-
tion algorithm is a Gauss-Newton method similar to the one that will be
presented in Section 7.3. For the optimization all visible data points in
the plots, with slip lower than approximately 20% are used and equally
weighted in the cost function. The shown data points belongs to the first
part of the test sequences where the braking torque is increasing.
The measured signals contain biases which effect the fit of the brush
model. The bias on the longitudinal force sensor is mostly small and need
no further correction. The bias on the longitudinal slip may be explained
by error in the estimated free rolling wheel-radius, R e , defined as the
rolling radius of the wheel when no force is transmitted by the tire, see
Section 2.4. Before start up of the tests R e is identified. There are, how-
ever, many factors that affect R e and changes in the radius may alter the
slip bias. Therefore, R e has to be identified and accounted for in each
test sequence. The slip bias, often called horizontal shift, sh , is related to
changes in the wheel radius as
∆ Re
sh ( (7.2)
Re
127
Only a few of the available measurements are shown here, but more
resulting plots and results can be found in [Svendenius, 2007a] and [Sven-
denius, 2007b].
Winter tire Winter tires, generally, have relative soft rubber compound
getting its optimal stiffness in a lower temperature region than the sum-
mer tire, which can be seen on lower values on C0x . The rubber texture is
rough to provide good grip in loose foundations such as snow.
Measurements together with the adaptations of the brush model to the
winter tire on wet asphalt are shown in the force-slip-plane to the left in
Figure 7.2. The right plot shows the test sequences generating the data.
The force excitation is similar to the other tests, but the number of appli-
cations varies. The results from dry asphalt and from basalt are shown
in Figure 7.3. On both wet and dry asphalt the agreement of the brush
model is very good. The measurement deviations are symmetrical around
the model realisation and can be explained by temporary disturbances.
The cluster of points in the low-force region (p f x p < 0.3) are explained
by the vibrations in the measurement rig when the braking torque is
released. See the force measurements in the upper plot to the right in
Figure 7.2 between, for instance, 2.5 and 3 sec. For the low-friction foun-
dation the temporal disturbances largely affects the measurements. The
splices between the basalt bricks introduce extra road unevennesses and
the disturbances are slightly larger than for asphalt, but their relatively
impact become larger due to the lower excitation force needed to lock the
wheel. It is difficult to verify the model from these measurements, but no
consistent deviation from the model realisation seems to be present.
Results in winter conditions from Arjeplog are presented in Figure 7.4
for snow and ice. The snow measurements are noisier due to the uneven-
nesses and traces in the randomly packed snow. In spite of this, it can
be seen from the data that the performance on snow diverge from the
brush model particularly at normalized forces above 0.2. The tire curve
has more substantial curvature and no pronounced force peak. The tire
force increases with the slip, at least up to λ = 40%. This is probably an
effect of the snow deformation.
Due to the large spread in the data it is difficult to draw any clear
conclusions of the ice measurements.
Summer tire Summer tires are designed to achieve good grip in sum-
mer condition. Compared to the winter tire, the somewhat simpler task
increases the possibilities to improve properties such as rolling resistance
and wear resistance to provide a more cost-effective driving. Slick tires
without texture is said to give best grip in good weather conditions on
asphalt [Haney, 2003]. Grooves in the tire are, however, inevitable for
128
Braking −Winter tire −Wet asphalt −H0511_107.dat

1.2
− Long. force [kN]

10
5
1
0
0.8 −5
Vert. force [kN]

0.6
5
0.4 4
3
0.2 1
2
Cx=28.3 µ=1.02 sh=1.72 2
3 1
Slip
0 4
5 0
brush
−0.2 −1
−5 0 5 10 15 20 0 2 4 6 8 10
Slip % Time [s]
Figure 7.2 Measurement data from Hällered showing results from the winter
tire on wet asphalt with a vertical load of 4 kN. To the left: Adaptation of the brush
model. The normalized brake force is shown as a function of the slip, λ . To the right:
Recorded signals showing the test sequences in the measurement file.
Braking −Winter tire −Dry asphalt −H0511_149.dat Braking −Winter tire −Low friction −H0511_124.dat
1.4 0.4
1.2 0.35
0.3
1
0.25
0.8
0.2
0.6 0.15
0.1 Cx=14.6 µ=0.312 sh=0.807

0.4
1 0.05
0.2 2
Cx=24.9 µ=1.19 sh=1.91 0
3
0 4 1
5 −0.05 2
brush brush
−0.2 −0.1
−5 0 5 10 15 20 −5 0 5 10 15
Slip % Slip %
Figure 7.3 Measurements of the winter tire at Hällered together with a brush
model adaptation at the vertical load 4 kN. To the left: On wet asphalt. To the right:
On basalt. Note, that scaling on the axes differs between the figures.
evacuating water from the contact patch to prevent friction losses due to
a water film when driving on a wet road. The texture on the tested sum-
mer tire seems to efficiently cope with the water, since the difference in
performance between the dry and wet asphalt is small, which can be seen
in Figure 7.5. A difference is that the tire force for the wet asphalt has a
slightly less curvature up to the peak, which comes at an lower slip then
predicted by the brush model. The force peak is more pronounced in the
wet condition and the friction force reduces quicker for higher slip. The
poor frictional properties of the summer tire on snow and ice, can be seen
129
rak broms −Vinterdäck −Snö −Arj0603_100.dat rak broms −Vinterdäck −is −Arj0603_106.dat
0.6 0.12
0.5 0.1
0.08
0.4
0.06
0.3 C =7.16 µ=0.0723 s =0.0996
x h
0.04
0.2
0.02
0.1 1 1
Cx=14.7 µ=0.382 sh=0.571 2 0 2
3 3
0 4 4
−0.02
5 5
BM BM
−0.1 −0.04
−5 0 5 10 15 20 25 30 35 40 −5 0 5 10 15
Slip % Slip %
Figure 7.4 Adaptation of the brush model towards measurements data from Ar-
jeplog with winter tire and with the vertical load 4 kN. To the left: On snow. To the
right: On Ice. Note, that scaling on the axes differs between the figures.
Braking −Summer Tire −Wet asphalt −H0511_154.dat Braking −Summer Tire −Dry asphalt −H0511_146.dat
1.2 1.4
1 1.2
1
0.8
0.8
0.6
0.6
0.4
0.4
0.2 1 1
Cx=41 µ=1.1 sh=1.17 2 0.2 2
Cx=37.2 µ=1.2 sh=1.02
3 3
0 4 0 4
5 5
brush brush
−0.2 −0.2
−5 0 5 10 15 20 −5 0 5 10 15 20
Slip % Slip %
Figure 7.5 Adaptation of the brush model towards measurement data from
Hällered with the summer tire and with a vertical load of 4 kN. To the left: On
wet asphalt. To the right: On dry asphalt. Note that scaling on the axes differs
between the figures.
in Figure 7.6. The tire do not get a good grip in the snow and the charac-
teristics clearly differs from both the winter and studded tire, in that the
friction is lower and do not increase for higher slip. The ice properties do
not significantly diverge from the winter tire.
Studded tire Studded tires are gripping the road foundation with
spikes. The tire force is build up with approximately 10% from the rubber
friction and 90% from the spikes. A drawback with studded tires is that
the abrade on the roads is extensive, and that micro particles are spread
130
rak broms −Sommardäck −Snö −Arj0603_157.dat rak broms −Sommardäck −is −Arj0603_167.dat
0.6 0.14
0.5 0.12
0.1
0.4
0.08
0.3
0.06
0.2 C =4.14 µ=0.0854 s =1.32
x h
0.04
0.1 1
Cx=33.3 µ=0.29 sh=0.874 2 0.02
3 1
0 4 2
0
5 3
BM BM
−0.1 −0.02
−5 0 5 10 15 −5 0 5 10 15
Slip % Slip %
Figure 7.6 Adaptation of the brush model towards measurement data from Arje-
plog with summer tire and with a vertical load of 4 kN. To the left: On snow. To the
right: On Ice. Note that scaling on the axes differs between the figures.
in the environment to a larger extent than when using normal tires. The
attention to this kind of pollution problem have gained a larger interest
recently. Studded tires have a positive effect on the road friction in that
they tear away the polished surface caused by normal tires. On the other
hand, the traces from the tires in the road get deeper and collect more
water, which require more extensive road maintenance [Gustafsson et al.,
2006b]. The studded tire is, however, outstanding in friction on ice, which
can be seen from the results in Figure 7.7. Particularly on ice, the mea-
surement signals contain more noise than the other tires. The grip of the
spikes seems to be varying and unpredictable. The measured behavior on
snow is very similar to the winter tire, but with a larger noise level. The
latter is most probably a result of the snow packing when running back
and forth on the test lane or of shifting of the snow properties by other
reasons.
In Figure 7.8 the footprints from the studded tire on ice in a braking
phase can be seen. The increasing length on grooves from the spikes is a
consequence of the increasing tire slip caused by ramping up the braking
torque. Since a bristle, according to the brush-model theory from Sec-
tion 4.1 both enters and leaves the road undeformed, the sliding distance
of a bristle onto the ground is ls ( vsx tc (−a) where vsx is the relative
velocity between the tire carcass and the road and tc (−a) is the time
that a bristle spends in the contact patch. The time for the road contact
can also be expressed as tc (−a) = 2a/vc . The groove length can then be
calculated as ls ( vsx 2a/vc = σ x 2a. Hence, the slip could theoretically
be derived from measurements of the groove length if the length of the
contact patch, 2a, is known.
131
rak broms −Dubbdäck −Snö −Arj0603_171.dat rak broms −Dubbdäck −is −Arj0603_175.dat
0.7 0.3
0.6 0.25
0.5 0.2
0.4 0.15
0.3 0.1
0.2 0.05 Cx=4.34 µ=0.158 sh=0.68
1 1
0.1 2 0 2
Cx=10.6 µ=0.458 sh=0.663 3 3
4 4
0 −0.05
5 5
BM BM
−0.1 −0.1
−5 0 5 10 15 20 25 30 35 40 −5 0 5 10 15 20 25 30 35 40
Slip % Slip %
Figure 7.7 Adaptation of the brush model towards measurements data from Ar-
jeplog with studded tire and with the vertical load 4 kN. To the left: On snow. To
the right: On Ice. Note that scaling on the axes differs between the figures.
Figure 7.8 Footprints of the studded tire on ice during brake application. The
brake torque is applied as a ramp and the increasing grooves are a result of the
increasing tire force. the wheel is rolling from the left to the right.
Conclusions From the shown figures it can be seen that the brush
model mostly fits the measurement data well. The optimized tire model
is well within the spread of the data points. The spread is, however, rela-
tively large on the low friction surfaces, which depends on that the force
disturbances on the measurement rig becomes more apparent when the
force excitation is low. The disturbances mainly come from low frequency
vibrations in the mechanical structure invoked by unevenesses in the road
and the force excitation on the wheel. The major question about the us-
ability of the brush model for friction estimation for the tested tires on
the available surfaces are treated further on.
The derived tire parameters for the different test setups are presented
in Table 7.1 and visualized in Figure 7.9. From the results it is stated that
132
• The braking stiffness for the summer tire is larger than for the win-
ter tire and the studded tire.
• The difference between the friction of the summer and the winter
tire on asphalt is small.
• The normalized braking stiffness seems to increase for increasing
vertical load for the winter tire on asphalt and snow, but not on
ice. This topic is important for the friction estimator and further
discussed later on.
• The studded tire has better friction on ice, that the other tires.
• The summer tire has poor grip on snow, when studded tire and the
winter tire performs similar.
• There seems to be a relation between the tire stiffness and friction
coefficient, see Figure 7.9.
• The braking stiffness is larger for wet asphalt than for dry. This is
further discussed later on.
An interesting observation is that the winter tire and the studded tire
behaves similar on snow. The force-slip curve has no obvious maximal
point and the tire force seems to be an increasing function of the slip.
The curve also bends off more than the brush model which might be an
effect of that the tires grip into the snow which yields at a particular
strain. The summer tire differs and behaves more as it would slide on the
packed snow surface. There are models on how to deal with deformable
foundations, e.g [Meschke et al., 1996] and [Saino, 2001] and they tend to
become very complex. One reason for not incorporating the snow behavior
in the brush model is the reluctance in introducing more parameters in
the model in this work.
The higher noise levels on the measurements on snow for both the
summer tire and the studded tire compared the winter tire might be con-
fusing. These are most probably results of the snow packing when running
back and forth on the test lane or of shifting of the snow properties by
other reasons. It is most unlikely that the different noise levels entirely
depends on the tire properties.
Brush model verification

The most relevant issue when evaluating the measurement data is the
ability to estimate the friction coefficient from the brush model realisa-
tion. It is obvious that the friction can be accurately estimated when data
from the full slip curve 0 < λ < 1 is available since the friction is com-
pletely given by the normalized tire force at high slip. At low slip the entire
133
Table 7.1 Results of the parameter optimization
Tire Foundation Load C0x µ sh [%]
Winter Wet asphalt 2 kN 23.8 0.98 1.2

Winter Wet asphalt 4 kN 27.6 1.0 1.7
Winter Wet asphalt 6 kN 28 1.1 1.9
Winter Dry asphalt 4 kN 25 1.2 1.8
Winter Basalt 4 kN 16 0.27 1.2
Winter Snow 2 kN 9.04 0.40 0.69
Winter Snow 4 kN 13.6 0.40 0.55
Winter Snow 6 kN 14.4 0.41 0.20
Winter Ice 2 kN 6.27 0.10 0.64
Winter Ice 4 kN 6.25 0.078 0.077
Winter Ice 6 kN 6.72 0.081 -0.25
Summer Wet asphalt 4 kN 42.8 1.1 1.1
Summer Dry asphalt 4 kN 37.3 1.2 1.0
Summer Snow 4 kN 22.8 0.26 1.1
Summer Ice 4 kN 3.6 0.077 1.35
Studded Snow 4 kN 11.4 0.51 0.77
Studded Ice 4 kN 5.6 0.16 -0.70
contact surface is adhesively bonded to the ground and according to the

brush model the force-slip relation do not contain any information about
the friction then. The eventual relation between the braking stiffness and
the friction is a different issue discussed further on in this section. To
be able to estimate the friction from the data using the brush model, the
curvature on the measured force-slip relation must be detectable. An im-
portant issue is the accuracy of the estimation if only low-slip data is used.
This section will investigate how the value of the highest available slip or
force affects the accuracy of the friction estimation. It might be possible
to detect a lowest slip value where the friction can be detected within a
certain accuracy from the slip and force measurements. Figures 7.10–7.11
show the results from the winter tire on snow and asphalt, where the
tire parameters are identified for data upto a certain slip. The obtained
tire characteristics are shown for the slip limits, 0.5, 1, 1.5...15%. The slip
limits are marked in the graph as vertical dashed-dotted lines. On re-
134
45 1.4
Winter tire 2 kN Winter tire 2 kN
40 Winter tire 4 kN Winter tire 4 kN
Winter tire 6 kN 1.2 Winter tire 6 kN
35 Summer tire Summer tire
Normalized Braking stiffness
Studded tire Studded tire

