IEEE TRANSACTIONS ON TRANSPORTATION ELECTRIFICATION, VOL. 1, NO.

3, OCTOBER 2015 245

Comparative Study of Interior Permanent Magnet,

Induction, and Switched Reluctance Motor Drives
for EV and HEV Applications
Zhi Yang, Fei Shang, Student Member, IEEE, Ian P. Brown, Member, IEEE,
and Mahesh Krishnamurthy, Senior Member, IEEE

AbstractWith rapid electrification of transportation, it is rare-earth material used in permanent magnet. Other com-
becoming increasingly important to have a comprehensive under- monly used motor topologies include induction motor (IM) and
standing of criteria used in motor selection. This paper presents switched reluctance motor (SRM). An IM has no magnet and
the design and comparative evaluation for an interior permanent
magnet synchronous motor (IPMSM) with distributed winding is characterized as robust. The limitation of this topology may
and concentrated winding, induction motor (IM), and switched lie in the cooling system since heat is generated both in rotor
reluctance motor (SRM) for an electric vehicle (EV) or hybrid and stator side. SRM does not depend on permanent magnets
electric vehicle (HEV) application. A fast finite element analysis and is exceptionally robust, making it suitable for harsh envi-
(FEA) modeling approach is addressed for IM design. To account ronments and fault-tolerant operation. However, high-acoustic
for highly nonlinear motor parameters and achieve high motor
efficiency, optimal current trajectories are obtained by extensive noise and low-power factor have been some of the major chal-
mapping for IPMSMs and IM. Optimal turn-ON and turn-OFF lenges. In addition, SRM drives may need customized inverter
angles with current chopping control and angular position con- and a higher number of power cables to enable independent
trol are found for SRM. Additional comparison including noise phase winding on the stator.
vibration and harshness (NVH) is also highlighted. Simulation and Comparisons have been proposed to indicate the merit of
analytical results show that each motor topology demonstrates its
own unique characteristic for EVs/HEVs. Each motors highest each motor topology for EV and HEV application. Researchers
efficiency region is located at different torque-speed regions for in [5] and [6] have presented efficiency maps of the IM, SRM,
the criteria defined. Stator geometry, pole/slot combination, and and PMSM with a general trend. In [7], the authors have ana-
control strategy differentiate NVH performance. lyzed second generation Prius-IPMSM and assessed alternative
Index TermsComparative study, electric vehicle (EV) and spoke-type IPMSM and IM. However, the analysis based on
hybrid electric vehicle (HEV), induction motor (IM), noise vibra- fixed phase current angle control over the full torque-speed
tion and harshness (NVH), permanent magnet motor, switched range may result in nonoptimal operating control. Pellegrino
reluctance motor (SRM).
et al. have presented a comparison of IPMSM and IM in terms
of output power and efficiency over the standard New European
I. I NTRODUCTION
Driving Cycle [8]. Similar work has been done in [9]. In [10],

D UE TO increased fuel efficiency and lower cost/mile fea-

ture, electric vehicles (EV) and hybrid electric vehicles
(HEV) have received increasing attention. To meet this demand,
authors have claimed that impact of efficiency distribution on
fuel consumption is small after comparing IM, PMSM, and
SRM. However, in literature, very few papers have provided a
EV and HEV motors, which form the core energy conversion comprehensive side-by-side comparison among different trac-
components, should not only satisfy specific requirements in tion motor topologies for EV or HEV applications, especially
performance and efficiency but also vibration, cost, etc. [1][4]. for noise vibration and harshness (NVH) point of view. This
Permanent magnet synchronous motors (PMSM) have dom- paper presents a comparative evaluation of the selected motors
inated the traction motor market for EV/HEV application topologies including IPMSM, IM, and SRM. Motor efficiency
recently. They can be designed to operate over wide torque- and NVH are the main focus of the comparison among all
speed range with superior torque density and power density. the candidates. A fast finite element analysis (FEA) modeling
The limitations of this topology are cost and availability of approach is adopted for IM over full torque-speed range con-
Manuscript received January 30, 2015; revised July 30, 2015; accepted sidering saturation effect and optimal current trajectory control
August 02, 2015. Date of publication August 19, 2015; date of current version is taking into account for highly nonlinear motor parameters.
October 15, 2015. This work was supported by the U.S. National Science In order to address the above-mentioned issues, it becomes
Foundation under Grant 1140772. necessary to design and conduct a thorough comparison and
Z. Yang is with GE Healthcare, Florence, SC 29501 USA (e-mail:
zyang22@hawk.iit.edu). assessment of candidate motor topologies, including IPMSM
F. Shang, I. P. Brown, and M. Krishnamurthy are with the Department of with distributed winding, IPMSM with concentrated winding,
Electrical and Computer Engineering, Illinois Institute of Technology, Chicago, IM, and SRM. However, it is impractical to build prototypes
IL 60616 USA (e-mail: kmahesh@ece.iit.edu).
for all the competitors and compared with experimental results.
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. Moreover, all the design and evaluation results are acquired
Digital Object Identifier 10.1109/TTE.2015.2470092 from FEA in ANSYS environment including RMxprt (magnetic

2332-7782 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
246 IEEE TRANSACTIONS ON TRANSPORTATION ELECTRIFICATION, VOL. 1, NO. 3, OCTOBER 2015

Fig. 2. Target torque-speed envelop for comparison.

