You are on page 1of 8

Mecatronics-REM 2012

November 21-23, 2012, Paris, France

Implementation of active Steering Systems into a


Multibody Vehicle Model by Co-Simulation
Volker Dorsch
Faculty of Mechanical Engineering
Ostfalia - University of Applied Sciences
Salzdahlumer Strae 46/48, 38302 Wolfenbuttel, Germany
Email: v.dorsch@ostfalia.de
AbstractBesides yaw rate control by braking intervention or
torque vectoring, vehicle dynamics can also be stabilized by rear
wheel steering. Alternatively the front wheel steering angle can
be adjusted.
This paper demonstrates the simulation process of combining
R
a validated multibody vehicle model with a Matlab/Simulink
controller model by co-simulation. The adjustments of the multibody model as well as the control algorithms of the active steering
systems are described.
Simulations of typical driving maneuvers show the improvements of vehicle handling by these systems.

modeled taking elastokinematic effects into account. Fig. 1


demonstrates the good representation of the camber angle.

I. I NTRODUCTION
There are many approaches to improve vehicle stability in
R
critical situations. The ESP
system [1] uses brake intervention. However, this results in a sudden velocity decrease and
thus is uncomfortable. If brake intervention is delayed too
much, the situation might get very critical.
One idea for a smoother intervention is to introduce rear
wheel steering. First pure mechanical solutions were introduced in the eighties of the last century. The rear wheel
steering angle was mechanically coupled to that of the front
wheels. Recently new active rear wheel steering concepts came
up using electrical or hydraulic actors [2].
A further idea to stabilize vehicle dynamics is to adjust
the drivers steering angle. These active front wheel steering
systems can also be found in some actual cars [2].
Both, active rear and front wheel steering systems modify
the steering angle and thus adjust the tire slip angle. The tire
side force will then change and so inuence the vehicle yaw
rate.
A model-in-the-loop approach is used to simulate active
steering systems implemented in a validated multibody vehicle
model. With this simulation model effects of both steering
systems can be studied and discussed. Also the comparison
with a brake intervention approach is possible, as there is also
a model including this system [3]. However, this paper will
focus on the active rear and front wheel steering angle control.

Fig. 1. Camber angle of the rear suspension due to vertical deection and
side force
R
For tire modeling MF-Tyre
is used. The vehicle model is
validated with test results [3]. Fig. 2 shows the accurate results
of a step steering input test as an example.
Active rear and front wheel steering are implemented by
additional rotational degrees of freedom of the corresponding
wheels. The angles of these degrees of freedom are controlled
R
by the external input of the co-simulated Matlab/Simulink
controller model.
Measured data from the multibody vehicle model are
the output into the control algorithm. These data are lateral
acceleration, yaw rate, steering angles and the velocity. The
latter which cannot directly be measured at a serial car can
also be estimated by an algorithm based on the acceleration
and the wheel speeds [4].

II. M ULTIBODY S IMULATION V EHICLE M ODEL


The multibody simulation model is performed with the
R
commercial software Simpack
. It includes the subsystems
body, powertrain, suspensions, brakes, steering system, driver
model and tires. Front and rear suspensions were thoroughly

978-1-4673-4772-3/12/$31.00 2012 IEEE

329

Fig. 2.

Yaw rate and roll angle response to a step steering input

B. Active Front Wheel Steering


The active front wheel steering concept superimposes a
steering angle increment to the drivers steering angle. Thus
the tire slip angle will be changed and the side force can be
adjusted to stabilize the vehicle.
IV. B ICYCLE M ODEL

Fig. 3.

The well known bicycle model (see Fig. 6) serves for


describing vehicle dynamics with an easy to handle linear
approach. Here it will be used to analyze the drivers request
due to the steering angle, the lateral acceleration and the
velocity of the vehicle. Additionally the bicycle model is used
to predict a rear wheel steering angle depending on the front
wheel steering angle and the velocity.

Oversteer and understeer

III. V EHICLE DYNAMICS

Oversteer and understeer are typical critical situations which


can be mitigated by yaw rate control, i.e. active steering
systems. Fig. 3 shows the situation and explains how a yaw
moment can be generated by additional side forces. These
side forces are generated by the adjusted tire slip angle which
results from the steering angle (see Fig. 4).
















   




MP

Fig. 6.

 

Fig. 4.



















