Journal of Asian Electric Vehicles, Volume 13, Number 2, December 2015

Efficient Design of a High Speed PMSM Drive for Electric Vehicle Application
Using Real Time Simulator

Amin Zabihinejad 1, Philippe Viarouge 2, Simon Roy 3, and Jéröme Cros 4

1
Department of Electrical Engineering, Laval University, [email protected]
2
Department of Electrical Engineering, Laval University, [email protected]
3
Department of Electrical Engineering, Laval University, [email protected]
4
Department of Electrical Engineering, Laval University, [email protected]

Abstract
In this paper, a novel approach has been employed in order to investigate the efficient design of a vector control
drive of a high speed PMSM for automotive application using analytical method and real time simulator. Most of
vector control drives set the d-axis current to zero to decrease the magnitude of the source current and working
with unit power factor. However, the optimal d-axis current is dependent to the motor specifications. An optimal
estimation scheme has been proposed and tested in order to work with maximum efficiency based on the opera-
tion point. In the high speed application, because of high losses density and low capacity of heat dissipation, the
importance of working in maximum efficiency point will be clear more and more. Also, measurement of inverter
losses is very complicate in the analytical approach. In this paper, the precise model of the switches has been
implemented using real time simulator. Real time simulator, is a complex electronic circuit which is composed
of CPUs, FPGA, etc.–is known as an efficient tool obtains very exact model of the IGBTs in order to have the
precise losses model of the motor and drive. Both of the analytical approach and real time simulator prove the
validity of the optimal estimation scheme.

Keywords ing conditions [Monga et al., 2012].

vector control drive, PMSM, loss minimization, real In compact electric vehicles, PMSMs are usually de-
time simulator, electric vehicle signed with maximum torque per volume. Therefore,
their power density is high. This characteristic leads to
1. INTRODUCTION high losses density in the machine. The heat dissipa-
Demands to use Permanent Magnet (PM) motors are tion is a complex issue in this type of motor. Decreas-
increasing more and more because of its high perfor- ing the losses and working at optimal point to generate
mance and ideal control characteristics [Ta and Hori, minimum losses will be an effective control strategy
2001; Blaabjerg and Pedersen, 1993]. The recent for this type of application.
development of rare-earth permanent magnet (PM) In this paper, a novel loss minimization aspect has
materials, power electronic components and simula- been introduced and evaluated in order to find the
tors have contributed to new energy efficient, high optimal losses point of vector control drive and PM
performance electric drives that use modern PM syn- motor for all operating points. The controller tries to
chronous motors [Ta and Hori, 2001; Blaabjerg and minimize the overall losses from the motor and in-
Pedersen, 1993; Joorabian et al., 2009]. Regarding to verter. The researchers proposed different minimum-
the appearance of rare-earth PMs, PM motors obtain loss control strategies previously. All of these methods
higher efficiency, power factor, power density and bet- are divided in three groups; Search method [Ta and
ter dynamic performance than other types of conven- Hori, 2001], Loss model method [Blaabjerg and Ped-
tional motors without sacrificing reliability [Blaabjerg ersen, 1993] and Power factor method [Joorabian et
and Pedersen, 1993]. The most important advantage al., 2009].
of PM motors is their ability to work in the wide con- The losses model approach has been employed in
stant power speed range [Joorabian et al., 2009] which order to estimate the losses energy in the different
causes a vast amount of attention to the PM motors operating conditions. The precise losses model of the
in a different types of automotive applications and PM motor and VSI has been derived using analytical
electric vehicles. For industrial applications such as analysis.
electric vehicles, high efficiency is very important to The validity of the controller performance has been
increase the range per battery charge under all operat- evaluated using analytical approach and real time

1733
 A. Zabihinejad et al.: Efficient Design of a High Speed PMSM Drive for Electric Vehicle Application Using Real Time Simulator