climbing hills, whereas the available torque at maximum speed

(point B) limits the vehicle speed at the highways. Transient
overload capability of the motor is limited by the inverter [8].
In this study, the 48/8 IPMSM used in the second generation
of Toyota Prius [14] was selected as the baseline motor. The
Fig. 1. Cross-section view of: (a) 48/8 IPMSM; (b) 12/8 IPMSM; (c) 48/36 IM; torque-speed envelope shown in Fig. 2 can be acquired from
and (d) 12/8 SRM. [15] where peak torque is 300 Nm up to base speed of 1500 rpm
and high torque of 60 Nm achieved at maximum speed of
equivalent circuit design), Maxwell (static and transient FEA 6000 rpm. Moreover, the maximum dc-link voltage for Prius
design), and Workbench (mechanical vibration analysis), which is 500 V. The other three motor candidates are designed under
can be treated as consistent and is a widely accepted method the same specification and requirements; thus, it generates a
by industrial [8]. This paper is organized into several sections. reasonable and relatively fair comparison between different
Section II identifies motor design inputs and control criteria. topologies. Other geometries such as stack length are optimized
In Section III, design results are compared. Section IV evalu- to meet this requirement.
ates different motor candidates in terms of efficiency and NVH. Special control strategy of different motor topologies should
Conclusion is finally made in Section V. be optimized to achieve a high efficiency. Motor control
relates to one fundamental question: what kind of current
should be applied to the winding to simultaneously satisfy the
II. M OTOR D ESIGN AND C ONTROL C RITERIA performance and efficiency requirement. Since traction motors
Four typical traction motor topologies have been selected for electric vehicle application operate in highly nonlinear
in this comparison, including an IPMSM with 48-slot 8-pole conditions [16], such as saturation and cross coupling, using
(referred to as 48/8 IPMSM), IPMSM with 12-slot 8-pole lookup table [17] may be the optimal solution. Building the
(referred to as 12/8 IPMSM), IM with 48-slot, 36-rotor bar lookup table to find the optimal current trajectories involves
(referred to as 48/36 IM), and a 12/8 SRM with cross-section several steps as following. The 48/8 IPMSM is used for
view shown in Fig. 1. Transient 2-D analysis is used for explaining the method.
achieving the performance curves of 48/8 IPMSM and 12/8 Step 1) Injecting currents into the winding, motor parame-
IPMSM while the 48/36 IM uses static 2-D FEA analysis. ters need to be extracted, especially the flux linkage.
For 12/8 SRM, transient 2-D analysis in Maxwell is used with Figs. 3 and 4 show Prius motors d- and q-axes flux
circuit-field coupling method. Vibration analysis and results are linkage at different current levels, respectively.
acquired through Workbench 3-D analysis. For a fair compari- Step 2) Based on the flux linkage information, optimal oper-
son, all the motors are designed to share the same outer stator ating plane needs to be generated, as shown in Fig. 5.
diameter, steel lamination. This plane is bounded by current limit circle, max-
Designing specific motor for EV and HEV application, imum torque per ampere (MTPA) curve, and max-
there are many considerations such as torque-speed profile, imum torque per voltage (MTPV) curve. Constant
torque ripple requirements, inverter output power capability, torque loci (black curves) and voltage ellipse (blue
dc-link voltage variations, total weight, and cost [8], [11]. The dotted curves) have also been shown.
required peak and continuous torque-speed envelope should be Step 3) For each given torque-speed requirement, optimal
defined according to different driving cycle requirements such current id and iq can be determined by using extrapo-
as UDDS and US06 with respect to mechanical constraints of lation and interpolation techniques. Fig. 6 shows the
the vehicle including vehicle mass, wheel inertia, gearbox ratio, unique combination of reference current id and iq
and efficiency. The detailed procedure can be found in [12] and for each torque-flux, i.e., torque-speed requirements.
[13]. As shown in Fig. 2, the maximum torque at base speed The step size for torque command is 15 Nm and for
(point A) determines the vehicle performance at starting or flux command is 0.007 Wb.
YANG et al.: COMPARATIVE STUDY OF INTERIOR PM, IM, AND SRM DRIVES 247

(a)

Fig. 3. Calculated d-axis flux linkage at different current levels.

(b)

Fig. 6. (a) Reference current iq versus torque and flux command. (b) Reference
current id versus torque and flux command.

TABLE I
S PECIFICATIONS OF M OTOR T OPOLOGIES

Fig. 4. Calculated q-axis flux linkage at different current levels.

Fig. 5. Optimal operating plane for the 48/8 IPMSM.

Estimated value.

III. D ESIGN R ESULTS AND C OMPARISON SRM is extended to 105 and 108 mm, respectively. It should be
Table I summarizes the specification of four candidate noted that the rated operating point of 48/8 IPMSM is selected
topologies. The outer diameter of stator is kept to be the same as by estimating the experimental results from [15] but it may
269 mm. The maximum dc-link voltage for all motors is 500 V. not be accurate because the data were protected by intellectual
To meet the torque requirement, the stack length of IM and property.
248 IEEE TRANSACTIONS ON TRANSPORTATION ELECTRIFICATION, VOL. 1, NO. 3, OCTOBER 2015

Fig. 8. Field orientation in rotor flux reference frame and slot information.
Fig. 7. Optimal operating plane for the 12/8 IPMSM.

