Beruflich Dokumente
Kultur Dokumente
To cite this article: Yinong Li , Wei Sun , Jingying Huang , Ling Zheng & Yanyang Wang (2013)
Effect of vertical and lateral coupling between tyre and road on vehicle rollover, Vehicle System
Dynamics: International Journal of Vehicle Mechanics and Mobility, 51:8, 1216-1241, DOI:
10.1080/00423114.2013.791395
To link to this article: http://dx.doi.org/10.1080/00423114.2013.791395
Yinong Li*, Wei Sun, Jingying Huang, Ling Zheng and Yanyang Wang
The State Key Laboratory of Mechanical Transmission, Chongqing University No. 174 Shazhengjie,
Shapingba, Chongqing 400044, Peoples Republic of China
(Received 20 February 2013; final version received 25 March 2013 )
The vehicle stability involves many aspects, such as the anti-rollover stability in extreme steering
operations and the vehicle lateral stability in normal steering operations. The relationships between
vehicle stabilities in extreme and normal circumstances obtain less attention according to the present
research works. In this paper, the coupling interactions between vehicle anti-rollover and lateral stability, as well as the effect of road excitation, are taken into account on the vehicle rollover analysis. The
results in this paper indicate that some parameters influence the different vehicle stabilities diversely
or even contradictorily. And it has been found that there are contradictions between the vehicle rollover
mitigation performance and the lateral stability. The direct cause for the contradiction is the lateral
coupling between tyres and road. Tyres with high adhesion capacity imply that the vehicle possesses a
high performance ability to keep driving direction, whereas the rollover risk of this vehicle increases
due to the greater lateral force that tyres can provide. Furthermore, these contradictions are intensified indirectly by the vertical coupling between tyres and road. The excitation from road not only
deteriorates the tyres adhesive condition, but also has a considerable effect on the rollover in some
cases.
Keywords: untripped vehicle rollover; anti-rollover stability; lateral stability; road excitation;
coupling effect
1.
Introduction
Vehicle rollovers are dangerous incidents and have a higher fatality rate than other kinds
of crashes. Of the nearly 9.1 million passenger car, SUV, pickup and van crashes in 2010,
only 2.1% involved a rollover. However, rollovers accounted for nearly 35% of all deaths
from passenger vehicle crashes. In 2010, more than 7600 people died in vehicle rollover
crashes [1].
According to the NHTSA-2003-14622 document [2], the vehicle rollover could be classified
as off-road rollover, on-road tripped rollover and on-road untripped rollover.
The tripped rollovers (on- or off-road) are going to occur when a vehicle leaves the roadway
and slides sideways, digging its tyres into soft soil or striking an object such as a curb or
guardrail. The high tripping force applied to the tyres in these situations can cause the vehicle
to roll over. To prevent this kind of rollover, rollover warning systems or collision avoidance
alarm systems are utilised.
*Corresponding author. Email: ynli@cqu.edu.cn
2013 Taylor & Francis
Table 1.
1217
Excitation type
shock
Steady-state vibration
Stochastic vibration
Fishhook manoeuvres
Slalom test
Road bumps
Washboarding road
Stochastic road surface
The vehicle on-road untripped rollover is mainly caused by the lateral acceleration during
cornering operations, e.g. high-speed collision avoidance manoeuvres. And the excessive roll
angle of vehicle body induced by lateral acceleration is the direct reason for rollover.
Rollovers, more so than other types of crashes, are complex and particularly violent in
nature, reflect the interaction of the driver, road, vehicle, and environmental factors. So while
vehicle type does play a significant role in rollover incidents, other factors such as driver
behaviour and road as well as environmental conditions can also cause vehicle rollover.
Based on the view of vibration and shock, the excitation to vehicle dynamics could be
classified as shock, steady-state vibration and stochastic vibration. Also, the excitation could
come from steering wheel or road according to different source. Table 1 lists the examples for
each kind of excitation. Generally, uneven road is used for testing the ride comfort of vehicles,
and the steering manoeuvres are designed mainly for testing the stability of vehicles. The conventional analyses usually treat these aspects separately, which are insufficient considerations
in some extreme situation.
In order to test the vehicle stability repetitively, many testing operations and standards are
specified to regulate the driver behaviours, e.g. Passenger car/trailer combinations lateral
stability test of ISO 9815, Elk or Moose test of ISO 3888-2:2011. Among these standards
and tests, fishhook specified by NHTSA-2003-14622 is most commonly used for vehicle
rollover tests.
To eliminate the possibility of on-road untripped rollovers, the common solution is to
increase vehicle stability by tuning vehicle parameters (passive method) [3,4] or equipping
with active systems such as ESP or Active Suspensions (active method) [57]. These existing
researches mainly focus on the vehicles anti-rollover stability, which leave other factors, such
as the general lateral stability and road conditions, out of consideration.
Besides the anti-rollover performance, the vehicle stability involves many other aspects,
e.g. the lateral stability such as oversteering tendency and sideslip, yet the current research
works [811] generally treated these features independently.
Sadri and Wu [12,13] utilised the Lyapunov function to analyse the lateral stability and
investigated the rollover stability region within the phase diagram of lateral velocity and yaw
rate. Nam et al. [14] constructed an observer to estimate the roll angle through lateral tyre force
and thus proposed a direct roll moment controller for four- wheel steering vehicle. Similarly,
Du et al. [15] developed a yaw moment controller, whereas the controller parameters are
dependent on the vehicle forward velocity and the controller is designed by solving finite
numbers of linear matrix inequalities (LMIs). Both Nam and Dus methods could achieve
adequate responses of sideslip angle and yaw rate. Lu et al. [16] discussed control authority
and the effective working region of different chassis key subsystems for rollover prevention,
and further utilised MR semi-active suspension as key actuator to implement integrated control
with braking and steering [17,18]. These existing researches can improve the steering or antirollover stability, however the couplings between lateral stability and roll dynamics are seldom
considered integrally. The role and effect of lateral stability such as sideslip in rollover analysis
need to be considered further.