1
30
Friction coefficient
25 0.8
20 0.6
15
0.4
10
0.2
5
0 0
D. Asph. W. Asph. Basalt Snow Ice D. Asph W. Asph Basalt Snow Ice
Figure 7.9 Visualization of the results from Table 7.1. The normalized tire stiff-
nesses and friction coefficients of the tested tires at different loads are plotted as
functions of the surface.
spective line there is a asterisk denoting the friction estimate and a ring
prescribing the braking stiffness. The parameters are derived by using
only the data to the left on the actual slip-limit value. The tire character-
istics belonging to the different parameter results are drawn with solid
thin lines. A small asterisk in the intersection between the vertical slip
limit line and the tire characteristic line shows to which parameter set the
line belongs. In some cases the small asterisk is obscured by the larger
friction estimate marking. An observation is that the friction is underes-
timated in both the shown examples. The more apparent deviation from
the brush model for snow makes this particularly obvious. An advantage
is that the estimation early says that the friction is low. The drawback is
that it might too early state that the friction peak is reached. For asphalt
it is clear that there is more available friction, for instance, at 2% slip the
estimation is 0.6 and the normalized tire force is around 0.4. When the
normalized tire force is around 0.9 the algorithm states that the friction
limit is reached. This is within the tolerances.
In Table 7.2 the need of tire force excitation necessary to provide an
friction estimation within the specified accuracy of ±0.15 is presented.
The tabular values prescribes the degree of utilization of available fric-
tion force. Generally more than 65% of the available friction force must be
generated before an accurate estimate can be derived. Ice can not produce
an estimate before the limit is reached. The available friction is, however,
very low and the friction is estimated at a low slip value. The friction is
given before the locking phase of the wheel, which might be good infor-
mation for the ABS-system. The poor result for the summer tire on wet
asphalt is explained by deviation between the tire behavior and the brush
135
rak broms −Vinterdäck −Snö −Arj0603_099.dat

0.6 30
0.5 25
Normalized Longitudinal force (−f )

x
Longitudinal Tire Stiffness

0.4 20
0.3 15
0.2 10
0.1 5
Cx=19.7 µ=0.355
0 0
−1 0 1 2 3 4 5 6 7 8
Slip %
Figure 7.10 Parameter estimates as functions of the highest used slip value for
the winter tire on snow. The estimated tire stiffness (dashed line with circles) and
friction coefficient (full line with asterisks) by using only the data points to the left
of the slip limits [0.5,1,1.5 .. 8] (dashed-dotted vertical lines) are plotted. Also the
tire characteristics for the estimated parameters at each slip limit are plotted.
Table 7.2 Required utilization of friction (in percent) to achieve a friction estimate
within an accuracy of ±0.15. Only tests at 4 kN is used.
Road foundation Winter Summer Studded tire
Dry asphalt 75 69 -
Wet asphalt 74 95 -
Basalt 87 - -
Snow 66 81 74
Ice 100 100 68
model around the force peak which is mentioned earlier.

The investigation shows on the possibilities and restrictions for an fric-
tion estimator relying on the brush model. The study has to be considered
as very approximate, since some of the disturbances that are present will
not affect the signals in a car. One example is the sensitivity towards
road unevenesses. The hydraulic piston generating the pressure between
the tire and the road has stiff and slow suspension. The vertical load on
the tire therefore fluctuates more than what is expected in a passenger
car. The sensors in a car do, on the other hand, not deliver that precise
136

1.4 42
1.3 39
1.2 36
Normalized Longitudinal force (−f )

1.1 33
Longitudinal Tire Stiffness

1 30
0.9 27
0.8 24
0.7 21
0.6 18
0.5 15
0.4 12
0.3 9
0.2 6
0.1 Cx=28.8 µ=0.949 3
0 0
−1 0 1 2 3 4 5 6 7 8
Slip %
Figure 7.11 Parameter estimates as functions of the highest used slip value for
the winter tire on asphalt.
measurements. The optimization algorithm used to derive the examples

is not fully developed and further enhancements, such as filter and cor-
rect data weighting may improve the result significantly. Overall, from
the experiences from the brush model verification, the conclusion is that
an estimation algorithm based on the brush model will work, but high
friction utilization is required for an accurate friction estimate.
Parameter sensitivity
This section will discuss the sensitivity of the tire behavior to a few con-
ditional factors. It is also important to know in what ranges and circum-
stances a parameter might be expected to vary when designing an estima-
tor. An aim with the investigation is to find out whether further signals,
to cover for parameter changes, have to be regarded within the algorithm.
The friction coefficient depends on the tire-road interaction, while the
braking stiffness is assumed to depend on more tire related properties as
the tread stiffness and the size of the contact patch between the tire and
road, see (4.31). As been noticed previously and in, for example, [Gustafs-
son, 1997] and [Müller and Uchanski, 2001] there is a co-variation between
the friction coefficient and the braking stiffness. The relation is physically
difficult to explain and does not always hold. Tests have confirmed that
very rough asphalt can provide high friction, but a low stiffness, due to the
smaller effective contact patch compared to a smooth surface. The braking
stiffness is sensitive to many other factors and a change of the stiffness
137
1 1
Normalized Brake Force
Normalized Brake Force

0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2
0 0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
Slip Slip
Figure 7.12 Plot of the force-slip relation of the original brush model. To the left:
Varying tire stiffness Cx = [6, 8, 10, 12]. To the right: Varying road friction coefficient,
µ = [0.6, 0.8, 1, 1.2].
might not guarantee a change in friction. The effects on the brush model
characteristics when varying the parameters are shown in Figure 7.12.
The rolling radius of the tire is another parameter that is important to
estimate accurately. Small changes from its nominal value will give an off-
set on the slip calculation, see (7.2), which largely effects the estimation
of Cx and µ .
Vertical load The vertical load on the tire affects the size of the contact
patch. If the pressure distribution in the tire-road contact was uniform the
area would be easily calculated as Fz/ pt . This is not, normally, the case, see
[Svendenius, 2003] or [Pauwelussen et al., 1999]. The pressure distribution
varies within the patch and with the magnitude of the vertical load, [Kim
and Savkoor, 1996]. At larger deformations of the tire, the pressure in the
tire increases due to the shape change of the carcass. The braking stiffness
is dependent of the contact patch size, but a simple physical model of the
relation between the stiffness and the vertical load is difficult to obtain.
The load on the tire is an important factor since it varies widely, due to
the load transfer during braking and acceleration. In trucks and trailers
the variation range of the tire load is particularly large, since the weight
of the cargo is the major share of the total load. There are, however, good
possibilities to estimate or measure the load on each tire.
Figure 7.13 shows the load sensitivity of the normalized tire stiffness
C0x = Cx / Fz and the friction coefficients from the previously presented
measurements. In Figure 7.14 the tire stiffnesses from the test on a Jeep
Cherokee, with Goodyear Wrangler tire, see Section 5.4 and Figure 5.4
is shown. In [Nordström, 1983] measurements of a Firestone 10.00x20/F
138
30 1.4
25 1.2
Normalized Braking Stiffness
1
20
Friction coefficient
Snow 0.8 Snow
15 Ice Ice
Wet Asphalt 0.6 Asphalt
10
0.4
5
0.2
0 0
2 4 6 2 4 6
Vertical load on tire [kN] Vertical load on tire [kN]
Figure 7.13 Diagrams showing the load dependence in the previously presented
measurements, see Table 7.1. To the left: The normalized braking stiffness, C0x . To
the right: The friction coefficient, µ .
truck tire shows an slightly decreasing normalized stiffness, C0x =[14.3,

9.8, 7.6] at the loads [10,25,40] kN. The available measurements regard-
ing the load dependence of the braking stiffness clearly states that dif-
ferent tires behaves differently. The braking stiffness increases with in-
creasing load, but not linearly since the normalized braking stiffness is
not, generally, constant and behaves differently from tire to tire. Within
the observed load range and the studied tires it seems reasonable and
less wrong to assume the normalized braking stiffness as constant rather
than the braking stiffness as constant.
Figure 7.15 shows the load dependence on the rolling wheel radius from
the measurements in Table 7.1. The rolling radius changes up to 1 mm/kN
or 0.3%/kN from the nominal values. In [Carlson and Gerdes, 2003], an
article diligently cited in the following, the tire stiffness and rolling radius
is estimated for the ContiWinterContact TS790 and the Goodyear Eagle
F1 tires under various conditions. The reported load sensitivity of, R e
from that article is in good agreement to this results.
Impact of water film on road A layer of water on the road affects the
size of the contact patch, since the layer can be assumed to carry a part of
the vertical load without any frictional force contributions. This is verified
in [Pauwelussen et al., 1999] where it also is stated that the decrease of
the dynamic contact size, i.e. for a rolling wheel, strongly depends on
wheel velocity.
A conclusion is that the braking stiffness, characterized as 2cpx a2 in
the brush model, see Equation (4.31), decreases since the utilization of
the contact length, 2a decreases. Result presented in Table 7.1 shows
the opposite, that the braking stiffness increases for the wet road. An
139
Load dependency of tire stiffness

120
100
80
Stiffness
Braking stiff. [kN]
60
Norm. Br. stiff.
40
20
0
1 2 3 4 5 6
Vertical load [kN]
Figure 7.14 Comparison of the load dependence between the braking stiffness and
the normalized braking stiffness. Longitudinal data from vehicle test, described in
Section 5.4 and [Salaani et al., 1999].
Rolling radius of the winter tire

0.316
Snow
Ice
0.314 Asphalt
VTI ref
0.312
Rolling radius [m]
0.31
0.308
0.306
0.304
2 4 6
Vertical load on tire [kN]
Figure 7.15 Diagram showing the load dependence of the rolling radius
explanation is that the water cools the tire and the lower temperature
will raise the braking stiffness. An observation in [Carlson and Gerdes,
2003] where wet and dry asphalt surfaces have been compared is that the
impact, on the braking stiffness, due to the used water layer, was smaller
than any other influencing factor, such as temperature, tire pressure, and
wear.
The friction force on a wet surface mainly decreases by two reasons.
140
Table 7.3 Braking stiffnesses for two different inflation pressures. Test results
from measurements performed at Haldex.
Pressure C0x
Asphalt Snow
6 bar 24 18
8 bar 16 14
ratio 0.75 0.67 0.78
First, the lower share of load on the frictional carrying contact surface and
then the lubricating effect of the water in the effective area in the contact.
In [Gothie et al., 2001] the friction for different tires, asphalt surfaces, and
water depths is further examined.
Inflation pressure The tire pressure is an important factor for the

contact patch properties. The size of the patch depends on the relation
between the pressure and the vertical load. The complex geometry of the
tire makes the pressure distribution non-uniform and changing with tire
deformation, hence it is difficult to derive this relation. Often a decrease
of the inflation pressure has similar effect on the tire properties as an
increase of the vertical load. In a test performed by Haldex at the test
facility in Arjeplog with a Scania test truck, see [Svendenius, 2003], the
impact of the inflation pressure on braking stiffness was measured. The
result for the tested truck tires is shown in Table 7.2 which shows a
nearly linear relation between the tire pressure and the braking stiffness.
Large variation of the dependence on the inflation pressure of the tire
stiffness for different tires has be noticed in [Schmeitz et al., 2005]. Out of
five tested tires three showed similar behavior as above, but two showed
almost the opposite, where the tire stiffness increased for higher pressure.
In [Carlson and Gerdes, 2003] an increase of 30% on the braking stiff-
ness for a pressure decrease of 20% from nominal tire pressure is reported.
The rolling radius decreased 0.5–1 mm (nominal 309 mm) for the tested
tires under the same conditions. The pressure dependence on the rolling
radius is e.g. used in a tire pressure monitoring device, described in [Pers-
son et al., 2002], as one of the factors for detection of low tire pressure.
Effect of tire wear When the depth of the tread decreases the tire stiff-
ness increases. The decreasing amount of rubber also makes the rolling
radius smaller. These observations are verified in [Carlson and Gerdes,
141
2003] where shaved tires are tested. The friction properties are mainly ef-
fected on wet road or on loose foundation such as snow and gravel, where
the tread pattern is important.
Velocity dependence The velocity mainly affects the friction in the

meaning that the sliding velocity between the tire and the road is higher
for a higher vehicle speed at a particular slip. The friction is often refereed
to as sliding speed dependent, see e.g. [Guo et al., 2006]. Different road
foundations and different tires certainly exhibits different interactions. A
water film on the road increases the velocity dependence and increasing
velocity drastically increases the risk for aquaplaning. An example of the
velocity dependence of the friction on a 3 mm thick water layer on the road
can be found in [Do and Delanne, 2003], where the locked wheel friction
drops from 0.7 at stand still to around 0.05 at 100 km/h.
In [Pirelli, 2005] the velocity dependence on the rolling radius was
tested for a few type of tires. The result showed that the rolling radius
increases for higher speed, but first above a (in Sweden) non-legally speed
limit it became noticeable.
Temperature sensitivity Rubber is normally very sensitive to tem-

perature changes. Therefore, also the tire parameters will change with
the temperature. In [Carlson and Gerdes, 2003] a decrease of 20% on the
braking stiffness between the first and the fifth braking in a number
of braking sequences is reported which is explained by the temperature
increase. For the same condition the increase in rolling radius was less
0.1% (0.3 mm). In [Mizuno et al., 2005] the braking stiffness was mod-
elled, based on measurement data, to decrease 5% within a temperature
change from 40 to 60 [deg] C.
Summary The dependence of the parameters in the brush model and

the rolling radius towards the examined factors are summed up in Ta-
ble 7.4
Conclusions
This section has treated the issue, whether it is possible to describe any
tire at any common road surface with the brush model containing only two
parameters. The answer is of course, that it depends on the requirements
on the accuracy of the model. In this context the accuracy of the model
towards the data affects the possibilities to estimate the friction coeffi-
cient. A conclusion is that there are deviations between the data and the
model, but friction estimation is possible, even though rather high utiliza-
tion of the friction is required for an accurate estimate. The utilization of
the friction for estimating the friction coefficient within ±15% is in the
142
7.3 Tire Parameter Estimation
Table 7.4 Summary of parameter dependence. A, ’+’, signifies that the factor
increases if the factor increases and vice versa. – means the opposite. A small and
normally neglectable relation is denoted by ∼. No sign means that no results are
available.
Parameter Pressure load Wear Water Temp. Vel.
Re + – – ∼+ ∼+
Cx – + + –
µ – – –
best case 66%, i.e if µ = 0.5 a normalized tire force of 0.5 ⋅ 0.66 = 0.33
is required. In the worst case the utilization has to be 95%. The used
parameter optimization is a pure minimization of the squared error be-
tween the data and the model. It is emphasized that a more advanced
estimation algorithm might reduce the need of friction utilization. The
measurements performed by VTI incorporate discrepancies not present
in a personal car. High demands on the signal processing that provides
the required signals, Fx , Fz , λ from the existing sensors, might further
improve the estimation results. It is also concluded that the parameters
included in the brush model varies with many factors. Particularly impor-
tant for a friction estimator is the load dependence of the rolling radius
and the braking stiffness. Since the application of an acceleration or re-
tardation force causes a load transfer on the vehicle, the vertical force on
the tires will change. Changes in the rolling radius will then affect the
slip calculation and cause a distorted slip-force curve. The variation of the
braking stiffness for varying loads differs between the tires. Generally, the
normalized braking stiffness is less sensitive to load fluctuations than the
braking stiffness. Another important factor is the temperature sensitiv-
ity of the braking stiffness. The braking stiffness can change around 20%
within a couple of brake applications, due to warming, without any change
in friction.

