Downloaded from SAE International by Imperial College London, Saturday, September 08, 2018

SAE TECHNICAL
PAPER SERIES 950311

Tire Modeling for Low-Speed and

High-Speed Calculations
J. E. Bernard and C. L. Clover
Iowa State Univ.

Reprinted from: New Developments in Vehicle

Dynamics, Simulation, and Suspension Systems
(SP-1074)

International CongressDetroit,
and Exposition
Michigan
February 27 - March 2, 1995
400 Commonwealth Drive, Warrendale, PA 15096-0001 U.S.A. Tel: (412)776-4841 Fax:(412)776-5760
 Downloaded from SAE International by Imperial College London, Saturday, September 08, 2018

The appearance of the ISSN code at the bottom of this page indicates SAE's consent
that copies of the paper may be made for personal or internal use of specific clients.
This consent is given on the condition, however, thatthe copier pay a $5.00 per article
copy fee through the Copyright Clearance Center, Inc. Operations Center, 222
Rosewood Drive, Danvers, MA 01923 for copying beyond that permitted by Sections
107 or 108 of the U.S. Copyright Law. This consent does not extend to other kinds
of copying such as copying for general distribution, for advertising or promotional
purposes, for creating new collective works, or for resale.
SAE routinely stocks printed papers for a period of three years following date of
publication. Direct your orders to SAE Customer Sales and Satisfaction
Department.
Quantity reprint rates can be obtained from the Customer Sales and Satisfaction
Department.

To request permission to reprint a technical paper or permission to use copyrighted

SAE publications in other works, contact the SAE Publications Group.

No part of this publication may by reproduced in any form, in an electronic retrieval

system or otherwise, without the prior written permission of the publisher.
ISSN 0148-7191
Copyright 1995 Society of Automotive Engineers, Inc.
Positions and opinions advanced in this paper are those of the author(s) and not
necessarily those of SAE. The author is solely responsible for the content of the
paper. A process is available by which discussions will be printed with the paper if
it is published in SAE transactions. For permission to publish this paper in full or in
part, contact the SAE Publications Group.
Persons wishing to submit papers to be considered for presentation or publication
through SAE should send the manuscript or a 300 word abstract of a proposed
manuscript to: Secretary, Engineering Activity Board, SAE.
Printed in USA 90-1203D/PG
 Downloaded from SAE International by Imperial College London, Saturday, September 08, 2018

950311

Tire Modeling for Low-Speed and

High-Speed Calculations
J. E. Bernard and C. L. Clover
Iowa State Univ.

ABSTRACT There has been little interest in the low speed behavior of
tires on smooth roads by the vehicle dynamics community.
Vehicle dynamics simulations typically use semi-empirical This is understandable - free rolling low speed behavior is not
tire models. The input to these models are normal load, very interesting to the vehicle dynamicist because lateral forces
sideslip angle and longitudinal slip, and the output are shear approach zero at very low speeds, and the trajectory of the
forces, aligning moment, and overturning moment. vehicle becomes a kinematics problem. From a braking point
Since the longitudinal speed is in the denominator of both of view, at very low speeds the vehicle is almost stopped, and
sideslip angle and longitudinal slip, the calculation of sideslip hardly any more distance is to be travelled.
angle and longitudinal slip at very low longitudinal velocities Furthermore, traditional simulations face numerical
leads to numerical problems. This has not been a particular problems at low speeds because the longitudinal speed of the
stumbling point in the past because vehicle dynamics wheel hub appears in the denominator of the expressions for
calculations were largely concerned with high speed analysis. both longitudinal and lateral slip. Thus zero speed becomes an
In situations wherein the vehicle was braked to a stop, impossible calculation, and some sort of ad hoc technique is
patchwork techniques sufficed for calculations at low speeds. invariably used for braking maneuvers which bring simulated
Now, however, with the advent of serious attention to vehicles to a stop [2]. In our experience, we found valuable
driving simulators, low speed tire modelling has become more ad hoc measures of the form:
important. This paper presents a new formulation which
differs from previous work in that both longitudinal slip and
slip angle become state variables rather than kinematic
functions of wheel spin rate and wheel velocity. The result is
a model which yields traditional solutions at high speed, and
which comes to a stop in a reasonable way without numerical
problems. The calculated longitudinal slip and slip angle can where u is the longitudinal speed of the wheel hub.
be paired with any semi-empirical tire model.
Several challenging examples, including braking-induced This brings the vehicle to a stop in a reasonable way, and
spinout ending in a stop on a severely inclined surface, errors resulting from the ad hoc technique have no significant
illustrate the power of this new technique. effect on calculations of trajectory. Of course, we had no
reason to simulate a vehicle which would stop and then start
INTRODUCTION again. Our interest focussed on the higher speed behavior of
the vehicle.
There is a body of published literature which deals with The recent interest in driving simulators has led to interest
transients in tire force buildup. Particular interest has been in in low speed simulation. In particular, since the purpose of
lags in the buildup of lateral force. The motivation for driving simulators is attaining better understanding of the
studying these lags and including them in vehicle simulations performance of the driver/vehicle system using real drivers and
has been either to gain a better understanding of tires or to simulated vehicles, the system needs to feel right to the driver.
better simulate the behavior of the tire/vehicle system. In It needs to be able to stop and start again, and to drive in
every case that we are aware of, this literature deals with reverse. It needs to be able to be stopped in any orientation on
analysis at high speeds. Reference 1 presents a good summary a hill or super-elevated road and started again. Such
of the history of these matters, leading to a call for further challenges are obviously well beyond the reach of low speed
study of "the shear forces and moments generated under time solutions of the genre of equation (1).
varying lateral and longitudinal slip".
85
 Downloaded from SAE International by Imperial College London, Saturday, September 08, 2018