Moreover, though road condition is not mandatory in the definition of on-road untripped
rollover, the excitation from uneven road is inevitable, especially that the uneven road profile
1218
Y. Li et al.
is an important reason for the vehicle body roll motion. The contributions of road excitation
to rollover have not been studied thoroughly yet.
In light of the above opinions, the lateral coupling of tyre and road, namely the tyre lateral
adhesive capacity is discussed comprehensively in this paper for the conjoint analysis of
vehicle anti-rollover and other stabilities. Meanwhile, the vertical coupling of tyre and road,
i.e. the attachment of tyres to the road is integrated into the analysis of vehicle rollovers by
the combination of the fishhook manoeuvres and uneven road excitation.
This paper is organised as follows: First of all, the vehicle models as well as nonlinear tyre
model are derived, followed by a preliminary analysis on the model characteristics. Subsequently, based on the fishhook manoeuvre method and the comparison of different rollover
critical factors (RCFs), the relationships between vehicle lateral and anti-rollover stability are
investigated. Then the effect of uneven road on vehicle rollover is investigated. And finally,
the conclusions drawn from this research are presented.
2. Vehicle models
The vehicle model consists of three sub-models: the vibration and roll model that emphasises
on sprung mass roll motion, which is the direct cause of the vehicle rollover; the vehicle
steering model that is used for the calculation of the lateral acceleration, which is the main
reason for the vehicle body roll motion; the tyre model that provides the lateral force for
steering model to calculate the lateral acceleration.
2.1.
Tyre model
The primary purpose of the tyre model is to work out the lateral force that tyre can provide.
When the vehicle is cornering, the roll motion of vehicle body will arouse lateral load transfer
on tyres, and this transfer leads to the variation of tyre force. To take account of this effect,
tyre load is considered with the slip angle. The Magic-Formula (MF) tyre models [19] are
adopted here because they are easy to implement, and have been widely proved accurate in
many vehicle dynamics simulation. Though they may not fit for low-speed conditions, in this
rollover study, the vehicle speed is usually high enough. For pure slip conditions, the nonlinear
MF equations can be described as
Fy = D sin{C arctan{B E[B arctan(B)]}},
b3 sin(2 arctan(Fz /b4 ))
, C = b0 ,
CD
b1 Fz2 + b2 Fz
D=
, E = b5 Fz + b6 ,
1000
B=
(1)
where the unit of lateral force Fy and tyre load Fz is kilo Newton, the unit of slip angle is
radian. B, C, D and E are the empirical parameters, respectively.
Two types of tyres are chosen to discuss the relationships between the vehicle anti-rollover
and lateral stability performance. Tyre A stands for high adhesive capacity tyres, and tyre B
stands for low adhesive capacity tyres, as shown in Table 2. Subjected to the same tyre load,
the lateral force responses of the two types of tyres are shown in Figure 1(a). It can be seen
that the extreme slip angle of tyre A is smaller than tyre B when the maximum lateral force
is achieved, whereas the lateral force tyre A can provide is larger than tyre B. The visualised
nonlinear force responses of the MF tyres models to tyre load and slip angle are shown in
Figure 1(b).
Tyre type
Tyre A
Tyre B
Maximum adhesive
coefficient
Note
8
12
0.9064
0.6564
(b)
8000
(a)
4000
0
4000
8000
1219
32 24 16 8 0 8
slip angle ()
tyre A
tyre B
16 24 32
tyre A
tyre B
8000
4000
0
8000
4000
Tyre load (N)
0 0
20
10
slip angle ()
Figure 1. Comparison of the two MF tyre models: (a) lateral force of the two tyres and (b) force response of the
two tyres.
vx
vy
(2)
(3)
The wheel steering angle is usually small enough, and only considering the front wheel
steering, the geometry relations are
vy b
vy + a
vy + a
vy b
1
1 , 2 = arctan
.
1 = arctan
vx
vx
vx
vx
1220
Y. Li et al.
The above Equations (2) and (3) could be rewritten in a differential equation form:
[(Fy1 + Fy3 ) cos(1 ) + (Fy2 + Fy4 ) cos(2 )]
vx ,
mv
[a(Fy1 + Fy3 ) cos(1 ) b(Fy2 + Fy4 ) cos(2 )]
=
.
Iz
v y =
(4)
(5)
(6)
m
n
+ (Fs2 Fs4 ) + ms ay d + ms g d,
2
2
(8)
Figure 3.
n
,
2
n
zs4 = z + b
.
2
zs2 = z + b +
(9)
1221
Then the vertical vibration and pitch motion of sprung mass are
ms z = Fs1 + Fs2 + Fs3 + Fs4 ,
Iy =
My = (Fs2 + Fs4 )b (Fs1 + Fs3 )a.
(10)
(11)
(12)
As an available solution [20,24], this model is chosen to discuss the effect of the vertical and
lateral coupling between tyre and road on vehicle rollover and stabilities.
The tyre load can be calculated by summing the static and dynamic load, where the static
load for each tyre is
s
s
Fz1
= Fz3
=
bmv g
,
2(a + b)
s
s
Fz2
= Fz4
=
amv g
.
2(a + b)
(13)
The trigger condition of termination for the test is when the inner tyres lift off the ground:
Fz1 0, Fz2 0
or
Fz3 0, Fz4 0.
m
1 a
2
n
1 b
0 0 0
2
T
H=
m , L = 0 0 0 , u1 = [0 0 ay ] , u2 = [zg ].
1 a
0 0 1
n
1 b
2
Then the above equations can be written in a matrix form, separated into sprung and unsprung
mass part:
[z] = Ms1 (ms dgL H T Ks H)[z] M 1 H T Cs H[z]
+ Ms1 H T Ks [zu ] + Ms1 H T Cs [zu ] + ms dMs1 u1 ,
[zu ] =
Mu1 Ks H[z]
(14)
Mu1 Cs H[z]
(15)
As can be seen, there are two inputs for this model: the vehicle lateral acceleration and
vertical road excitation. The lateral acceleration influences the lateral coupling of tyres and
road, while the vertical road excitation has an influence on the vertical coupling of tyres and
road. The excitation from road mainly deteriorates the ride comfort, impacts the attachments
of tyres to road, disturbs the vehicle body roll motion and reduces the manoeuvring stability.
