This document describes a simulation study of a vehicle model with four independent electric motors and active anti-roll bars on both axles. The vehicle model is composed of sub-models for the vertical dynamics, horizontal dynamics, and tire model. Simulation results show that coordinated control of the electric drive train and active suspension components can improve the vehicle's ride, stability, and handling. A complex cascade controller using PID and fuzzy logic techniques governs the integrated system based on sensor data from the virtual vehicle model.

1. 11
Number of article: 503 Mechanical Engineering – Scientific Journal, Vol. 34, No. 1, pp. 11–18 (2016)
CODEN: MINSC5 In print: ISSN 1857–5293
Received: January 13, 2016 On line: ISSN 1857–9191
Accepted: February 29, 2016 UDC: 629.33.027–83 : 004.942
Original scientific paper
IMPROVING VEHICLE PERFORMANCE USING INDEPENDENT ELECTRIC DRIVE
AND ACTIVE ANTI-ROLL BARS
Aleksandar Zahariev1
, Igor Gjurkov2
1
Jonče Murdžeski 3/16, 7000 Bitola, Republic of Macedonia
2
”Ss. Cyril and Methodius” in Skopje, Faculty of Mechanical Engineering,
Karpos II bb, P.O. Box 464, 1001 Skopje, Republic of Macedonia
aleksandar.zahariev@gmail.com // igor.gjurkov@mf.edu.mk
A b s t r a c t: This paper presents a simulation study of a vehicle model with four independent electric motors
drive and built-in active anti-roll bars on both axles. The proposed control strategies and the coordinated action of the
drive-train and the active suspension components clearly show improvement in ride, stability and handling of the ve-
hicle. For building and simulation of the vehicle model, including the controllers, Matlab/Simulink platform was
used. The vehicle is structured and presented as a combination of several sub-models which are highly nonlinear due
to the use of a nonlinear tire model, as well as the nonlinear suspension elements. The operation of the system is gov-
erned by a complex cascade controller, using modern control techniques, such as PID and fuzzy logic. The tuning of
the controller is performed using simulation data.
Key words: electric motor drive; active anti-roll bars; ride and handling; simulation; fuzzy logic; PID
ПОДОБРУВАЊЕ НА ПЕРФОРМАНСИТЕ НА ВОЗИЛО СО НЕЗАВИСЕН ЕЛЕКТРИЧЕН ПОГОН
И АКТИВНИ ТОРЗИОНИ СТАБИЛИЗАТОРИ
А п с т р а к т: Овој труд претставува симулациско проучување на однесувањето на возило (претставено
преку модел) со независен погон на четири електромотори и активни торзиони стабилизатори во системот за
потпирање на предната и на задната оска. Претставените стратегии на управување и координираното дејство
на независниот погон и активните стабилизатори покажуваат подобрувања во комфорот, стабилноста и
управливоста на возилото. За моделирањето на возилото, како и на контролерите е користена програмата
Матлаб/Симулинк. Комплексниот модел е составен од модули. Тој е нелинеарен како резултат на соодветни-
те карактеристики на елементите во системот за потпирање и на моделите на пневматиците (модел „волшеб-
на формула“ на Пацејка). Управувањето на интегралниот систем е направено со каскаден контролер и корис-
тење на современи методи како фази логика (fuzzy logic) и ПИД управување и регулација.
Клучни зборови: електричен мотор диск; активни шипки против превртувањето; возење и управување;
симулација; фази логика; PID
INTRODUCTION
Improving vehicle ride and handling using
one or more active or adaptive systems has always
been a challenging for automotive engineers. The
purpose of this paper is to examine the potential of
two separate active vehicle systems operating in
coordinated action, in order to improve vehicle
performance in aspects such as ride, handling and
stability. Those two active systems were chosen
because of their potential to influence both hori-
zontal and vertical dynamics.
The study is presented through simulation of
a vehicle model with four electric motors inde-
pendent drive and active anti-roll bars. The pres-
ence of independent drive provides all-wheel drive,
differential steering without using conventional
differential, management of additional speed and
torque to the wheels depending on road conditions,

