P. Bandel ~ and C.

Di Bernardo ~

A Test for Measuring Transient

Characteristics of Tires 2

REFERENCE: Bandel, P. and Di Bernardo, C., "A Test for Measuring Transient
Characteristics of Tires," Tire Science and Technology, TSTCA, Vol. 17, No. 2, April-
June, 1989, pp. 126-137.

ABSTRACT: A new method is described for measuring tire cornering stiffness and
relaxation length under transient conditions. The hub of the test tire, which is riding atop
a rotating drum, is attached to a horizontal beam that is hinged at its other end to permit
free lateral motion of the tire. When the tire is given a lateral pulse it oscillates at its
natural frequency about the position of zero steer angle. Both cornering stiffness and
relaxation length of the tire can be calculated from the frequency and logarithmic decay
of the resulting oscillations. Results of this "pendulum" test are compared with those
from steady-state tests and from the sinusoidal steer angle test, which has been the
conventional one for measuring transient tire characteristics. Advantages of the pendu-
lum test are pointed out.

KEY WORDS: transient tire characteristics, tire modeling, sinusoidal steer angle, "pen-
dulum" test, relaxation length, cornering stiffness.

C o m p u t e r simulation o f vehicle handling on flat surfaces requires knowl-

edge o f the forces a n d m o m e n t s transmitted b y the g r o u n d t h r o u g h the fires
to the vehicle body. T h e g r o u n d - t i r e - b o d y system is often split into parts a n d
only the b o d y is simulated, the ground-tire subsystem being replaced by its
forces a n d m o m e n t s acting on the wheel hub. Such inputs d e p e n d o n slip
angle, c a m b e r angle, load, a n d t o r q u e applied to the tire. T h e y also d e p e n d
on time if, as usual, the effects o f transitory conditions are examined.
W e l l - k n o w n testing m a c h i n e s a n d m e t h o d s are available for m e a s u r i n g
such tire characteristics as the relation o f lateral force to the operational
variables m e n t i o n e d . However, they are n o t always sufficiently accurate for
detecting small differences between fires, especially w h e n time d e p e n d e n c e
o f the forces is a p r i m a r y concern. A better m e t h o d for evaluating the
dependence o f lateral force o n slip angle a n d time is described here for the

Pirelli Coordinamento Pneumatici, Milan, Italy.

2 Presented at the seventh annual meeting of The Tire Society at The University of Akron,
Akron, OH, March 22-23, 1988.
126
 BANDEL ON TRANSIENT TIRE CHARACTERISTICS 127

range of tire linearity and for oscillation frequencies typical of a car's re-
sponse.
At moderate lateral acceleration the pertinent tire characteristics are the
cornering stiffness (rate of change of tire lateral force per unit change of slip
angle) and relaxation length ( approximately the tire travel distance required
to move from one equilibrium lateral force to another when a step change is
made in the slip angle). These should be measured at the same range of
oscillation frequency and tire speed as those to be used in the mathematical
model. Except for problems related to motorcycle instability, the frequency
is usually not high enough to introduce any significant problem in the
testing equipment; speed might be a limiting factor.
In the most widely used test for measuring a tire's transient characteris-
tics, a slip angle is applied sinusoidally while the gain and phase lag between
slip angle and lateral force are being measured. The cornering power and
relaxation length can be calculated from these. Unfortunately, the phase
difference at low frequencies and high speeds is quite small, which impairs
the accuracy of results. This problem is overcome by the "pendulum"
method described here, which has the following practical advantages:
1. The moment of inertia and length of the beam can be adjusted to
operate in the required frequency range and to give the most suitable
damping index for accurate measurements at speeds as high as desired.
2. Changes in tire characteristics due to heating or wear are avoided since
this is a transient test of very short duration.
Some results are shown and discussed.

Tire Modeling

As described later, the model uses the natural frequency and damping of
the simple mechanical system to which the tire is fitted. The model configu-
ration need not be an accurate replica of the "physical" tire if it can fit the
experimental data and can be used in the vehicle simulation.
As a first stage, the following well-known implicit relationship among
force, slip angle, and time was considered:
dF V
-- - (Ca -F) (1)
dt a
where F = lateral force
V= longitudinal speed
= relaxation length
C= cornering power
= slip angle
The validity of Eq 1 is not discussed here; it may be presumed that the
relaxation length a depends slightly on both the speed and the rate of change
128 TIRE SCIENCE & TECHNOLOGY

of slip angle a within reasonable limits. Equation 1 can be seen as a simple

and reliable model for the tire. Moreover, for small values of a, C can be
assumed to be constant, thus giving a linear model.
Equation 1 describes a tire having a lag of force response to the applied
slip angle; it is well known that this equation is consistent with the model in
Fig. 1, which is a tire without response lag, where
F=C.a

and a spring having a rate K exists between the tire and the hub
K= C/a

If a dashpot having a viscosity of cl is added in parallel to the spring as

shown in Fig. 2, it is possible to take into account the hysteretic losses in the
tire material while the tire is deflected by the force F. A more general form
of Eq 1 can then be worked out:

d~(l+f~-~)= V ( Ca F ) + cI V d a
a dt (2)

Equation 2 was eventually used as the model.