1222
Y. Li et al.
Table 3.
Order of
modal
zu1
zu2
zu3
zu4
1.11
1.27
1.37
11.96
11.96
11.96
11.96
78.62
21.29
0.00
0.00
0.07
0.02
0.00
21.29
78.60
0.00
0.00
0.02
0.09
0.00
0.00
0.00
99.87
0.00
0.00
0.00
0.13
0.04
0.00
0.03
24.95
46.91
3.05
25.02
0.00
0.05
0.03
25.05
3.05
46.90
24.92
0.04
0.00
0.03
24.95
46.91
3.05
25.02
0.00
0.05
0.03
25.05
3.05
46.90
24.92
1
2
3
4
5
6
7
Natural
frequency (Hz)
(16)
where
Ms
M=
043
034
,
Mu
T
H Ks H
K=
Ks H
H T Ks
,
(Ks + Kt )
[z]
Z=
.
[zu ]
The modal coupling results are given in Table 3. The bold values indicate the principal mode
of the vehicle vibration system. As can be seen, the vehicle body roll motion is decoupled
with other degree of freedom, it composes 99.87% of the vibration energy in the third modal,
and the natural frequency is 1.37 Hz. The resonance of vehicle body roll motion may induce
rollover when the vehicle is excited at 1.37 Hz. The excitation may be from road disturbances
as well as steering manoeuvres. Moreover, the bounce of wheels converges at 11.96 Hz. The
bounce of wheels may lead to the detachment of tyres from the road, which will deteriorate
the adhesion conditions of tyres and finally make the vehicle lose stability.
The transfer responses of the vehicle body roll motion and wheel bounce to different excitations are shown in Figure 4 and Table 4. As can be seen, except the response of wheel
bounce to road excitation, the resonance frequencies of other motions are at 1.15 Hz, which
is approximately equal to the vehicle body roll motion frequency of 1.37 Hz. And the little
difference is due to damping effect as mentioned earlier.
Both the lateral acceleration and road excitation can arouse the vehicle body roll motion as
well as the wheel bounce motion. The vehicle is likely to roll over due to the excessive roll
angle, or is easy to lose stability due to the uneven tyre load distribution induced by wheel
1223
(a)
0
phase ()
0.02
0.01
100
0
0.2
10
200
0.2
20
phase ()
0.5
0
0.2
0.5
0.5
1
2
5
Frequency (Hz)
10
20
0.5
1
2
5
Frequency (Hz)
10
20
0.5
1
2
5
Frequency (Hz)
0.5
1
2
5
Frequency (Hz)
1
2
5
Frequency (Hz)
10
100
200
300
0.2
20
(c)
0
0.6
phase ()
mag (rad/m)
0.8
0.4
0.2
0
0.2
0.5
1
2
5
Frequency (Hz)
10
100
200
300
0.2
20
(d)
10
20
3
2
phase ()
mag
1
2
5
Frequency (Hz)
1.5 10
mag (s )
(b)
0.5
1
0
0.2
0.5
1
2
5
Frequency (Hz)
10
20
100
200
0.2
10
20
Figure 4. FRs of vehicle model: (a) (1) magnitude FR of roll angle to lateral acceleration, (a) (2) phase FR of roll
angle to lateral acceleration, (b) (1) magnitude FR of wheel bounce to lateral acceleration, (b) (2) phase FR of wheel
bounce to lateral acceleration, (c) (1) magnitude FR of roll angle to road excitation, (c) (2) phase FR of roll angle to
road excitation, (d) (1) magnitude FR of wheel, and (d) (2) phase FR of wheel bounce to road excitation.
Table 4.
Excitation
Lateral acceleration
Road excitation
Roll angle
Figure 4(a) (1)
Figure 4(c) (1)
Wheel bounce
Figure 4(b) (1)
Figure 4(d) (1)
bounce. As a result of these consequences, the lateral and anti-rollover stability should be
considered as a whole response to the action of both steering and road excitations. The impact
of these coupling effects on the vehicle stabilities will be further discussed in the following
section.
1224
Y. Li et al.
Figure 5.
3.
1225
And the lateral load transfer rate (LTR) is also chosen as RCF [31] in some studies:
Fz1 + Fz2 Fz3 Fz4
.
RCF3 = LTR =
F +F +F +F
z1
z2
z3
(19)
z4
The larger RCF2 or RCF3 is, the more likely that the vehicle rolls over, and if the RCF2
or RCF3 is greater than 1, it is regarded that rollover has already occurred. To normalise the
numerical range of the three RCFs, a transformation of RCF1 is needed:
1
1,
0.5RCF1 / max(RCF1 ) + 0.5
(20)
where obviously
max(RCF1 ) =
g(m + n)
4
then
RCF1 =
1
1.
0.5RCF1 /[g(m + n)/4] + 0.5
(21)
A comparison of the three RCFs subject to the same fishhook manoeuvre is shown in Figure 6.
In the simulations of this research, in order to make a comparison between different RCFs,
and investigate the effect of road fluctuation on rollover, a RCF value less than 1 is beneficial
for the purpose of continuity in simulations. So, by trial and error, the simulation speed is
set as 54 km/h, which is enough to arouse a significant RCF value whereas the termination
conditions will not be triggered. As can be seen, the RCF1 is more sensitive to the rollover, and
have a higher safety threshold at the beginning of T2 in fishhook. The RCF2 is less sensitive
but can give a higher safety threshold in the entire manoeuvring. The RCF3 mainly focuses
on the asymmetric load on different tyres, and its rollover safety threshold is much lower than
that of RCF1 .
Besides the sensitivity to vehicle stability, the cost of the three RCFs should also be noticed.
RCF2 requires the fewest parameters, which means fewer sensors are needed. The RCF1 needs
more parameters and computation, but is more sensitive to the rollover detection.
According to these comparisons, the RCF1 is more sensitive to the rollover and have a higher
safety threshold, besides, it can express more effect of different factors, such as the oscillation
of roll acceleration, the movement of both roll inertia and vehicle mass, so it is more superior
in the rollover analysis [20].