2. 12 A. Zahariev, I. Gjurkov
Mech. Eng. Sci. J., 34 (1), 11–18 (2016)
additional steering without changing the angle of
the steering wheel, etc.
On the other hand, the possibility of control-
ling the moment or the torsional stiffness of the
active anti-roll bars (or Active Torsion Stabilizers
– ATS) can produce and maintain minimum roll
angle of the vehicle body in curves thus improving
ride. They can also influence the handling of the
vehicle by vertical force control, which is a key
factor to tire lateral stiffness and side-slip angle.
VEHICLE MODEL
Real vehicles are exceptionally complex sys-
tems which consist of numerous components with
their own mass and inertia characteristics. For the
vehicle modeling, the number of components was
reduced to a limited number of system elements
with specific characteristics and organized in sub-
models (the vertical dynamics sub-model shown on
Figure 1).
In this case the vehicle is represented by a
dynamic model composed of three sub-models (for
both horizontal and vertical dynamics and tire
model) that are mutually related and coupled. The
model is nonlinear due to the nonlinear compo-
nents in the suspension system (springs, dampers,
anti-roll bars) and the implemented nonlinear tire
model (Pacejka’s “ Magic Tire Formula”).
Fig. 1. 3D vehicle model
All of the sub-models are integrated into
complete vehicle model with the following 14 de-
grees of freedom (dof):
 Vertical displacement of the four wheels;
 Longitudinal, lateral and vertical displace-
ment of the centre of mass;
 Roll, pitch and yaw of the vehicle body;
 Rotational motion of the four wheels.
The inputs of the model among others include
the angle of the steering wheel, vertical displace-
ment of each wheel due to road profile variation,
etc. This model allows calculation and presentation
of displacement (linear and angular) for each dof.
The main displacements of the vehicle body at the
centre of gravity are considered.
It is worth mentioning that while modelling
the vehicle model the following assumptions were
taken into account:
 The vehicle body is rigid with the mass con-
centrated at the centre of gravity;

3. Improving vehicle performance using independent electric drive and active anti-roll bars 13
Маш. инж.науч. спис., 34 (1), 11–18 (2016)
 Roll and pitch centers are at the same loca-
tions;
 Suspension geometry and wheel-lift phe-
nomena are not modeled.
The full-vehicle model structure and the in-
terconnection of the sub-models is depicted on
Figure 2.
Fig. 2. Complete vehicle model
The data used for the simulations is repre-
sentative for a small family car (B-segment car).
ACTIVE SYSTEMS MODELING
As previously mentioned, the vehicle is driv-
en by four separate electric motors. These motors
are connected directly to each wheel and are pow-
ered by a battery pack. They are governed by a
controller that operates in dependence of the meas-
ured values of various vehicle parameters such as
the accelerator command, longitudinal and lateral
acceleration, yaw rate and rotational speed of each
wheel. With accurate management of these inde-
pendent motors, desired torque and angular veloci-
ty for the wheels for different driving conditions
are achieved. The advantages of using separate
motors are rapid response, compact design, rota-
tion reversal, etc.
Fig. 3. Placement of the electric motors
The active systems, the independent electric
drive train and the two active anti-roll bars have
the possibility of regulation depending on the esti-
mated road conditions. Main areas of regulation
include delivering additional torque and/or angular
velocity for the electric motors and stiffness regu-
lation through assigning additional torsional mo-
ment values for each axle.
For further consideration, only the following
values, rules of distribution and calculation will be
used:
Total additional angular velocity for
all wheels, equally distributed left and right (see
Figure 4);
Additional electric voltage, equally
distributed left and right;
Total active moment for both torsion
bars and respective coefficient of distribution front
and rear .
Fig. 4. Additional angular velocity and electric voltage for
each wheel
(3.1)
– Total assigned additional
angular velocity for each wheel and for left and
right wheels accordingly;

4. 14 A. Zahariev, I. Gjurkov
Mech. Eng. Sci. J., 34 (1), 11–18 (2016)
– Coefficient of distribution for the total
angular velocity, left to right; for equal
distribution.
Similarly:
(3.2)
– Additional electric voltage;
– Coefficient of distribution for the total
voltage, left to right; for equal distribution;
The active anti-roll bars (active torsion stabi-
lizers, ATS) are placed on both axels. Thus they
are composed of steel rod that creates torsion –
passive part, and an activation system or active
part. It is assumed that it can introduce additional
torsional moment in both directions.
(3.3)
– Total active moment of torsion for the
two stabilizers;
– Coefficient of distribution front to rear
for the total active moment (0 – the total moment is
transferred to the rear ATS, 0.5 for equal distribution, 1
– the total moment is transferred to front ATS, see Fig-
ure 5).
Fig.5. Placement of the active anti-roll bars
ACTIVE SYSTEMS CONTROLLER
The required data which describes the current
state of vehicle operation is collected from the vir-
tual sensors in the model. The signals are pro-
cessed and control signals are then sent to the actu-
ators (drive-train and stabilizers). Operation of the
actual vehicle (complex model) is compared with a
reference model (two-track model with Pacejka’s
tire model). The initial desired angular velocity for
each wheel is calculated according to the Acker-
mann-Jeantaud’s geometry, and some adjustments
are done depending on the estimated road condi-
tions. Those additional adjustments are done by
assigning different stiffness coefficients for the
stabilizers or different angular velocities or torques
for each motor/wheel.
The additional required data (adjustments) is
calculated by complex cascade controller, using
modern control techniques such as fuzzy logics
and PID. The set-up of the controller is done by
previously gathered data from series of performed
simulations.
The fuzzy logic controller is composed of
three sub-controllers depending on the three calcu-
lated values using expressions 3.1, 3.2, and 3.3.
The next Figure 6 depicts typical membership
functions for one input, while on Figure 7 the inte-
gral fuzzy controller is shown.
Fig. 6. Typical membership functions for specific input (linear
velocity)
Fig. 7. Inputs and outputs of the fuzzy-logic controller