The literature contains many different methods to identify a relation that
describes the behavior of an observed system, see for instance [Johansson,
2002]. In some cases the identification aims at finding the set of equations
that best describes the relation between the input and output signal of an
unknown, so called black-box system. In the actual case the system is
partly known and better explained as a grey-box model, where the behav-
143
ior is described by a particular equation structure, which includes known

and unknown parameters. The latter are to be estimated. Normally, the
procedures for parameter estimation aim at finding the set of parameters
that minimizes the error between the adopted model and the measured
output for a known input signal. The calculation of the error differs be-
tween the methods. In off-line methods the accumulated error for a set
of data is used for the minimization. In the parameter computation all
values in the data set, which can be measurement during a certain time
period or observations collected by other means, are regarded without any
demands on causality. In real-time systems often recursive methods are
used to on-line update the estimation at each sampling instant. Mostly,
the estimates at each time instant are based on a weighted sum of the
previous estimate and information that can be extracted from the present
measurements, see [Åström and Wittenmark, 1994]. Previous input data
are only remembered implicitly in the estimated parameters and are for-
gotten as time passes.
A major problem in the present application is the issue of persistence
of excitation. For the further description, the excitation is separated into
space and time domains. Having persistence of excitation in the space
domain requires that measurements are available in a sufficiently large
region in the force-slip plane, providing an accurate stiffness and friction
estimation. The aim of the method is to, as good as possible, estimate the
friction coefficient at a low force excitation. The lower friction utilization,
the smaller amount of information about the friction is contained in the
data. Hence, the sensitivity towards measurement errors and model de-
viations toward real conditions increases. Further, it is required that the
data is spread over sufficiently large region to be able to get an accurate
and robust estimate. If the data points are clustered the cost function will
not have a distinct minimum and the quality of the results decrease.
Having persistence of excitation in the time domain requires that the
inputs vary sufficiently irregularly and often. When the conditions have
changed, the data information should be fresh, to provide an accurate
estimation that reflects the actual surface and tire condition.
Traditional estimation techniques are time depending which makes
it hard to formulate the space-distributional requirement on the data. A
typical situation where those methods fail is a brake application starting
with a tire-force ramp which provides evenly spread data in the space
domain. If the final tire-force value is large enough, this excitation is
sufficient for a proper parameter estimation. When the ramp has settled
the tire force is kept constant for the major part of the event. A recursive
algorithm calibrated to be fast enough to catch the parameters during
the ramp phase will soon forget the information collected in the low slip
region and adapt to the measurements around the reached stationary
144
point. The measurements around this point will not contain information
sufficient to estimate two parameters and the parameter observability and
accuracy decreases drastically. As long as the estimated curve intersects
the stationary point the parameters will change in between according to
the present noise conditions and lose the relation to the actual friction
condition and braking stiffness. Proposals on how to supervise adaptive
controllers in related problems at limited excitation situations are given
in, e.g [Hägglund and Åström, 2000]. A different approach is proposed in
this work.
In the first phase, invariant surface conditions are assumed, such that
there is no need to distinguish between old and new data. Information
from the force and slip measurements are stored in data bins distributed
over the force and slip-axis. The optimal tire parameters are then calcu-
lated using the stored data. Techniques to treat detection of a new surface
and adaptation to slow changes of the braking stiffness and friction can
be implemented together with the data storage and will be discussed af-
terwards.
Formulation of the estimation problem

The tire characteristic according to the normalized brush model, compare
to Equation (7.1) used for the estimation is given as
2 3
− C σ + 1 C0xσ x pσ x p − 1 ( C0xσ x )


0x x
Ψ(σ x , C0x , µ ) = 3 µ 27 µ2
−µ sign(σ )

otherwise
⋅ x
(7.3)
where the slip, σ x and the normalized tire force, f x = Fx / Fz = Ψ(σ x , C0x , µ )
are the measured signals. The included parameters θ = [ C0x , µ ] are to be
estimated.
Data filtering by use of storage-bins

The motivation for data-filtering based on storage-bins is the problem of
estimating the parameters of a scalar nonlinear function.
y = ( x, θ) (7.4)
The specific application at hand is the estimation of the normalized brak-

ing stiffness, C0x , and the friction coefficient, µ , from measurements of tire
slip, σ x , and normalized tire force, f x , using the brush tire-model (7.3).
The problem is divided into the construction of a point-wise approximation
of Ψ described by pairs (σ ′xi , fi′ ) and (σ ′′xi , f xi
′′
) constructed from the mea-
surements, and the estimation of C0x and µ is performed by minimizing
145
the cost:
X 2 X ′′ ′′ 2
′
w′i f xi − Ψ(σ ′xi , Ĉ0x , µ̂ ) + w j f x j − Ψ(σ ′′x j , Ĉ0x , µ̂ ) (7.5)
i j
using parameter optimization methods.

The storage-bin concept is introduced to offer input data of a suit-
able format for the optimization algorithm. It is a method to store, in
a compact form, information from different operating points with persis-
tence over time. By this, it is easier to ensure sufficient excitation and
observability for parameter estimation. For example, it may prevent loss
of observability when driving at constant slip for a long time. It also offers
the possibility of convenient post-processing and confidence classification
of data before delivered to the optimization algorithm. As a result, change
detection and handling of outliers can be done intuitively. Additionally,
the method reduces effects of measurement noise, which is particularly
important for the slip. The storage-bin filtering should be regarded as a
practical method and we have not yet supported it with thoroughly theo-
retical analysis.
Definition Storage-bins are used to estimate an approximation of a

function : x → y at discrete points along the ranges of x or y, from
imperfect samples x̂ and ŷ. The ranges of x and y are partitioned into
disjoint intervals X i and Y i defined by points xi and yi such that:
X i = { xp xi−1 < x ≤ xi }
(7.6)
Y i = { yp yi−1 < y ≤ yi }
A basic form of a data bin X i is a tuple X i = (X i , xi′ , yi′ ) where xi′ and yi′
gives the local estimation of for x ∈ X i . Likewise, a data bin Yi is a
tuple Yi = (Y i , xi′′ , yi′′ ) where xi′′ and yi′′ gives the local estimation of for
y ∈ Y i . Monotonicity of is obviously required for unique mapping of y to
the Yi -bins. For non-monotonic functions still the X i mapping is valid.
Estimation The data-bin estimation can be regarded as event based

rather than time based, since the estimate in any particular bin is up-
dated only when measurement values fall within the corresponding range.
Basic estimation is done by computing ( xi′ , yi′ ) and ( xi′′ , yi′′ ) as the recursive
average of samples x̂ and ŷ. To quickly get an accurate bin value at an
initialisation a sample-count limit Ni and a sample count ni ∈ [0, Ni ] to
the bin tuples are introduced, such that the bin-value is an average of
all values up to Ni , by having the forgetting factor λ i = 1 − 1/ ni . After
that, the forgetting factor is kept constant such that the estimates mainly
146
Figure 7.16 Illustration of data-bin concept. The dots represent samples that have
been stored in the bins, and the crosses the resulting estimates. The staples indicate
the present number of samples in the bin.
depend on the last Ni samples. Formally, the recursive-average estimates

are given by:
(
λ i ( k) xi′ ( k − 1) + (1 − λ i ( k)) x̂( k), when x̂( k) ∈ X i
xi′ ( k) = (7.7)
xi′ ( k − 1), otherwise
and
(
λ i ( k) yi′ ( k − 1) + (1 − λ i ( k)) ŷ( k), when x̂ ( k) ∈ X i
yi′ ( k) = (7.8)
yi′ ( k − 1), otherwise
The ( xi′′ , yi′′ ) are analogously formed by the ŷ( k) ∈ Y i condition. Fig-
ure 7.16 illustrates some properties of the bin estimator for a function
that relates slip and force in presence of measurement noise. Note how
the Yi (relating to the tire force) and X i (relating to the slip), give better
resolution in x at small and large x-values, respectively, because of the
character of the brush-model function. This motivates the use of both sets
of bin estimates ( xi′ , yi′ ) and ( xi′′ , yi′′ ) in the cost function (7.5), using x = σ x
and y = f x .
Statistical properties The averaging in the bin estimates computation

reduces the effects of measurement noise. In theory, since the estimates
( xi′ , yi′ ), ( xi′′ , yi′′ ) are formed by sums of the samples, they will approach
a normal distribution as stated by the Central Limit Theorem. That is,
147
the estimate xi′ will be aligned to the center of the bin X i , and the esti-
mate yRi′ will go toward the mean value of function in the actual bin, i.e.
pX i p−1 x∈X i ( x) dx, under the assumption of equal and uniform distribu-
tion of the samples x̂. Note that the local estimate only fulfills yi′ = ( xi′ )
if the length on the interval pX i p is infinite. The error depends on the de-
viation from linearity of in the bin range. The variance of the estimates
decrease with the rate 1/ ni . In practice, it is unlikely that the Central
Limit Theorem convergence will have any significant visible effects, since
the sample-count limit Ni will truncate the averaging sum.
Confidence assessment The data-bin tuple can also be extended with

measures of the content quality. In particular, a recursive variance es-
timate of the samples can be introduced. Likewise, a freshness measure
can be introduced to keep track of which data bins that contain the most
recent information. Measures like this can be used to form weights wi to
be used in the cost function (7.5) for parameter estimation.
Optimization routine
The estimator is an optimization S algorithm that from a set of data points,
Ω = {Ω i = [ xi , yi ], [ x, y] = [ x′ , y′ ] [ x′′ , y′′ ]}, calculates the parameters, θ,
included in a function, , such that the cost function
N
X wi e2 (Ω i , θ)
V (Ω , θ) = (7.9)
2
i=1
is minimized. The number of coupled observations in Ω is N. The weight-

ing of each data point for the estimation is specified in the vector w =
{wi ∈ [0, 1]}. The error is derived as
ei = e(Ω i , θ) = yi − ( xi , θ) (7.10)
Two different optimization algorithms are found suitable in the actual

case. The least-squares method, a robust method, limited to linear param-
eter relations and therefore used to detect the slope curvature at low slips
where the third order term, from Equation (7.3) is neglectable. The it-
erative and more computer demanding Gauss-Newton algorithm catches
non-linear behavior and is used to reach a more accurate and reliable es-
timation when data points closer to the friction limit are available. The
methods and their implementation using the brush model are described
in the following.
148
The least-squares method The least-squares method is a linear-re-

gression analysis method based on the model formulation
yi = φT θ + ei (7.11)
where the output, yi and the input, xi are measured, calculated, or, in
other way derived signals. The elements in the regressor vector φ corre-
spond to the parameters for estimation in the vector θ. The linearised
approximation, φT θ should not largely deviate from the real parameter
relation for proper work of the least squares. The method in this form is a
typical offline-method, which means that a batch of data, is necessary for
the calculation of the parameters. For on-line use, the method is rewrit-
ten, as the recursive least-squares, see for example [Ljung and Söderström,
1983]. The result of the least-squares, derived for example, in [Gelb et al.,
1977], is the estimate that minimizes the sum of the square of the errors
(7.9) and can be calculated as
θˆ = (ΦT W Φ)−1 ΦT W y (7.12)
The regressor matrix, Φ contains the corresponding regressor vectors as

[φ1 φ2 .. φ N ]T and the weight matrix is formed as W = Iw.
Using the least-squares method together with the brush model, the
demand on linearity in the parameters, restricts the working range. For
low slip, the tire stiffness can be calculated by choosing θ = C0x , with
φi = − xi . For estimation of the friction when a curvature of the force-slip
relation is detected, the parameters are, θ = [ C0x θ 2 ]T and the regressors
φi = [− xi xi p xi p]T . The friction coefficient can be derived from the result
2
as µ = C0x /(3θ 2 ). The neglection of the third-order factor of the slip from
the brush model will give an error at higher slip and the method is not
reliable if the data set includes slip values above σ ○x . The input and output
signals, x, y are given by the σ x and f x values derived and stored using
the storage-bin algorithm.
Gauss-Newton optimization In the area of convex optimization the

Gauss-Newton method is used as a substitute to the least-squares method
when non-linear problems have to be solved. The general problem is to
minimize the sum of errors (7.9). A non-linear problem can in most cases
not be solved directly and the solution has to be found by iteration. The
amount of iterations depends, e.g. on the properties of the function, the ini-
tial values of the parameters and the required accuracy on the estimates.
A common way is to update the parameters in the direction of steepest
descent of the cost, V (Ω , θ), hence θk+1 = θ k − ∇ V (Ω , θk ). In Newton’s
method the quadratic term, corresponding to the change of the gradient,
149
is used to enhance the step length and direction of the search. The up-
date becomes θk+1 = θk − H −1 (Ω , θk )∇ V (Ω , θ k ) where H is the Hessian,
a quadratic matrix, whose entries are the second order derivatives of V
with respect to θ, defined as
2 V
H (Ω , θ) = (Ω , θ) (7.13)
θ i θ j
In most cases the calculation of the Hessian is not convenient and in the
Gauss-Newton method it is approximated as H (Ω , θ) ( 2J (Ω , θ)W J T (Ω , θ).
The Jacobian matrix of e is defined as, J (Ω , θ) = ( ei /θ j ) with i = 1...N
denoting the row number and j the column number. The update scheme
becomes
θk+1 = θk − ( J (Ω , θk )T W J (Ω , θk))−1 J (Ω , θk)W e)) (7.14)
For the Gauss-Newton method to converge it is required that the initial

parameters are sufficiently close to the optimal values and that the ap-
proximation of H is valid, which is the case if the residual errors e are
small enough [Böiers, 2004].
To adapt the method to the brush model, θ will denote the parameters
[ C0x µ ]. The error vector, e is calculated from (7.10) with = Ψ as the
brush model from (7.3). The model output, y, and the input, x, are replaced
by the contents in the data bins, the tire force f x and the tire slip σ x
respective. The Jacobian matrix becomes
e1 e2 eN T
 
 C0x ⋅⋅⋅
C0x C0x 
J (Ω , θ ) = 
 e
 (7.15)
1 e2 eN 
⋅⋅⋅
µ µ µ
where respective partial derivative is given by
2 3
−σ + 2 C0xσ x pσ x p − 1 C0xσ x


e x
= 3 µ 9 (µ )2 (7.16)
C0x
0 otherwise


and
2 3
− 1 C0xσ x pσ x p + 2 ( C0xσ x )


e
= 3 µ2 27 µ3 (7.17)
µ −sign(σ ) otherwise

x
150
with σ ○x = 3µ / C0x . The method works on a batch of data just as the method
of least-squares. The benefit with Gauss-Newton is that it catches the non-
linearity in the brush model, which makes the result more accurate at slip
values close to and above the limit slip. A drawback is that the method
is more computer demanding, due to the many operations necessary to
derive J and the risk of numerous iterations for an accurate result.
Properties and further functionalities of the estimator