The first part is electrometric torque and the second part is

A. 48/8 IPMSM
reluctance torque. For IPMSM with fractional slot concentrated
Prius 48/8 IPMSM has several unique design features. winding such as 12/8 IPMSM, it exhibits low saliency which
1) Stator teeth are deep and wide so as to avoid saturation means Lq /Ld is low so that it is easier to saturate at high current
and increase saliency. region. Therefore, reluctance torque cannot be fully utilized and
2) Permanent magnets are arranged in V shape optimally to electromagnetic torque dominates. In contrast, IPMSM with
take advantage of reluctance torque and reduce no load integer slot distributed winding, such as the 48/8 IPMSM used
iron loss. in Prius, has a higher saliency and utilizes reluctance torque
3) Prius motors dominant vibration mode order (which is more effectively as shown in Fig. 5.
equal to the greatest common divisor (GCD) between the
number of slots and the number of poles) is pretty high
so that the resonance with stators low mode orders is C. 48/36 IM
successively avoided. Compared to the 48/8 IPMSM, 48/36 IM has longer stack
length (108 mm) to achieve large torque and high-efficiency
design. FEA transient analysis of induction machine requires
B. 12/8 IPMSM
significant computation time, especially when the excitation is
The designed 12/8 IPMSM has the same stack length as provided by a voltage source. The current needs several peri-
that of Prius motor. Following points are addressed during the ods to reach steady state. To reduce computation time, method
design process. presented in [19] has been adopted as reference. Instead of
1) Nominal operating point was chosen to be 120 Nm at applying voltage in stator winding while keeping rotor copper
2500 rpm. Flux density in the stator teeth and yoke bar short-circuited, both stator and rotor currents are injected
was designed to be lower than the knee point of BH into stator winding and rotor copper bar individually. In the
curve of lamination. This would help avoid saturation at rotor flux reference frame, when the field-oriented condition
overload torque. Current density is kept around 6 A/mm2 (FOC) is satisfied, which means only the d-axis rotor flux exists
for nominal operating point.


while q-axis rotor flux is zero, i.e., r = rd as shown in
2) More permanent magnets were utilized to increase elec- Fig. 8. In other words,
tromagnetic torque component and PMs were arranged in
V shape optimally to minimize torque ripple. LM
irq = isq . (2)
3) Stator slot geometry was optimized to lower overall loss Lr
by balancing copper and iron losses.
4) Number of turns per coil was optimized to meet peak The fast FEA modeling approach based on current excitation
torque-speed envelope requirement. consists of two parts.
Fig. 7 shows the optimal operating plane for the 12/8 1) Magnetostatic FEA Modeling: Both stator and rotor cur-
IPMSM. Its characteristic current is about 100 A, which is close rents are injected into the stator winding and rotor copper
to Prius motors 105 A. For 12/8 IPMSM, MTPA line shifts bar separately. The q-axis currents are adjusted iteratively
away into iq > id region, as shown in Fig. 7, which means until FOC is satisfied as shown in (2). Machine parameters
that the electrometric torque dominates. The torque equation including stator inductance Ls , rotor inductance Lr , and mutual
of IPMSM is given by [18] inductance Lm can be calculated based on current and flux
information. Slip speed related to stator currents (isq /isd ) and
mP rotor time constant (Lr /Rr ) can also be obtained. Obviously,
Tp = (d,P M iq + (Ld Lq )id iq ) . (1)
2 2 the calculation takes saturation effect into account.
YANG et al.: COMPARATIVE STUDY OF INTERIOR PM, IM, AND SRM DRIVES 249

Fig. 9. Optimal operating plane for the 48/36 IM.

Fig. 10. Flux linkage and static torque profile at different rotor position.

2) Transient FEA Modeling: Both the stator and rotor are

kept stationary. Stator currents with frequency fe are injected designed to be lower than the knee point of BH curve
into the stator winding. Rotor currents with frequency fe are of lamination. Stack length was increased to 105 mm to
also injected into rotor copper bar with amplitude indicated in meet torque requirement.
(1). Inductances (Ls , Lr , Lm ) varying with currents have been 2) Number of turns per coil was optimized to meet torque-
calculated in magnetostatic FEA. speed envelope requirement such as peak torque up to
Several iterations are needed to extract IMs parameters as base speed (1500 rpm) and high toque at maximum speed
described in magnetostatic FEA. However, this fast FEA mod- (6000 rpm) [24].
eling method saves significant computation time in predicting 3) The turn-ON angle, turn-OFF angle, and current ampli-
IMs performance over the entire torque-speed range. tude can be used to optimize SRM efficiency over full
At the maximum speed of 6000 rpm, the magnetizing reac- torque-speed range [24], [25]. To keep it simple, the
tance Xm >> Rs and stator reactance Xls >> Rs , by neglect- control strategy was designed as follows. At low speed,
ing the stator resistance Rs , the induction machines pull-out current chopping control was adopted with fixed turn-ON
torque can be approximately expressed as [20] angle and fixed dwell angle; at medium speed, current
3P V2 chopping control is adopted with fixed dwell angle and
Tp =  variable turn-ON angle; at further high speed, angular
22 2 + (X 2
(RTH + RTH TH + Xlr ) ) position control (single pulse operation) with advanc-
(3)
3P V2 ing turn-ON angle and fixed turn-OFF angle is adopted.
Several iterations and sweeps were performed in the
2 2 2 (Lls + Llr )
RMxprt and Maxwell environment to find the maximum
where V is the supply voltage, Lls is the stator leakage induc- efficiency for each torque-speed point. Fig. 10 shows the
tance, Llr is the rotor leakage inductance, and P is the number flux linkage and static torque profile at different rotor
of poles which equals 4. positions.
According to (3), as the speed approaching the maximum
speed, the pull-out torque will be inversely proportional to the
IV. C OMPARATIVE E VALUATION
square of the speed. Thus, to improve torque performance at
high-speed, rotor slot is designed to be wider and shallower and A. Efficiency Map
number of turns per coil is decreased reasonably [21][23]. This The efficiency maps are presented in Fig. 11. The efficiency
results in reduced rotor leakage inductance and enhanced pull- calculation is defined as
out torque. Optimized rotor slot shape and dimension are shown
Pout
in Fig. 8. Rotor slot height, bottom width, top width, and rotor = (4)
slot open width are 18.5, 6.4, 8, and 3.9 mm, respectively. Pout + Ploss
Fig. 9 shows the optimal operating plane for the 48/36 IM. where Ploss include iron loss, copper loss, and permanent mag-
Optimal stator current trajectories can be obtained following net (PM) losses but exclude frictional or mechanical loss. Iron
the procedure for IPMSM. core loss can be calculated by means of Steinmetz equation [26]
Ploss (W/kg)=Kh (fe )B 2 fe +Kc (fe )B 2 fe 2 +Ke (fe )B 1.5 fe 1.5
D. 12/8 SRM (5)
Design optimization is focused on the following parts. where Kh , Kc , and Ke represent frequency-dependent coeffi-
1) Nominal operating point was chosen to be 100 Nm at cients of hysteresis, eddy current, and excess losses. B is the
3000 rpm. Flux density in stator teeth and yoke was amplitude of flux density and fe is the electrical frequency.
250 IEEE TRANSACTIONS ON TRANSPORTATION ELECTRIFICATION, VOL. 1, NO. 3, OCTOBER 2015