1
RCF1
0.8
RCF
RCF1 =
RCF2
0.6
RCF3
0.4
0.2
0
0
Figure 6.
4
6
Time (s)
10
1226
Y. Li et al.
As mentioned previously, the existing researches treat the lateral stability [8,9] and anti-rollover
stability [4,20] independently or partially. In this paper, the different stabilities are intended
to be analysed correlatively.
Some indicators are chosen to evaluate these stabilities. For the anti-rollover stability, it
is featured by the RCF. For the over or under steering tendency, it is easy to be evaluated
according to the vehicle trajectory. As for the lateral stability, which means whether there is
sideslip of vehicle, many researches [32,33] utilise the tyre friction ellipse to carry out the
analysis. However, a single parameter, the tyre slip angle, is used to estimate whether the
vehicle is stable [34] in this paper. When the slip angle of tyre exceeds the extreme value,
the sideslip is then regarded as occurring. The benefit of this method is that only one simple
parameter is needed.
The responses of the four tyres subjected to a fishhook 1a manoeuvre are shown in Figure 7,
where the vehicle speed is 54 km/h. The 3D curve of tyre response includes tyre lateral force,
the corresponding slip angle and tyre load. It shows that the curves of tyres 1 and 3 cover
a larger range than their counterparts. And the load of tyre 4 is near zero, which means that
tyre 4 is about to lift off the ground.
The slip angles are extracted and shown in Figure 8, the slip angle of the rear wheels, i.e.
2 , is approximately smaller than 8 (extreme slip angle mentioned in Table 2) during the
whole manoeuvre period. However, the front tyre slip angle, 1 , reaches approximately 15
in the first stage of fishhook manoeuvre, which means the sideslip occurs on the front tyres.
Moreover, the 1 rises even to 17 in the second stage, i.e. the sideslip is more serious.
tyre 1
tyre 2
tyre 3
tyre 4
MF tyre
7500
2500
2500
7500
10
0
10
20
5000
tyre load (N)
slip angle ()
Figure 7.
18
1
slip angle ()
9
2
0
9
18
0
6
time (s)
Figure 8.
10
1227
20
pole 1
0
pole 2
pole 3
20
40
-10
0
Roll angle ()
10
40
60
Forward direction (m)
80
20
Lateral direction (m)
Figure 9.
-5
0
20
40
60
80
100
20
The phase diagram of vehicle body roll response is shown in Figure 9. There are three poles.
Pole 1 represents the vehicle entrance speed into fishhook manoeuvre, pole 2 and pole 3 stands
for the first and second steering stage in fishhook manoeuvre, separately. The closure of the
roll phase diagram implies that the vehicle can return to steady state eventually. The vehicle
trajectory is also shown in Figure 10.
3.3. Evaluation of vehicle stabilities
Quite a number of vehicle parameters could be adjusted to improve vehicle dynamics, e.g.
vehicle mass, location of the CG, the stiffness, characteristics of suspension and tyre, etc. The
influence of these parameters on the single stability performance of vehicle has already been
illustrated by many researches [3,4,35], and will not be repeated here. In this paper, only the
cross effect on different stabilities is discussed.
The evaluation between two types of tyre is made, with high and low adhesive capacity
respectively, as listed in Table 2. The original reference vehicle is equipped with high adhesive
tyres (tyre A), and the contrast vehicle is equipped with low adhesive tyres (tyre B). The
comparison results shown in Figure 11 indicate that the sideslip tendency of contrast vehicle
increases significantly, especially in the second stage of the manoeuvre, from less than 15 to
more than 45 . So tyres with high adhesive capacity possess reliable lateral stability.
However, according to the roll phase diagram shown in Figure 12, the roll angle and roll
rate decrease slightly. This means that the roll motion of the vehicle body is more stabilised.
As a result, the rollover risk of contrast vehicle is greatly reduced as shown in Figure 13, the
RCF drops from 0.76 to 0.48. We can say the anti-rollover stability is increased by 36.8%.
So tyres with high adhesive capacity possess a high risk to rollover. However, there is also a
1228
Y. Li et al.
45
(tyre A) 1
(tyre B) 1
Slip angle ()
30
15
(tyre A) 2
(tyre B) 2
0
15
30
45
0
6
Time (s)
10
45
Roll rate (deg/s)
tyre A
30
Figure 12.
tyre B
15
0
pole 1
15
pole 2
30
pole 3
45
10
0
Roll angle ()
10
(tyreA)RCF1
0.8
RCF
Figure 11.
(tyreB)RCF1
0.6
0.4
0.2
0
0
Figure 13.
6
Time (s)
10
new problem. The phase diagram stays in the third pole and cannot get closed. This means
the vehicle is keeping side slipping in the second stage of fishhook manoeuvre and thus the
lateral stability of this vehicle is decreasing.
These two phenomena reveal that there is contradiction between different vehicle stability
performances, and tyre adhesive capacity is one key reason for this contradiction. Tyres with
high adhesive capacity possess reliable lateral stability, but a high risk to rollover. It can be
explained with the smaller force that low adhesive tyres can provide. The lateral force that low
adhesive tyres can provide is too small to afford a large lateral acceleration. And the smaller
lateral acceleration can only induce a smaller roll angle on vehicle body, and thus reduce the
rollover risks ultimately. On the other hand, tyres with high adhesive capacity guarantee larger
lateral force to achieve a higher steering capability, but the larger lateral force is easy to induce
vehicles rollover.
1229
20
0
20
40
60
80
100
tyre A
tyre B
0
25
50
75
Forward direction (m)
100
The vehicle trajectory also changes very differently as shown in Figure 14. It becomes quite
sharper than the original trajectory due to sideslip of tyres both in the first and second stage
of fishhook manoeuvre.
4.
As analysed previously, not all the extreme steering manoeuvre will end in rollover, and
sometimes the side slip should be the major concern. However, the excessive roll angle of
sprung mass do play a role of direct reason for vehicles rollover, and the roll motion of sprung
mass is mainly caused by the lateral acceleration induced by cornering on vehicle body. In
addition to these factors, another noticeable and common reason for vehicle body roll motion
is the excitation from uneven road, and its effect is inevitable and should not be ignored. Next,
we will look into this factor.
