OPTIMISATION OF SWITCHED RELUCTANCE MOTORS FOR HYBRID ELECTRIC VEHICLES

W Wu, H C Lovatt, J B Dunlop

CSIRO Telecommunications & Industrial Physics, Australia

ABSTRACT
A switched reluctance motor (SRM) is well suited
to the electric propulsion of a hybrid electric vehicle
(HEV), due to its simple and rugged construction,
low cost, and ability to operate over a wide speed
range at constant power. This paper presents the
design optimisation of SRMs for HEV applications,
by using a combination of 2D electromagnetic finite
element (FE) analysis, experience from previous
designs, 3D correction factors, simple lumpedparameter thermal models, and computer search
techniques. A prototype SRM was built and fully
tested. The experimental results validate the
design methodology. Based on the optimal
lamination profile of the prototype, five motors were
constructed for two HEVs: two for a parallel HEV
and three for a series HEV. The two cars have
demonstrated excellent performance with the
optimised SRMs.
INTRODUCTION
Currently there is considerable interest in the
development of hybrid electric vehicles (HEVs)
since they offer a promising compromise solution
to fill the gap between the conventional internal
combustion engine (ICE) vehicle and the future
purely electric vehicle (EV) powered by fuel cells.
HEVs generally have an ICE and an electrical
energy storage system. The use of supplemental
electrical energy leads to a smaller ICE than that in
a conventional automobile. The basic principle of
the HEV is that the electrical energy cuts in and the
ICE switches off when the latter would otherwise
have to operate inefficiently, eg at traffic lights or at
low part-loads. The combination of the best
aspects of both power sources improves fuel
consumption with consequential reduction in
emissions of pollutants. Compared with a battery
electric vehicle, HEVs overcome range limitations
and battery recharging problems.
HEVs are usually classified into two types: parallel
and series. In a parallel HEV, the ICE may be
directly connected to the wheels and the electric
motors provide extra power (or all the power) when
required. In the series configuration, the ICE is
connected to a generator which, either separately
or jointly with the stored electrical energy source,

powers the electric motors that drive wheels. The

key difference between the two types of HEVs is
that the parallel HEV has a mechanical parallel
connection, whereas the series HEV has an
electrically parallel connection. The most notable
examples of parallel arrangements are the Toyota
Prius and Honda Insight from Japan that are
currently available in most countries around the
world.
In Australia, the national research agency, the
Commonwealth Scientific and Industrial Research
Organisation (CSIRO) has developed HEVs in
conjunction with the Australian automotive
industry. The resultant two vehicles are the parallel
hybrid GM Holden ECOmmodore and the series
hybrid aXcess. An overview of the latter vehicle
was presented in [1] and its drive train in [2]. The
ECOmmodore and aXcess cars use a number of
key CSIRO technologies that include light-metal
casting, modelling of energy usage and pollution,
an electrical storage system using a unique
combination of advanced lead-acid batteries and
super capacitors, and a novel electrical topology
using switched reluctance motors (SRMs). The
unique drive train and power transfer of multiple
energy storage units are discussed in [3]. This
paper presents the optimised high-efficiency SRMs
for the novel drive train of both vehicles.
The advantages of a SRM for HEV electric
propulsion include its simple and rugged
construction, low cost, ability to operate over a
wide speed range at constant power, and hazardfree operation, as presented in [4]. In contrast to
traditional motors such as induction motors, the
SRM is intended to operate in deep magnetic
saturation to increase the output power density.
Therefore, because of saturation effects and the
variation of magnetic reluctance, all the relevant
characteristics, i.e. flux-linkage, inductance and
torque, of the machine are highly non-linear
functions of both rotor position and phase current.
This paper discusses design optimisation by using
a combination of 2D electromagnetic finite element
(FE) analysis, experience from previous designs,
3D correction factors, simple lumped-parameter
thermal models, and computer search techniques.
After a brief description of the motor ratings for
HEVs, the performance analysis of a SRM by

using finite element analysis and thermal models is

given. Then an optimisation routine for the design
of SRMs for HEV applications is presented and
followed by the experimental results of an
optimised 25 kW prototype motor to validate the
design approach. Finally, the design details of five
motors for two HEVs, two for the ECOmmodore
and three for the aXcess, are given.
MOTOR RATINGS