Combining the storage-bin concept with one of the presented optimization
algorithms gives a parameter estimator that copes with space-domain per-
sistence in a consistent manner. The bin-filtering will have consequences
on the signal properties, which may affect the reliability of the resulting
estimates. This will be discussed in the following. The low-pass filtering
of the data in the bin-storage will only partly cover the ability of the algo-
rithm to adapt for changing parameters. Proposals on how to enhance such
behavior and how to minimize the impact of outliers are also discussed
below.
Effects on the optimization due to the storage-bin classification

Methods derived in the area of parameter optimization mainly aims at
minimizing the measurement error of a model output, hence the plant is
assumed to be on the form y = G (θ , x) + e, where the input, x, is cal-
culated by the controller before fed into the system and therefore known
with good accuracy. In the present case, the system is given an excitation
from where it is possible to obtain two signals from measurements on the
vehicle. Due to the relation between these signals information can be ex-
tracted such that estimation of the friction coefficient is possible. Hence,
there are measurement errors on both included signals. In the process of
computation and estimation of the particular signals used for the param-
eter optimization further errors and uncertainties are encountered. Both
the input and output are corrupted by noise.
There are estimation and optimization methods that cope with sys-
tems having noise on both the input and the output signals. One example
is the total least-squares-method described in [de Groen, 1996], where the
orthogonal distance between the data points and the adopted function is
minimized. In [Carlson and Gerdes, 2003] this method is applied estimat-
ing the braking stiffness and rolling radius of a tire. Image-processing
methods treat similar problems when locating obstacles in a picture. The
most probable incline, shape or curvature of an edge are to be identified
out of a number of points with the correct color. Methods for this have to
deal with noise in both directions.
An obvious effect of the use of storage-bins is that the noise in the two
signals are differently treated. The signal deciding which bin to activate,
151
further on called the leading signal, is filtered in a way that decreases

the covariance if it stays in one bin, while it increases the covariance
if the noise spreads over numerous bins, which it normally does. The
uncorrelated noise on the other, further on called the trailing signal, is
effectively filtered and the bin memory will provide a recursive average of
that signal. Correlated noise or signal dependency with the leading signal
are averaged within the bin, but are captured by the bin changes provoked
by variations of the leading signal.
The main purposes with the storage-bin concept is to provide an appro-
priate way to remember recorded data and take control over the weighting
of data such that a lot of measurement data in one region will not over-
shadow the behavior in another region where a lower amount of data
is available. A secondary effect, which has to be treated correctly not to
arise erroneous inferences on the optimization, is the influence of the data
filtering.
To summarize, in the leading direction the noise properties will spread
out to a uniform distribution over the affected bins and the variance of
the signal-noise increases. In the trailing direction the value in the bin
converges to the average of the including data points. Hence, the noise
covariance in the trailing signal will be efficiently diminished when the
bins fills up with data points. In the leading direction the noise properties
will be corrupted and the covariance is enlarged and hard to predict, which
will effect the optimization.
Example To further study the effect of the classification of the signals

into disjoint intervals, two signals ( x, y) corresponding to the slip and the
normalized tire force are created with realistic assumptions on magnitude
and noise variance. Since the effect of the bin-storage is dependent on the
shape of the tire curve, two situations, that reflect the tire behavior at
low respective high slips are analysed. In the low slip region the linear
slope of the curve is to be estimated. The high slip region is in the interval
around the peak force where the tire force is saturated and the level of a
horizontal line will be estimated.
In the first case, shown in Figure 7.17, the y-signal is obtained as
ym = kx + ey and the x-signal as xm = x + ex . The deviations are set
to σ ex = 0.25 and σ ey = 0.0125. The value of σ ex depends on the wheel
speed magnitude and quality and the choice of realistic variance of ey
depends on how y is derived. In this case the variance roughly corresponds
to obtaining y from a filtered acceleration signal. If the engine torque is
used, as described in Section 7.4 σ ey decreases further. Using a force sensor
similar to the equipment used by VTI, see Section 7.2 would give a larger
noise level.
The x-values for generating the data are 300 points evenly distributed
152
Figure 7.17 Illustration of the deviation in result using the different bins com-
pared to the real data for least-squares optimization at low slip . The larger filled
points are Y-bin values and the large circles are X -bin values. The dotted line de-
notes the real slope k = 0.2. The solid line shows the result for use of all data points.
For the dashed and dashed-dotted lines the Y-bins and X -bins, respective, are used.
The error in the left-most plot is minimized in the vertical direction and to the right
in the horizontal direction.
in the interval x = [0, 1]. Performing a least-squares regression on all

included data points assuming ym = k̂xm + ey will give a result biased
as k̂ = k xi2 /( xi2 + Nσ 2ex ). It is also clear from the left-most plot in
P P
Figure 7.17 that the Y-bins consisting of the [ xi′′ , yi′′ ]-values, with y as
leading signal, are averaged in the x-direction and gives a better estimate,
but the X -bins ([ xi′ , yi′ ]), with x as leading signal, reinforces the mismatch
and shows an even more biased result. Since σ ex > σ ey a better estimate
of the slope can be reached by using the relation xm = ym / k̂ + ex for the
least-squares estimation, see right-most P 2 plot 2in Figure
P 7.17. The result
is a less biased estimate, k̂ = ( k2 xi + Nσ ey)/( k xi2 ) using all data
points. Minimizing the error in the x-direction for the bin-values do not
largely change the result for the Y-bins, since they already are derived as
an average in the x-direction. The estimation from X -bins are, however,
enhanced but still not as good as using the Y-bins.
The estimation of the friction level at high slip is illustrated in Figure
7.18. Here, it is only possible to minimise the error in the y-direction. Since
the brush-model is not strictly monotone in this area the Y-bins can not
be placed such that they resemble the tire model. Therefore the X -bins
have to be used in this region.
As a result of this discussion the weight in the cost-function in the real
153
Figure 7.18 Illustration of the bin classification at high slip. The bin-values are
here used to estimate the level of a straight horizontal line. The left plot shows the
storage of [ x′i , y′i ] and the right plot [ x′′i , y′′i ].
implementation should be reduced on the X -bins (related to the slip) for

low slip.
Time varying tire parameters It is assumed to this point that the

estimated function Ψ is time invariant. The tire parameters are assumed
to vary in two different manners. Either the variation is small and slow,
signifying temperature variations during driving, tire wear etc. or they
are changing abruptly as when the vehicle rolls in on a different surface.
If Ψ depends on time-varying parameters, θ(t), it is, in general, diffi-
cult to maintain the storage-bin contents in a consistent state. But com-
bining the low-pass filter in the storage-bins with freshness information
it can be guaranteed that the latest updated bins are most significant in
the optimization. Also old data not necessary for the persistence of the
excitation can be disregarded.
If θ(t) changes at a discrete event in time, the changes can be detected
by using information on bin variance or from the parameter optimization.
A sudden change of θ(t) at time t1 such that θ(t) = θ0 for t < t1 and
θ(t) = θ1 for t ≥ t1 will lead to a corresponding increase in the estimated
variance v in the storage-bins at the present operating point. This may be
used to trigger operations, e.g. a reset in the oldest updated storage-bins,
to be performed at a change detection. Suddenly occurring inconsistencies
in the bin contents may also lead to a corresponding increase of the cost
function (7.5) in the parameter optimization, which may also be used as
a trigger. Other sensors available on the vehicle can also be employed for
this purpose. Starting the windscreen wipers may, for example, be one
154
7.4 Algorithm Implementation
event that signifies a change in surface condition. The a sudden change

of the variance on the wheel speed may also denote a potential surface
change.
Outliers To reduce the effects of noise on bin-values that contain a

limited number of measurements, the weight of each bin will depend on
the number of collected data points as
0 if


 n < Nlow

 n− N
low
wi (n) = if n < Nhigh (7.18)

 N high − Nlow

1 otherwise

The variance measure can be used to eliminate outliers, for example

by allowing only new samples that lies within a variance-dependent range
from the current estimate. The effect of outlying bin-values can also be
reduced by two steps in the optimization. After a successful optimisation
the residuals of all bin-values (7.10) are analysed and any bin having a
residual larger than a factor times the standard deviation of all residuals
will have its weight reduced. Care must be taken so that such mechanisms
do not interfere with change detection.
7.4 Algorithm Implementation

The algorithm was implemented in a dSPACE AutoBox using code gen-
erated by Real-Time-Workshop from a Simulink model for use in a front-
wheel driven Volvo S40. The data-bin storage and the parameter opti-
mization were directly coded in C. The dSPACE-box contained a 1 GHz
PowerPC with the possibilities of floating-number operations and the pos-
sibility of data acquisition at 1 kHz with an A/D-conversion of 16 bit res-
olution. The sampling time for the estimator code is 10 ms. This section
shortly describes the implementation and the computations required to
derive the signals used by the friction estimator. The algorithm in this
implementation is only active during acceleration, since the required ac-
curacy on the vehicle speed estimation was not reached by other means
than using undriven or unbraked wheels. The implementation is only in-
tended for straight driving without any cornering.
Implementation structure
The model structure is described in Figure 7.19. The input signals to the
algorithm, the wheel speeds, the engine torque and rotation speed are
155
already available signals in the car. Hence, no extra sensors are needed.
The operations required to derive the normalized tire force and the slip
that are the inputs to the estimator are described in the following.
Input signal conditioning All signals are low-pass filtered.
Acceleration estimator The purpose of this block is to derive an ac-

celeration signal for the calculation of the vertical load on each wheel.
The vehicle acceleration is calculated by filtering and differentiation of
the speed of the undriven rear wheels.
Vertical force estimator The vertical wheel load, Fz, on each wheel
has to be determined to calculate the normalized tire force and the rolling
wheel-radius. The tire load is calculated from the force and moment equi-
librium of the car, concerning only longitudinal motion changes.
2( Fz f + Fzr ) − m = 0 (7.19a)
2Fzr l − ma x h − ml f = 0 (7.19b)
where l is the length between the wheel axles and l f is the horizontal
distance from the front axle to the vehicle center of gravity. The height of
the latter is denoted by h.
Gear estimator The different gear ratios of the gear-box are stored in a
table. By comparing the relation between the wheel speed and the engine
rotation the most appropriate ratio is chosen.
Driving force estimator The tire force is generated by multiplying the

engine torque with the actual gear ratio and the approximate effective
radius, given as a vehicle parameter. Energy losses in the power line are
accounted for.
Normalized force estimator The force input to the friction estima-

tor is derived by normalizing the tire force on the front wheels with the
corresponding vertical tire load.
Wheel radius correction For good optimization result it is important

that the rolling radius, R e , of each wheel is accurately estimated. The
rolling radius affects the slip calculation and a deviation from its nominal
value causes a horizontal shift on the force-slip relation, such that it do not
cross the origin. There are many factors that affect R e and the deviation,
δ R, has to be continuously updated by comparing the wheel speeds of
each tire with a reference speed when no forces are transmitted by the
156
estimator.
Figure 7.19 Wheel
ω Wheel Slip σx
Radius δR Estimator
Correction
Schematic block diagram of the implementation of the road friction
ω Long itudinal ax Fz
Vertical Force DataBins
Acceleration
ω Estimator
Estimator
fx Friction
σx Parameter
Data-Bin
Estimation
Storage
Normalized Algorithm
w Stiffness
Force
ω eng Input Signal Teng fx
Estimator
Condtioning
7.4
Algorithm Implementation
Driving force
Teng ω Gear ear Estimator Fx
ω eng Estimator
157
tire. If the periods of free rolling are not sufficient the estimation can
be complemented by adding one extra parameter, denoting the horizontal
shift, in the optimization routines. The resulting Jacobian is derived in
Appendix C.
Since the undriven rear wheels are used as reference for the vehicle
speed, only the differences in-between the radii of each tire are necessary
estimates for the slip computation. The deviations, δ R, are derived by
comparing rotational speed of each wheel, ω i , with the rotational speed of
the front-left wheel, ω f l and computing correction factors for the wheels
as
δ Ri = δ Ri + K ( R e f l ω f l − ( R ei )ω i ) (7.20)
where R ei = Rnom (1 + δ Ri + ∆ Ri (⋅)) is the rolling radius for the wheel i.

Known dependencies on other factors, such as the vertical load on the tire
can be included in ∆ R(⋅). Correction is only performed when no horizontal
forces are transmitted by the tire.
In many cases the dependency of the vertical load on the rolling radius
is small and almost neglectable. Investigations performed in the project
have shown that incorporating the assumed load effects on the radius,
reduces the braking stiffness estimate on asphalt about 25%, during ac-
celeration. Measurements performed within the project, see Figure 7.15,
and tests in the Pirelli laboratory [Pirelli, 2005] show an approximate
relation between the increment of rolling radius and the vertical load as
∆ R = kl ( Fnomz − Fz)/ Rnom (7.21)
which should be included in (7.20). The relation and certainly the value of
kl is tire specific and dependent on many other factors, but at an average
the optimization is enhanced, by using this correction and setting kl =
0.7 ⋅ 10−6 m/N.
Wheel slip estimator The tire slip on each of the driven wheels is
calculated according to the σ -definition, see also (2.5) as
vx − R e f l ω f l
σ xfl = (7.22)
Re f lω f l
and corresponding for the right wheel. The reference velocity is formed as
R erlω rl + R errω rr
vx = (7.23)
2
158
7.5 Results
Implementation of the friction estimator

The bins in the storage algorithm are filled with slip and force values
according to the description in Section 7.3. The x̂-values are replaced by
measurements of the slip σ x and ŷ-values are replaced by the normalized
tire force. Consequently the X - and Y-intervals are refereed to as S- and
F-bins. The estimator implementation reported in this thesis contains no
change-detection. The impact of outliers is reduced by (7.18). From the
experiences derived in the discussion concerning the reliability of the S-
and F-bin values, the weights corresponding to S-bins with pσ x p < K s are
set to zero to enhance the estimation of the tire stiffness at low slip.
The parameter estimation routine works as described in Section 7.3
with one iteration per sample. Some rules are set to ensure that the bins
contain enough information for a valid estimate. Attempting to estimate
the friction based on an insufficient amount of data may produce com-
pletely erroneous results and conclusions about the reliability. The rules
for the estimation scheme are made clear in Algorithm 1. The condition
on Line 1 states that there must be reliable values (at least Nlow mea-
surement points) in K 1 number of bins for estimation of the tire stiffness.
Before trying to estimate the friction, there must be even more reliable
bins ( K 2 > K 1 ) together with a minimum demand on the magnitude of
the lowest available slip and force value in the bins. Implicitly, this also is
a check of the ability to invert the matrices H T H and Φ T Φ , even if each
determinant has to checked before any inversion procedure.
7.5 Results
This section presents the results of the implemented estimator for the
winter tire on asphalt, snow and ice. The properties of the estimator are
discussed together with an analysis of the possibility to affect the behavior
by the tuning parameters. The results are further exemplifications on the
behavior and function of the algorithm. To give a clear picture of the
algorithm properties, the excitation procedure is an acceleration ramp,
proceeded by a free rolling sequence used for calibration of the wheel radii.
The estimator is reset and initialised before each test and the storage-bins
contains no prior information.
The measurements were performed on a Volvo S40 at Colmis Prov-
ing Ground in Arjeplog. The presented results are derived from off-line
computations of measured data. Though, the implementation was success-
fully running on-line and the execution of computational operations was
verified to be well within the processing limits for the dSPACE AutoBox,
it was decided to record the sampled signals to be able to elaborate the
159
Algorithm 1 Pseudo-code implementation of the rules of the friction es-

timator.
Old
Require: Ω ∈ {Ω i = [σ xi , f xi , wi ]}, Ĉ0x , µ̂ Old
Produces: Ĉ0x , µ̂
1: if Ω (wi > 0) ≥ K 1 then
P
L1
2: Calculate C0x using the Least-Squares method as described in (7.12) with
φi = −σ xi
3: Compute the corresponding cost function, Jcx from (7.9)
4: if ( Ω (wi > 0) ≥ K 2 ) & (max(p f xi p) > K f ) & (max(pσ xi p) > Kσ ) then
P
L2
5: Estimate C0x , µ L2 using the Least-Squares method as described in (7.12)
with φi = [−σ xi σ xi pσ xi p]T
6: Compute the corresponding cost function, Jcxµ from (7.9)
7: if µ L2 ∈ [0, K µ ] then
8: if Jcxµ < K j Jcx then
GN
9: Estimate C0x , µ GN by one iteration of the Gauss-Newton algorithm
(7.14).
GN
10: Ĉ0x , µ̂ Z C0x , µ GN
11: else
L2
12: Ĉ0x , µ̂ Z C0x , µ̂ L2
13: end if
14: else
L1
15: Ĉ0x , µ̂ Z C0x , µ̂ Old
16: end if
17: else
L1
18: Ĉ0x , µ̂ Z C0x , µ̂ Old
19: end if
20: else
Old
21: Ĉ0x , µ̂ Z C0x , µ̂ Old (Algorithm inactive)
22: end if
calibration of the algorithm afterwards.