To calculate copper loss, it is assumed that stator windings

are set at 100 C and rotor bar of IM is set at 140 C. The
PM temperature is assumed to be lower than 150 C where
there is little irreversible demagnetization effect of the perma-
nent magnet [8]. The experiment results of 48/8 IPMSM in [15]
also prove this assumption. Additional thermal design would
be necessary to maintain the PM temperature. To maximize the
efficiency, the control strategy is carefully chosen as described
in Section II.
For IPMSM and IM, efficiency calculation is done by tran-
sient 2-D-FEA analysis for every 10 Nm and 500 rpm. For
SRM, circuit-field coupling method [27] is used to perform
current chopping control and angular position control.
Efficiency maps show the region where the highest effi-
ciency locates at. For 48/8 IPMSM, the highest efficiency is
97% and locates around 2000 rpm and between 20 and 50
Nm. As stated in [28], the peak torque-speed envelop shows
a good consistency with the result provided in [15] with error
ranging from approximate 5% to 10% between simulated val-
ues and the experimental results at different speed points. For
12/8 IPMSM, the highest efficiency locates in similar region
but with expanded area. Due to its concentrated winding struc-
ture, 12/8 IPMSM with lower copper loss has 0.5%1% higher
efficiency than that of 48/8 IPMSM in most regions. This is
favored to EV and HEV application, as most frequently oper-
ation lies between low- and medium-speed range. However, at
speeds above 5000 rpm, 12/8 IPMSM loses this advantage due
to the penalty from PM eddy current loss. For 12/8 IPMSM
topology, the dominant MMF space harmonic orders and cor-
responding per unit amplitudes are 1st (100%), 2nd (50%), 4th
(25%), 5th (20%), and 7th (14.29%).
Higher order harmonics with relatively large amplitude rotate
relative to the rotor, thereby contributing to high-eddy current
loss at high speed. Regarding 48/8 IPMSM topology, the dom-
inant MMF space harmonic orders and corresponding per unit
amplitudes are 1st (100%), 5th (5.39%), 7th (3.83%), and 11th
(9.09%).
48/36 IM offers efficiency over 96% at high-speed
(>3000 rpm) benefiting from: 1) low copper loss by decreas-
ing d-axis current; 2) low core loss due to low pole number.
It is important to keep in mind that we use fast FEA method
to model induction machine. Any rotor or stator slot harmonic
relating to the motion is not considered as the rotor is kept
stationary. Therefore, iron loss calculation is underestimated.
Below base speed and near peak torque region, efficiency is as
low as 60%. Thus, attention should be paid on how to dissipate
this loss.
Fig. 11(d) shows the efficiency map of 12/8 SRM. The
torque-speed envelop is shaped to represent power less than
60 kW only. However, SRM does have the ability to oper-
ate at higher speed by advancing turning-ON angle or con-
tinuous conduction operation. The maximum efficiency 95%
appears only after 3800 rpm in medium torque (100 Nm)
region.
Detail comparison of loss and efficiency at typical operating
Fig. 11. Efficiency map from top to bottom. (a) 48/8 IPMSM. (b) 12/8 IPMSM. points is given in Appendix I.
(c) 48/36 IM. (d) 12/8 SRM.
YANG et al.: COMPARATIVE STUDY OF INTERIOR PM, IM, AND SRM DRIVES 251

Fig. 13. Radial force density distribution and its spectrum of 48/8 IPMSM at
60 Nm 2500 rpm (MTPA point).