∞
simulator. Analytical approach estimates the precise Ph = K h ∑ nωs λtβ, n (3)
value of the losses, but it is very difficult to implement n =1

in the simulation and usually it passes up. Real time Similarly, for the eddy current losses, we will have the
simulators have resolved this problem. equation below:
In this research, a real time simulator has been em- ∞
ployed to verify the simulation results. Real time Pe = K e ∑ (nωn ) 2 λt2,n (4)
simulators are composed of software and hardware n =1

sections. OPAL-RT Company introduces the RT-Lab

as the software which compiles the model in order to
implement in the hardware section. It is fully integrat- 3. THE LOSSES IN THE VOLTAGE SOURCE
ed with Matlab/Simulink and compiles the generated INVERTER
model in the sympower library. OP4500 is a real time The inverter losses are often separated in conduction
hardware simulator which is used to execute the pow- losses and switching losses [Casanellas, 1994]. The
er electronic models. OP4500 utilizes the exact IGBT conduction losses (Pcon) are the losses of the IGBT or
and Diode and power system elements model using diode in the ON position.
FPGA technology. Moreover, other parts of the simu- The switching losses are the most important propor-
lation are executed by the powerful CPUs of OP4500. tion of the inverter losses. They are directly related to
The Real time simulator helps us to execute the very the switching frequency. Also, in order to calculate the
complex power electronic model in real time to evalu- precise value of the switching losses, it is necessary to
ate the dynamic setting of the controller and power consider the ON/OFF fitting curve of the switch. Most
quality of the converter [Monga et al., 2012]. This is of the previous researches employed the approxima-
a valuable step in order to design the high power con- tion method to calculate the switching losses. In this
verters which are very expensive and time consuming research, the exact model of the IGBT has been used
to implement. by the real time simulator.
The total inverter losses will be written as below
2. THE HARMONIC LOSSES MODEL IN [Casanellas, 1994; Zabihinejad and Moghani, 2008],
PMSM MOTORS fc
 n n 
One of the advantages of PMSMs is their higher ef- Pinv = 6 f s ∑  K IGBT i ( ) β IGBT + K diode i ( ) βdiode ) 
ficiency than induction motors when used in adjust- n =1  fc fc  (5)
able speed drives [Monga et al., 2012]. G. Slemon
(1990) proposed an analytical approach in order to
+6 f s {∫ V T

0 IGBT , on
T
(t )i (t )dt + ∫ Vdiode, on (t )i (t )dt
0 }
calculate the fundamental losses of PMSM [Slemon
and Liu, 1990]. In 2008 a theoretical study has been 4. DYNAMIC MATHEMATICAL MODEL OF
done to calculate the effect of PWM parameters on the PMSM LOSSES
iron losses [Ruifang, 2008]. Regarding to the copper
losses, measuring the effective value of the current is The equivalent circuit of q-axis and d-axis of the
acceptable. PMSM are shown in Figure 1 respectively. These are
∞
the most precise circuit model which are convention-
PCu = 3Rs ∑ I n2 (1) ally used to calculate the copper and iron losses in
n =1 the same platform. It is clear that the iron losses and
The core losses in the machine consist of two com- mechanical behavior of PMSM are not independent
ponents, i.e., hysteresis and eddy current losses. Both parameters. The armature current which is passed
types of core losses are due to time variation of the through Rs, is converted to electromagnetic torque (Iqm,
flux density in the core. The value of flux density is Idm) and iron losses (IqFe, IdFe).
not equal in the yoke and tooth section. In the steady-state condition, the currents of IdFe, IqFe
will be as below:
 1 ∞

 Pht = γ K ht ∑ nωs λt , n
β

 Fe n =1 (2) Rs Rs Ld
 Lq
P = 1 K
∞

 hy γ Fe hy ∑
Id m
nωs λ yβ, n Iq Iqm Id
RFe ωΨ q
Vd

RFe ωΨ d
Vq

n =1
IqFE IdFE

We consider that λy,n = mλt,n where m is dependent to

the mass proportion of tooth to yoke. Therefore, Fig. 1 q/d-axis equivalent circuit of PMSM

1734
 Journal of Asian Electric Vehicles, Volume 13, Number 2, December 2015

 −ω Lq I qm
 I dFe =
 RFe
 (6)
 I = ω Ld I dm + ωλm
 qFe RFe

where IdFe = Id – Idm, IqFe = Iq – Iqm.