To be able to verify the functions and limits for the main friction esti-
mator, i.e. line 9 in Algorithm 1, the tuning is set such that it is activated
at a low amount of available force and slip values. Valid information in at
least 6 bins is the demand for starting the Gauss-Newton algorithm.
Estimation results on asphalt

The measured and estimated signals on asphalt are shown in Figure 7.20.
It can be seen that the friction estimate quickly reaches the maximum al-
lowed value of 1.5. Without this limit the estimator would show an infinite
friction, since the available friction utilization of the tire force is too low
to be able to detect any curvature from the data. An exception is the
time slot between 6.5 and 7.2 s, where low friction is predicted. This is a
non-acceptable result, since the algorithm would not be able to distinguish
160
7.5 Results
between asphalt and snow at that period. It should, however, be noted that
the slip values in the bins are very low and the high sensitivity towards
wheel speed noise at low slip is one factor for the discrepancy. By a more
conservative tuning, for example using Kσ > 0.005 (see Algorithm 1) no
estimation would be available at the actual time slot and the misjudge-
ment is avoided. Also the disturbance on the braking stiffness estimation
in this period, invoked by the start of the Gauss-Newton iteration would
then be prevented. The plots of the bin-contents in Figure 7.21 explains
the reason for the erroneous estimation at t=6.7 s. The group of bins sur-
rounded by the dashed circle have a disturbance that gives an apparent
curvature in the tire-force relation. It is also clear from the right-most
plot that this group deviates from the others. The linearity of the bin-
values in the right-most plot also shows the difficulty of estimating the
friction at the low tire-force utilization. The remarkable difference of the
tire stiffness compared to previously reported VTI-measurement will be
discussed later.
Estimation results on snow

In the test on snow, see Figure 7.22, the estimator stabilizes at t (7.7 s.
The maximal f x -value in any bin at that time is below 0.2, approximately
50% of the available friction. The precision of the delivered friction esti-
mate is ±0.2 at that time, but improves with the increase of the available
force measurements. The underestimation of the friction, explained by
the error of the brush model on snow, also discussed in Section 7.2, is
an accuracy problem, limiting the precision of the method. The mismatch
between the model and the real tire behavior is clear from the plots in
Figure 7.23. It is, however, obvious that the difference in tire behavior is
clearly distinguished from asphalt at low tire force excitation.
Estimation results on ice

It is difficult to catch the friction coefficient on ice before reaching full uti-
lization of the friction. Directly when starting the real estimation, at the
condition of having data in 6 bins (t (5.8 s), see Figure 7.25, the friction is
fully utilized, but a good estimate of both tire parameters is provided. The
conclusion is comparable to the result presented in Table 7.2, describing
friction estimation from the VTI-measurements, where it was not possi-
ble to catch a correct estimation before full sliding started. On the other
hand, the algorithm certainly provides a good estimate directly when the
sliding starts. This is a great benefit compared to conventional stabiliza-
tion systems that waits for a certain wheel retardation before detecting a
wheel-lock situation where the friction might be estimated directly from
the corresponding actual tire force. The measurement sequence on ice is
shown in Figure 7.24.
161
Measurements and estimation results on asphalt.
Veh Spd [m/s]

30
25
20
0.4
0.2
x
f
0
Slip [%]
−0.5
−1
60
C0x
40
20
1.5
1
µ
0.5
0
2 3 4 5 6 7 8 9 10
Time [s]
Figure 7.20 Plot of measured vehicle velocity, the calculated normalized tire force,
and wheel slip and resulting estimations of the normalized braking stiffness and
friction coefficient on asphalt. Solid line denotes left wheel and dashed line right
wheel.
Discussions
As mentioned previously the algorithm is tuned to be fast and rather
volatile in the presented results. The tuning parameters are set according
to Table 7.5. In a real application it is important to tune it slower and more
robust towards noise and disturbances, since it is favourable to prevent
the deliverance of an erroneous estimate. The algorithm should, however,
be fast enough not to unnecessarily delay any reliable prediction, even on
ice. The parameters deciding when the storage-bins contain sufficient data
for friction estimation are K 2 , Kσ , and K f . Where K 2 describes the lowest
number of bins containing reliable force and slip values and Kσ and K f
determine the minimum requirement of the magnitude of the available
slip and force values, respectively. These parameters denote fix limits on
the bin-contents, independent on the road foundation. If the data contents
are below these limits only the normalized braking stiffness is estimated.
162
7.5 Results
Figure 7.21 Illustration of the bin contents from the winter tire test on asphalt,
shown together with the estimated tire characteristics for the right wheel during
the time instants 6.7 s (left plot) and 10 s (right plot). The size of the marker is
relative to the weight of the bin in the cost function. Only F-bins are present in the
plots, since the weight of the S-bins are reduced for slip below K s = 0.02. The group
of points surrounded by the dashed circle contains a disturbance that enforces the
bad estimation during the time period t =6.5–7.1 s.
The estimations performed by the least-squares method are indepen-

dent of the prior estimations. The choice of initial value on the estimate,
Cxinit and µ init , at the algorithm initialization is therefore small. But be-
fore starting the Gauss-Newton solver it should be secured that the ac-
tual estimates are in a valid range, not to risk an unstable iteration.
The choice of K j < 1 guarantees that a curvature is discerned from the
available bin-values, before starting Gauss-Newton iterations. The least-
squares algorithm on Line 5 in Algorithm 1 should then have provided
descent initial estimates. The parameter K j determines how much the
cost function should decrease when using a second order relation instead
of a first order relation. If the curvature is small the estimation of two
parameters in a linear relation, see (7.11) is sufficient. A difficulty with
the least-squares estimation is that negative values on the friction can be
obtained, if a negative curvature is detected. This is viewed as an tempo-
ral error and only the stiffness from the one-parameter linear estimation
should be delivered in such case.
The quickness of the algorithm to adapt to a changing condition is
determined by Ni , which should be set large enough such that tempo-
ral disturbances do not have a too large effect the estimation, but small
enough such that smaller deviations in tire characteristics, i.e. due to tem-
perature changes, can be correctly accounted for. Larger changes should
be detected by a change-detection algorithm, enforcing a complete reini-
tialisation of the bin-memory. The estimation robustness is adventured
if some bins contains measurements from new conditions and other from
163
Measurements and estimation results on snow.
Veh Spd [m/s]

20
15
10
0.5
fx
0
Slip [%]
−10
−20
40
0x
20
C
1.5
1
µ
0.5
0
5 6 7 8 9 10 11 12
Time [s]
Figure 7.22 Measured signals and result of friction estimation of the winter tire
on snow. Solid line denotes left wheel and dashed line right wheel.
older conditions.
The parameters Nlow and Nhigh are limits for the reliability of the bin-
value, stating that a certain number of data points are required for full
weight of a bin. It is logically to set Nhigh = Ni , since Nhigh points out that
the bin-value has full confidence.
There is a large difference in the normalized braking stiffness between
the estimation on the car and the VTI measurements on asphalt. The dif-
ference is almost a factor two, 50 compared to 25. One reason for the stiff-
ness deviation is the load dependency of the wheel radius. The test wheel
on the VTI-truck is not exposed to any load transfer during the force ramp,
which is inevitable for the personal car. Without adjustments for the load
dependency, the stiffness would be about 70 and a further increase of the
impact of load dependency, by raising ki , can not be motivated. Another
explanation might be an existing difference in the motion of the front and
rear axle, e. g. due to longitudinal deflection of the suspension caused
164
7.5 Results
Bin contents at t=8 s Bin contents at t=12 s
0.5 0.5
Normalized Tire Force

0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
−2.5 −2 −1.5 −1 −0.5 0 −20 −15 −10 −5 0
Slip [%] Slip [%]
Figure 7.23 Bin contents from measurement of the winter tire on snow together
with estimated tire characteristics for the right wheel during the time instants 8 s
(left plot) and 12 s (right plot). The size of the marker denote the weight of the bin
in the cost function. The F-bins are marked with ’o’ and the s-bins with ’*’.
by the applied acceleration torque. Simulations show, however, that this

has a small and neglectable effect. The derivation of the normalized force
introduces error sources, as efficiency losses, gear ratios, and uncertain-
ties in torque measurements and load transfer etc that might affect the
results. A distinct difference between the measurements are that the VTI
measures while braking, when the estimator implemented in the personal
car only uses driving data. The potential error sources will be further pen-
etrated in the future work. Any physical reason for the tire to be stiffer
on the car relative the test vehicle has not yet been found.
It is very important that the tire radius is accurately determined. For
foundations such as e.g. asphalt that gives a high stiffness on the tire the
accuracy on δ R/ Rnom must be lower than one per thousand. It is not en-
tirely verified that it is possible to obtain this high accuracy. As previously
mentioned, the radius estimator can be combined with the estimation of
the horizontal shift as a third parameter in the Gauss-Newton algorithm.
The expression for the extended Jacobian is described in Appendix C.
In Figure 7.26 the resulting estimates are shown for the measurement
case from Figure 7.20 when starting with incorrect wheel radius, but cor-
recting its value by inclusion of the horizontal shift as an estimated pa-
rameter in the Gauss-Newton iteration. It is clear from the results that
the estimates stabilize and that the stiffness value is identical to the re-
sult with correct and fixed wheel radius. This ensures that the horizontal
shift is correctly estimated. The period of an uncertain friction estimation
is, however, longer which indicates, not surprisingly, that the estimation
of three parameters makes the algorithm more sensitive towards noise
165
Measurements and estimation results on ice.
Veh Spd [m/s]

11
10
0.2
0.1
x
f
0
Slip [%]
−20
−40
40
0x
20
C
1.5
1
µ
0.5
0
4 5 6 7 8 9
Time [s]
Figure 7.24 Measured signals and result of friction estimation of the winter tire
on ice. Solid line denotes left wheel and dashed line right wheel.
and errors, but on the other hand gives a better estimate if the data ex-
citation is sufficient and an accurate wheel radius is hard to obtain by
other means.
7.6 Conclusions
This chapter has described a new type of friction estimator based on force
and slip measurements. The tire behavior is assumed to follow the brush
model prescribing the force and slip relation, depending on the braking
stiffness of the tire and the road-tire friction. From the measurements
these two parameters can be derived. A major invention of the proposed
method is the way to collect the sampled measurements into bins, such
that the available data used for optimization is evenly spread and weighted
along the force and slip axis.
166
7.6 Conclusions
Bin contents at t=5.9 s Bin contents at t=9 s

0.2 0.2
0.18 0.18
0.16 0.16
0.14 0.14

0.12 0.12
0.1 0.1
0.08 0.08
0.06 0.06
0.04 0.04
0.02 0.02
0 0
−3 −2.5 −2 −1.5 −1 −0.5 0 −35 −30 −25 −20 −15 −10 −5 0
Slip [%] Slip [%]
Figure 7.25 Bin contents from measurement of the winter tire on ice together
with the estimated tire characteristics for the right wheel during the time instants
5.9 s (left plot) and 9 s (right plot). The size of the marker denote the weight of the
bin in the cost function. The F-bins are marked with ’o’ and the s-bins with ’*’.
100
Cx
50
1.5
1
µ
0.5
0
S [%]
−0.5
h
−1
2 3 4 5 6 7 8 9 10
Time [s]
Figure 7.26 Plot of parameter estimation with the horizontal shift as as third es-
timated factor. The measurement sequence is the same as one shown in Figure 7.20.
The algorithm switches between different optimization methods de-

pending on the contents in the bins. If only low slip and forces are avail-
able the braking stiffness is estimated as the sole parameter using the
least-squares method. For slightly higher values of the signals both the
friction and the stiffness are derived by the same method. When the curva-
ture of the force-slip relation becomes evident the Gauss-Newton method
167
Table 7.5 List of tuning parameter for the actual implementation of road friction
estimator
Parameter Description Nom. value
Cxinit Initial value on Cx 40

