Control Approach Based on Equivalent Vehicle Rotational
Inertia Suitable for Motor-Dynamometer Test Bench
Emulation of Electric Vehicles
P. Fajri, R. Ahmadi, and M. Ferdowsi

Abstract – In this paper, a test bench setup consisting of a
drive motor connected to a dynamometer is modeled using
MATLAB/Simulink for the purpose of emulating an electric
vehicle drive system. Two different approaches of electric
vehicle emulation are discussed which are based on predefined
drive cycle and unpredictable driving behavior. The effect of
total vehicle inertia is considered for both scenarios and a
control scheme is developed for each case based on equivalent
vehicle rotational inertia. Simulations are carried out for each
case. Experimental test bench results are used to validate the
overall effectiveness of this method.
Index Terms- Dynamometer, electric vehicle, equivalent
vehicle rotational inertia, MATLAB/Simulink, test bench.
I. INTRODUCTION
E LECTRIC VEHICLE (EV) emulation by means of an
electric motor-dynamometer test bench is a timely
research topic that is receiving increased attention in recent
years. This method of vehicle emulation is considered to be
a low cost and flexible approach which enables vehicle
designers to evaluate and analyze power train performance
and energy consumption in the early development stages of
EV design. Additionally, it allows researchers to evaluate
and study electric drive performance and develop new
control strategies in academic environments.
The hardware of an EV test bench consists of a
dynamometer coupled to an electric motor that serves as a
drive motor (DM) of a real EV. A dynamometer is an
energy-absorbing device capable of applying a controllable
load in which the load is varied under computer control to
simulate driving scenarios. In an electric dynamometer, this
resistive load is produced by an electric motor and is applied
as torque or rotating force to resist the motion of the DM.
A properly designed test bench for EV emulation can
decrease development time and result in more thorough tests
at a cost that may be significantly less than the cost of using
traditional system test methods. However, the development
of such test platform itself can be a complex process since it
needs a real-time simulation model capable of controlling
both electric motors at the same time for different driving
scenarios. In an EV emulation test bench, the aim is not only
to follow the drive cycle profile but to control the motors in
such a way that the contribution of torque from the
dynamometer resembles the resistive forces acting on the
actual vehicle while the generated torque by the DM closely
matches the drive system torque of a real vehicle for
Poria Fajri is with the Electrical and Computer Engineering Department,
Missouri University of Science and Technology, Rolla, MO 65409 USA
(telephone: 573-612-9605; e-mail: pfkr7@mst.edu).
Reza Ahmadi is with the Electrical and Computer Engineering
Department, Missouri University of Science and Technology, Rolla, MO
65409 USA (e-mail: rahd6@mail.mst.edu).
Mehdi Ferdowsi is an Associate Professor with the Department of
Electrical and Computer Engineering, Missouri University of Science and
Technology, Rolla, MO 65409 USA (e-mail: Ferdowsi@mst.edu).
978-1-4673-4974-1/13/$31.00 ©2013 IEEE
different scenarios.
Different test benches with different emulation
capabilities have been proposed in the literature [1-5].
However, not much attention has been paid to two main
aspects. One is that in most cases the effect of vehicle inertia
is either absent or it is replaced with the inertia of the
wheels. Secondly, vehicle emulation with a real driver and
undefined driving profile has rarely been discussed and it is
always assumed that a predefined drive cycle is to be
simulated. In this paper, a control approach for controlling a
motor-dynamometer set is proposed which considers both
modes of EV emulation on a test bench and takes into
account the effect of vehicle inertia by accurate
dynamometer control.
II. VEHICLE AND TEST BENCH EQUATIONS
In order to emulate the behavior of an EV on a test bench,
knowledge of physical aspects of the vehicle and test
platform and the dynamic equation governing them should
be available. The dynamic equation of a vehicle in the
rotational context can be expressed as

  (1)
where ωw is the rotational speed of the wheels and Tv is the
vehicle torque calculated at the wheels. TR is the total
resistive torque calculated at the wheels and can be
calculated from the resistive forces acting on the vehicle by
knowing vehicle specifications [6]. Jew is the equivalent
rotational inertia of the vehicle calculated at the wheels
which is discussed in detail in [7] and implies to the amount
of flywheel inertia that has the same stored energy as a
moving vehicle with a known mass when rotating at the
same rotational speed as the vehicle’s motor. Based on this,
the equivalent rotational inertia of a vehicle with a wheel
radius of rd is given by
  (2)
where me is the equivalent mass of the vehicle and is defined
as the mass increase due to the angular moments of the
rotating components of the vehicle [8]. For a vehicle with a
mass of m and total gear ratio of G this term can be
estimated by
1 0.04 0.0025   (3)
The equivalent dynamic equation of a motor-dynamometer
test bench can be expressed by