This paper will briefly review the literature concerning The spirit of Equation (2) has been implemented in other
computer simulation of transients in lateral force buildup. It ways. For example, earlier versions of the well known
will then extrapolate ideas underpinning these models to a VDANL simulation [4] use an equation similar to
more robust model of the tire road interface which has utility
for both high and low speed. This leads to differential
equations for slip angle and longitudinal slip which can be
used in conjunction with traditional models of the shear forces
at the tire/road interface, yielding shear forces useful for
simulation in any smooth road situation.
LITERATURE REVIEW

An early vehicle simulation which modelled the effect of Equations (2) and (3) yield transient results which are
lateral force buildup on vehicle directional response was given identical in the linear range and very similar in the nonlinear
by Owen and Bernard [3]. They were attempting to simulate range. Steady state results are identical.
a tractor loader backhoe with very soft and very large rear tires Recent work by Heydinger et al. [5] calls for a second
which led to very important delays in the buildup of lateral order relationship between kinematics and forces. Allen et al.
force. They suggested and used a straightforward first order [6] use a model based on Heydinger' s work which is of the
lag: form

This was shown in [5] to lead to an improved match

between frequency response tests and simulation of frequency
response tests. ([5] and [6] differ slightly in that [5] applies the
2nd order lag to &#x0 3B1; rather than to Fy. Thus [5] lags aligning
moments, while [6] does not.)
An important contribution to this discussion was given by
Shapley [7]. He suggested a formulation for the sideslip angle
which derived from the rate of change of lateral displacement
of a representative point in the contact patch. Shapley's
Thus a step change in &#x0 3B1; , caused, for example, by a step formulation can be stated as:
steer input, would lead to lateral force as shown in Figure 1.
It is routine in such a semi-empirical model to choose b to
match the lag measured in tire tests. The parameter b is often
called the relaxation length. Setting b to some percentage of
the circumference of the tire was thought to lead to reasonable
results.
Equation (2) breaks down at low speeds because u appears This equation is remarkable in the following sense for
in the denominator of &#x0 3B1;. small angles, it has the appropriate steady state solution

and the solution to equation (5) is well behaved at u = 0.

The next section of this paper will present a derivation of
a differential equation for slip angle which is closely related to
Shapley's work. A similar derivation from a longitudinal point
of view will lead to a differential equation for the longitudinal
slip. Both these equations are well behaved as the longitudinal
velocity becomes small. The solution to these equations yields
slip angle and longitudinal slip which may be used in semi
empirical tire models.

DERIVATION OF THE EQUATIONS FOR SLIP ANGLE

AND LONGITUDINAL SLIP

Figure 2a presents a plan view of the tire's contact patch.