µ init Initial value on µ 0.5
Ks Slip limit for using S-bin 0.02
K1 Required filled bins for C0x - 3
estimation
K2 Required filled bins for 6
C0x , µ -estimation
Kj Residual difference to en- 1
able Gauss-Newton algo-
rithm
Kσ Slip limit for enable µ - 0
estimation
Kf Limit on normalized force to 0
enable µ -estimation
kl Load dependence on rolling 0.7 ⋅ 10−6
radius
Nlow Threshold for measurement 2
points in bin
Nhigh Number of measurement 20
points in bin to maximize
weight
Ns Number of slip bins 150
Smax Maximum slip bin value 0.5
Nf Number of force bins 150
Fmax Maximum force bin value 1.2
Ni Number of points in bin 100
memory
is used for better accuracy. The evaluation of the result has focused on
the Gauss-Newton method.
The estimator works well in the presented situations. From no prior
information it distinguishes between snow and asphalt for an available
normalized tire force below 0.25. The estimate on snow available at that
time is lower than expected, which depends on that the tire behavior
on snow slightly differs from the model. The estimator detects and pre-
scribes the low friction on ice when the tire force reaches its peak level
168
7.6 Conclusions
which is before the wheel starts its locking phase and before a high ac-
celeration/retardation can be noticed. The accuracy on asphalt has not
been further validated, since larger accelerations, than could be achieved
during the measurement occasion, are needed to reach sufficient friction
utilization. The output from the estimator, in that test, only indicates
that there are much more friction available. One of the task for the fu-
ture work is to optimize the signal quality and estimation performance to
reduce the need of friction utilization for a sufficiently accurate friction
determination. The estimator has been validated through force excitation
as an acceleration ramp. Its structure has been constructed to effectively
cope with weighting problems when the tire force is kept constant on a
certain level for a long time or varying in a way such that the data is not
well spread in the force-slip plane. The accuracy for estimation during
this kind of unfavourable excitations needs further validation. The esti-
mator implementation is so far restricted to acceleration with two-wheel
driven vehicles. The major reason for this restriction is the insufficient
accuracy of the estimate vehicle reference velocity if not measured by the
undriven wheel-rotations. Further work incorporates enlargement of the
usable area of the estimator to work for both braking and acceleration
during simultaneous cornering.
169
8
Conclusions
The thesis has presented research performed mainly in the related areas
of tire modeling and friction estimation. The theory behind the brush tire
model developed by the pioneers in tire modeling has been a central theme
throughout the work. The thesis contains an extensive review of the brush
model including the effects of camber. It also presents a validation of the
model towards measurements and an evaluation of the variation of the
included parameters, due changes in conditional factors, such as load and
road surface.
A new method to derive the tire forces for simultaneous braking, cor-
nering and camber, by combining empirical models for pure braking, cor-
nering and cambering has been presented. Based on brush-model me-
chanics, the combined-slip forces may be described by a scaling of corre-
sponding empirical pure-slip forces. The way to derive the scale factors by
dividing the expression for the combined slip force with the pure slip force
in the appropriate direction is an unique approach. The generation of the
camber force has been described and included in a simple, but physically
motivated manner. The pure-slip tire model can be given as a empirical
model or as raw tabular data, as long as the horizontal and vertical shift
are zero, i.e. that there is no tire force for zero slip.
A major aim of the semi-empirical modeling has been to extract as
much information from the available pure-slip curve as possible. The phys-
ical approach has made it possible to include velocity dependency and
conversion between braking and driving data in the model. The proposed
model is simple to use, since it does not introduce any new parameters
and all necessary information is given by the chosen pure-slip model.
A dynamic extension of the steady-state combined-slip model is pro-
posed, based on qualitatively realistic tire relaxation behaviour. The model
includes cross-couplings between longitudinal and lateral dynamics, that
are not commonly found in other models of similar complexity. The issue
of stability is handled by including a physically motivated damping.
170
The tire model was successfully implemented in Matlab and Modelica
code and tested together with the VehicleDynamics Library in the multi-
body simulation environment, Dymola. The implementation was verified
to be well-working and computationally sound on both platforms. Good re-
sults are obtained in validation with the available empirical data. Experi-
mental validation indicates that the proposed model and previous models
perform similarly. The efficiency in computational means compared to
other models has not yet been evaluated. The presented model is advan-
tageous in that it combines the data-fitting accuracy of empirical pure slip
models with the stringency of physical modeling. This claims descent be-
havior, according to the physical constraints, in the defined working area
without the need for calibration, at the cost of quite large expressions. The
methodology to derive the scale-factors based on first principles makes it
straight forward to include new functionality to the model.
The wheel speed signal is a most important signal in many system
applications. Based on problems observed during measurements, a few
ways to reduce noise from the signal are discussed in the thesis. One
method to compensate the wheel-speed signal for disturbances caused by
axle and suspension deflections during transients in brake applications is
proposed. The disturbance is predicted by a first order feedforward filter
of the brake torque signal. From the validation it can be seen that the
signal error can, by this simple means be reduced by at least 50%. This
is probably sufficient to significantly increase the performance of vehicle
systems relying on the wheel-speed signal during brake-force transients.
The thesis also describes the development of a new type of friction
estimator based on the assumption that the tire behavior follows the brush
model. Since the brush model predicts the tire characteristic, as depending
on the braking stiffness and friction coefficient, these parameters can be
estimated from the force and slip measurements. Experimental data has
been collected and evaluated to ensure the validity of the brush model
during certain conditions. The shortcomings and limitations of the model
accuracy are discussed. A major invention in the proposed method is the
way of collecting the measurements into bins, such that the available data
used for optimization is evenly weighted along the force and slip axis. The
least-squares and the Gauss-Newton methods are used for finding the
most accurate values of the stiffness and friction.
The estimator has been implemented and validated on a passenger
car. For a sequence in the form of an acceleration ramp it has shown to
work well in the tested situations.
As expected, the amount of friction utilization is strongly connected
to the accuracy of the estimation and the usability of the method will
strongly relate to the requirement regarding accuracy and possibilities
for utilization.
171
Chapter 8. Conclusions
The structure of the estimator is developed to effectively cope with

weighting problems when the tire force is kept constant on a certain level
for a long time or varying in a way such that the data is not well spread
in the force-slip plane. The accuracy for estimation during this kind of
unfavourable excitations needs further validation. It can be concluded that
this way of approaching the problem of friction estimation is promising.
8.1 Ideas of Further Research

All of the presented methods and ideas can be further developed, analysed,
and tested. The particular needs have been described in the respective
sections, from where the most urgent and important issues are concluded
in the following list:
• The wheel speed signal is essential for many vehicle systems and
further work to improve the quality of that signal is important. The
proposals mentioned in the thesis have to be further evaluated and
developed by testing in real implementations.
• The proposed semi-empirical tire-model is, in its present state, use-
ful for a number of applications, but the addition of effects of turn-
slip and flexible carcass would further enhance the applicability. The
physical foundation of the model is expected to make this possible
with reasonable effort. Further validation of, particularly, the effects
of camber are need.
• It has to be emphasised that the proposed friction estimator is de-
veloped in an ongoing project. The presented result should therefore
be seen as the first glimpse of the possibilities to use this type of
methodology for this purpose. A large amount of work is still required
to achieve good response of the estimate when changing surface and
to ensure the ability of catching slow variations of the parameters.
Solutions for change detection and outlier management have to be
implemented and calibrated, such that a maximum of robustness
and accuracy is achieved. Improvement of signal quality and esti-
mator performance to minimize the need of friction utilization are
other needs. Further, a lot of tests and validations are necessary
really evaluate the performance in action and to reveal not yet dis-
covered weak points.
• A demand for developing a slip-based friction estimator for use dur-
ing braking or for “All Wheel Drive” vehicles is to have a good es-
timate of the vehicle velocity or to succeed in removing this depen-
dence from the estimator structure. The area of usability is also to
172
8.1 Ideas of Further Research
be enlarged to enable friction estimation during cornering, when a

lateral slip affects the longitudinal tire properties.
An important conclusion of a research project is often how the work

should proceed in the development of an idea or a product. Since the pos-
sible fields of improvements in vehicle related issues have been expanded,
by the newly available techniques and instrumentation possibilities, it is
important to put the efforts where they have the best effect. Hopefully,
this thesis may serve as a guide in this purpose for the related topics.
173
9
Bibliography
Analog Devices (2001): Wheel speed sensor for ABS systems, AD22157.
Åström, K. J. (1970): Introduction to Stochastic Control Theory. Academic
Press.
Åström, K. J. and B. Wittenmark (1994): Adaptiv Control. Addison-Wesley
Pub Co.
Bakker, E., L. Nyborg, and H. B. Pacejka (1987): “Tyre modelling for use
in vehicle dynamics studies.” SAE Technical Paper 870421.
Bakker, E., H. B. Pacejka, and L. Lidner (1989): “A new tire model with an
application in vehicle dynamics studies.” SAE Technical Paper 890087.
Bayle, P., J. F. Forissier, and S. Lafon (1993): “A new tire model for vehicle
dynamics simulations.” Automotive Technology International, pp. 193–
198.
Bernard, J. E., P. S. Fancher, R. Gupta, H. Moncarz, and L. Segel (1975):
“Vehicle-in-use limit performance and tire factors - the tire in use.
appendix a, b. c.” Technical Report UM-HSRI-PF-75-1-2. Highway
Safety Research Institute, Ann Arbor, Michigan.
Böhm, F. and H.-P. Willumeit, Eds. (1996): Tyre Models for Vehicle
Dynamics Analysis. Swets & Zeitlinger Publishers. Supplement to
Vehicle System Dynamics Volume 27.
Böiers, L. C. (2004): Lectures on Optimization. Center for Mathematical
Science, Lund University.
Brach, R. M. and R. M. Brach (2000): “Modeling combined braking and
steering tire forces.” SAE Technical Paper 2000-01-0357.
Carlson, C. and C. Gerdes (2003): “Nonlinear estimation of longitudinal
tire slip under several driving conditions.” In Proceedings of the
American Control Conference, vol. 6, pp. 4975–4980.
174
Casselgren, J., M. Sjödahl, and M. S. Sara Woxneryd (2007): “Classifica-
tion of road conditions – to improve safety.” In 11th International Fo-
rum on Advanced Microsystems for Automotive Applications. Berlin.
To be published.
Clark, S. K., R. N. Dodge, and G. H. Nybakken (1971): “An evaluation of
string theory for the prediction of dynamic tire properties using scale
model aircraft tires.” Technical Report. University of Michigan.
Colmis AB (2006): “Colmis proving ground, Arjeplog, Sweden.” Internet.
http://www.colmis.com.
Continental (2003): “History of the tire.” Internet. http://www.conti-
online.com/generator/www/us/en/continental/transport/themes/
about_continental/tire_history/tire_history_en.html.
de Groen, P. (1996): “An introduction to total least squares.” Nieuw Archief
voor Wiskunde, pp. 237–253.
Do, M.-T. and Y. Delanne (2003): “Prediction of tire/wet road friction and
its variation with speed from road macro- and microtexture, and tire-
related properties.” Technical Report. French Public Works Research
Laboratory, France.
Dugoff, H., P. S. Fancher, and L. Segel (1969): “Tire performance char-
acteristics affecting vehicle response to steering and braking control
inputs. final report.” Technical Report. Highway Safety Research In-
stitute, Ann Arbor, Michigan.
Dynasim (2006): “Dymola, dynamic modeling laboratory.” Internet.
http://www.dynasim.se/.
Eichhorn, U. and J. Roth (1992): “Prediction and monitoring of tyre/road
friction.” In XXIV FISITA Congress: Safety, the Vehicle, and the Road,
pp. 67–74. London.
Ellis, J. R. (1994): Vehicle Handling Dynamics. Mechanical Engineering
Publications Limited, London.
Fancher, P. S., L. Segel, C. C. MacAdam, and H. Pacejka (1972):
“Tire traction grading procedures as derived from the maneuvering
characteristics of a tire-vehicle system. volume i and ii. final report.”
Technical Report HSRI-71-129. Highway Safety Research Institute,
Ann Arbor, Michigan.
Gäfvert, M. (2003): Topics in Modeling, Control, and Implementa-
tions in Automotive Systems. PhD thesis ISRN LUTFD2/TFRT--
LUTFD2/TFRT–1066–SE--SE, Lund University.
175
Chapter 9. Bibliography
Gäfvert, M. and J. Svendenius (2003): “Construction of semi-empirical tire

models for combined slip.” Technical Report ISRN LUTFD2/TFRT--
7606--SE. Department of Automatic Control, Lund University, Sweden.
Gäfvert, M. and J. Svendenius (2005): “A novel semi-empirical tire model
for combined slip.” Vehicle System Dynamics, 43:5.
Gäfvert, M., J. Svendenius, and J. Andreasson (2006): “Implementation
and application of a semi-empirical tire-model in multi-body simula-
tion of vehicle handling.” In Proceedings of the 8th International Sym-
posium on Advanced Vehicle Control. Taipei, Taiwan.
Gelb, A., J. Kasper, R. Nash, C. Price, and A. Sutherland (1977): Applied
Optimal Estimation. The MIT press.
Gim, G. and P. E. Nikravesh (1991): “An analytical model of pneumatic
tires for vehicle dynamics simulations. Part 2: Comprehensive slips.”
International Journal of Vehicle Design, 12:1, pp. 19–39.
Gipser, M. (2005): “FTire homepage.” Internet. http://www.ftire.com.
Gothie, M., T. Parry, and P. Roe (2001): “The relative influence of
the parameters affecting road surface friction.” In 2nd International
Colloquium on Vehicle Tyre Road Interaction. Florence, Italy.
Goyal, S. (1989): Planar Sliding of a Rigid Body with Dry Friction: Limit
Surfaces and Dynamics of Motion. PhD thesis, Cornell University,
Department of Mechanical Engineering.
Guo, K., Y. Zhuang, S. Chen, and L. Willlam (2006): “Experimental
research on friction of vehicle tire rubber.” Frontiers of Mechanical
Engineering in China, 1:1, pp. 14–20. DOI 10.1007/s11465-005-0001-z.
Gustafsson, F. (1997): “Slip-based tire-road friction estimation.” Automat-
ica, 33:6, pp. 1087–1099.
Gustafsson, F., J. Nilsson, and T. Gustavsson (2006a): “Method and
system of determining the absolute velocity of a vehicle.” Patent no.
EP1642140.
Gustafsson, M., C.-M. Berglund, B. Forslund, and I. Forslund (2006b):
“Effekter av vinterdäck – en kunskapsöversikt.” Technical Report. VTI.
Report 543.
Hägglund, T. and K. J. Åström (2000): “Supervision of adaptive control
system.” Automatica, 36, pp. 1171–1180.
Haney, P. (2003): Racing & High Performance Tire: Using Tires to Tune
for Grip and Balance. SAE.
176
Hirschberg, W., G. Rill, and H. Weinfurter (2002): “User-appropriate
tyre-modelling for vehicle dynamics in standard and limit situations.”
Vehicle System Dynamics, 38:2, pp. pp. 103–125.
Isermann, R., P. Scheerer, O. Nelles, and R. Schwarz (2000): “Method for
compensating variations of a wheel speed sensor.” Patent no. 6,446,018.
Assignee: Continental Teves.
ISO 8855 (1991): “Road vehicles — Vehicle dynamics and road-holding
ability — Vocabulary.”
Johansson, R. (2002): System Modeling and Identification. Prentice Hall.
Kageyama, I. and S. Kuwahara (2002): “A study on tire modeling for
camber thrust and camber torque.” JSAE Review Paper No. 20024252.
Kiencke, U. and L. Nielsen (2000): Automotive Control Systems: For
Engine, Driveline, and Vehicle. Springer-Verlag.
Kim, S. and A. Savkoor (1996): “The contact problem of in-plane rolling of
tires on a flat road.” In Böhm and Willumeit, Eds., Tyre Models for Ve-
hicle Dynamic Analysis, pp. 189–206. Swets & Zeitlinger. Supplement
to Vehicle System Dynamics Volume 27.
Levy, G. and N. Fangeat (2004): “Automatic stability control system for
a vehicle using an invariant charaterizing any tire.” Patent no US
2004/0032165 A1.
Ljung, L. and T. Söderström (1983): Theory and Practice of Recursive
Identification. The MIT Press.
Lugner, P. and P. Mittermayr (1991): “A measurement based tyre char-
acteristics approximation.” In Pacejka, Ed., Tyre Models for Vehicle
Dynamics Analysis, pp. 127–144. Swets & Zeitlinger. Supplement to
Vehicle System Dynamics Volume 21.
Matschinsky, W. (1997): Road Vehicle Suspension. John Wiley & Sons,
Inc. ISBN 1-86058-202-8.
Meschke, G., C. Liu, and H. Mang (1996): “Large strain finite-element
analysis of snow.” Journal of Engineering Mechanics, 122:7, pp. 591–
602.
Michelin (2004): “Annual report.” Internet. www.michelin.com.
Mizuno, M., H. Sakai, and K. Oyama (2005): “Developement of a tyre force
model incorporating the influence of the tyre surface.” In Lugner and
Plöchl, Eds., Tyre Models for Vehicle Dynamic Analysis, pp. 395–402.
Taylor & Francis. Supplement to Vehicle System Dynamics Volume 43.
177
The Modelica Association (2005): ModelicaF