(4)

 
1155
where ωm is the motor rotational speed, Jcoupling is the rotating torque signal to be calculated. Therefore this method cannot
inertia of the coupling, TDM and JDM are the torque and be used to simulate the behavior of an EV for an unknown
rotating inertia of the DM, and TDyno and JDyno are the torque drive cycle.
and rotating inertia of the dynamometer’s electric motor The simulation circuit used in MATLAB/Simulink to
which is responsible for applying a resistive torque. model the test bench system is shown in Fig. 1. A simulation
model of a permanent magnet synchronous motor (PMSM)
III. EV TEST BENCH SIMULATION MODEL and a DC motor have been used to model the DM and the
One of the most significant sections in the development electric dynamometer respectively. A mechanical shaft
of an EV test bench is the design of its control system which model is also used to simulate mechanical coupling between
must be capable of controlling both the DM and the the two motors. For this mode of operation, the PMSM is
dynamometer electric motor at the same time for different speed controlled and its drive receives the difference
driving scenarios. Two different modes of EV emulation are between the reference and actual speed and generates a
considered; one using a predetermined speed profile in torque signal that is fed to the motor. On the other hand, the
which resistive torque values are calculated for each step and dynamometer is torque controlled and receives its torque
the other using a driver model where the demanded speed is reference directly from the controller block and generates the
obtained through driver pedal command. desired resistive torque. The controller block receives the
speed reference from the driving schedule and converts the
A. Mode 1: Drive cycle mode vehicle linear speed into a desired rotational speed. The
This mode of operation is used for simulating the EV controller is also responsible for resistive torque calculation
behavior on a predetermined drive cycle where the speed at and it calculates the desired dynamometer resistive torque
each instance is known and given as a drive cycle. The from (5). Figs. 2-4 show the simulation results of this
required resistive torque signal can be calculated from the method for the duration of one cycle of the Urban
speed trace using test bench and vehicle dynamic equations Dynamometer Driving Schedule (UDDS). The vehicle and
at each instance. In this case, the reference for vehicle’s test bench parameters used for this simulation are presented
translational speed signal which is obtained from the drive in Table I.
cycle is translated into a required rotational speed and given
TABLE I
to the DM. On the other hand, the dynamometer resistive VEHICLE AND TEST-BENCH SPECIFICATIONS
torque can be calculated from (1) and (4) and by taking into
account that in a real vehicle, total gear ratio governs the Parameter Value
transformation of torque from the wheels to the DM
Vehicle mass (m) 400
Air Density(a) 1.22 kg/m3

(5) Aerodynamic drag coefficient (Cd) 0.19
Frontal area (Af) 1.6 m2
In this approach, it is assumed that the trace is always Rolling resistance coefficient 0.01
met. Hence, in order to accomplish this in a real test bench, Wheel radius (rd) 0.28m
the DM and the dynamometer must both be capable of Alpha 0
following the speed and torque profiles with minimum error
Overall Gear ratio (G) 2.3
and delay. Only in this case, the method allows the test
DM inertia (JDM) 0.016 kgm2
bench to accurately simulate the dynamic changes of the
vehicle for a certain driving schedule. The main drawback of Dynamometer motor inertia (JDyno) 0.016 kgm2
this method is that since the torque signal is calculated from Coupling inertia ( Jcoupling) 0.003 kgm2
the speed profile, the speed profile must always be known Equivalent Vehicle Rotational Inertia (Jew) 33.029 kgm2
and given as a drive cycle, allowing the required resistive