PERFORMANCE ANALYSIS OF SRM

Torque, Power and Efficiency
Assuming each phase winding of the SRM is fed
by a constant current with 180 (electrical) width,
two static 2D FE analyses were conducted for the
rotor in the unaligned and aligned positions. Figure
1 shows the resulting flux plots of the SRM.
The average electromagnetic torque is given by

As described in [3], CSIRO developed a dual

electrical storage system for HEVs consisting of
lead-acid batteries and super capacitors. To
eliminate the requirement of an expensive DC/DC
converter for the two storage units with different
voltages, it was proposed to split the motor and its
controller into two units, one associated with each
storage type. In the series HEV, a third motor is
required to connect to the generator. Therefore,
there are two motors for a parallel HEV and three
motors for a series HEV. The sum of the ratings of
each of these small motors equals the rating of the
original single motor.

Tav = m

k aWca k uWcu
2 / n r

(1)

where m is the number of phases and nr the

number of rotor teeth. Wca and Wcu are the coenergies when the rotor is at the aligned and
unaligned position, respectively, which can be
calculated in every element and integrated over the
volume of the entire finite element mesh. ka and ku
are the 3D correction factors for the aligned and
unaligned co-energy, respectively, which account
for the 3D end effects including winding end-turns
and the axial field fringing [5][6]

The ratings of each motor in terms of power

delivery and speed were determined using
mathematical models of HEVs that were developed
by CSIRO. The ratings of the required SRMs for
the GM Holden ECOmmodore and the aXcess
cars are give in Table 1, where the capacitor motor
represents the motor powered by the capacitor,
battery motor powered by battery, and the main
motor is connected to the generator.
The speed ratios for constant-power operation of
the motors are less than three. This can be readily
achieved in a SRM, since an optimally designed
and controlled SRM demonstrates a speed ratio of
up to seven to one for constant power [4].
Therefore the design optimisation focuses on the
ratings at base speed, i.e. the nominal speed in
Table 1. The required performance at speeds
greater than the base speed was achieved by an
optimally designed controller.

(a) Unaligned position

TABLE 1 Summary of SRM ratings

HEV

ECOmmodore

SRM

Capacitor Battery Main Capacitor Battery

Motor
motor
motor motor
Motor

AXcess

DC voltage
(V)

180

55

350

180

55

Peak power
(kW)

32

10

50

33.1

11.2

Continuous
power (kW)

16

25

16.6

5.6

Nominal
speed (rpm)

3820

3820

3440

3440

3440

(b) Aligned position

Maximum
speed (rpm)

5730

5730

7640

7640

7640

Figure 1: Finite element flux-plot for a SRM with 12

stator and 8 rotor teeth (12/8)

The output torque of the SRM is the average

electromagnetic torque minus the bearing friction
torque of the motor. The output power is the
product of the average output torque and the
mechanical speed, which is determined by the
fundamental switching frequency in one phase and
the number of rotor teeth.
The losses that affect the efficiency of the SRM are
the winding copper loss, stator and rotor core
losses, mechanical loss, and stray loss. The
copper loss is the principal loss in a SRM under
most operating conditions. The stray loss is
assumed to be 1% of the output power in the
design.
The prediction of the core loss in SRMs is
extremely difficult since the flux waveforms are
non-sinusoidal and have different shapes with
various frequencies for different portions of the
SRM. The flux waveforms are also varying with the
degree of saturation. Therefore, a more accurate
estimation of core loss in SRMs would require
frequency decomposition in each portion [7].
Based on a pre-calculated Fourier series of the flux
density distribution for each part in a SRM with
typical three phase currents, the core losses are
calculated from the stator tooth fluxes that are
obtained from a finite element analysis.
The efficiency of the SRM can be calculated from
the resultant losses and the output power.
It was found that the power factor, cos, of a SRM
is related to the stored energy in the unaligned and
aligned position. The power factor is estimated by