R — A Unified Object-Oriented
Language for Physical Systems Modeling: Language Specification
Version 2.2.
Modelon (2006): “VehicleDynamics Library for Modelica.” Internet.
http://www.modelon.se/index.php?did=25&level=2.
Müller, S. and M. Uchanski (2001): “Slip-based tire-road friction estima-
tion during braking.” In Proceedings of IMECE01. ASME.
Müller, S., M. Uchanski, and K. Hedrick (2003): “Estimation of the maxi-
mum tire-road friction coefficient.” Journal of Systems, Measurements
and Control.
Nguyen, P. K. and E. R. Case (1975): “Tire friction models and their effect
on simulated vehicle dynamics.” In Proceedings of a Symposium on
Commercial Vehicle Braking and Handling, number UM-HSRI-PF-75-
6, pp. 245–312.
Nicholas, V. T. and T. Comstock (1972): “Predicting directional behavior
of tractor semitrailers when wheel anti-skid brake systems are used.”
In ASME Winter Annual Meeting, number Paper No. 72–WA/Aut-16.
Nielsen, L. and L. Eriksson (1999): “Course material vehicular systems.”.
Linköping Institute of Technology, Vehicular Systems, ISY.
Nordström, O. (1983): Antilåssystem för tunga fordon, VTI Rapport nr
257. Klintland grafiska.
Nordström, O. and H. Åström (2001): “Upgrading of VTI friction test vehi-
cle BV12 for combined braking and steering tests under aquaplaning
and winter conditions.” In 2nd International Colloquium on Vehicle
Tyre Road Interaction. Florence, Italy.
Oertel, C. and A. Fandre (1999): “Ride comfort simulations and step
towards life time calculations: RMOD-K and ADAMS.” In International
ADAMS Users’ Conference, Berlin.
Olsson, H. (1996): Control Systems with Friction. PhD thesis ISRN
LUTFD2/TFRT--1045--SE, Department of Automatic Control, Lund
University, Sweden.
Optical Sensors (2006): “Road eye – a laser sensor for monitoring road
state.” Internet. http://opticalsensors.se/roadeye.htm.
Pacejka, H. B. (1988): “Modelling of the pneumatic tyre and its impact
on vehicle dynamic behavior.” Technical Report i72B. Technische
Universiteit, Delft.
178
Pacejka, H. B. (2002): Tyre and Vehicle Dynamics. Butterworth-
Heinemann.
Pasterkamp, W. R. and H. B. Pacejka (1997): “The tyre as a sensor to
estimate friction.” Journal of Vehicle System Dynamics, 27, pp. 409–
422.
Pauwelussen, J. P., G. Anghelache, C. Theodorescu, and A. Schmeitz
(1999): European Tyre School, chapter 10 — Truck tyre behavior
in use and testing methods. Nokian Tyres plc. http://www.tut.fi/
plastics/tyreschool.
Pavkovi, D., J. Deur, J. Asgari, and D. Hrovat (2006): “Experimental
analysis of potentials for tire friction estimation in low-slip operating
model.” SAE Technical Paper 2006-01-05561.
Persson, N. (2002): “Event based sampling with application to spectral
estimation.” Technical Report. Department of Electrical Engineering,
LiU, Sweden. Lic. Thesis No 981.
Persson, N. and F. Gustafsson (2001): “Event based sampling with ap-
plication to vibration analysis in pneumatic tires.” In IEEE Interna-
tional Conference on Acoustics, Speech, and Signal Processing, vol. 6,
pp. 3885–3888. IEEE, Salt Lake City, UT, USA.
Persson, N., F. Gustafsson, and M. Drevö (2002): “Indirect tire pressure
monitoring using sensor fusion.” SAE Technical Paper 2002-01-1250.
Pirelli (2005): “Measurement of tire rolling radius.” Confidential.
Pohl, A., R. Steindl, and L. Reindl (1999): “The "intelligent tire" utilizing
passive saw sensors-measurement of tire friction.” IEEE Transactions
on Instrumentation and Measurement, 48:6.
Pottinger, M. G., W. Pelz, and G. Falciola (1998): “Effectiveness of the slip
circle: "COMBINATOR", model for combine tire cornering and braking
forces when applied to a range of tires.” SAE Techical Paper 982747.
ProAuto Limited (2006): “Tech talk - anti-lock braking systems – Part 3.”
Motor Industry Magazine, March, pp. 38–39.
Radt, H. (1995): “Tire data treatment.” In Milliken and Milliken,
Eds., Race Car Vehicle Dynamics. Society of Automotive Engineers,
Warrendale, U.S.
Ray, L. (1997): “Nonlinear tire force estimation and road friction identi-
fication: Simulation and experiments.” Automatica, 33:10, pp. 1819–
1833.
179
SAE Recommended Practice J670e (1976): “Vehicle dynamics terminol-

ogy.”
Saino, P. (2001): “Snow surface model for tyre performance simulation.”
In 2nd International Colloquium on Vehicle Tyre Road Interaction.
Florence, Italy.
Salaani, K., D. Guenther, and G. Heydinger (1999): “Vehicle dynamics
modeling for the national advanced driving simulator of a 1997 Jeep
Cherokee.” SAE Paper No. 1999-01-0121.
Schmeitz, A., I. Besselink, J. de Hoogh, and H. Nijmeijer (2005): “Ex-
tending the magic formula and swift tyre model for inflationpres-
sure changes.” In Wies and Wallentowitz, Eds., Reifen, Fahrwerk,
Fahrbahn, pp. 201–225. Hannover, Germany.
Schofield, B., T. Hägglund, and A. Rantzer (2006): “Vehicle dynamics con-
trol and controller allocation for rollover prevention.” In Proceedings of
the IEEE International Conference on Control Applications. Munich,
Germany.
Schuring, D. J., W. Pelz, and M. G. Pottinger (1996): “A model for combined
tire cornering and braking forces.” In Investigations and Analysis in
Vehicle Dynamics and Simulation, pp. 61–83. SAE International. SAE
Technical Paper 960180.
Schwarz, R. (1999): Rekonstruktion der Bremskraft bei Fahrzeugen mit
Elektromechanisch Betätigten Radbremsen. PhD thesis, Technische
Universität Darmstadt.
Schwarz, R., O. Nelles, P. Scheerer, and R. Isermann (1997a): “Increasing
signal accuracy of automotive wheel-speed sensors by on-line learning.”
In Procedings of the American Control Conference, pp. 1131–1135. New
Mexico.
Schwarz, R., M. Willimowski, R. Isermann, and P. Willimowski (1997b):
“Improved wheel speed and slip determination considering influences
of wheel-suspension dynamics and tire dynamics.” SAE-paper No.
971117.
Sharp, R. (2004): “Testing and improving a tyre shear force computation
algorithm.” Journal of Vehicle System Dynamics, 41:3, pp. 223–247.
Sharp, R. and M. Bettella (2003): “Shear force and moment descriptions
by normalisation of parameters and the "magic formula".” Journal of
Vehicle System Dynamics, 39:1, pp. 27–56.
Solyom, S., A. Rantzer, and J. Lüdemann (2004): “Synthesis of a model-
based tire slip controller.” Vehicle System Dynamics, 41:6, pp. 477–511.
180
Sugai, M., K. Asano, and T. Umeno (2003): “Physical quantity estimating
apparatus and tire state determining aparatus, and estimating method
of same and determination of same.” Patent no. US 2003/0192375 A1.
Svendenius, J. (2003): “Tire models for use in braking applications.”

Licentiate thesis ISRN LUTFD2/TFRT--3232--SE. Department of
Automatic Control, Lund University, Sweden.
Svendenius, J. (2007a): “Validation of the brush model towards VTI-

measurement data recorded at Hällered 2005.” Technical Report ISRN
Svendenius, J. (2007b): “Validation of the brush model towards VTI-

measurement data recorded in Arjeplog 2006.” Technical Report ISRN
Svendenius, J. and M. Gäfvert (2004a): “A brush-model based semi-

empirical tire-model for combined slips.” SAE Technical Paper 2004-
01-1064.
Svendenius, J. and M. Gäfvert (2004b): “A semi-empirical tire-model for

transient combined-slip forces.” In AVEC ’04.
Svendenius, J. and M. Gäfvert (2005): “A semi-empirical tire model for

combined slips including the effects of cambering.” In Lugner and
M.Plöchl, Eds., Tyre Models For Vehicle Dynamics Analysis, vol. 43,
pp. 317–328. Taylor & Francis Group, Glasgow. Supplement to Vehicle
System Dynamics Volume 43.
Svendenius, J. and M. Gäfvert (2006): “A semi-empirical dynamic tire

model for combined-slip forces.” Vehicle System Dynamics, 44:2,
pp. 189 – 208. Special Issue: AVEC ’04: 7th International Symposium
on Advanced VEhicle Control, 23-27 August 2004 HAN University,
Arnhem, The Netherlands.
Svendenius, J. and B. Wittenmark (2003): “Brush tire model with

increased flexibility.” In ECC. European Union Control Association,
Cambridge, UK.
Thomson, J. (2003): “What we should know about tires.” Internet.

http://www.jags.org/TechInfo/2001/05May/tires.
Thorvald, B. (1998): On truck tyre modelling. PhD thesis ISRN

LUTFD2/TFRT--KTH/FKT/DA–98/29–SE--SE, KTH.
181
Tielking, J. T. and N. K. Mital (1974): “A comparative evaluation of five

tire traction models.” Technical Report UM-HSRI-PF-74-2. Highway
Safety Research Institute, Ann Arbor, Mich.
TYDEX-Working group (1997): “Tydex-format, description and reference
manual.”
Umeno, T., E. Ono, and K. Asano (2002): “Estimation of tire-road friction
by tire rotation vibration.” SAE paper no. 2002-01-1183.
Vägverket (2006a): “EU-project: INTRO – Intelligent Roads.” Internet.
http://www.vti.se/templates/Page____5777.aspx.
Vägverket (2006b): “Homepage of IVSS.” Internet. http://www.ivss.se/.
Vägverket (2006c): “IVSS-project: Slippery Road Information System.”
Internet. http://www.ivss.se/templates/ProjectPage.aspx?id=206.
Velenis, E., P. Tsiotras, and C. C. de Wit (2002): “Extension of the
LuGre dynamic tire friction model to 2D motion.” In Mediterranean
Conference on Control and Automation (MED2002), vol. 10th. Lisbon,
Portugal.
Volvo AB (2006): “Volvo proving ground Hällered.” Internet. Available
from www.volvo.com/NR/rdonlyres/F7CB2A13-6BD6-495E-898E-
D491 A2683BE9/0/hallered_proving_ground_rsp_83364_eng.pdf.
Wallman, C.-G. and H. Åström (2001): “Trafiksäkerhet, vägkonstruktion,
drift och underhåll, vägunderhåll, vinterunderhåll.” Technical Report.
Statens väg-och transportforskningsinstitut. VTI meddelande 991 A.
Wong, J. Y. (2001): Theory of Ground Vehicles, 3rd edition. John Wiley &
Sons, New York.
Yasui, Y., W. Tanaka, Y. Muragishi, E. Ono, M. Momiyama, H. Katoh,
H. Aizawa, and Y. Imoto (2004): “Estimation of lateral grip margin
based on self-aligning torque for vehicle dynamics enhancement.” SAE
Technical Paper 2004-01-1070.
182
A
Nomenclature
The tables below describe the nomenclature and the notations used in the
thesis. If there is a different ISO-definition for an entity the corresponding
ISO notation is prescribed in the second column. The description, together
with a reference to the defining equation of the entity, is given in the third
column.
Nomenclature ISO Description