Sinusoidal Steer Angle Test

An opportunity for a different method of testing arises from an examina-
tion of the one most used, namely the sinusoidal steer angle test. In this test
a slip angle (steer angle) with angular frequency o~is applied sinusoidally to
a tire that is fitted to a fixed rig and is rotating atop a drum. The amplitude
and phase lag of the resulting lateral force are measured. The experimental
values can be quite well fitted by Eq 1 in restricted ranges of frequency and
speed. The resulting solution is given by the vectorial diagram of Fig. 3.
The phase lag ~b = arc tan ( w a / V ) can be measured without difficulty for
the usual values of a encountered in car tires (0.5-1 m) and for high values

rtUB
~

FIG. 1 - - Tire transient model without hysteretic losses.

 BANDEL ON TRANSIENT TIRE CHARACTERISTICS 129

C 1

~E [ ~ ~ ~ ~ - ~ HUB

dF ciV V d~
(i+ - - ) = (C~- F) + clV
dt C ~ dt
FIG. 2 - - Tire transient model with hysteretic losses.

of w and low values of V by using available Fast Fourier Transform ana-

lyzers.
Unfortunately, in car handling problems, ~0 is usually less than 10 r a d / s
while V c a n be higher than 40 m / s . These give ~ values o f 0.12-0.25 rad
= 6-12 degrees. Since m a n y analyzers cannot give an accuracy better than
+ 1o, the total accuracy of the test is + 10%-+20%, which is quite poor for

~coF

C=

600"
= arctg ( )
V
FIG. 3 - - Vectorial diagram for sinusoidal response.
130 TIRE SCIENCE & TECHNOLOGY

many purposes. A test is required in which the reponse lags at high speeds
are magnified.

The Pendulum Test

A trailer-like arrangement was adopted in which a tire rotating on a drum
having a peripheral speed Vis fitted to a beam that is free to oscillate in both
vertical and lateral directions, as shown in Fig. 4.
The equation of movement of the beam is:
F = - c 2 e ~ - (J/e)D" (3)
where F is the force transmitted to the hub by the tire, J is the moment of
inertia about the swivel axis, and c2 takes into account the friction of the
hinge. Its derivative is:
P = -c2gD"- (J/e)'fl (4)
The slip angle and its derivative are:
= ~ + ~(elV) (5)
A =/~ + ~'(e/V) (6)
Substituting Eqs 3-6 into Eq 2, the expression in Eq 7 is obtained.

g l+---~--)/J+ 1+-~(c,+c21+ ~,--ff j~

ce ( .
+---~- 1 + - - - ~ + - - - ~ /#+C/3:0 (7)

A substantial simplification of Eq 7 can be obtained by assuming some

typical values for its parameters:

g
L

2ZA
F I G . 4 - - P e n d u l u m test s c h e m e .
 BANDEL ON TRANSIENT TIRE CHARACTERISTICS 131

J = 500 kg. m
g=lm
a=0.5m
V= 20 m / s
Cl = 600 N . s / m
C = 35 k N / r a d
c2 = 30 N . s / m

The value for c2 was found by swinging the pendulum with an air bearing
in place o f the tire; the lateral force of the fire was replaced by the restoring
force of a steel spring. The value for Ca was calculated from measurements of
the damping coefficient of the oscillating system. F r o m these determina-
tions it was found that:

cl V / C = 0.35 cannot be neglected relative to unity.

o-g2(C q- c 2 ) / V J = 0.3 can be neglected, especially at higher speeds.
ffClC2~2/Cj = 5 o 10 -4 can be neglected.

12 oJn (rad/sec)

10

2;
i Speed (Km/h )
0t" I I I
h
v

0 50 i00 150
FIG. 5 - - Effect o f s p e e d on vibration f r e q u e n c y o~n; - - correct values, - - - approximate
values.
132 TIRE SCIENCE & TECHNOLOGY

c2V/C = 0.017 can be neglected for speeds up to say 50 m / s (180

k m / h ) ; at the highest speed its value is 0.04.
aclV/gC = 0.17 cannot be neglected, especially at higher speeds.

Therefore, for speeds of 20-50 m/s, which are quite reasonable for han-
dling problems, Eq 7 may be simplified to:

g 7 fl + ---~- 1 + - - - ~ } ~ + C/3 = 0 (8)

The physical significance of each of the terms in Eq 8 is given below:

C(g) / g/3 elastic force

c e 1+ aclV)( e . ~ ) damping force

7"T -TU

.2

.1

Speed (Km/h)
0 I 9
0 50 i00 150
FIG. 6 -- Effect of speed on damping i n d e x ~; - - correct values, - - - approximate values.
 BANDEL ON TRANSIENT TIRE CHARACTERISTICS 133

j(g~') / # 2 mass force

J cr 1 + (e. ~) amplifying force

e~" v --C-
where the forces are supposed to act at the hub, the displacement of which
is e~.
The natural frequency of the system is
/--~. :
~o--o~o Vl _~2=~
/c/e ~/_~_.~_ ~2 (9)
V J
and for small values of ~"
(.On 2 ' ~ Cg/J (lO)
while the damping coefficient is

D = C ( 1 +9- c- r-q~V}\ - w . 2 J cr {
~-2"-~ ~1 + - ~ )

J
AAAAA
VVVVVVV- i
A
t

n
J
2.
t

1 A .-I
[ log
2H (n-l) A~

FIG. 7 - - Free oscillation o f the system.