Bicycle model

Linearization of angles and tire behavior and neglecting the


inuence of braking and driving forces yields to the equations
of motion for yaw and sideslip angle of the bicycle model [5]:

Tire side force versus slip angle

2
CH lH CV lV
CH lH
+ CV lV2

(1)

Jz
Jz v
CV lV
CH lH
+
V
H ,
Jz
Jz 

CV + CH
CH lH CV lV

1
(2)
=
+
mv
mv 2
CH
CV
V +
H ,
+
mv
mv
with C as cornering stiffness of the tire side force linearization, so that FS = C holds. Jz is the moment of inertia
around the vertical axis, is the steering angle. The index V
stands for front, H for rear.
= 0 and = 0)
In case of steady-state cornering (i.e.
without a rear wheel steering angle (H = 0) one yields

A. Rear Wheel Steering


Rear wheel steering here is used to improve handling
stability at cornering with high speeds. In this case mostly
higher side forces are needed and the rear wheel angle will be
parallel to that of the front wheels. In case of slow driving,
e.g. parking, rear wheel steering can also be used to decrease
the turning circle of the car. Then the steering angle will be
opposite to the front one, see Fig. 5.

stat = 1 v V 2 .

l 1 + vv2

(3)

ch

The characteristic velocity


Fig. 5.

vch =

Direction of rear wheel steering angle

330

CV CH l2
m(CH lH CV lV

(4)

can also be identied as a result of steady-state cornering tests.


This has been done with the multibody vehicle model during
the validation process [3].
The computed yaw rate of the bicycle model from (3) will
des .
be used in the control algorithm as desired yaw rate
For prediction of the rear wheel steering angle H (1) and
(2) can be used. Provided that the sideslip angle is minimized,
i.e. = = 0, the relationship between front and rear angle
can be computed [7]:
G(s) =

H (s)
V (s)
with






Tz =

"




  


 
!






1 + Tz s
(5)
1 + T1 s
CV CH lh l CV lV mv 2
=
,
CV CH lV l + CH lH mv 2

= K H
K H



"! 








!

Jz v
,
CH lH l lV mv 2


!

Jz v
T1 =
.
CV lV l + lH mv 2

Fig. 8.

Fig. 7 shows the predicted rear wheel steering angle from


(5) for the steady-state case. At low velocities, the rear wheel
steering angle has another algebraic sign than the front wheel
steering angle. This corresponds to the left case in Fig. 5. At
higher speeds both steering angles have the same algebraic
sign which is the right case in Fig. 5.
V. C ONTROL S TRATEGY
A. Rear Wheel Steering
Fig. 8 shows the control structure which has been proR
grammed in Matlab/Simulink
. The bicycle model computes
the desired yaw rate from (3). The input data needed are
the front wheel steering angle, the velocity and the lateral
acceleration. These are taken from the multibody simulation
vehicle model. As the vehicle cannot follow sudden changes
of the desired yaw rate, there is a delay element included.
Furthermore the maximum yaw rate is limited due to the
maximum lateral acceleration:
max = aymax

(6)
v

Control structure

The desired yaw rate then is compared with the measured


one. A PID controller computes the imposed increment of the
actual rear wheel steering angle. In our actual model only the p
cof f )
term is active. In case of small changes (cof f
controller action is avoided, see Fig. 9. Between cof f and cin
the controler activity will smoothly be activated.
In the next block the decision of the steering angle orientation is drawn. Table I shows the analysis of the driving
situation.
Fig. 10 explains the process of determining the angle orientation. The computed steering angle is added to the predicted
one (see Fig. 7) and serves as a kind of corrector. The resulting
rear wheel steering angle is limited to 3 degrees and delayed to
take the inertia of a real actor into account. This angle serves
as input for the multibody simulation vehicle model.
B. Active Front Wheel Steering
This control strategy works similar. However, the front
wheel steering angle is a direct input in case of open loop
maneuvers without any driver model or is the angle chosen
R
by the driver model of Simpack
. Therefore the prediction
block (see Fig. 8) is not needed.
VI. S IMULATION R ESULTS
All simulation results demonstrate the effect of active rear
wheel steering and active front wheel steering compared to



Fig. 7.

Predicted rear wheel steering angle [6]

Fig. 9.

331







Tolerance function to smoothly activate controller action

TABLE I
A NALYSIS OF THE DRIVING SITUATION
des | ||
< 0: oversteer
|
des | ||
= 0: neutral
|
des | ||
> 0: understeer
|
< 0: oversteer
Left Turn

> 0)
= 0: neutral
(

> 0: understeer

< 0: understeer
Right Turn

< 0)
= 0: neutral
(

> 0: oversteer









 



     
     

     
     






   
  

Fig. 10.

 
  
  





  
  

Decision process for rear wheel steering angle orientation

a vehicle without any control system. All simulation results


base on models established by [8].
A. Co-Simulation
R
For the simulations the MatSim Interface of Simpack
was
used. Both software packages run parallel and exchange their
data at the comunication time points. These were chosen to
each millisecond.