Coordinates and
Positions
x x Longitudinal direction, in the front di-
rection of the vehicle
y −y Lateral direction
z −z Vertical direction, positive downwards
xr Longitudinal coordinate for bristle po-
sition at the road contact
yr Lateral coordinate for bristle position
at the road contact
Movements
v̄ Horizontal wheel hub velocity vector
vx Longitudinal wheel hub velocity
vy Lateral wheel hub velocity
vsx Longitudinal tire slip velocity (2.3)
vsy Lateral tire slip velocity (2.3)
vc Carcass velocity (2.2)
ω Wheel rotational velocity
ω0 Wheel rotational velocity (free
rolling)
183
Appendix A. Nomenclature
γ ǫV Camber angle
Forces and
Torques
F̄ Force vector
Fx Longitudinal tire force
Fy -Fy Lateral tire force
Fz -Fz Vertical tire force
f¯ µ̄ W Force vector normalized by vertical
force
Mz Self-aligning torque
Mz′ Self-aligning torque
Mz′′ Second order torque addition, due to
deformation
My Rolling resistance
Mx Overturning torque
d F̄ Force vector on infinitesimal bristle
Deflections
δx Longitudinal deflection of tire, fig
4.15
δy Lateral deflection of tire
δ xs Longitudinal tire deformation due to
slip (4.3)
δ ys Lateral tire deformation due to slip
δ xb Longitudinal deflection of bristle
δ yb Lateral deflection of bristles
δ yc Lateral deflection of carcass relative
leading edge of contact patch
δ yctot Total lateral deflection of carcass
δ y,cam Lateral deflection of due to camber
Slip
α α Slip angle (2.1)
λ −S X w Longitudinal slip (2.5– 2.6)
184
sH Horizontal shift (7.2)
σx Longitudinal physical slip (2.5)
σy Lateral physical slip (2.5)
β Slip velocity angle (2.4)
βf Angle for calculation of friction force
direction (4.28)
β′ Suggestion for β f to achieve collinear-
ity (4.20)
β SPM Suggestion for β f, slip-projection
(4.23)
β MDR Suggestion for β f , MDR-principle
(4.24)
ψ Normalized slip (4.49)
Parameters
R0 Unloaded radius of tire
R Tire radius (distance from hub to con-
tact patch). Force lever arm
Re Rolling radius
σ ax Longitudinal relaxation length
σ ay Lateral relaxation length
Cx Braking stiffness
Cy Cornering stiffness
Cy′ Cornering stiffness, calculated under
the assumption of a flexible carcass.
Cγ Camber stiffness
Cγ′ Camber stiffness, derived from (4.53),
(4.45) and (4.42)
Ccx Longitudinal carcass stiffness
Ccy Lateral carcass stiffness
Cc Lateral carcass bending-stiffness
cp Rubber bristle stiffness
µ Friction coefficient
Jw Wheel inertia
185
Semi-Empirical
Scale Factors
G ax Longitudinal scale factor for adhesion
force (5.9)
G ay Lateral scale factor for adhesion force
(5.9)
G sx Longitudinal scale factor for sliding
force (5.13)
G sy Lateral scale factor for sliding force
(5.13)
G camy Lateral scale factor for camber force
(5.11)
G mz Torque scale factor for pure-slip
torque (5.20)
Gfz Torque scale factor for lateral force
(5.21)
G camz Torque scale factor for camber (5.22)
Misc.
ϒ Force expression (5.10)
Λ Normalisation of the normalized slip
(5.15)
Λz Normalisation of the normalized slip
(5.23)
Γ x,y Sliding force scale factor without
friction ellipse limitation (5.14) and
(5.16)
186
Wheel Speed Correction
θ Nominal angle between teeth
δθ Angular deviation
N Number teeth on tooth wheel
i Tooth number
j Revolution number
ϕa Angular deflection of axle
MB Braking torque
Ja Axle inertia
Da Damping in axle
Ka Axle stiffness
Friction Estimation
θ Parameters for estimation
V Cost function (7.9)
wi Weight of bin i in cost function
X Bin intervals, S-bin (slip) (7.6)
Y Bin intervals, F-bin (force) (7.6)
J Jacobian matrix (7.15)
H Hessian matrix (7.13)
λi Forgetting factor for bin i
187
Subscripts
Notation Valid on Description
a
F, M , G, µ Contribution from adhesive part of contact
patch
s
F, M , G Contribution from sliding part of contact
patch
s
µ Static friction
k
µ Kinetic friction
0
F, M Pure-slip entity
0
C Normalized entity, C0x = Cx / Fz
x
Vectors Longitudinal direction
y
Vectors Lateral direction
z
Vectors Vertical direction, positive downwards
cam
F, M , δ y Contribution from camber
v
F, M , δ y Vehicle entity
ij
v̄, λ , α , F̄, M̄ Wheel identifier, where i = f , r (front or
rear) or 1.. number of axles on the vehicle
and j = l, r (left or right).
m
... Measured quantity
Superscripts
Notation Valid on Description
ˆ F, M , C, ... Measured or estimated quantity
○
λ , α , σ̄ Slip where transition to full sliding occurs
∗
λ , α , σ̄ , F, M Point for friction maximum. Corresponds to
subscript crit and max in ISO 8855
¯ Vectors Vector ( x, y, z)
′
v, λ , α Value, measured at the rim under the as-
sumption of a flexible carcass
′
σ x , f x , x, y S-bin value (the slip signal is leading)
′′
σ x , f x , x, y F-bin value (the force signal is leading)
reg
σ 0x,y Region-based pure slips, Eq (5.18)
vel
σ 0x,y Velocity-invariant pure slips, Eq (5.12)
188
B
Scale Factors and Slip
Conversion
The derivation of the scale factors is only performed in the lateral direction, but
the procedure for the longitudinal scale factors is similar.
Adhesive slip scale factors
Fay,slip(σ x , σ y, γ )
Gay (σ x , σ y, γ ) =
F0y(σ 0ya )
− Cyσ y (1 − ψ (σ x , σ y , γ ))2
=
− Cyσ y (1 − ψ (0, σ y, 0))2 − Fz µ y sin(β )ψ 2 (0, σ y, 0) (3 − 2ψ (0, σ y, 0))
− Cyσ y (1 − ψ (σ x , σ y , γ ))2
= σ
− Cyσ y (1 − ψ (0, σ y, 0))2 − Fz µ yσ ○y σ ○yy ψ (0, σ y, 0) (3 − 2ψ (0, σ y, 0))
2
3 (1 − ψ (σ x , σ y , γ ))
= (B.1)
ϒ(0, σ y , 0)
when ψ < 1, otherwise Gay (σ x , σ y , γ ) = 0.
J
ϒ( x, y, z) = ψ 2 ( x, y, z) − 3ψ ( x, y, z) + 3 (B.2)
Camber scale factor
Fcamy (σ x , σ y, γ )
Gcamy (σ x , σ y, γ )) = = 2ψ 3 (σ x , σ y , γ ) − 3ψ 2 (σ x , σ y , γ ) + 1 (B.3)
F0cam (γ )
189
Appendix B. Scale Factors and Slip Conversion
Sliding scale factors
Fsy(σ x , σ y, γ )
Gsy (σ x , σ y, γ ) =
F0y(σ 0ys )
− Fz µ y sin(β ′ )ψ 2 (σ x , σ y, γ ) (3 − 2ψ (σ x , σ y , γ ))
= “ ”
vel vel vel vel
− Cyσ 0y (1 − ψ (0, σ 0y , 0))2 − Fz µ yψ 2 (0, σ 0y , 0) 3 − 2ψ (0, σ 0y , 0)
σ ○y ψ 2 (σ x , σ y, γ )(3 − 2ψ (σ x , σ y, γ ))
= sin(β ′ ) q vel
vel
σ x2 + σ 2y sgn(σ 0y ) ϒ(0, σ 0y , 0)
v
u v0 2 `
u !
(1 + σ x )2 + σ 2y − (σ x2 + σ 2y ) (B.4)
´
⋅ t
v
remark that
ψ (σ x , σ y, γ )
Λ= q
σ x2 + σ 2y
0 s„ 1
«2 „ «2 „ «2
γ ○2 @ γ cos ( β ) sin( β ) cos ( β )γ
sin(β )σ ○y ○ + + − A (B.5)
γ ○2 − γ 2 γ σ x○ σ ○y σ x○γ ○
using
σy σx
sin(β ) = q ; cos(β ) = q (B.6)
σ x2 + σ 2y σ x2 + σ 2y
Scale factors for the self aligning torque
Mz(σ x , σ y, γ ) = ( M0z(σ reg reg reg

y ) − t0a (0, σ y , 0) Fay,slip (0, σ y , 0)) sin(β )
′
+ ta (σ x , σ y, γ ) Fay,slip(σ x , σ y, γ ) + Mz,cam (σ x , σ y, γ )
= G z (σ x , σ y, γ ) M0z(σ reg reg
y ) + G f z (σ x , σ y , γ ) F0y (σ y ) + Gcamz F̂0cam (γ ) (B.7)
where
Gmz = sin(β ′ ) (B.8)
and
ta (σ x , σ y , γ ) Fay,slip(σ x , σ y, γ ) t0a (0, σ reg reg

y , 0) Fay,slip(0, σ y , 0)
Gfz = reg − reg sin(β ′ )
F0y(0, σ y , 0) F0y (0, σ y , 0)
t0a (σ x , σ y, γ )
= ( Fay,slip(σ x , σ y , γ ) − Fay,slip (0, σ reg
y , 0) sin(β ))
′
F0y(0, σ reg y , 0)
a (4ψ − 1)(1 − ψ )2 a (4ψ − 1)(1 − ψ )2 sin(β )

„ « „ «
σy
= − sin ( β ′
) = − sin ( β ′
)
3 ϒ σ ○yψ 3 ϒ σ ○y Λ
(B.9)
190
where ta is the pneumatic trail for the adhesive region defined as
Maz(σ x , σ y, γ )
ta (σ x , σ y, γ ) = (B.10)
Fay,slip (σ x , σ y, γ )
Finally
Mz,cam (σ x , σ y, γ )
Gcamz = (B.11)
Fcam (0, σ reg
y , 0)
simply derived from Equation (4.56) and (4.52).
Computation of velocity invariant slips

The relative velocity vs at the combined slip (σ x , σ y) and the wheel-travel velocity
v is given by q
q v σ x2 + σ 2y
vs = v2sx + v2sy = q (B.12)
(1 + σ x )2 + σ 2y
For the pure longitudinal slip at the wheel-travel velocity v0x = v0 and the slip
velocity v0sx = vs sgn(σ x ) the slip definition gives:
q
v0sx vs sgn (σ x ) σ x2 + σ 2y sgn(σ x )
vel
σ 0x = = = q
v0x − v0sx v0 − vs sgn(σ x ) v0 q
(1 + σ x )2 + σ 2y − σ x2 + σ 2y sgn (σ x )
v
(B.13)
q
and for the lateral case with v0sy = vs sgn(σ y), v0x = v20 − v20sy and v0sx = 0,
q
v0sy vs sgn(σ y) σ x2 + σ 2y sgn(σ y)
vel
σ 0y = = q = v (B.14)
v0x − v0sx v20x − v20sy u v0 2 `
u !
(1 + σ x )2 + σ 2y − (σ x2 + σ 2y )
t ´
v
Use of slip expressed in λ and α

In many occasions the slip expressed in λ and α is more convenient to use. The
pure-slip tire forces are also mostly expressed as functions of these slips. Here
follows guidelines for necessary conversions if λ and α are preferred. Calculate
ψ (σ x , σ y, γ ) by use of the following formulas
λ tan(α )
σx = ; σy = (B.15)
1−λ 1−λ
The limit slips are calculated as
σ x○
8
>
<1 +σ○
> if λ > 0; Braking
x
λ○ = (B.16)
σ x○
if λ < 0; Driving
>
>
1 − σ x○
:
191
Appendix B. Scale Factors and Slip Conversion
α ○ = atan(σ ○y ) (B.17)
For the adhesive region the deformation-invariant pure-slips are calculated as
λ 0a = λ (B.18)
!
tan(α )
α 0a = arctan (B.19)
1−λ
and for the sliding region the velocity-invariant pure-slips as
v
q
2
λ 0s = (λ cos (α )) + sin2 (α ) sgn (λ ) (B.20a)
v0
v
q
sin(α 0s ) = (λ cos (α ))2 + sin2 (α ) sgn (α ) (B.20b)
v0
The region-invariant slips as derived as
λ ○ψ (σ x , σ y, γ )
λ 0z = sgn(λ ) (B.21a)
1 − λ ○ + λ ○ψ (σ x , σ y, γ )
α 0z = arctan(tan(α ○ )ψ (σ x , σ y , γ ) sgn(α )) (B.21b)
The slip-velocity angle is derived as
sin(α )
tan(β ) = (B.22)
λ cos(α )
192
C
Extension of Optimization
Algorithm to Include
Horizontal Shift
To include the horizontal shift in the parameter estimations the brush

model definition has to slightly redefined
2
1 ( C0x (σ x + sh ))p(σ x + sh )p)

 − C0x (σ x + s h ) +
3

µ




3
Ψ(σ x , C0x , µ , sh ) = 1 ( C0x (σ x + sh )) (C.1)
− 2
27




 µ
−µ ⋅ sign(σ x ) otherwise

Using the least-squares method for low-slips the regressor vector can be
set as φ i = [1 xi ] estimating the parameters θ = [θ 0 C0x ] from where
the horizontal shift is derived as sh = θ 0 / C0x . In fact, it is the vertical shift
that is estimated, but in the linear region the relation between horizontal
and vertical shift is proportional. The same relation do not hold when
extending slip area to include detection of the curvature of the force-slip
relation. Using the regressor φ i = [1 xi xi p xi p]T with θ = [θ 0 C0x θ 2 ]
2
and sh = θ 0 / C0x and µ = C0x /(3θ 2 ) might provide erroneous result.
Using the Gauss-Newton algorithm, the Jacobian matrix has to be
extended with the derivative of the shift factor as
193
Appendix C. Extension of Optimization Algorithm . . .
e1 e2 eN T
 
 C0x ⋅⋅⋅
 C0x C0x 

 e1 e2 eN 
 
J (Ω , θ ) =  ⋅⋅⋅  (C.2)
 µ µ µ 
 
 e1 e2 e 
N
⋅⋅⋅
sh sh sh
with
2 2

e  C − 2 C0x pσ x + sh p + 1 ( C0x (σ x + sh ))

0x
= 3 µ 9 (µ )2 (C.3)
sh 
0 otherwise

194

Tire Modeling and Friction Estimation - Jacob Svendenius

Uploaded by

Copyright:

Available Formats

Tire Modeling and Friction Estimation - Jacob Svendenius

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Tire Modeling and Friction Estimation - Jacob Svendenius

Uploaded by

Copyright:

Available Formats

Tire Modeling and Friction

Department of Automatic Control

c 2007 by Jacob Svendenius. All rights reserved.

New system technologies are continuously improving the performance of

development of the semi-empirical tire model and of the friction estima-

5.4 Validation and Results of the Semi-Empirical Model . . 74

1.1 Background and Motivation

of the vehicle, is a great issue for improvement of today’s systems. The

Improvement of wheel speed measurement

Road friction estimation

1.2 Main Results

• A filter that stores representatives of the sampled data in a compact

1.4 Related Publications

Gäfvert, M. and J. Svendenius (2005): “A novel semi-empirical tire model

Svendenius, J. (2007b): “Validation of the brush model towards VTI-

2.2 Design and Construction

The circumferential velocity of the wheel is

where ω is the wheel angular velocity, and R e the effective rolling-radius

The tire slip is defined by normalizing the slip-velocity with a reference

v̄s v̄s v̄s

2.5 Tire Mechanics

Direction of Wheel Travel v̄

up to a certain slip value (σ ∗x ) and then diminishes slightly when σ x goes

Combined slip Combined slip signifies simultaneous longitudinal and

2.6 Tire Deformation

stretched string method gives, however, a good visualization of the physics

3.1 Steady-State Tire Models

F0y = a0 + a1 vx + a2 v2x + a3α + a4α 2 + a5α 3 + a6 R + a7 P (3.7)

where P is a tire-pattern constant and R is a tire-tread constant. The

F = (a1 (1 − e−a2 sRes ) − a3 sRes ) e−c4 sRes v (1 − c5 Fz2 ) (3.8)

A disadvantage in the approach is that the tire characteristic is assumed

where ȳ(⋅) is the region-separated function-approximation. The adjusted

where ( x, y) is (λ , F0x ), (α , F0y), or (α , M0z). The four coefficients have

3.2 Combined-Slip Semi-Empirical Tire-Models

where F0x∗ is the maximum achievable longitudinal force, and F (α ) de-

An objection to this model is the assumption (3.16), which is not true,

A drawback with this model is that longitudinal and lateral characteristics

Some early efforts to model tire forces under combined-slip conditions

is an entity that is less than one for non-sliding conditions. It is based on

Fx = − cos(β ○ ) F0x (σ ∗xσ N ) and Fy = −ǫd (σ N ) sin(β ○ ) F0y(σ ∗yσ N )

Fx = cos ((1 − ϑ )β ○ + ϑ β ) F0x

The variables ϑ , F0x′

Fx = F cos(β ) and Fy = F sin(β ) (3.28b)

3.3 Transient Tire-Models

3.4 Finite Element Models

Much research is devoted to tire-modeling related issues and the brief

4.1 Basics of the Brush Model

Road Slide Adhesion

region into small brush elements. Each element stretches laterally ( y-

Adhesive bristle forces Regard the specific infinitesimal bristle which

Insertion of (4.1) and the expression for tc ( x) yields

As a result, the magnitude of the available static friction force is depen-

The solution for the break-away point xs is then

Since xs is a point in the contact patch it must belong to the interval

Equation (4.10) may now be rewritten as

xs (σ x , σ y) = (2ψ (σ x , σ y) − 1)a (4.13)

It is clear that partial sliding occurs when ψ (σ x , σ y) < 1. At full sliding

Adhesion force Knowing the size of the adhesive region, xs (σ x , σ y),