Fig. 1. Test bench simulation model for the drive cycle mode

1156
 It can be seen from Fig. 4 that the vehicle follows the
100
UDDS cycle with minimum errors. As expected, the
resistive torque generated by the DC motor is found to be
50 opposing the applied torque with the same deviations and
Torque (N.m)

almost the same amplitude. The maximum required torque to

0 complete this cycle is found to be around 92 N.m for this
vehicle.
-50
B. Mode 2: Driver mode
This method is useful when the behavior of the vehicle is
-100 analyzed directly from the torque command given by the
0 200 400 600 800 1000 1200 1400
Time (s) driver. In this case, the input reference signal is a varying
torque signal which is directly applied to the DM and the
Fig. 2. Simulation results of PMSM torque for the drive cycle mode aim is to produce a resistive torque such that the speed
profile of a real vehicle is obtained. For this case, the DM is
100 torque controlled and the dynamometer is speed controlled.
The desired speed command can be calculated from the
50
vehicle dynamic equation in the rotational context given
in (1). By replacing Tv with GTDM and also taking into
Torque (N.m)

account the relation between the rotational velocity of the

0
motor and tires, one can write

-50
   (6)
-100
0 200 400 600 800 1000 1200 1400
Time (s) The calculated speed is then used as a command to the
dynamometer which is operated in speed mode, resulting in
Fig. 3. Simulation results of DC motor torque for the drive cycle mode vehicle driving condition simulation. The simulation circuit
used in MATLAB/Simulink to implement this method is
shown in Fig. 5, where PMSM is torque controlled and the
2000
DC motor is speed controlled.
UDDS Reference
In order to validate and compare the results of simulating
1500 Shaft Speed this method to the results obtained from the previous
Speed (rpm)

method, the DM torque input reference is reconstructed

1000 using the same values obtained from the drive cycle method
given in Fig. 2. The simulation is carried out for one UDDS
500
cycle using the test bench model of Fig. 5 with the same
vehicle and test bench parameters as Table I. The plots for
this case are shown in Figs. 6-8.
0 From Fig. 8, it is observed that the actual shaft speed is
0 200 400 600 800 1000 1200 1400
Time (s) the same as the shaft speed profile of the UDDS cycle of
Fig.4. Reference and shaft speed simulation results for the drive cycle
Fig. 4. This implies that if the torque profile of Fig. 2 is
mode applied to this model, the resulting vehicle speed will be the

Fig. 5. Test bench simulation model for the driver mode

1157
 Ref. Torque
Torque (N.m) 100 PMSM Torque

50

-50

-100
0 200 400 600 800 1000 1200 1400
Time (s)

Fig. 6. Simulation results of reference and actual PMSM torque for the
driver mode

Fig. 9. Experimental test platform

100

Matlab/Simulink is used to perform real-time simulation

50 modeling by building the vehicle model and executing the
Torque (N.m)

files in real time using an onboard PC processor. To emulate

0 on-road operation conditions for the PMSM, speed and
torque commands are calculated from mathematical models
of vehicle dynamics, road load, and vehicle inertia effect
-50
according to the drive cycle profile or torque command
received from the pedal. The controller changes the
-100 operating point of the dynamometer and the DM
0 200 400 600 800 1000 1200 1400
Time (s) continuously and sends synchronous speed and torque
signals to the two drives at each instance allowing exact
Fig. 7. Simulation results of DC motor torque for the driver mode interaction between them.
The need of a real-time communication is satisfied by
2000 means of a Controller Area Network (CAN) bus which
assures synchronous distribution of reference speed and
1500
torque commands as well as reading back actual speed and
torque from the two drives.
Speed (rpm)

LabVIEW is used as a graphical user interface (GUI) for

1000
500
0
0 200 400 600 800 1000
Time (s)
Fig. 8. Simulation results of shaft speed for the driver mode
same as UDDS speed profile. Also, by comparing Fig. 3
with 7 it is concluded that the resistive torque produced by
the dynamometer for both methods is the same. This
indicates a duality between these two systems where one
emulates vehicle resistive forces from the speed profile input
and the other produces a similar speed profile from the same
torque reference as input.
IV. EXPERIMENTAL TEST BENCH
The experimental test platform used in this paper is
shown in Fig. 9. The architecture is based on real-time
simulation and consists of two motors on a common shaft.
The experimental arrangement comprises of a 15 kW, 6-pole
PMSM, which represents the DM and is connected to a
15 kW DC machine fed by a 4-quadrant chopper acting as
the dynamometer to emulate road load and vehicle inertia.
real time system monitoring. It allows the user to define the
mode of operation between driver and drive cycle at the start
of simulation and also determine the configuration of the
vehicle to be simulated.
1200 1400
V. EXPERIMENTAL RESULTS
To verify the overall performance of the system and
validate the effectiveness of the implemented control
approach, two experiments were carried out using the test
platform of Fig. 9. The experimental tests were performed
with two aims. First, to validate that the test bench can
indeed meet real-time performance constraints using this
approach. Secondly, to compare the experimental results
with the results obtained from simulation. Consequently,
vehicle parameters chosen in LabVIEW for both
experiments were the same as Table I. The first test was
performed in drive cycle mode and the drive cycle was
chosen to be UDDS. The second test was executed in driver
mode and the PMSM torque obtained from the first test was
used as the torque reference input instead of a driver
generated torque command. The experimental results for
both modes of operation are illustrated in Figs. 10 and 11.
Fig. 10 shows the reference and actual shaft speed for a
complete drive cycle while the test bench is operated in drive
cycle mode. It is clearly shown that the speed command is
followed throughout the driving cycle with minor deviations.