We assume a hypothetical element of the tire which is
currently a distance b back from the front of the contact patch
sticks to the road, and that a line from the front of the contact

86
 Downloaded from SAE International by Imperial College London, Saturday, September 08, 2018

patch through the element defines the slip angle &#x0 3B1;. The
velocity vector V of the hub of the wheel is

now however, while we retain

Where i and j are unit vectors attached to the hub.
The tangent of the slip angle is given by
we have the new relationship

Differentiating with respect to time yields Using equations (14) in equation (13) yields

Neglecting motion of the contact patch due to steering about Note that, although our steady state solution is now
an offset kingpin, it is straightforward to note that u is negative, tanα is positive, and positive v
again yields lateral forces in the negative j direction.
Now since equation (11) is valid only for u = 0 and equation
and (15) is valid for u = 0, we can combine the two to get

Equation (9) yields

Implementation of this model yields the new state variable,
tanα, as a function of the velocity of the wheel hub. Taking
the inverse tangent yields a which can be used with traditional
For small angles, equation (11) matches equation (5). It semi-empirical tire models to compute the shear forces at the
also shares desirable features of equation (5), namely, the tire road interface.
appropriate steady state solution for constant u and v, and good Now consider a similarly-reasoned derivation for
behavior for small u. Note that positive v yields lateral forces longitudinal slip. Figure 3a again presents the contact patch.
in the negative j direction. The classic definition of longitudinal slip (see, for example,
Another step is needed to use this model in driving [8]) is
simulators, namely, we need to deal with negative u, which
comes up in a variety of ways including spin out and driving
in reverse.
Figure 2b presents the kinematics for negative u. Clearly, where B locates a hypothetical element which follows the road,
we retain and A locates a hypothetical element which is undeformed due
to longitudinal shear forces. Differentiation with respect to
time yields
and

Again, ignoring the kinematics associated with steering about

87
 Downloaded from SAE International by Imperial College London, Saturday, September 08, 2018

an offset kingpin yields

and Now since equation (20) is valid only for u = 0, and

equation (25) is valid only for u = 0, we can combine the two
to get
where R is the rolling radius and ? is the spin velocity of the
hub. Using equations (19) in equation (18) and rearranging
yields
Note that steady state for slip from equations (25) and (26)
is equal to the classic definition of slip given in equation (21).
However, in the u = 0 case, positive slip yields forces in the
Equation (20) has features in common with equation (11). positive i direction, the opposite direction from the solution for
First of all, for constant u and ? , the steady state solution u = 0. This requires that we multiply longitudinal forces by
yields the classic definition of slip, in this case, the sign of u and that when u switches signs during a spin out
maneuver or a stopping maneuver, we change the sign on slip
for implementation in the ongoing differential equation.
Figure 4 presents a schematic diagram of the slip algorithm.
Note that positive slip yields longitudinal forces in the negative Implementation of this model requires the use of the spin
i direction. And like equation (11), equation (20) is well velocity, ? , which we calculate based on a moment
behaved for small u. Furthermore, as was the situation with summation about the wheel spin axis. Then equation (26) can
the formulation for slip angle, we need to reformulate for be solved for the additional state variable, s. The longitudinal
negative u. Consider Figure 3b, which presents the kinematics slip, s, can be used with traditional tire models to compute
for negative u. Again we have shear forces at the tire/road interface.

Equations (16) and (26) share the following features:

and differentiation again yields 1) The appropriate steady state solution.

2) First order lag characterized by distance dimensions b and

B. In particular, given a step change from zero to any
But in this case, kinematically defined &#x0 3B1; at constant v and u, the slip
angle, &#x0 3B1;, will rise to about 63% of &#x0 3B1; in b/u seconds.
Similarly, given a step change from zero to some
and kinematically defined longitudinal slip, s at constant u,
the value of longitudinal slip, s, will rise to about 63 % of s
in B/u seconds.
Using equations (24) in equation (23) and rearranging yields
3) Good behavior for small u, including u = 0.

88
 Downloaded from SAE International by Imperial College London, Saturday, September 08, 2018

The next section presents numerical results computed

based on a simulation of a tire/vehicle system coupled with
equations (16) and (26).

NUMERICAL RESULTS

The results presented in this section were calculated using

a vehicle model described in Reference 9. In brief, the model
includes lateral and longitudinal motion and pitch and roll of
the sprung mass, and a wheel spin degree of freedom for each
wheel. The tire model is presented in Reference 10. It
includes wheel spin and a well founded relationship between
sideslip angle and longitudinal slip for calculations of shear
forces in maneuvers involving braking and turning. Table 1
gives details of the passenger car model used for the
simulation.

with these relationships. Rather, it is in calculating the sideslip

angles, &#x0 3B1;, and the longitudinal slip, s, to be used in the tire
model.
For all the simulations here we use b = 0.91 m. We
believe this is a fairly typical value for a passenger car tire.
Figure 1 was computed using b = 0.91 m, C&#x0 3B1; (cornering
stiffness) = 51,000 newtons/rad, and u = 15.24 m/sec.
As far as we know, there is no experimental data available
to guide the choice of the longitudinal relaxation length, B.
Since our intuitive sense is that the lag times for longitudinal
force are much shorter than the lag times for lateral force,
throughout this section we use B = 0.091 m. We have
established to our satisfaction that the calculated results
presented here do not change with small changes in B. It is
our view that strict attention to the details of the B value will
become necessary only when lags in the buildup of
longitudinal force become important, perhaps for the simulation
of antilock brakes which operate at high frequency.