4.1.
A harmonic and periodical excitation from road is uncommon; however, its effect on vehicle
body roll motion is concise and intuitive. So it is necessary to look at the effect of this kind
of excitation. The periodical road is a kind of steady-state sine excitation. The magnitudefrequency response (FR) of this excitation has already been explained in Section 2.4. This
section focuses on the effect of the road phase difference between the left and right tyres.
Supposing that the vehicle wheel base is much smaller than the wavelength of the road
fluctuation, which means that the front tyres are in the same phase as that of the rear tyres,
then the vertical excitation from road applied on left and right tyres are
zg1 = G0 sin(2 ft ),
(22)
(23)
where f is the frequency of the road fluctuation, the phase difference between left and right
side tyres and G0 stands for the degree of roughness.
Then the road excitation on the left tyres relative to the right is
zg1 zg3 = G0 sin(ft ) G0 sin(ft)
= G0 sin(ft ),
where
tan( ) =
sin( )
,
cos( ) 1
G0 = 2G0
sin
.
2
(24)
(25)
1230
Y. Li et al.
mag (rad/m)
1.5
1
0.5
0
720
540
360
phase
difference ()
0 0.2
0.5 1
10 20
Frequency (Hz)
1.5
1
RCF
Figure 15.
180
0.5
0
0
Figure 16.
6
Time (s)
10
Drawn from the Equations (24) and (25), the conclusion is that the magnitude of the relative
road excitation is determined only by the phase difference, and obviously the magnitude is
maximal when the phase difference is 180 . Combining Equation (24) with Figure 4(c), it is
easy to obtain the results in Figure 15, which shows the synthesis response when vehicle is
subjected to different frequency and phase excitation. As can be concluded, the extreme case is
when the excitation amplitude is maximal and the excitation frequency is inducing resonance
of vehicle body, i.e. the phase difference is 180 and the frequency is approximately 1.15 Hz.
The comparison of fishhook manoeuvres between with and without road excitation is shown
in Figure 16. The extreme case is chosen to see how severe the road effect is: the phase
difference between the two sides of tyres is set to be 180 ; the fluctuating magnitude of road
is 0.02 m (approximately equivalent to the B class road). And the road excitation frequency is
about 1.15 Hz with the condition that the wavelength of road is 13 m and the vehicle speed is
54 km/h (54 km/h divided by 13 m is 1.15 Hz).
As can be seen, when excited by road in the extreme condition, the RCF1 reaches 1.309,
which exceeds 1 significantly, i.e. the vehicle rolls over. However, the RCF1 is only 0.7227
without the road excitation. The effect of this kind of road makes the risk of rollover increase
by 81.13%. According to this result, the uneven road is an important factor that cannot be
neglected in the vehicle rollover analysis.
4.2. Effect of the stochastic road on vehicle stabilities
The stochastic road is a more common excitation than the periodical road. Its effect on vehicle
rollover is investigated as follows. First a model of stochastic road excitation for each tyre
is established. Then the coherence analysis of this road model is carried out, which is an
important factor for the vehicle body roll motion.
1231
Lower limit
Geometric mean
Upper limit
32
128
512
2048
8192
32,768
131,072
16
64
256
1024
4096
16,384
65,536
262,144
32
128
512
2048
8192
32,768
131,072
524,288
A
B
C
D
E
F
G
H
(26)
Zg3 (s)
a0 + a1 s + a2 s2
.
=
Zg1 (s)
b0 + b 1 s + b 2 s 2
(27)
And the transfer function of the road excitation from tyre 1 to tyre 2, i.e. the road excitation
from front tyres to rear tyres, is a pure time delay:
zg2 (t) = zg1 (t td ).
(28)
Then the transfer function from front to rear could be written in 2-order Pade approximation
form:
G21 (s) =
Zg2 (s)
1 (td /2)s + (td2 /12)s2
,
=
Zg1 (s)
1 + (td /2)s + (td2 /12)s2
(29)
(30)
With Equations (26)(30), the uneven road excitation model at four tyres has been accomplished.
1232
Y. Li et al.
Tyre 4
Tyre 3
Transfer Path 1
Transfer Path 2
Tyre 2
Tyre 1
Figure 17.
4.2.2.
The difference between the road excitations on left and right tyres arouses a significant portion
of the vehicle body roll motion, and this difference is determined by the coherency between
the left and right road excitation. As a result, the coherence analysis of road excitation is an
essential part in this rollover analysis.
According to the control theory, for the single input single output system, the FR function
from input to output is defined as follows:
H(f ) =
Sxy (f )
,
Sxx (f )
(31)
where Sxy (f ) is the cross-spectrum between input and output, and Sxx (f ) is the self-spectrum
of input.
And the coherence function between input and output is
coh2xy (f ) =
|Sxy (f )|2
.
S2xx (f )Syy (f )
(32)
For the stochastic road, the statistic characteristics of left and right side of road should be
the same, i.e. Sxx (f ) = Syy (f ). Then from the above Equations (31) and (32), there is
cohxy (f ) = |H(f )|.
(33)
(34)
Four typical types of road are exhibited in Figure 18. The coh1 shows low coherency in the
broadband frequency domain. The coh2 shows low coherency in the low-frequency domain.
The coh3 show slow coherency in the high-frequency domain. And the coh4 shows high
coherency in the broadband frequency domain.
1233
1.5
coh1
coh2
coh3
coh4
coh
0.5
0
0
10
Frequency (Hz)
15
20
3
10
2
PSD (m /Hz)
Figure 18.
5
10
7
10
9
10
0.2
Figure 19.
coh1
coh2
coh3
coh4
0.5
1
2
5
Frequency (Hz)
10
20
Low coherency means that the coherency between the road excitation at left and right tyres
is small, normally is smaller than 0.2 as can be seen from Figure 18; high coherency means
the aforementioned coherency is large, normally is great than 0.8 as shown in the Figure 18.