where Fm,n is the amplitude; m is the space harmonic order

which determines force distribution shape; n is the time har-
monic order which determine the number of times per second
the force repeats itself; e = 2fe is the rotor angular velocity
Fig. 12. Modal shapes. (a) 48/8 IPMSM. (b) 12/8 IPMSM. (c) 48/36 IM.
in electrical degree.
(d) 12/8 SRM. Fig. 13 shows the evolution of radial force density over time
and space both in one period for 48/8 IPMSM. The motor
TABLE II
operates at MTPA point (60 Nm, 2500 rpm, id = 30 A, iq =
NATURAL F REQUENCIES (H Z ) OF S TATOR
39 A). The horizontal line represents variation in time domain
at one random location, whereas the vertical line represents
variation in space domain at one random time. By applying
bidimensional fast Fourier transformation (FFT), its spectrum
in terms of space and time harmonic order can be identi-
fied which is shown on the right. The amplitude (unit: N/m2 )
is represented by color block. For better visualization, only
amplitudes larger than 2000 N/m2 are shown by color block.
Space harmonic order and frequency under load condition
are in the form as follows (Appendix II):

[m, nfe ] = [2kp p ks Ns /2, 2kp fe ],

Number with italic and underline is from Hoppes equation number without
kp = 0, 1, 2, . . . ; ks = 0, 1, 2, . . . (8)
italic and underline is from ANSYS.

B. NVH Clearly, the space harmonic order is modulated by the combi-

nation of pole and slot numbers. Negative space harmonic order
1) Modal Analysis: Fig. 12 shows the mode shapes of the means the deformation rotates in opposite direction [28].
stator calculated in ANSYS workbench environment [28], [29]. Fig. 14 shows the radial force density spectrum under load
Only circumferential modes [29] are plotted in Fig. 12. Table II conditions: 50 Nm at 1000 rpm and 50 Nm at 5000 rpm.
shows the comparison of natural frequencies obtained from For 48/8 IPMSM, radial force density [0, 0fe ] has the largest
ANSYS and analytical calculation based on Hoppes equation amplitude. However, this excitation only causes a time invariant
[28]. The two results agree well for low mode orders. deformation over stator outer periphery, thus vibration velocity
Natural frequencies (unit: Hz) of 12/8 IPMSMs stator are induced is negligible. Radial force density [m, nfe ] with m > 8
relatively low due to its thinner yoke thickness and short stack can be neglected.
length. As from a vibration point of view, higher vibration mode
2) Radial Force: Electromagnetic radial force density act- order (which is equal to space harmonic order) contributes little
ing on stator teeth causes deformation of stator yoke. Based to the stator deformation due to the fact that vibration ampli-
on Maxwell stress tensor method, it is calculated as follows tude is inversely proportional to the fourth power of mode order
[28][32]: [28]. For 48/8 IPMSM, dominant radial force components are
Br2 (, t) Bt2 (, t) [8, 2fe ] [0, 12fe ]. For 12/8 IPMSM, dominant radial force
frad (, t) = (6) components are [4, 2fe ] [0, 6fe ]. For 48/36 IM, dominant
20
radial force component is [4, 2fe ]. Component with space har-
where Br and Bt are the radial and tangential components of monic order 0 cannot be taken into account in Fig. 14(c) as the
the air gap flux density, 0 is the permeability of air, is the rotor is kept stationary in the FEA calculation. For 12/8 SRM,
angular position, and t is the time. Radial force density can dominant radial force component is [4, fe ].
also be expressed by Fourier series as shown in the following For IPMSM and IM at high speed, radial force amplitude is
equation: attenuated as the field is weakened for speed extension.

frad (, t) = Fm,n cos(m + ne t + mn ) (7) For the SRM, harmonic components are much more than the
m,n other motors, especially relating to space harmonic order 0.
252 IEEE TRANSACTIONS ON TRANSPORTATION ELECTRIFICATION, VOL. 1, NO. 3, OCTOBER 2015

Fig. 15. Stator core deformation under different load conditions. (a) 48/8
IPMSM. (b) 12/8 IPMSM. (c) 48/36 IM. (d) 12/8 SRM.

TABLE III
M AXIMUM D EFORMATION (m) OF S TATOR

Fig. 15 shows the total deformation of stator core under dif-

ferent load conditions. Under the load 50 Nm at 1000 rpm,
48/8 IPMSM has the minimum stator core deformation with
5.84e7 m, whereas 12/8 IPMSM and 12/8 SRM have the max-
imum deformation with 5.17e6 m. Under the load 50 Nm
at 5000 rpm, 12/8 SRM has the maximum deformation with
4.72e6 m due to single pulse operation. For the other three
motors, deformation is smaller than those under 50 Nm at
1000 rpm condition due to field weakening control at high
speed. Table III compares the maximum deformation of stator
under different load conditions.
Fig. 14. Radial force density spectrum under different load conditions. Left
column: 50 Nm 1000 rpm. Right column: 50 Nm 5000 rpm. (a) 48/8 IPMSM. V. C ONCLUSION
(b) 12/8 IPMSM. (c) 48/36 IM. (d) 12/8 SRM.
This paper performs the side-by-side comparison and assess-
This is due to discontinuous current operation which yields ment of four candidate motor topologies for traction application
much more nonsinusoidal flux in the air gap. in for EVs and HEVs. The topologies considered include inte-
FEA results in Fig. 14 show that the lowest space harmonic rior permanent magnet synchronous motor (IPMSM), IM, and
orders except zero for different machines are 8 (48/8 IPMSM), SRM. Comparison criteria include performance, efficiency, and
4 (12/8 IPMSM), 4 (48/36 IM), and 4 (12/8 SRM). This agrees vibration. Also, a fast FEA modeling approach has been devel-
well with analytical analysis shown in Appendix II. The conclu- oped to predict the IMs performance over the full torque-speed
sion is also valid for both low and high speeds, which indicates range and the control strategy is carefully designed to meet
that control strategy for high speed will not change the lowest performance and efficiency requirement simultaneously. NVH
space harmonic order, but only amplitude. From a vibration study including modal analysis and transient analysis is also
point of view, higher space harmonic order (which is equal to provided. Comparative evaluation indicates the following.
vibration mode order) contributes little to the stator deformation 1) 48/8 IPMSM and 12/8 IPMSM offer efficiency as high
due to the fact that vibration amplitude is inversely propor- as 97%. 12/8 IPMSM has lightly higher efficiency at
tional to the fourth power of mode order [33][35]. Therefore, low speed, but above 5000 rpm, this advantage is lose
48/8 IPMSM should have the quietest operation. It is impor- as PM eddy current losses increase by 50 times, about
tant to keep in mind that additional radial force with low space 1000 W. 48/36 IM delivers the highest efficiency 96% at
harmonic order would be introduced once the excitation mode high speed but it has the widest low efficiency region at
changes, e.g., during six step operation [28]. low speed due to copper loss. 12/8 SRM has concentrated
YANG et al.: COMPARATIVE STUDY OF INTERIOR PM, IM, AND SRM DRIVES 253