Figure 5 presents a carpet plot giving lateral force as a Example 1

function of tire sideslip angle as computed by the tire model
used in this section. Figure 6 presents several mu-slip curves Figures 7a through 7c present calculated results for
which indicate the relationship between longitudinal slip and Example 1, which simulates a straight line stop from about 25
brake force. The contribution of this paper is not in dealing m/sec at about .7 g. The maneuver takes place on a 5%
89
 Downloaded from SAE International by Imperial College London, Saturday, September 08, 2018

downslope. Note that although the numerical aspects of the

analysis are well behaved, we are left with an oscillation about
u = 0. This is an expected outcome of the model - the
relationship between vehicle velocity and longitudinal slip
allows the storage of potential energy. The frequency of the
oscillations, which is about 3.5 hz, can be shown to be
appropriate based on the values for the mass of the vehicle, the
longitudinal stiffness of the tires, and the relaxation parameter
B.
Although we believe the oscillations are appropriate for
this model, they are clearly not appropriate for vehicle
simulation. In particular, although we expect the hubs to
overshoot the contact patch at the stopping point, we do not
expect lightly damped oscillations about the stationary contact
point. Clearly, some important damping mechanism of the
vehicle/tire system has been left out.
Rather than trying to develop a model of the necessary
damping based on a detailed examination the details of the tire
mechanics (and perhaps the suspension mechanics), we have
elected to insert damping into the system in a convenient
manner. The damping is included only for very low speeds,
and it is initiated only by the first zero crossing of the
longitudinal velocity of the hub. We set the sum of the
damping at all four wheels equal to 0.7 critical for the entire
tire vehicle system, apportioning the damping to each wheel as
a function of the dynamic normal load. Appendix A gives
details of the damping model. The damping is used in all the
remaining simulations presented here.
Example 2

Figures 8a through 8c present the straight line braking

results from the damped model on the 5% downslope. Note
that the vehicle stops smoothly. Also note that there are
residual brake/traction forces after the vehicle stops. The sum
of these forces is 1001 newtons, 5% of the weight of the
vehicle, which is the amount needed to hold the vehicle
stopped on the hill.
It is worth noting that traditional kinematic relationships
between longitudinal slip, lateral slip, and the shear forces at
the tire road interface will not, in general, yield equilibrium
when the vehicle stops. Furthermore, the solution for the shear
forces acting on a vehicle at rest is statically indeterminate.
The new formulation provided in this paper yields a solution
for one possible static equilibrium condition for a vehicle at
rest. For example, Figure 8c shows that the at-rest longitudinal
forces, which sum to 5% of the weight of the vehicle, have a
brake force in front and a slightly smaller tractive force in the
rear.

90
Downloaded from SAE International by Imperial College London, Saturday, September 08, 2018

Example 3

Figures 9a through 9d present a braking in a turn

maneuver. Starting from an approximately .4g turn, the
vehicle is braked to a stop. The road surface is flat. Note that
the vehicle stops smoothly, and the residual shear forces and
the moments resulting from those forces sum to zero.

91
 Downloaded from SAE International by Imperial College London, Saturday, September 08, 2018

Example 4

Finally, Figures 10a through 10c illustrate a very

complicated maneuver, severe braking in a turn on a 10%
sideslope leading to a spin out before stopping. These figures
illustrate the power of the techniques presented here to handle
formerly intractable scenarios.

