same limitation stands for the tyres used in straddle carriers and tractors which dont fit the
typical test rig because of their size.
Measurement method of normalized cornering stiffness was introduced in [5]. They exploited
a phenomenon that the sign of vehicle slip angle is reversed at certain velocity in steady-state
cornering. They utilized the method in analysing the vehicles steering response. This paper
implements the same method for tyre parameter evaluation. Simulation studies are followed
by proving ground experiments. The test rig results are compared to the vehicle measurements
on tarmac surface to analyse accuracy of the method. Some winter testing results are also
presented.
2. MEASUREMENT METHOD FOR NORMALIZED CORNERING STIFFNESS
The method is based on bicycle model (Figure 1), thus it can only handle linear behaviour of
the vehicle under steady state conditions. Since the tyre cornering stiffness is considered, we
prefer to measure the rear axle cornering stiffness, instead of front axle, due to the smaller
influence of suspension compliance (no steering compliance).
The rear axle cornering stiffness can be calculated from vehicle slip angle. There is an
equation [2] for the velocity v when the vehicle slip angle is zero:
C
(1)
v = 57.3 g c r
Wr
where g is acceleration of gravity, c is rear axle distance from the centre of gravity, Wr is rear
axle wheel load and C r is rear axle cornering stiffness. It is simple to solve cornering stiffness
from equation 1 if v is measured.
A method which is based on same reversing slip angle at certain velocity is introduced in [5],
but slip angle measurement position is not limited for centre of gravity. The following
derivation is based on it with a slight modification. Rear axle lateral force can be written:
Fr = C
l1
m a
l
(2)
where is tyre slip angle, m is mass of the vehicle, l1, 2 is distance axle from the centre of
gravity l is axle length and subscript 1 refers to front axle and 2 to rear axle. Normalized
cornering stiffness can be defined:
C
C2 n = l1 2
(3)
l mg
If combining previous equations 2 and 3, the tyre slip angle is:
ay
u2
=
2 =
(4)
C2 n g C2 n g R
where u is velocity of the vehicle and R is turning radius. From turning geometry of the
vehicle (Figure 1):
d
d + 2 = 2
(5)
R
where d 2 is distance of the optical sensor from rear axle and d is vehicle slip angle at sensor
position.
Optical velocity
sensor
f
d
d
90
l2
Steering
angle at wheel
velocity
when slip
angle is
zero
~ 42 km / h
1
3
5
10
-5
-10
10
20
30
40
50
60
70
80
90
100
110
120
Velocity [ km/h ]
The rear axle cornering stiffness at certain lateral acceleration is depicted in Figure 3. The
cornering stiffness is not decreasing linearly, thus the limit for the lateral acceleration during
the measurement could be 3-4m/s2.
12.5
12
11.5
11
10.5
10
9.5
C2 =
Fy 502,1N
= 0, 6188 = 46494
57 , 3
N
rad
(8)
C2n =
C2
1
N
= 46494 rad
= 11,37
W2
4090 N
1
rad
(9)
3500
3000
Lateral force [ N ]
2500
2000
1500
W hen = 0,6188
F = 502,1 N
1000
500
10
12
14
16
18
20
Slip angle [ ]
If one calculates backwards the velocity at which the sign of slip angle is changed:
u 0 = C 2n g d 2 = 11,88 m / s 42,8 km / h
This can be compared to the previous value, 42km/h, defined Figure 2.
(10)
There is also another way to measure changing of the slip angle instead of increasing velocity
circle. Figure 5 shows the vehicle slip angle for step steers at different velocities. In this
manoeuvre, the vehicle is running straight ahead at constant velocity and after one second
steering wheel is turned rapidly to the certain value. This steering angle is maintained and
when the vehicle is settled down for steady-state cornering the slip angle value is recorded.
0.8
0.6
0.4
0.2
Velocity
0
5 m/s
10 m/s
15 m/s
-0.2
-0.4
0.5
1.5
2.5
Time [ s ]
The step steers are repeated with different velocities and Figure 6 shows the steady-state
values of slip angle to evaluate u 0 (12 m/s). Note that the velocity u 0 is squared in the
equation 7, thus the error in the measured value is also squared as well. Thus, it is necessary
to pay attention for planning the measurements routines to minimize the error. Table 1
combines the results achieved by simulations.