TABLE IV 1) IP M SM
C OMPARISON OF L OSS AND E FFICIENCY AT T YPICAL O PERATING P OINTS
m=vu
v = (3k1 + 1)P/d (q = b + c/d, d = even, k1 = 1, 2, . . .)
v = (6k1 + 1)P/d (q = b + c/d, d = odd, k1 = 1, 2, . . .)
u = (2k2 + 1)P/2 (k2 = 1, 2, . . .)

where q is the slot number per pole per phase, v is the

harmonic order due to stator magneto-motive-force (MMF),
and u is the harmonic order due to rotor MMF. The lowest
mode order except 0 for 48/8 IPMSM is 8. The lowest mode
order except 0 for 12/8 IPMSM is 4.
2) IM
m=vu
v = (6k1 + 1)P/2 (k1 = 1, 2 . . .)
u = k2 z2 + P/2 (k2 = 1, 2 . . .)

where z2 is the rotor slot number. Keeping pole number P

to be 4, the lowest mode number except 0 for 48/42, 48/38,
48/36 topologies are 2, 2, and 4, respectively. Comparing
with the other two topologies, 48/36 design exhibits rela-
tively lower noise and vibration due to higher mode order.
3) SRM

m = z1 /r

where z1 is the number of stator poles and r is the number of

phases. The lowest mode order except 0 for 12/8 SRM is 4.

R EFERENCES
[1] M. Zeraoulia, M. Benbouzid, and D. Diallo, Electric motor drive selec-
tion issues for HEV propulsion systems: A comparative study, IEEE
winding with low resistance but may need more current Trans. Veh. Technol., vol. 55, no. 6, pp. 17561764, Nov. 2006.
to deliver the same torque. 12/8 IPMSM has concentrated [2] I. Boldea, L. N. Tutelea, L. Parsa, and D. Dorrell, Automotive electric
winding with low resistance but may need more PM mate- propulsion systems with reduced or no permanent magnets: an overview,
IEEE Trans. Ind. Electron., vol. 61, no. 10, pp. 56965711, Oct. 2014.
rial to meet torque requirement. IM and SRM have lower [3] A. Chiba et al., Torque density and efficiency improvements of a
peak power density, i.e., 50 kW/48 kg and 50 kW/42 kg switched reluctance motor without rare-earth material for hybrid vehi-
respectively, comparing with 50 kW/30 kg for IPMSM. cles, IEEE Trans. Ind. Appl., vol. 47, no. 3, pp. 12401246, May/Jun.
2011.
2) Stator geometry, pole/slot combination, and control strat- [4] Z. Q. Zhu and C. C. Chan, Electrical machine topologies and technolo-
egy differentiate NVH performance. Higher mode order gies for electric, hybrid, and fuel cell vehicles, in Proc. IEEE Veh. Power
will generate smaller vibration. For the same 12/8 topol- Propul. Conf., 2008, pp. 16.
[5] S. S. Williamson, S. M. Lukic, and A. Emadi, Comprehensive drive train
ogy, the SRM has severe vibration deformation than that efficiency analysis of hybrid electric and fuel cell vehicles based on motor
of IPMSM, as the former operates in discontinuous cur- controller efficiency modeling, IEEE Trans. Power Electron., vol. 21,
rent mode. Research is currently underway including no. 3, pp. 730740, May 2006.
[6] T. Finken, M. Felden, and K. Hameyer, Comparison and design of dif-
optimization of the lamination material for the candidate ferent electrical machine types regarding their applicability in hybrid
motor topologies. electrical vehicles, in Proc. 18th Int. Conf. Elect. Mach., 2008, pp. 15.
[7] D. G. Dorrell, A. M. Knight, L. Evans, and M. Popescu, Analysis and
design techniques applied to hybrid vehicle drive machines -Assessment
A PPENDIX I of alternative IPM and induction motor topologies, IEEE Trans. Ind.
Electron., vol. 59, no. 10, pp. 36903699, Oct. 2012.
Table IV shows a comparison of loss and efficiency at typical [8] G. Pellegrino, A. Vagati, B. Boazzo, and P. Guglielmi, Comparison of
induction and PM synchronous motor drives for EV application including
operating points. design examples, IEEE Trans. Ind. Appl., vol. 48, no. 6, pp. 23222332,
Nov./Dec. 2012.
A PPENDIX II [9] J. Goss and M. Popescu, A comparison of an interior permanent magnet
and copper rotor induction motor in a hybrid electric vehicle application,
Stator slot number, rotor slot number, and pole number in Proc. IEEE Elect. Mach. Drives Conf. (IEMDC), May 2013, pp. 220
play an important role in NVH. Space harmonic order m or 225.
[10] H. Neudorfer and N. Wicker, Comparison of three different electric pow-
mode order of radial force density causing the vibration can be ertrains for the use in hybrid electric vehicles, in Proc. IET Conf. Power
calculated as follows. Electron. Mach. Drives, 2008, pp. 510514.
254 IEEE TRANSACTIONS ON TRANSPORTATION ELECTRIFICATION, VOL. 1, NO. 3, OCTOBER 2015