Fig. 11. Response to the step steering input (1 - no control, 2 - active rear
wheel steering, 3 - active front wheel steering)

B. Step Steering Input


Table II reports the parameters of this open loop maneuver
without any driver inuence.
TABLE II
S TEP S TEERING I NPUT
road surface
velocity
front wheel
steering angle

dry, = 0.9
80 km/h
steady-state lateral acceleration is 0.4 g,
steering wheel velocity is more than 200 deg/s

Fig. 11 shows the results. Both, active rear wheel and front
wheel steering, reduce the peak of the yaw rate without increasing the delay between steering input and vehicle response.
This will improve vehicle handling.
The rear wheel steering control chooses a negative steering
angle which is opposite to that of the front wheels. That will
reduce the rear tire slip angle and thus decrease the side force.
A reduction of the understeer behavior follows from that.

332

The active front wheel steering system enlarges the drivers


steering angle. Thus the understeer behavior is avoided by
generating larger front side forces.
The sideslip angle is increased by the rear wheel steering
system. This might be a lack of the control algorithm: The
actual version does only consider the yaw rate without taking
the sideslip angle into account.
C. Single Sine with Dwell
This is also an open loop test mostly used to analyze vehicle
dynamics control systems. Parameters are given in Table III,
Fig. 12 shows the drivers steering wheel input function.
TABLE III
S INGLE S INE WITH DWELL
road surface
velocity
front wheel
steering angle

Fig. 12.

dry, = 0.9
80 km/h
single sine of 0.7 Hz
with dwell of 500 ms after 3/4 of period, see Fig. 12

Single sine with dwell: steering wheel input function [8]

Fig. 13 demonstrates that the vehicle cannot be controlled


without the active systems: The yaw rate, the lateral acceleration and esp. the sideslip angle increase to extreme values
without control.
This effect disappears for both active steering systems. The
additional steering angle of rear and front wheels are opposite.
This is easy to understand: If there is e.g. understeer, the front
steering angle must increase to generate larger side forces. The
rear angle, however, must decrease to make rear side forces
smaller. Both effects will reduce understeer.
Overall rear wheel steering seems to be a bit more effective
for this maneuver.

Fig. 13. Response to the single sine with dwell steering angle input (1 - no
control, 2 - active rear wheel steering, 3 - active front wheel steering)

D. Double Lane Change


This is a closed loop maneuver where the driver model of
R
Simpack
is involved. Table IV reports the conditions, Fig.
14 shows the track of this test.
TABLE IV
D OUBLE LANE CHANGE
road surface
velocity
front wheel steering angle
driving torque

dry, = 0.9
80 km/h
closed loop, by driver model
controlled, so that v stays constant

Fig. 14.

333

Track of the double lane change maneuver [8]

Although the different tracks measured at the front axle center look quite similar (see Fig. 15), Fig. 16 and 17 demonstrate
the benet of both control systems. Without them the driver
will have problems to go straight ahead after the lane change.
The steering angle, the yaw rate, the lateral acceleration and
the sideslip angle are oscillating. Rear wheel and active front
wheel steering avoid that. The response peaks of the yaw
rate and the sideslip angle are reduced. This will increase the
driving safety.

Fig. 15. Vehicle track of the double lane change (1 - no control, 2 - active
rear wheel steering, 3 - active front wheel steering)

Fig. 17. Lateral acceleration, additional steering angle and sideslip angle
response to the double lane change (1 - no control, 2 - active rear wheel
steering, 3 - active front wheel steering)

E. Slalom
R
This also is a closed loop maneuver with the Simpack
driver model involved. Table V reports the conditions, Fig. 18
shows the track.

TABLE V
S LALOM

Fig. 16. Steering wheel angle and yaw rate response to the double lane
change (1 - no control, 2 - active rear wheel steering, 3 - active front wheel
steering)

road surface
velocity
front wheel steering angle
driving torque

Fig. 18.

334

dry, = 0.9
100 km/h
closed loop, by driver model
controlled, so that v stays constant

Track of the slalom maneuver [8]

Fig. 19 shows the different track results of the front axle


center. The amplitudes in case of no control systems tend to
be higher. Both control systems again reduce oscillations of
the yaw rate and the sideslip angle at the end of the track, see
Fig. 20 and Fig. 21. The active front wheel steering system,
however, is not capable of reducing the amplitudes of the yaw
rate and the sideslip angle. It also needs large steering angles
by the driver which is uncomfortable. The sudden changes of
the drivers steering wheel angle at 10 s and 12 s are caused
R
by a bug of Simpack
output which will change the algebraic
sign for angles larger than 180 degrees.
The rear wheel steering system is advantageous. The yaw
rate and sideslip angle amplitudes are minimized.