W
cos 1 su
Wsa

(2)

where Wsa and Wsu are the stored energy when the
rotor is at the aligned and unaligned position,
respectively.

model. In the model, the heat generated by the

rotor core loss flows radially to the stator
lamination stack through the air gap. The heat
generated by the copper loss in the stator windings
flows radially to the stator lamination stack through
conduction of the enamelled copper winding and
slot insulation. All the heat in the stator lamination
stack flows to its surface through conduction and
then dissipates by convection into the water. The
heat dissipation to the end-caps from the end
windings is ignored.
The copper loss and core loss vary with the duty
cycle of the SRM operation. In one cycle the SRM
operates 20 minutes at continuous power and 20
seconds at peak power.
The thermal parameters, particularly the thermal
conductivity of the copper winding, slot wedge and
slot insulation, and the heat transfer coefficient at
the motor surface, are the most critical for the
prediction of the temperatures in the motor.
Experiments have been conducted on a steel
bobbin wound with enamelled copper to measure
the thermal parameters. The thermal parameters
used in the lumped-parameter model are given in
Table 2. The water temperature is 60 C and the
maximum winding temperature rise is specified as
180 C.
TABLE 2 Thermal parameters
Component

Density
3
(kg/m )

Air
Lamination
(stator and rotor)
Mild steel (shaft)
Copper winding
Slot wedge
Slot liner

1
7650

Stator lamination
outer surface

Thermal
conductivity
(W/(mK))
0.026
20

Specific
heat
(J/ (kgK))
10000
438

7800
43
473
8900
0.83
383
2150
0.29
1172
2150
0.076
1172
2
Film coefficient (W/(m K))
2000.0

DESIGN OPTIMIZATION AND RESULTS

The rms voltage of the SRM can be calculated by

V rms =

Pin
mI rms cos

(3)

where Pin is the input power and Irms the phase rms
current. The designed rms voltage of the SRM at
the base speed should match the specified
maximum DC bus voltage by adjusting the number
of turns of the stator windings.
Cooling and Winding Temperature
All the SRMs for the HEVs are shrunk-fit into an
aluminium water jacket. Temperatures in the SRMs
are modelled by a lumped-parameter thermal

Based on the above analysis, an optimisation

routine for the design of SRMs was set up, which
includes three FE analyses (one each for a onephase energisation when the rotor is at unaligned
and aligned positions to calculate the average
torque, one for a three-phase energisation to
calculate core loss) and a lumped-parameter
model. The FE analyses and design optimisation
searches were implemented in a commercial
comprehensive package, ANSYS.
The dimensional variables were rotor tooth width at
top and at bottom, stator tooth width, tooth height
of the stator and rotor, yoke thickness of the rotor,
and lamination stack length. The thickness of the