134 TIRE SCIENCE & TECHNOLOGY

and, i f w , 2 - 60o 2 ~-- Cg/J

D= T

from which the damping index f can be calculated:

D.g 2 1
f - 2w0~ - 2 V ~176 - ~) (12)

The validity of Eq 10, hence that o f E q s 11 and 12, depend on ~', which is
generally smaller than unity. For V = 20 m / s , and the values of the other
parameters being those already reported, the correct result is

w. 2 = 69.2 rad 2 and ~" = 0.1062.

while the simplified equation gives

W n 2 ~--- 0902 = 70 tad 2

~" = 0.1045

~.(rad/sec)
i0
%

. %

Speed (Km/h )
h
0 ! - | | v

0 50 100 150
FIG. 8 - - Experimental values of natural system frequency o~ as a function of speed.
 BANDEL ON TRANSIENT TIRE CHARACTERISTICS 135

The error is quite acceptable, and it becomes even smaller at higher

speeds. A more comprehensive comparison of the correct and simplified
values of ~0n and ~ are reported in Figs. 5 and 6.
For V = 0 Eq 7 becomes

g2
(c, + c2) + caa = 0

from which, for small values of ~',

2 Cg2
O,)n ,~'
crJ

D = c1+ c2

De2 (r + r e2"
2WoJ 2r
Only in this situation can the tire parameter c2 be measured, but nothing
can be known about its dependence on speed. Its previously reported order

.4

%
%
.3
I
I

~ ~

o1~

Speed (Km/h)
0 | I |
I
r

0 50 i00 150
~G. 9 - - Experimental values ~ ~mping index ~ as a ~nction ~ e e t
136 TIRE SCIENCE & TECHNOLOGY

of magnitude should not change, however, thus justifying the simplification

of Eq 7.
From the free oscillation of the system, initially displaced from its equilib-
rium position, it is therefore possible to measure the values ofo~nand ~"(Fig.
7), and to calculate the tire parameters C and a through Eqs 10 and 12.
Because of the decreasing values of ~" as the speed increases, the beam
motion lasts for many cycles, thus allowing accurate measurements to be
made of both frequency and damping, hence of C and a. This achieves the
principal aim of the test.

Results
Values of o~nand ~"measured for one tire at different loads and at speeds
ranging from 0 to 160 k m / h are shown in Figs. 8 and 9. Accurate measure-
ments are not possible from 10 to about 60 k m / h because the damping
coefficient of the system is too high; this agrees with the trend of Eq 7. Above
80 k m / h ~" decreases with increase in speed, while ~0, is nearly constant.
From Eqs 10 and 12, speed should not have much effect on cr and C.

(m)
.8

.6

.4

.2
Speed (K m/h)

0 5O i00 150
FIG. 10 -- C a ~ u ~ t ~ values ~ rel~ation length ~ as ~nctions ~ ~ e e d and ~ a t
 BANDEL ON TRANSIENT TIRE CHARACTERISTICS 137

TABLE 1 - - C o r n e r i n g power, kN/rad.

Steady state test Pendulum test

30 k m / h 1O0 k m / h 130 k m / h 160 k m / h

49.7 52.6 52.1 50.5

32.3 32.4 31.8 31.6
51.3 54.3 53.3 52.3
46.8 50.3 49.4 48.9

Figure I0 shows the effect of speed on the relaxation length a in the

pendulum test and results at one speed on the sinusoidal steer angle test.
Results of the two tests are in good agreement in that they lie on the same
interpolation curve, but only the pendulum test can accurately predict the
increase of a with increase in speed.
The cornering power C changed only slightly with change in speed for the
tire tested. The values of C for four different tires measured in a steady state
test are compared with those from the pendulum test in Table 1. The
differences between the results of the two tests can be ascribed to differences
in the tire temperature, inflation pressure, etc., resulting from test condi-
tions. The small but consistent variation of C observed for the pendulum
test indicates a linear effect of speed.

Discussion
The pendulum test is preferable to the sinusoidal steer angle test for
practical reasons such as accuracy, simplicity, and no transducer calibration
problems. Although the pendulum test can measure a at any (high) speed,
its range of working frequencies is limited by the beam length and moment
of inertia. Frequency variation requires a change in at least the moment of
inertia, by a change in beam mass, which is laborious. This may not be a
serious drawback, however, since the frequencies involved in car handling
problems are low and therefore have little effect on a. Moreover, a reference
frequency could be chosen that is typical of the yaw mode of a vehicle. Even
with such a limitation, this test can provide quick and useful information
under service conditions.