0.8
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
4
10
12
14
16
Velocity [m/s]
Figure 6 Vehicle slip angle vs. velocity u 0 from step steer manoeuvres
Table 1 Summary of simulation studies
Measurement method
Tyre model
11,37
10,97
11,60
4. MEASUREMENTS
4.1 Test rig
Two different types of tyres were measured in a test rig over the dynamometer drum (Figure
7) [6-7].
The lateral force against tyre slip angle is shown in Figure 8 for two different tyres. The wheel
load is the same as in the test vehicle and velocity is 50km/h. The most interesting area in the
curves, regarding to this paper, is the linear region near the origin. The cornering stiffness is
calculated from the 0-2deg region (Table 2).
3500
3000
2500
2000
Tyre 1
Tyre 2
1500
1000
500
0
0
10
Figure 8 Lateral force vs. tyre slip angle (wheel load 2867N, 50km/h)
46 2
3, 6
1,11 9,81
= 14,99
1
rad
(10)
1,2
0,8
0,4
-0,4
-0,8
-1,2
25
30
35
40
45
50
55
60
65
lateral acceleration that tyre doesnt operate in the linear region anymore on ice surface. That
is supposed to explain the remarkably lower cornering stiffness values for ice surface. It is
also worthwhile to mention that ice was extremely slippery during the measurements. Similar
measurements were also done on snow surface, but the results are not reliable due to the
uneven track and continuously changing conditions. However, similar trend was occasionally
seen than on the ice measurements.
Table 2 Estimated and measured cornering stiffness
Estimated
Error
(1/rad)
(1/rad)
(%)
Tyre 1
14,99
14,87
0,81%
Tyre 2
15,99
15,63
2,30%
Tyre 3 (ice)
7,95
Tyre 2 (ice)
9,35
Tyre
5. DISCUSSION
The results from the test rig and vehicle tests correlate accurately. However, there are many
factors that can cause divergence to the results. Furthermore, even if the results were precise
in these particular tests, it is possible that some of the possible errors were compensating each
other.
Possible sources of error in vehicle measurements could be:
(i) axle compliance, roll steer and roll-camber,
(ii) weight transfer to the outer tyre,
(iii) how to define the maximum lateral acceleration (to stay within linear region)
and in test rig measurements:
(iv) small diameter of the drum and different surface material in test rig,
(v) how to define linear behaviourof the tyre
First three sources of errors stay small if the lateral acceleration is low during the
measurement. The influence of steering compliance is avoided, because the rear axle is
measured. The curvature of the drum in the test rig is 1219 mm and the contact material is
safety walk paper. This is maybe the most difficult error to be evaluated, because curvature of
the drum changes the pressure distribution of the tyre contact area and safety walk paper is
totally different material than a tarmac road.
The best aspect of the method is that lateral acceleration or yaw rate is not needed to calculate
normalized cornering stiffness. This is minimizing the measurement error, because only one
additional sensor is needed to measure lateral velocity. Equally, the value of vehicle slip angle
is not needed, only the sign of it.
The simulation model showed that the method is really accurate, if the measurements are
done at low lateral acceleration. Especially important is to stay under 4m/s2 acceleration, but
accuracy is improved if lower values are reached.
The comparison of test rig and vehicle measurements confirms the method to be pretty
accurate.
When planning the measurements, it is worthwhile to keep in mind that the position of the
lateral velocity sensor can be adjusted. If the sensor is located really near the rear axle, the
velocity at which the slip angle is reversed is small. However, the values of slip angle are also
small, so it may be difficult to observe the signal from noise. The better way is to install
sensor as far as possible from the rear axle, to guarantee that the slip angle values are big
enough. Unfortunately, the needed forward velocity u 0 increases at the same time and really
huge circle is needed to keep the lateral acceleration sufficiently low. Thus, it is possible to
reduce errors by planning the measurements carefully.
This research is confirming the method to measure normalized cornering stiffness presented
in [5]. It adapts the method for tyre parameter evaluation for the simulation models. The
presented method is simple and cost-effective way to measure the tyre parameters, if there are
no suitable test rigs or measurement vehicles available.
REFERENCES
(1)
(2)
(3)
(4)
(5)
(6)
(7)