[11] J. Wang, X. Yuan, and K. Atallah, Design optimization of a surface- Zhi Yang received the B.S. degree from Huazhong
mounted permanent-magnet motor with concentrated windings for elec- University of Science and Technology, Wuhan,
tric vehicle applications, IEEE Trans. Veh. Technol., vol. 62, no. 3, China, the M.S. degree from Washington State
pp. 10531064, Mar. 2013. University, Pullman, WA, USA, and the Ph.D. degree
[12] P. Morrison, A. Binder, B. Funieru, and C. Sabirin, Drive train design for from Illinois Institute of Technology, Chicago, IL,
medium-sized zero emission electric vehicles, in Proc. 13th Eur. Conf. USA, in 2005, 2010, and 2014, respectively, all in
Power Electron. Appl. (EPE09), Sep. 810, 2009, pp. 110. electrical engineering.
[13] G. Choi and T. M. Jahns, Design of electric machines for electric vehi- He is currently a Lead Engineer with GE
cles based on driving schedules, in Proc. IEEE Int. Elect. Mach. Drives Healthcare, Florence, SC, USA. His research inter-
Conf. (IEMDC13), May 1215, 2013, pp. 5461. ests include motor control and electromagnetic device
[14] M. Olszewski, Report on Toyota Prius motor torque capability, torque development.
property, no-load back EMF, and mechanical losses, Oak Ridge Nat.
Lab., U.S. Dept. Energy, Oak Ridge, TN, USA, Tech. Rep. ORNL/TM- Fei Shang (S14) was born in Tianjin, China, in 1981.
2004/185, 2007. He received the B.S. degree in electrical engineering
[15] M. Olszewski, Evaluation of the 2004 Toyota Prius hybrid synergy drive from Beihang University (BUAA), Beijing, China,
system, Oak Ridge Nat. Lab., U.S. Dept. Energy, Oak Ridge, TN, USA, in 2003, and the M.Sc. degree in power electron-
Tech. Rep. ORNL/TM-2006/423, 2005. ics and drives from the University of Birmingham,
[16] K. M. Rahman and S. Hiti, Identification of machine parameters of a Birmingham, U.K., in 2004. He is currently pursu-
synchronous motor, IEEE Trans. Ind. Appl., vol. 41, no. 2, pp. 557565, ing the Ph.D. degree in electrical engineering at the
Mar. 2005. Illinois Institute of Technology, Chicago, IL, USA.
[17] B. Cheng and T. R. Tesch, Torque feedforward control technique for From 2005 to 2010, he was the Product
permanent-magnet synchronous motors, IEEE Trans. Ind. Electron., Development Manager with Beijing Accuenergy
vol. 57, no. 3, pp. 969974, Mar. 2010. Technology Co., Ltd., Beijing, China. From 2010 to
[18] S. T. Lee and L. M. Tolbert, Analytical method of torque calculation 2013, he joined Infineon Technologies, Beijing, China, as the Senior Engineer
for interior permanent magnet synchronous machines, in Proc. Energy in failure analysis of IGBT module. Since 2013, he has been a Research
Convers. Congr. Expo. (ECCE09), Sep. 2024, 2009, pp. 173177. Assistant with Electric Drives and Energy Conversion Laboratory, Illinois
[19] L. Alberti, N. Bianchi, and S. Bolognani, Variable-speed induction Institute of Technology. His research interests include application of wide band
machine performance computed using finite-element, IEEE Trans. Ind. gap devices, high-gain dcdc converter design, and electric motor control and
Appl., vol. 47, no. 2, pp. 789797, Mar./Apr. 2011. design.
[20] S. Jurkovic, J. C. Morgante, K. M. Rahman, and P. J. Savagian,
Electric machine design and selection for General Motors e-Assist Ian P. Brown (S00M09) received the B.S.
Light Electrification Technology, in Proc. Energy Convers. Congr. Expo. degree in engineering from Swarthmore College,
(ECCE12), Sep. 1520, 2012, pp. 906913. Swarthmore, PA, USA, in 1999, and the M.S. and
[21] A. Harson, P. H. Mellor, and D. Howe, Design considerations for induc- Ph.D. degrees in electrical engineering from the
tion machines for electric vehicle drives, in Proc. 7th Int. Conf. Elect. University of Wisconsin-Madison, Madison, WI,
Mach. Drives, 1995, pp. 1620. USA, in 2003 and 2009, respectively.
[22] Z. M. Zhao and C. C. Chan, A novel induction machine design suitable Since 2012, he has been with the Illinois Institute
for inverter-driven variable speed systems, IEEE Trans. Energy Convers., of Technology, Chicago, IL, USA. Previously, he was
vol. 15, no. 4, pp. 413420, Dec. 2000. with the Corporate Technology Center, A.O. Smith
[23] T. Wang, P. Zheng, Q. Zhang, and S. Cheng, Design characteristics of Corporation, Milwaukee, WI, USA. His research
the induction motor used for hybrid electric vehicle, IEEE Trans. Magn., interests include high-performance electrical drives
vol. 41, no. 1, pp. 505508, Jan. 2005. and the design of electric machines.