5. Improving vehicle performance using independent electric drive and active anti-roll bars 15
Маш. инж.науч. спис., 34 (1), 11–18 (2016)
Fig. 8. Linguistic variables for choosing in the sub-
controller
The linguistic variables of the fuzzy controller
(Figure 8) are programmed according to previously
gathered data from many performed simulation.
They are written using the operators (IF, AND,
THEN). Here is an example:
– “if the linear velocity is ten meters per sec-
ond, the lateral acceleration is one meter per sec-
ond squared, for error of one meter per second
squared, total moment is hundred seventy five, ful-
ly prescribed to the rear stabilizer”, or:
IF v10 AND a1 AND e1.0,
THAN =M175 AND k0.0
Accordingly, the rest of the linguistic
variables are created. The linguistic variables for
the first fuzzy sub-controller are shown on the
diagram on Figure 8.
The PID controllers are used to calculate the
requred voltage for each electric motor according
to the desired refference angular velocity.
This full controller determines the overall
electric voltage for the motors and the additional
torsional moment for both stabilizers based on sev-
eral inputs such as: desired vehicle speed and steer-
ing wheel angle (given by the driver), vehicle be-
havior, as well as estimated road conditions. The
full controller structure is depicted on Figure 9.
Fig. 9. Scheme of the full controller
SIMULATION RESULTS
The scheme on Figure 10 shows the structure
and the connections of the integral full simulation
model depicting the full vehicle model and the
controller. This includes the road-profile modeling
as well.
The driver’s commands for the desired vehi-
cle motion such as desired vehicle speed (pressing
of the accelerator pedal) and steering wheel angle
are taken as inputs. As previously mentioned,
modeling and simulation was performed using
Matlab/Simulink.
Fig. 10. Scheme of the full simulation model
In order to perceive the features of the vehicle
with the two active systems, firstly a simple stand-
ardized constant radius cornering maneuver is sim-
ulated and the results are compared to a passive
vehicle. It can be easily seen on Figure 11 that the

6. 16 A. Zahariev, I. Gjurkov
Mech. Eng. Sci. J., 34 (1), 11–18 (2016)
“active” vehicle maintains longer linear range and
has the ability to achieve higher lateral acceleration
before reaching the limit during steady-state cor-
nering.
Fig. 11. Lateral acceleration vs steering wheel angle
The following Figure 12, valid for the same
steady-state maneuver shows the roll angle of the
vehicle body as a function of lateral acceleration of
the centre of gravity. The ATS can easily prevent
rolling the body in desired or proposed lateral ac-
celeration range (up to around 6 ). For higher
lateral accelerations the vehicle body is deliberate-
ly allowed to slightly tilt (roll motion) in order to
give the driver information that he is driving near
the edge of the grip.
To show the difference of the actual angular
speeds, as well as the reference speeds (from the
controller) for each wheel, a simple J-turn maneu-
ver is simulated (Figure 13).
Fig. 12. Roll angle versus steering wheel angle (80 km/h)
Fig. 13. Desired and measured angular velocities for each of
the motors (wheels) during J-turn maneuver (120 km/h)
With regard to the vehicle trajectories while
performing double lane change maneuver (see Fig-
ure 14: dashed line for the passive vehicle), it is
notable that the “active” vehicle needs lees lateral
space to complete the maneuver. After reaching
steady state, both vehicles maintained the direction
of driving. The one without additional control en-
tered in the adjacent lane by about one meter fur-
ther to the left, which in real case scenario could be
potentially dangerous.
Fig. 14. Vehicle trajectories during double lane change
at 80 km/h
In order to show the potential of the system
under external weather disturbances, for example a
vehicle being subject to crosswind (side wind on
the vehicle while exiting a tunnel) simulation test
was carried out without steering intervention. It is
clear that the vehicle with the active systems reacts
with a lesser lateral deviation from the desired
straight trajectory (see Figure 16).