Fig. 19. Vehicle track of the slalom maneuver (1 - no control, 2 - active


rear wheel steering, 3 - active front wheel steering)

Fig. 21. Lateral acceleration, additional steering angle and sideslip angle
response to the slalom maneuver (1 - no control, 2 - active rear wheel steering,
3 - active front wheel steering)

F. Double Lane Change with low Friction


This is the same maneuver as desribed in Table IV and Fig.
14. However, the friction coefcient between the tires and the
road is reduced to = 0.4 which represents a slippery road.
Fig. 22 proves that the driver will not be able to keep the
vehicle on the track without control systems.

Fig. 20. Steering angle and yaw rate response to the slalom maneuver (1 no control, 2 - active rear wheel steering, 3 - active front wheel steering)

Fig. 22. Vehicle track of the double lane change with low friction (1 - no
control, 2 - active rear wheel steering, 3 - active front wheel steering)

335

Fig. 23 and Fig. 24 show that in spite of large steering


angles the driver cannot stabilize the vehicle. The large yaw
rate and sideslip angle at the end of the maneuver make
this apparent. The results show that especially in case of
low friction situations control systems like active steering are
crucial.
The abrupt changes of the drivers steering wheel angle are
R
caused by a bug of Simpack
output which will change the
algebraic sign for angles larger than 180 degrees.
Fig. 24. sideslip angle response to the double lane change with low friction
(1 - no control, 2 - active rear wheel steering, 3 - active front wheel steering)

VII. C ONCLUSION
This paper demonstrates the simulation of active steering
systems by a model-in-the-loop approach. Whereas the control
R
R
systems are programmed in Matlab/Simulink
, Simpack
is used for the validated multibody simulation model of the
vehicle. Both programs are combined by co-simulation.
The active rear wheel steering system is controlled by a
simple PID controller and takes the yaw rate of a bicycle
model as a basis. Same does the active front wheel steering
system which superimposes an additional steering angle to the
drivers one. Both systems were tested in several maneuver
simulations and offered an improvement of vehicle handling
and driving safety.
Future work should enhance the algorithms by additionally
taking the sideslip angle into account. The actors for steering
wheel angle change should be modeled more realistic and
should be included into the suspension of the multibody
vehicle model.
ACKNOWLEDGMENT
The author would like to thank Yangfang Yu for doing a lot
of excellent modeling and simulation work.
R EFERENCES

Fig. 23. Steering wheel angle, yaw rate, lateral acceleration and additional
steering angle response to the double lane change with low friction (1 - no
control, 2 - active rear wheel steering, 3 - active front wheel steering)

[1] R. Isermann (ed.), Fahrdynamik-Regelung - Modellbildung, Fahrerassistenzsysteme, Mechatronik. Vieweg&Sohn, Wiesbaden, 2006.
[2] P. Pfeffer and M. Harrer (eds.), Lenkungshandbuch. Vieweg+Teubner,
Wiesbaden, 2011.
[3] V. Dorsch and A. Holm, Implementation of electronic control systems
into multibody automotive models, in J. C. Samin, P. Fisette (eds.), Proceedings of Multibody Dynamics 2011, ECCOMAS Thematic Conference,
Brussels, Belgium, 2011.
[4] V. Dorsch, Estimation of velocity and side slip angle for a stability
control algorithm of a vehicle simulation model, in M. Gulec (ed.), Proceedings of the 12th International Workshop on Research and Education
in Mechatronics 2011, Kocaeli, Turkey, 2011.
[5] A. von Vietinghoff, Nichtlineare Regelung von Kraftfahrzeugen in querdynamisch kritischen Fahrsituationen, PhD. Thesis, University Karlsruhe,
Universitatsverlag Karlsruhe, 2008.
[6] A. Holm, Y. Yu and V. Dorsch, Simulation eines Fahrzeugs mit geregelter Hinterradlenkung, in X. Liu-Henke, R. Buchta and F. Quandtmeyer
(eds.) ASIM Mitteilung AM 140, Simulation technischer Systeme, ASIM
Workshop, Wolfenbuttel, pp. 121 - 130, 2012.
[7] C. Woernle, Fahrmechanik, unpublished lecture notes, University Rostock, Rostock, 2002.
[8] Y. Yu, Erweiterung eines Fahrzeugmodells um eine Hinterradlenkung
und eine aktive Lenkung, unpublished Masterthesis, Ostfalia University
of applied Sciences, Wolfenbuttel, 2012.

336

You might also like