Using these definitions, the power equations are:

PCu = Rs ( I d2 + I q2 ) (7)
ω LI
2 2 2
q qm (ω Ld I dm + ωλm )
2
PFe = + (8) Fig. 2 The Inverter Losses versus line current
RFe RFe
Pmotor = Pout + PCu + PFe (9)
Company, has been used. The switching losses and
Therefore, the input power versus the current is writ- conduction losses of the inverter versus the phase
ten as below: current is shown in Figure 2. The parameters of the
semiconductor switches in the IGBT pack have been
Pin = K m Ω3 + 3Rs ( I q2 + I d2 ) + K hωλ 2 + K eω 2 λ 2 (10) extracted from the data sheets of Fuji Electric module.
These values are true when the switching frequency
where, is 10 kHz, DC link voltage is 400 V and the ambient
temperature is 25 °C. Increasing the line current leads
λ 2 = Ld I dm
2 2
+ Lq I qm + Ld I dm λ f + λ f2 (11) to increase the inverter losses.
The parameters of a PMSM that has been prepared for
a Formula Student Electric Vehicle Competition are
2
RFe I q − ω Ld RFe I d + ωλ f RFe
I qm = (12) listed in Table 1. The field oriented control algorithm
2
RFe − ω 2 Ld Lq including the loss minimization algorithm is imple-
ω Lq ( RFe I q − ω Ld I d + ωλ f ) mented in MATLAB/SIMULINK and Real-Time
I dm = I d − (13) Simulator. The FOC algorithm with loss minimization
RFe2
− ω 2 Ld Lq
algorithm is shown in Figure 3. The electromagnetic

Table 1 PMSM Parameter

5. FINDING THE OPTIMAL OPERATING POINT
PMSM Parameter Value Units
In PMSM motors, the electromagnetic torque is de- Number of pole pairs 4
pendent to Iqm. Therefore, in each operating point of Stator resistance 0.0533 Ω
the motor, there is an optimum value for Idm which Core loss resistance 150 Ω
generates the minimum motor losses. In order to find Rated speed 12000 RPM
the optimum point, we need to have the total value of
Rated torque 8 Nm
power from the inverter input. The total input power
Rotor flux linkages 0.0239 Wb
will be:
Rated peak voltage per phase 400 V
PTotal = Pmotor + Pinverter (14) Ld 0.00017 H
Lq 0.00017 H
The minimum value of the total power is determined
by the equation below.
dPTotal
=0 (15)
dI d
The optimum value of Id will be determined solving
the above equation. It a nonlinear equation which
needs a numerical approach to solve.

6. SIMULATION AND RESULTS ANALYSIS Fig. 3 FOC strategy for PMSM with loss minimiza-
In this research, a 6-IGBT module, from Fuji Electric tion algorithm

1735
 A. Zabihinejad et al.: Efficient Design of a High Speed PMSM Drive for Electric Vehicle Application Using Real Time Simulator

design of the PMSM has been done using Flux-3D

software. It is possible to calculate the iron losses ver-
sus mechanical speed using electromagnetic software.
The inverter losses are directly calculated using the
SIMULINK system.
The losses of the PMSM and drive versus d-axis cur-
rent in the nominal condition are shown in Figure 4.
It is clear that optimal control is not happening in the
zero value of Id. The value of optimal Id is calculated
based on the value of speed, q-axis current and motor
parameters. Figure 5 shows the optimal Motor losses
for different value of the speed and Id. The optimal
Id has been calculated by solving the nonlinear equa- Fig. 6 Step response of the speed and torque
tions of the loss minimization algorithm which was
explained in the previous section.