7. Improving vehicle performance using independent electric drive and active anti-roll bars 17
Маш. инж.науч. спис., 34 (1), 11–18 (2016)
Fig. 15. Vehicle approaching crosswind
Fig. 16. Vehicle trajectories passing crosswind at 80 km/h
Usually the real road surface is far from per-
fect and the grip for each side of the vehicle or
even for each wheel may be different. In this case a
vehicle which already drives through a curve, en-
counters wet or iced road on one side (mu-split;
see Figure 17). For a short period of time, the
wheels on the right vehicle side travel on a signifi-
cantly lower coefficient of friction road surface. In
such a situation, unpredictable yaw motion may
occur due to unequal traction and lateral forces.
The change in vehicle side-slip angle for both ve-
hicles with and with no additional control (dashed
line) is shown on Figure 18. The “active” vehicle
demonstrates superior handling in this particular
transient state of motion.
Fig. 17. Vehicle approaching wet/frozen surface
Fig. 18. Vehicle side-slip angle for a mu-split passage while
cornering at 120 km/h
Finally, to show that this system is not con-
fronting the operation of other systems (in this case
the suspension system), a test of climbing on the
sidewalk with the left or the right wheels and con-
tinued straight driving, was carried out (see Figures
19 and 20).
Fig. 19. Vehicle approaching a sidewalk
Fig. 20. Vertical displacement of the centre of gravity while
climbing a sidewalk with the wheels on one side of the vehicle
CONCLUSION
This paper presents a simulation study of a
vehicle model with four electric motors independ-
ent drive accompanied with active anti-roll bars for

8. 18 A. Zahariev, I. Gjurkov
Mech. Eng. Sci. J., 34 (1), 11–18 (2016)
each axle in the suspension system. The topic is
challenge by itself because of the modeling ap-
proach and the formulation of the control strate-
gies.
There are notable effects and benefits for the
vehicle design simplification and the vehicle dy-
namics (for example no necessity for some con-
ventional parts and assemblies, such as gearbox
and differential) with the applied concept of the
drivetrain.
By means of control of the active anti-roll
bars and the achieved partial or full reduction of
the vehicle body roll angle regardless of the road
conditions and the driving maneuver, the comfort
is highly improved. The simulation results con-
firmed the potential for additional directional con-
trol of the vehicle (control of the direction of
movement) by assigning and distribution of addi-
tional angular speed and torque in the electric mo-
tors and/or additional torsion moments in the ATS.
The study demonstrated that it is also possible to
influence the response speed with assigning differ-
ent torque to the individual electric motors.
From the simulated standardized steady-state
and transient-state maneuvers undertaken in the
study, it can be concluded that the coordinated ac-
tion by the active systems can improve vehicle
handling and stability by shortening the response
time, reducing the overshoot and increasing the
highest achievable lateral acceleration and yaw rate
by the vehicle for given road conditions.
REFERENCES
[1] Abe, M.: Vehicle Handling Dynamics. Butterworth-
Heinemann, Oxford, 2009.
[2] Alberer, D., Hjalmarsson, H., del Re, L.: Identification
for Automotive Systems. Lecture Notes in Control and
Information Sciences 418, 2012.
[3] Brown, L.: Improving Performance Using Torque
Vectoring on an Electric All-Wheel-Drive., (2013).
[4] Genta, G.: The Automotive Chassis, vol. 2., SAE, 2009.
[5] Gillespie, T. D.: Fundamentals of Vehicle Dynamics.
SAE 1999.
[6] Hartani, K., Miloud, Y., Bourahla, M., Sekour, M.:
Electronic Differential with Direct Torque Control,
2009.
[7] Ivanov, V., Augsburg, K.: Fuzzy Control for Vehicle
Propulsion System. Turk J Elec Eng & Comp Sci, Vol.
17 (2009).
[8] Obialero, E.: A Refined Vehicle Dynamic Model for
Driving Simulators. Master Thesis - Chalmers, 2013.
[9] Pacejka, H.: Tire and Vehicle Dynamics, SAE, 2013.
[10] Schäfer, M.: Computational Engineering – Introduction
to Numerical Methods. Heidelberg, Germany, 2012.
[11] Schaltz, E.: Electrical Vehicle Design and Modeling.
Aalborg University, 2011.
[12] Ѓурков, И.: Симулација на динамиката на возилата.
интерна скрипта, Машински факултет – Скопје,
2012.
[13] Захариев, А.: Подобрување на управливоста и ком-
форот на возило со независен погон на четири
електромотори и активни торзиони стабилизато-
ри преку симулациски модел. Магистерска работа,
Машински факултет – Скопје, (2015).