[24] K. M. Rahman, B. Fahimi, G. Suresh, A. V. Rajarathnam, and M. Ehsami,
Advantages of switched reluctance motor applications to EV and HEV: Mahesh Krishnamurthy (S02M08SM13)
Design and control issues, IEEE Trans. Ind. Appl., vol. 36, no. 1, received the M.S. degree in electrical engineering
pp. 111121, Jan./Feb. 2000. from the Missouri University of Science and
[25] A. Chiba, M. Takeno, N. Hoshi, M. Takemoto, S. Ogasawara, and Technology, Rolla, MO, USA, in 2004, and the Ph.D.
M. A. Rahman, Consideration of number of series turns in switched degree in electrical engineering from the University
reluctance traction motor competitive to HEV IPMSM, IEEE Trans. Ind. of Texas at Arlington, Arlington, TX, USA, in 2008.
Appl., vol. 48, no. 6, pp. 23332340, Nov./Dec. 2012. Currently, he is an Associate Professor of Electrical
[26] J. M. Mille, Propulsion Systems for Hybrid Vehicles. Stevenage, U.K.: Engineering and the Director of the Electric Drives
IET, 2010. and Energy Conversion Laboratory and Grainger
[27] P. Kuo-Peng, N. Sadowski, N. J. Batistela, and J. P. A. Bastos, Coupled Power Electronics and Motor Drives Laboratory,
field and circuit analysis considering the electromagnetic device motion, Illinois Institute of Technology (IIT), Chicago, IL,
IEEE Trans. Magn., vol. 36, no. 4, pp. 14581461, Jul. 2000. USA. Before joining IIT, he worked as a Design Engineer with EF
[28] Z. Yang, M. Krishnamurthy, and I. P. Brown, Electromagnetic and vibra- Technologies, Arlington, TX, USA. He has coauthored over 60 scientific arti-
tional characteristic of IPM over full torque-speed range, in Proc. IEEE cles, book chapters and technical reports and has one US patent and three
Elect. Mach. Drives Conf. (IEMDC), May 2013, pp. 295302. pending. His research interests include design, analysis, and control of power
[29] J. F. Gieras, C. Wang, and J. C. Lai, Noise of Polyphase Electric Motors. electronics, electric machines and adjustable speed drives for automotive and
Boca Raton, FL, USA: CRC, 2006. renewable energy applications.
[30] Y. S. Chen, Z. Q. Zhu, and D. Howe, Vibration of permanent magnet Dr. Krishnamurthy is a Distinguished Lecturer with the IEEE-Vehicular
brushless machines having a fractional number of slots per pole, IEEE Technology Society. He was the General Chair for the 2014 IEEE-
Trans. Magn., vol. 42, no. 10, pp. 33953397, Oct. 2006. Transportation Electrification Conference and Exposition. In the past, he
[31] M. Boesing and R. W. De Doncker, Exploring a vibration synthesis pro- has served as the Technical Program Chair for the 2011 Vehicle Power
cess for the acoustic characterization of electric drives, IEEE Trans. Ind. and Propulsion Conference and 2013 IEEE-Transportation Electrification
Appl., vol. 48, no. 1, pp. 7078, Jan. 2012. Conference. He has served as the Guest Editor or Associate Editor for several
[32] T. Kobayashi, F. Tajima, M. Ito, and S. Shibukawa, Effects of slot com- IEEE journals including the Special Section of the IEEE T RANSACTIONS
bination on acoustic noise from induction motors, IEEE Trans. Magn., ON V EHICULAR T ECHNOLOGY ON S USTAINABLE T RANSPORTATION
vol. 33, no. 2, pp. 21012104, Mar. 1997. S YSTEMS, Special Issue of the IEEE T RANSACTIONS ON P OWER
[33] C. Lin and B. Fahimi, Prediction of acoustic noise in switched reluctance E LECTRONICS ON T RANSPORTATION E LECTRIFICATION AND V EHICLE
motor drives, IEEE Trans. Energy Convers., vol. 29, no. 1, pp. 250258, S YSTEMS , and Special Issue of the IEEE J OURNAL OF E MERGING
Mar. 2014. AND S ELECTED T OPICS IN P OWER E LECTRONICS ON T RANSPORTATION
[34] M. Arata, N. Takahashi1, M. Fujita, M. Mochizuki, T. Araki, and E LECTRIFICATION. He is currently serving as the Deputy Editor-in-Chief for
T. Hanai, Noise lowering for a large variable speed range use perma- the IEEE T RANSACTIONS ON T RANSPORTATION E LECTRIFICATION and
nent magnet motor by frequency shift and structural response evaluation as the on the steering committee of IEEEs Transportation Electrification
of electromagnetic forces, J. Power Electron., vol. 12, no. 1, pp. 6774, Community. He was also the Advisor for the Formula Electric Racecar Team
Jan. 2012. at IIT, which won the prestigious Fiat Chrysler Innovation Award at the SAE
[35] K. Rahman, M. Anwar, and S. Schulz, The Voltec 4ET50 electric drive Formula Hybrid Competition. He was the recipient of the 20062007 IEEE
system, SAE Int. J. Engines, vol. 4, no 1, pp. 323337, Apr. 2011. VTS-Transportation Electronics Fellowship Award for his contributions.