92
 Downloaded from SAE International by Imperial College London, Saturday, September 08, 2018

CONCLUSIONS 9.Bernard, J.E., Bhatnagar, A., Clover, C.L., "Evaluation of

Select Vehicle Dynamics Models (Phase II)", MVMA
The recent interest in driving simulators has led to an Contract No. IOWA 9114-C11302, June 1992.
interest in low speed simulation. This presents a numerical
challenge to traditional formulations for longitudinal slip and 10.Szostak, H.T., Allen, R.W., and Rosenthal, T.J.,
slip angle. This paper derives differential equations for "Analytical Modeling of Driver Response in Crash
longitudinal slip and slip angle which are numerically well Avoidance Maneuvering, Volume II: An Interactive Tire
behaved at low speed and at zero speed. These can be used Model for Driver/Vehicle Simulation", DOT-HS-807-271,
with traditional formulations relating longitudinal slip and slip Systems Technology, Inc., April 1988.
angle to the shear forces at the tire/road interface. Examples
illustrate the utility of the model. APPENDIX A: Damping near zero velocity
ACKNOWLEDGEMENTS Equations (16) and (26) can yield oscillatory vehicle
motions which are very lightly damped near zero velocity. We
This work was supported in part by the National Highway expect these oscillations to damp out due to mechanisms not
Traffic Safety Administration. Discussions with Chris Shapley included in the model. Thus, we have added auxiliary
were instrumental in helping us develop the models for lagged damping which is only used at very low speeds.
sideslip angle and longitudinal slip. Auxiliary damping is triggered if u changes sign while v
is small, i.e., if
REFERENCES

1.Segel, L., "An overview of developments in road-vehicle and

dynamics: past, present, and future", Proceedings of The
Institution of Mechanical Engineers, Automobile Division,
1993. We have found 0.15 m/s to be a reasonable value for &#x0 3B5;.
Under these conditions, damping should be added in both the
2.Bernard, J.E., Winkler, C.B., and Fancher, P.S., "A longitudinal and lateral directions.
Computer Based Mathematical Method for Predicting the We apportion the damping to each wheel according to the
Directional Response of Trucks and Tractor-Trailers normal load on that wheel. We do this by treating each wheel
(Phase II Technical Report)", UM-HSRI-PF-73-1, June as a single degree of freedom system of mass m¡, where
1973. m¡g = Fzi, the normal load on the ith wheel.
Damping for each wheel is given in terms of percent
3.Owen, R.H. and Bernard, J.E., "Directional Dynamics of critical damping. Thus, we use our single degree of freedom
a Tractor-Loader-Backhoe", Vehicle System Dynamics, model to calculate a natural frequency. From the natural
Vol. 11, December 1982. frequency and the mass, we calculate the appropriate damping.
4.Allen, R.W., Rosenthal, T.J., and Szostak, H.T., A) Lateral Natural Frequency at Zero Velocity
"Analytical Modeling of Driver Response in Crash
Avoidance Maneuvering, Volume I: Technical Using equation (16) and assuming tan(a)=oc near zero
Background", DOT-HS-807-270, Systems Technology, velocity yields
Inc., April 1988.

5.Heydinger, GJ., "Improved Simulation and Validation of

Road Vehicle Handling Dynamics", Ph.D. Dissertation, Differentiating this equation gives
Ohio State University, Columbus, Ohio, 1990.
6.Allen, R.W., Szostak, H.T., Rosenthal, T.J., et al.,
"Vehicle Dynamic Stability and Rollover", Systems Newton's law yields
Technology, Inc, TR-1268-1, NHTSA Contract DTNH22
88-C-07384.

7.Shapley, C.G., "The Influence of Tire Modeling in where C&#x0 3B1; is cornering stiffness, so that
Commercial Vehicle Simulation", Proceedings of a
Symposium on Commercial Braking and Handling, UM
HSRI-PF-75-6, May 5-7, 1975.
Assuming
8.Bernard, J.E., Segel, L., Wild, R.E., "Tire Shear Force
Generation During Combined Steering and Braking
Maneuvers", SAE Transactions, SAE Paper No. 770852,
1977. and that equation (A-6) is of the form
93
 Downloaded from SAE International by Imperial College London, Saturday, September 08, 2018

where ? is the damping coefficient. Under the conditions of

equations (A-1) and (A-2), Fyi(DAMPING) and Fxi(DAMPING)
implies that the lateral natural frequency near zero velocity is are added to the lateral and longitudinal forces computed for
approximated by each tire.

B) Longitudinal Natural Frequency at Zero Velocity

Using equation (26) and assuming that the wheel spin

velocity is zero when the vehicle is stopped yields

Again, by differentiating we obtain

Also

where Csi is longitudinal stiffness.

Thus

Assuming slip is small,

and that equation (A-13) is of the form

implies that the longitudinal natural frequency near zero

velocity is approximated by

C) Auxiliary Damping Terms

Now assume

where

Inserting the natural frequencies from equations A-9 and A-16

yields

and

94