Thus, the coherency can be a presentation of road quality. The higher the coherency is, the
flatter the road is in lateral section.
Low-frequency domain represents the frequency domain that covers the natural frequency
of vehicle body roll motion, which is of a relatively low frequency. And in this case, it is about
1.37 Hz in Table 3. Whereas high-frequency domain represents the frequency domain that
covers the natural frequency of wheel bounce motion, which is of a relatively high frequency.
And in this case it is about 11.96 Hz in Table 3. And broadband frequency domain represents
the frequency domain that covers both of the two aforementioned natural frequencies. The
frequency here can be a presentation of the road scale in the longitudinal section. The high
frequency signifies micro scale, and low frequency signifies the macro scale.
These four different types of stochastic roads exhibit different coherency at the natural
frequencies of vehicle body roll motion as well as wheel bounce; they also reflect the excitation
of the road with different qualities in a different scale. It is efficient to see the effect of road
excitation on the response of sprung and unsprung mass according to these four types of road.
Though the coherencies and the time histories of these roads are different, the power spectral
density (PSDs) of these roads are the same, as shown in Figures 1923, which is a class C
road.
The time histories of the coh1 and coh2 roads are almost the same, and so do the coh3 and
coh4 . The conclusion is that the time history of the road is mainly determined by coherency
in the low-frequency domain. So the following discussion will focus on the roads of coh2 and
coh3 because they are more representative.
1234
Y. Li et al.
tyre1
0.06
tyre2
tyre3
10
Time (s)
15
tyre4
Road (m)
0.04
0.02
0
0.02
0.04
20
tyre1
tyre2
tyre3
tyre4
Road (m)
0.06
0.04
0.02
0
0.02
0.04
10
Time (s)
15
20
tyre1
tyre2
tyre3
tyre4
Road (m)
0.06
0.04
0.02
0
0.02
0.04
0
10
Time (s)
15
20
tyre1
tyre2
tyre3
tyre4
0.06
Road (m)
0.04
0.02
0
0.02
0.04
10
Time (s)
15
20
RCF
1
4
6
Time (s)
10
D class road
C class road
B class road
without road
0.5
0
0
Figure 25.
RCF
1
coh3
coh2
without road
0.5
0
0
Figure 24.
1235
4
6
Time (s)
10
The fishhook manoeuvre is conducted on the class C road, with coherency of coh2 and coh3 ,
respectively. The results are shown in Figure 24. It seems that there is no apparent distinction
between the two conditions, only that the coh2 road is more likely to arouse the rollover risk
at time about 3 or 6 s, while the most dangerous time should be at about 2.5 s when there is
no road excitation.
Among these roads, coh3 is more common in reality. The RCF results of this coherency on
different road classes are shown in Figure 25. As can be seen, the risk of rollover increases
proportionally to road class. And the road excitation contributes to the RCF more at nonextreme time, e.g. at time about 3 or 6 s, and especially in the first stage of fishhook manoeuvre,
e.g. at time about 1.8 s, whereas the extreme time should be at about 2.5 s. In extreme cases,
a class D road is enough to induce vehicle rollover. More comparison details are summarised
in Table 6. We can see that class C road already arouses a noticeable increase in rollover risk,
more than 12%.
Table 6.
Road type
RCF
Without road
coh2 C Class
coh3 B Class
coh3 C Class
coh3 D class (peak 1)
coh3 D class (peak 2)
0.7227
0.8144
0.7658
0.8105
1.068
0.8957
Relative change
+12.69%
+5.96%
+12.15%
+47.78%
+23.94%
Y. Li et al.
1236
without road
with road
MF tyre
7500
2500
2500
7500
10
0
slip
10
20
angle ()
Figure 26.
0
5000
Tyre load (N)
Without road
With road
Relative change
4.2.4.
5040 N
5523 N
+9.58%
1653 N
1075 N
34.97%
4664 N
5241 N
+12.37%
2439 N
3327 N
34.41%
As mentioned previously, the vertical coupling state of tyres and road determines the condition
of how tyres are attached to the road, i.e. the road hold performance of tyres. If the vertical
coupling state between tyres and road is bad, the tyres easily jump off the road. This phenomenon is not good for the vehicle stability because no adhesive force (both lateral and
longitudinal) could be provided by tyres. In some high-speed conditions, vehicle is likely to
lose control if the tyres separate from the road. So a symmetrical, continuous contact between
the four tyres and road is essential for vehicle stability. In some extreme case, the tyre load may
drop to zero, which is an even worse condition. And according to the example of Figure 26,
this is a possibility. Figure 26 shows an example of deteriorative vertical coupling state of tyres
and road, the force response of tyre 4 due to excitation of class C road with coh3 and fishhook
manoeuvre. More details are listed in Table 7, as can be drawn from Figure 26 and Table 7,
the minimum tyre load drops 34.97%, and the minimum lateral force that tyre can provided
decreased by 36.41%, this imply that the vehicle lateral stability decreases when considering
the road excitation.
To comprehensively evaluate the effect of stochastic road excitation on vehicle anti-rollover
and lateral stability, RCF1 is still a convenient indicator for vehicle rollover, and the RCF3
proposed in the previous section is just suitable to monitor the variation of tyre load and thus
vehicle lateral stability.
Because the combined condition of road excitation and fishhook manoeuvre is a nonstationary state, the timefrequency analysis such as short-time Fourier transform or wavelet
transform is more suitable for a thorough investigation. However, for simplicity, here the
road excitation is considered independently, and the final result of RCF is assumed to be the
linear superposition of the two excitations. Then the focus of RCF can be limited only to
the proportion aroused by road excitation and the conventional fast fourier transform (FFT)
method could be applied.
The RCF1 and RCF3 induced by the class C road with different coherencies are shown in
Figures 27 and 28, and the FFT results are shown in Figures 29 and 30. From the time domain
results, it can be seen that the coh2 road can arouse a higher RCF1 , while a relative lower RCF3
1237
RCF1
0.06
coh2
coh3
0.04
0.02
0
0
10
Time (s)
15
20
coh3
RCF
3
0.2
coh2
0.15
0.1
0.05
0
Figure 28.