stator was constrained to be at least 0.75 times the

stator tooth width to account for up to three phases
simultaneously conducting. The peak current
densities at peak power and continuous power
operations are also set as variables. The
constraints are the maximum size envelope, peak
torques, continuous torque, and maximum winding
temperature. The objective is to minimize total cost
of the motor including material (copper and
lamination) and power electronics. The cost of
power electronics is determined by the controllers
peak VA. The number of turns of the stator
windings is assumed to be one in the optimisation
process and is scaled to match the specified DC
link voltage in the final design stage.
The optimisation was done for SRMs with three
different numbers of stator and rotor teeth: 6/4,
12/8, and 18/12. It is found that the 12/8 SRM
gives superior performance in this application.
Therefore a 12/8 SRM design was chosen for the
HEVs.
It is found that the SRM size and active mass
(copper winding plus lamination stack) mainly
depend on the required torque during continuous
operation. Before finalising the SRM designs for
the two HEVs, a 25 kW SRM with a base speed of
2400 rpm was fully optimised. The design details
are given in Table 3. The lamination material is
high-grade electrical steel, 0.35 mm Ly-core 130.
Water-cooling of the prototype SRM was obtained
by gluing water pipes into semi-circular grooves on
the stator lamination surface.
Subject to minimal total cost, the prototype SRM
has an active mass of 33.7 kg. Its efficiency for 25
kW continuous power delivery at the base speed is
92.3%. The maximum output torque is 203.3 Nm.
TEST RESULTS OF THE PROTOTYPE SRM
The optimised 25 kW prototype SRM was built and
fully tested. Figure 2 shows the stator core with
windings. The grooves on the outer surface of the
stator lamination are for the water pipes.

Figure 2: Stator core with windings of the prototype

TABLE 3 Design details of the prototype SRM

Specifications
DC voltage (V)

650

Peak power (kW)

50

Continuous power (kW)

25

Nominal speed (rpm)

2400

Maximum speed (rpm)

6000

Main dimensions
Air gap, mm

0.5

Rotor lamination ID (mm)

144

Stator lamination OD (mm)

281.4

Lamination stack axial length (mm)

117

Windings
Number of phases
Number of turns per coil

3
46

Copper fill factor

0.594

Phase resistance at 20C (Ohm)

0.41

Predicted performance at 2400 rpm

Peak

Continuous

Phase rms current rms (A)

Phase rms voltage (A)

48.6
650

26.8
565

Core loss (W)

254.4

162.4

Copper loss (W)

4684.9

1426.8

Mechanical loss (W)

251.3

251.3

Stray loss (W)

511

251.9

Output power (W)

51098

25187

Output torque (Nm)

203.3

100.2

Input power (W)

56800

27279

Efficiency (%)

90.0

92.3

Power factor

0.58

0.60

Peak VA, kVA

95

Winding temperature rise (C)

120

To test the prototype, the SRM was connected to

an eddy-current dynamometer through a two-step
gearbox. A specially designed controller was used.
The turn-on and turn off angles and the phase
voltage were optimally controlled to maximize the
torque per ampere and hence efficiency, for a
given load point. Constant power operation over
the base speed was achieved by phase advancing.
Figure 3 shows the measured maximum torque
and power versus speed. The maximum torque is
210 Nm. The designed peak power of 50 kW was
not tested due the limited capacity of the laboratory
power supply and the load limit on the
dynamometer. For speeds up to 6000 rpm,
operation at a constant power of 25 kW was
achieved readily.
Figure 4 shows the torque as a function of stator
phase current at different speeds. Below the base
speed, the torque is proportional to the current for
high currents because of saturation of the
magnetic circuit, whilst for low currents the torque
is approximately proportional to the current
squared. Above the base speed the torque per
ampere decreases due to the phase advancing

control. Figure 5 shows the phase current

waveforms for a high torque output at low speed
and a low torque at high speed.
20A/div
1500rpm
120Nm

The efficiencies at different speeds are shown in

Figure 6. The efficiency for a 25 kW output at 2400
rpm was 89.0 %, which was 3% lower than the
predicted 92.3 %. The reason for the lower
efficiency is that the winding temperature was 10%
higher than the prediction due to inefficient heat
transfer between the stator lamination and the
water pipes in the prototype. The five SRMs for two
HEVs have a shrunk-fit aluminum water jacket
instead of pipes to improve the heat transfer.