1158

Fig. 10. Experimental results of reference and actual speed for the drive
cycle mode of operation
Fig. 11. Experimental results of actual speed for the driver mode of
operation
Fig. 11 shows the resulting shaft speed measured for the
driver mode of operation. By comparing the experimental
results with their respective simulation results, it can be
concluded that the system seems to follow the UDDS speed
profile for both cases and the overall profile of resulting
shaft speed is almost the same for both operating modes.
However slight speed fluctuations are observed for the driver
mode. Most of these speed fluctuations are due to the fact
that for this mode of operation the reference torque is
extracted from the drive cycle mode, which itself contains
small oscillations.
Based on these results, the overall effectiveness of the
1159
control approach is validated and this method of simulating
EVs can be considered as a flexible and accurate approach
that can be applied for different vehicles with different
specifications using a motor-dynamometer setup.
VI. CONCLUSION
In this paper, two different modes of emulating the
behavior of EVs on a test bench comprising of a motor-
dynamometer set was investigated. A control approach was
developed for each case based on equivalent vehicle
rotational inertia. This method of electric vehicle emulation
not only takes into account all the stress imposed on the DM
due to vehicle inertia effect, but also allows EV emulation
for any standard drive cycle as well as an undefined driving
scenario. The results obtained from the simulation showed
accurate calculation of vehicle resistive forces for both cases
and the experimental results validated this method of EV
emulation on a motor-dynamometer setup.
REFERENCES
[1] R. M. Schupbach and J. C. Balda, “A versatile laboratory test bench for
developing powertrains of electric vehicles,” IEEE 55th Vehicular
Technology Conference, 2002, pp. 1666–1670.
[2] Zhao Hui, Li Cheng, and Zhang Guojiang, “Design of a Versatile Test
Bench for Hybrid Electric Vehicles,” IEEE Vehicle Power and
Propulsion Conference (VPPC), September 3-5, 2008, pp. 1-4.
[3] A. Eiraku, T. Abe, and M. Yamaoka, “An Application of Hardware in
the Loop Simulation to Hybrid Electric Vehicle,” 15th Electrical
Vehicle Symposium, Bressels-Belgium, 1998, pp. 75-82.
[4] G. Lo Bianco, G. Pede, A. Puccetti, and E. Rossi, “Vehicle Testing in
ENEA Drive-train Test Facility,” SAE Society of Automotive Engineers,
2001, pp. 101-111.
[5] P. Baracos, G. Murere, C. Rabbath, and J. Wensi, “Enabling PC-Based
HIL Simulation for Automotive Applications”, 2001 International
Electric Machines and Drives Conference, Cambridge, June 17-20,
pp. 721-729.
[6] P. Fajri, R. Ahmadi, and M. Ferdowsi, “Test Bench for Emulating
Electric Drive Vehicle System Using Equivalent Vehicle Rotational
Inertia,” to appear in Proc. Power and Energy Conference at Illinois
(PECI), Champaign, IL, Feb. 22-23, 2013.
[7] P. Fajri, R. Ahmadi, and M. Ferdowsi, “Equivalent Vehicle Rotational
Inertia Used for Electric Vehicle Test Bench Dynamic Studies,” in Proc.
38th IEEE Industrial Electronics Society Annual Conference (IECON),
Montreal, Canada, October 25-28, 2012, pp. 4115 – 4120.
[8] M. Ehsani, Y. Gao and A. Emadi, Modern Electric, Hybrid Electric, and
Fuel Cell Vehicles: Fundamentals, Theory, and Design, Second Edition,
USA: CRC Press, 2009, Chap. 2.