10
Time (s)
15
20
x 10
-3
coh2
coh3
2
RCF
Figure 27.
Figure 29.
10
Frequency (Hz)
15
20
than the coh3 . The frequency domain results also illustrate the same phenomenon, though not
very significant.
These results indicate that road excitation with low coherency in the low-frequency domain
has a greater effect on vehicle rollover than lateral stability. Conversely, the road of high
coherency in the low-frequency domain has more effect on vehicle lateral stability than
rollover. Therefore, on the same section of road, the vehicle anti-rollover stability and lateral
stability are also in contradiction.
Table 8 summarises the increased percentage of RCF due to the excitation of different class
coh3 road. It reveals that the increase in RCF3 is greater than RCF1 , and the class C road is
already enough to have an obvious effect on vehicle stabilities, where the risk of rollover may
increase by 6% and the lateral stability decreases by 12%.
1238
Y. Li et al.
x 10
3
coh3
RCF
coh2
4
0
0
Figure 30.
10
Frequency (Hz)
15
20
Table 8.
5.
RCF
RCF1
RCF3
2.5%
6%
6%
12%
10%
26%
25%
180%
Conclusions
The vehicle stability involves many different performances, of which the anti-rollover stability
is an important one. The direct reason for rollover is the excessive roll angle, which is determined by both the road excitation and lateral acceleration. Moreover, the lateral acceleration
is determined exactly by the vehicle lateral stability, which refers to the ability to prevent
over steering tendency or sideslip. On the other hand, the vehicle body roll motion in turn has
crucial influence on the tyre load, which will affect the vehicle lateral stability significantly.
As a result, the relationships and interactions among these different stability performances
are considered integrally. The role and effect of lateral stability such as sideslip in rollover
analysis are investigated in this paper, and the important findings are as follows:
(1) There are contradictions between the vehicle lateral and anti-rollover stability. When
cornering, the vehicle with higher lateral stability possesses a higher risk of rollover.
Meanwhile, on the same section of the road, the effects of road excitation on lateral or antirollover stability are different. Road excitation with low coherency in the low-frequency
domain has a greater effect on vehicle rollovers than lateral stability.
(2) The coupling between tyres and road is a key reason for the above contradiction. The
force that tyres can provide, i.e. the lateral coupling between tyres and road, is the direct
cause of this contradiction. High adhesive capacity tyres mean that the vehicle has a higher
performance to maintain driving trajectory, while the greater lateral force that tyres provide
raises the risk of rollover. On the other hand, the interaction force and attachment between
tyres and road, i.e. the vertical coupling between tyres and road is an indirect factor that
intensifies this contradiction. The vertical coupling between tyres and road disturbs the tyre
lateral force and vehicle body roll motion, and thus the lateral and anti-rollover stability.
So the lateral coupling is regarded as a direct factor and vertical coupling is an indirect
factor. These results suggest that a coordinated control system is needed for the different
objectives of vehicle stability performances, and the key lies in the distribution of the tyre
load.
1239
(3) The effect of road excitation on vehicle rollover is considerable in some extreme situation.
Periodical road, though uncommon, will result in the vehicle resonant vibration, and
increases the rollover risk easily and greatly. Stochastic road has some similar effects, and
may further induce the rollover at the moment which was supposed to be safe if there is
no road excitation.
In this paper, the focus is limited to the tyre adhesive capacity and the road excitation, both of
which have explicit impact on the conclusions. However, the wheel alignment parameters, the
effect of vehicle acceleration or braking, and the yaw moment induced by tyre longitudinal
force distribution are not considered, since the action mechanism of these factors is implicit
and intricate. A more comprehensive research is needed to make a thorough investigation on
the contradictions of vehicle stabilities, and these factors will be taken into account in the next
work, which is going to focus on the proposition of the multi-objective coordinated control
system for the different objectives of vehicle stability performances.
Acknowledgements
This paper is supported by National Natural Science Foundation of China (Grant No. 51275541 and 51005256),
National Natural Science Foundation of Chongqing (Grant No. cstc2013jjB70001) and Foundation of state key
laboratory of Mechanical Transmission (Grant No. 0301002109165).
References
[1] Available at http://www.safercar.gov/Vehicle+Shoppers/Rollover/Fatalities
[2] G. Forkenbrock, W. Garrott, M. Heitz, and B. OHarra, An experimental examination of J-turn and fishhook
maneuvers that may induce on-road, untripped, light vehicle rollover, SAE Technical Paper 2003-01-1008,
2003, doi:10.4271/2003-01-1008.
[3] T. Shim and P.C. Velusamy, Influence of suspension properties on vehicle roll stability, SAE Technical Paper
2006-01-1950, 2006, doi:10.4271/2006-01-1950.
[4] T. Shim and P.C. Velusamy, Improvement of vehicle roll stability by varying suspension properties, Veh. Syst.
Dyn. 49 (2010), pp. 129152.
[5] S.B. Lu, Y.N. Li, and S.B. Choi, Contribution of chassis key subsystems to rollover stability control, Proc. Inst.
Mech. Eng. D, J. Automob. Eng. 226 (2012), pp. 479493.
[6] S. Yim, Design of a robust controller for rollover prevention with active suspension and differential braking, J.
Mech. Sci. Technol. 26 (2012), pp. 213222.
[7] S. Yim, K. Jeon, and K. Yi, An investigation into vehicle rollover prevention by coordinated control of active
anti-roll bar and electronic stability program, Int. J. Control Autom. Syst. 10 (2012), pp. 275287.
[8] K. Nam, H. Fujimoto, and Y. Hori, Lateral stability control of in-wheel-motor-driven electric vehicles based on
sideslip angle estimation using lateral tire force sensors, IEEE Trans. Veh. Technol. 61 (2012), pp. 19721985.
[9] S. Han and K. Huh, Monitoring system design for lateral vehicle motion, IEEE Trans. Veh. Technol. 60 (2011),
pp. 13941403.