225

45

200

40

175

35
30

Torque
Power

125

25

100

20

75

15

50

10

25

10A/div
6000rpm
20Nm

Power (kW)

Torque (Nm)

150

(a) At low speed

0
500 1000 1500 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500

Speed (rpm)

Figure 3: Measured maximum torque and power

versus speed

(b) At high speed

225

175
150

100

125

Speed
500 rpm
1000 rpm
2000 rpm
3000 rpm
5000 rpm

95
100

90

75
50
25
0

10

20

30

40

50

60

70

80

90

100

110

120

Phase rms current (A)

Efficiency (%)

Torque (Nm)

Figure 5: Phase current waveforms

Speed
500rpm
1000rpm
4000rpm
6000rpm

200

85

80

75
70

Figure 4: Torque against stator phase current at

different speeds

65

25

50

75

100

125

150

175

Torque (Nm)

Figure 6: Efficiency at different speeds

200

225

SRMS FOR TWO HEVS

Using the same lamination profile as in the
prototype, the stack lengths and windings were
optimised for five motors for two HEVs: two for the
ECOmmodore and three for the aXcess. Table 4
gives a summary of the SRM designs.
In each car, all the motors are mounted on a
common shaft within a common water jacket [3].
This makes construction economic because only
one lamination stamping is used for all the motors.
Figure 7 shows the SRMs situated in the HEVs.

5. Corda J, Stephsenson J M, 1979, Analytical

estimation of the minimum and maximum inductances
of a double-salient motor, Conf. Rec. Int. Conf. on
Stepping Motors and Systems, Leeds, UK, pp50-59.
6. Tang Y, 1997, Characteristization, numerical
analysis, and design of switched reluctance motors,
IEEE Transactions on Industry Applications, Vol. 33,
No. 6, pp1544-1552.
7. Materu P N, Krishnan R, 1992, Estimation of
switched
reluctance
motor
losses,
IEEE
Transactions on Industry Applications, Vol. 28, No. 3,
May/June 1992, pp668-679.

TABLE 4 Summary of SRMs for two HEVs

CONCLUSIONS

HEV
SRM

The design optimisation of SRMs for HEVs has

been presented in this paper, by using a
combination of 2D electromagnetic finite element
(FE) analysis, experience from previous designs,
3D correction factors, simple lumped-parameter
thermal models, and computer search techniques.
Although the design example was optimised for
minimal cost, it nevertheless exhibits a high
efficiency. Test results of the prototype SRM have
shown that the design goal was achieved and the
design optimisation routine validated. The
experience obtained from tests on the prototype
was used to improve the thermal design of the
SRMs for two HEVs.

ECOmmodore

aXcess

Capacitor Battery Main Capacitor Battery

motor
motor motor motor
motor

Stack length 58.5

(mm)

30

117

70

39

Active mass 18.6

(kg)

10.4

34.6

21.8

13.1

Efficiency
(%)

89.2

79.7

92.0

89.9

82.5

Maximum
180
winding
temperature
(C)

180

152

180

173

The two HEVs, GM Holden ECOmmodore and

aXcess, have demonstrated excellent performance
with the optimised SRMs and the potential of
SRMs for vehicle application.
REFERENCE
1. Lamb D, 2000, The aXcessaustralia Hybrid
Electric Car Project, Seoul 2000 FISITA World
Automotive Congress, Seoul, Korea, June 12-15,
2000, Paper No. F2000H260.

(a) ECOmmodore

SRMs

2. Lovatt H C, Dunlop J B, 2000, The

aXcessaustralia Hybrid Electric Car Project, Seoul
2000 FISITA World Automotive Congress, Seoul,
Korea, June 12-15, 2000, Paper No. F2000A035.
3. Lovatt H C, Dunlop J B, 2002, Power Transfer in
Hybrid Electric Vehicles with Multiple Energy Storage
Units, PEMD 2002, Bath, UK, April 2002.
4. Rahman K M, Fahimi B, Suresh G, Rajarathnam A
V, Ehsani M, 2000, Advantages of Switched
Reluctance Motor Applications to EV and HEV:
Design and Control Issues, IEEE Transactions on
Industry Applications, Vol. 36, No. 1, Jan/Feb 2000,
pp111-121.

(b) aXcess
Figure 7: SRMs situated in HEVs