Fig. 4 The PMSM and drive losses versus Id

Fig. 7 Step response of q and d-axis currents with

loss minimization algorithm

current.
Fig. 5 PMSM power losses with loss minimization
algorithm 8. SIMULATION WITH REAL TIME SIMULA-
TOR
Real-time simulator (RTS) is an effective tool of
7. STEP RESPONSE OF FOC WITH LOSS MIN- virtual prototyping used to study the dynamics of a
IMIZATION ALGORITHM physical system prior to start the actual hardware de-
The step response of the FOC algorithm with loss velopment [Zheng et al., 2009; Le-Huy et al., 2006].
minimization strategy has been investigated in Sim- Recently, RTS has been utilized by engineers in vari-
ulink/MATLAB. The step response is a suitable crite- ous industries such as aviation [Zheng et al., 2009],
rion to evaluate the performance of algorithm. power systems [Le-Huy et al., 2006], networking
Figure 6 shows the step response of the speed and [Xiaobo et al., 2009], automotive [Tavernini et al.,
electromagnetic torque. Control system keeps the 2009], traffic management [Maroto et al., 2006], and
electromagnetic torque at maximum value in transient medicine [Lerotic et al., 2009].
state. The variations of q-axis and optimal d-axis The implementation of power electronic converters
current in the step response with loss minimization is very expensive and time consuming, especially in
algorithm are shown in Figure 7. The proposed loss high power applications. In electric vehicle applica-
minimization algorithm chooses the optimal value of tions, the stability and reliability of the controller
d-axis current in transient and steady state condition. performance in the various operating conditions is an
The results prove that increasing the value of electro- important point of the design process. Real time simu-
magnetic torque increases the optimal value of d-axis lators are known as a very effective approach to verify

1736
 Journal of Asian Electric Vehicles, Volume 13, Number 2, December 2015

the model dynamics and controller settings validation.

In this research, OP4500, a real time simulator from
OPAL-RT Company, has employed to evaluate the
controller design. The RTSs often have two main
sections; software and hardware. The software is a
computer program which connects the simulation files
to the hardware. OPAL-RT introduces the RT-Lab
as a software that receives the Matlab/Simulink files
and translates them in order to load and execute. The
OPAL-RT hardware is composed of two main sec- Fig. 9 Step response of speed and Torque in RTS
tions; CPU and FPGA. The power system and power
electronic components are directly implemented in
FPGA section. The RT-LAB uses the precise model of
IGBT, diodes and other power electronic components.
The control system component is executed by the
CPU section. All simulation sectors must execute in
real time. Figure 11 shows the OP4500 and the setup
which was prepared for this research.
In order to implement the FOC drive in the real time
simulator, the components are divided in to two sec-
tions. Power electronic components are implemented
in FPGA section in order to generate fast response for
Fig. 10 Step response of q and optimal d axis currents
real time performance. PMSM machine and controller
in RTS
are implemented in the CPU section.

9. REAL TIME SIMULATION RESULTS IGBT model is used. The step response of q-axis cur-
Figure 8 shows the OP4500, the real time simulator rent and optimal d-axis current are shown in Figure 10.
which is used in this project. The OP4500 has two
independent FPGA modules supporting 24 power 10. CONCLUSION
electronic switches each. The FPGA modules execute In this paper, an effective design approach has been
the exact model of the IGBTs and diodes in real time, investigated in order to design an efficient vector con-
causing a real time simulation with precise results. trol drive of a high speed PMSM for electric vehicle
application. Losses minimization strategy for a high
speed PMSM has been simulated and evaluated us-
ing two different approaches. The concept of vector
control drive with losses minimization algorithm is
verified by the Matlab/Simulink and real time simu-
lator results. Real time simulator is an effective and
valuable approach to demonstrate the validation of the
controller setting in power electronic design, especial-
ly in high power applications [Tavernini et al., 2009;
Maroto et al., 2006; Lerotic et al., 2009]. The results
of the controller have been compared with conven-
tional drive which sets the d-axis current to zero. The
simulation results proved that the losses minimization
algorithm decreases the motor losses and increases the
Fig. 8 Implementation of the drive with OP4500 overall efficiency of the motor and drive. Implementa-
tion of high speed PMSM, especially in electric vehi-
Figure 9 shows the step response of the mechanical cle technology is very expensive and risky. Real time
speed and electromagnetic torque. The waveform of simulator helps us to evaluate the simulation results.
the voltage in RTS and Simulink are the same except The OP4500 - OPAL-RT simulator was used in this
for the high frequency spikes. RTS uses the more pre- research. The results with RTS proved the simulation
cise model for IGBTs. The oscillations which are seen results.
in the voltage waveform are visible when the exact