[10] R. Rajamani, D. Piyabongkarn, V. Tsourapas, and J.Y. Lew, Parameter and state estimation in vehicle roll
dynamics, IEEE Trans. Intell. Transp. Syst. 12 (2011), pp. 15581567.
[11] J. Song, Integrated control of brake pressure and rear-wheel steering to improve lateral stability with fuzzy
logic, Int. J. Automot. Technol. 13 (2012), pp. 563570.
[12] S. Sadri and C.Q. Wu, Lateral stability analysis of on-road vehicles using Lyapunovs direct method, Intelligent
Vehicles Symposium (IV), IEEE 2012, pp. 821826.
[13] S. Sadri and C. Wu, Stability analysis of a nonlinear vehicle model in plane motion using the concept of Lyapunov
exponents, Veh. Syst. Dyn. 51 (2013), pp. 119.
[14] K. Nam, S. Oh, H. Fujimoto, and Y. Hori, Direct roll moment control for electric vehicles based on roll angle
observer and lateral tire force control, 2011 IEEE 8th International Conference on Power Electronics and ECCE
Asia (ICPE & ECCE), The Shilla Jeju, Korea, 2011, pp. 26812686.
[15] H. Du, N. Zhang, and F. Naghdy, Velocity-dependent robust control for improving vehicle lateral dynamics,
Transp. Res. C, Emerg. Technol. 19 (2011), pp. 454468.
[16] S.B. Lu, Y.N. Li, and S.B. Choi, Contribution of chassis key subsystems to rollover stability control, Proc. Inst.
Mech. Eng. D, J. Automob. Eng. 226 (2012), pp. 479493.
[17] S.B. Lu, S.B. Choi, Y.N. Li, M.S. Seong, and J.S. Han, Global integrated control of vehicle suspension and
chassis key subsystems, Proc. Inst. Mech. Eng. D, J. Automob. Eng. 224 (2010), pp. 423441.
1240
Y. Li et al.
[18] S.B. Lu, Y.N. Li, S.B. Choi, L. Zheng, and M.S. Seong, Integrated control on MR vehicle suspension system
associated with braking and steering control, Veh. Syst. Dyn. 49 (2011), pp. 361380.
[19] H.B. Pacejka, Tyre and Vehicle Dynamics, SAE International, Waltham, 2012.
[20] N. Zhang, G.M. Dong, and H.P. Du, Investigation into untripped rollover of light vehicles in the modified
fishhook and the sine maneuvers, part I: Vehicle modelling, roll and yaw instability, Veh. Syst. Dyn. 46 (2008),
pp. 271293.
[21] C. March and T. Shim, Integrated control of suspension and front steering to enhance vehicle handling, Proc.
Inst. Mech. Eng. D, J. Automob. Eng. 221 (2007), pp. 377391.
[22] N.Yagiz and Y. Hacioglu, Backstepping control of a vehicle with active suspensions, Control Eng. Pract. 16
(2008), pp. 14571467.
[23] H.J. Kim, Robust roll motion control of a vehicle using integrated control strategy, Control Eng. Pract. 19 (2011),
pp. 820827.
[24] L. Segel, Theoretical prediction and experimental substantiation of the response of the automobile to steering
control, Inst. Mech. Eng., Proc. Automob. Div. 7 (1956), pp. 310330.
[25] Available at http://www.nhtsa.gov/Research/Vehicle+Dynamic+Rollover+Propensity
[26] A. Hac, Rollover stability index including effects of suspension design, SAE Technical Paper 2002-01-0965,
2002, doi:10.4271/2002-01-0965.
[27] Z.L. Jin, J.S. Weng, and H.Y. Hu, Rollover stability of a vehicle during critical driving manoeuvres, Proc. Inst.
Mech. Eng. D, J. Automob. Eng. 221 (2009), pp. 10411049.
[28] A. Hac, T. Brown, and J. Martens, Detection of vehicle rollover, SAE Technical Paper 2004-01-1757, 2004,
doi:10.4271/2004-01-1757.
[29] M.D. Dorohoff, A study of vehicle response asymmetries during severe driving maneuvers, M.E.diss.,
Department of Mechanical Engineering, The Ohio State University, 2003.
[30] R.V. Dukkipati, P. Jian, S. QatuMohamad, S. Gang, and S.G. Zuo, Road Vehicle Dynamics, SAE International,
Waltham, 2008.
[31] S. Solmaz, Switched stable control design methodology applied to vehicle rollover prevention based on switched
suspension settings, IET Control Theory Appl. 5 (2011), pp. 11041112.
[32] H. Nozaki, Effect of the front and rear weight distribution ratio of a formula car during maximum-speed
cornering on a circuit, Int. J. Automot. Technol. 9 (2008), pp. 307315.
[33] H. Ogura and T. Murakami, Improvement of vehicle stability by reaction force control on accelerator pedal and
steering wheel, Power Electronics Conference (IPEC), Sapporo, Japan, 2010.
[34] R.S. Vieira, L C. Nicolazzi, and N. Roqueiro, Four-wheel vehicle kinematic and geometric constraints for
definition of tire slip angle, Int. J. Automot. Technol. 13 (2012), pp. 553562.
[35] G.M. Dong, N. Zhang, and H.P. Du, Investigation into untripped rollover of light vehicles in the modified
fishhook and the sine maneuvers, part II: Effects of vehicle inertia property, suspension and tyre characteristics,
Veh. Syst. Dyn. 49 (2011), pp. 949968.
1241
Value
Unit
Expression
a
b
m
n
h
d
ms
mv
mui
Iz
1.178
1.464
1.458
1.455
0.506
0.256
1403.3
1559.2
m
m
m
m
m
m
kg
kg
kg
kg m2
Iy
Ix
Fyi
Fzis
Fsi
ksi
kti
csi
1
2
1
z
zui
zsi
zgi
ay
vy
vx
f0
g
T1
T2
CG
i
2131
522
2340
20,000
200,000
1500
54
0.01
10
0.25
3
1, 2, 3, 4
kg m2
kg m2
N
N
N
N/m
N/m
N s/m
radian
radian
radian
radian
radian
radian
m
m
m
m
m/s2
m/s
km/h
Hz
m/s2
s
s