1737
 A. Zabihinejad et al.: Efficient Design of a High Speed PMSM Drive for Electric Vehicle Application Using Real Time Simulator

Acknowledgment Vehicle Power and Propulsion Conference, 890-

The authors would like to acknowledge the active 895, 2009.
participation and financial support of the OPAL-RT Xiaobo, X., Z. Kangfeng, Y. Yixian, and X. Guoai, A
Company and the department of electrical engineering model for real-time simulation of large-scale net-
of Laval University. works based on network processor, Proceedings of
the Broadband Network Multimedia Technology,
References 237-241, 2009.
Blaabjerg, F., and J. K. Pedersen, An integrated high Zabihinejad, A., and J. S. Moghani, Design & imple-
power factor three-phase AC-DC-AC converter for mentation of a novel vector-controlled drive by di-
AC-machines implemented in one microcontroller, rect injection of random signal, Vehicle Power and
IEEE Power Electronics Specialists Conference, Propulsion Conference, 2008.
285-292, 1993. Zheng, S., J. He, and J. Han, An optimized distributed
Casanellas, F., Losses in pwm inverters using igbts, real-time simulation framework for high fidelity
IEE Proceedings-Electric Power Applications, Vol. flight simulator research, Proceedings of Interna-
141, No. 5, 235-239, 1994. tional Conference on Information and Automation,
Joorabian, M., A. Zabihinejad, and K. Malekian, Prac- 1597-1601, 2009.
tical aspects of constructing a wind Ironless axial
flux direct-drive generator, International Review (Received July 9, 2015; accepted October 14, 2014)
of Electrical Engineering, Vol. 4, No. 2, 235-241,
2009.
Le-Huy, P., S. Guerette, L. Dessaint, and H. Le-Huy,
Dual-step real-time simulation of power electronic
converters using an FPGA, Proceedings of IEEE
International Symposium on Industrial Electronics,
1548-1553, 2006.
Lerotic, M, S. L. Lee, J. Keegan, and G.-Z. Yang, Im-
age constrained finite element modelling for real-
time surgical simulation and guidance, Proceedings
of Biomedical Imaging: From Nano to Macro,
1063-1066, 2009.
Maroto, J., E., Delso, J. Felez, and J. Cabanellas,
Real-time traffic simulation with a microscopic
model, IEEE Transactions on Intelligent Transpor-
tation Systems, Vol. 7, No. 4, 513-527, 2006.
Monga, M., M. Karkee1, S. Sun, L. K. Tondehal, B.
Steward, A. Kelkar, and J. Zambreno, Real-time
simulation of dynamic vehicle models using a high-
performance reconfigurable platform, Procedia
Computer Science, 1-10, 2012.
Ruifang, L., C. C. Mi, and D. W. Gao, Modeling of
eddy-current loss of electrical machines and trans-
formers operated by pulsewidth-modulated invert-
ers, IEEE Transactions on Magnetics, Vol. 44, No.
8, 2021-2028, 2008.
Slemon, G.R., and X. Liu, Core losses in permanent
magnet motors, IEEE Transactions on Magnetics,
Vol. 26, No. 5, 1653-1655, 1990.
Ta, C. M., and Y. Hori, Convergence improvement of
efficiency-optimization control of induction motor
drives, IEEE Transactions on Industry Applica-
tions, Vol. 37, 1746-1753, 2001.
Tavernini, M., B. Niemoeller, and P. Krein, Real-time
low-level simulation of hybrid vehicle systems for
hardware-in-the-loop applications, Proceedings of

1738