Drive Motor Designs for Electric Motorcycles

David G. Dorrell

Mircea Popescu

University of Technology Sydney

Sydney, Australia
david.dorrell@uts.edu.au

Motor Design Ltd

Elsmere, UK
mircea.popescu@motor-design.com

AbstractThis paper addresses the design of motors for

electric motorcycle racing. These tend to be high torque-dense
brushless permanent-magnet machines (either DC or AC)
operating over a very wide speed range with an extended
maximum power range above base speed. The thermal design
and operation are key to their successful utilization and this is
a major focus of this paper. The motors can operate at very
high current density for a short time during acceleration.

I.

INTRODUCTION

Electric drives for vehicles are an important aspect of

recent transportation technology development. There have
been many studies of brushless permanent magnet AC
machines for automotive applications and [1] and [2] are
examples. There have been detailed descriptions of
commercial drives [3]-[5]. There are alternatives to the
brushless permanent magnet motor; alternative drives such
as switched reluctance and induction motors [6]-[9] have
been reported upon. These have attempted to produce
comparable machines without permanent magnets with high
efficiency and torque density. However, the brushless
permanent magnet machine (BPM) still represents the most
efficient and torque-dense machine. The fundamental issues
are torque density, thermal performance, efficiency and
torque ripple. Torque-dense machines can have substantial
torque ripple. At first glance it can be hypothesized that this
should not be an issue for a vehicle drive; the internal
combustion engine inherently has a lot of torque ripple and a
straight prime mover replacement should allow the same
amount of ripple. However, flywheels are employed to
smooth the ripple in internal combustion engines, which are
not usually employed in an electric motor drive (although the
rotor often has substantial inertia in its own right) and since
torque ripple can be designed out of an electric drive, it is
considered good practice to try and do so. This will reduce
wear on mechanical power train parts and also reduce tire
wear. In the context of an electric motorcycle, this will also
improve handling, which is obviously very important.
Many production automotive BPMs are of the 3-phase
AC type (sinusoidal current with 180 degree PWM
conduction and three bridge devices controlled
simultaneously). There are several reasons for this including
smoother torque and the utilization of phase advance and an

978-1-4673-0803-8/12/$31.00 2012 IEEE

interior magnet rotor (IPM) to allow for reluctance torque.

The sinusoidal current is controlled using an encoder (unless
a sensorless strategy is used) and the Toyota Prius motor has
an IPM; it runs with constant phase advance up to 60 deg
elec. As illustrated in [1], the reluctance torque approaches
the same magnitude as the excitation torque and this makes it
a very high efficiency machine without substantial torque
ripple. Automotive machines often use very high current
density, possibly up to 25 A/mm2 and they are cooled using
water, oil or other fluid cooling. Because the maximum
torque tends to be a transient during acceleration then duty
cycling within the race can be used to prevent overheating.
There has been an increase in the sales and use of electric
scooters in such countries as China [10][11]. These types of
electric vehicle tend to be more straightforward drive
systems with a brushless permanent magnet DC hub motor
and 3-phase DC inverter control (trapezoidal current with
only 120 degree PWM or 6-pulse control: two or one bridge
devices will be controlled simultaneously depending on the
mode of operation). These require only Hall sensors for rotor
position feedback at the 6-pulse switching positions rather
than a higher resolution encoder position feedback.
However, recently, there has also been increasing interest in
electric racing vehicles including small single-seat cars [12]
and motorcycles [13]. By the nature of the competition, the
battery packs and/or power are restricted. The Formula
Student racing uses a power limit [12] of 100 kW at any one
particular acceleration point with a maximum of 600 V DC
or AC RMS between any two connections, while the
motorcycle racing uses a restriction on battery size [13] of
7.5 kW-hours for the TTX75 class with a maximum voltage
of 700 V. By the nature of a battery pack, there will be a
natural point where it is unfeasible to mount more than about
10 to 12 kW-hours of batteries onto a motorcycle and there is
a weight restriction (between 100 and 250 Kg).
In this paper we will explore the design of motors to meet
the drive requirement for a motorcycle. The control is DC
initially, then extended to AC. Many of the DC controllers
can operate close to the DC rail voltage; however, accurate
control of the current becomes difficult close to the DC rail
voltage. A typical 800 A controller is given in [14] and this
can handle up to 800 A at 150 V. Waveforms under running

4354

light conditions are shown in Fig. 1 at a mid-speed and a

high speed. It can be observed that there is little PWM
switching at the high speed. These controllers limit the
current and hence the torque so there is no real field
weakening capability. This means there is a trade-off
between the torque constant and voltage constant and the
maximum power is only reached close to maximum torque.
There is also the issue with regenerative braking. Fig. 2
shows the DC braking technique. This is soft braking; hard
braking can be obtained by shifting the phase switching at
illustrated in [15]. In terms of AC regenerative braking, this
is a more straightforward control as described next. The
machine usually has encoder position feedback, or a
sensorless control to calculate the position and a PWM
switching strategy. It is a relatively simple procedure to shift
the inverter switching so that the inverter voltage is such that
the current (if motor convention is used) is in advance of the
motor phase back-EMF by more the 90 deg elec, where both
power and reactive power are flowing out of the machine
and it appears to be an overexcited generator, or it lags the
voltage be 90 deg elec, where power flows out of the
machine but reactive power flows in, and the machine
appears as an under excited generator. Different converter
topologies were reviewed in [16]. Many BPMs have neither
a sinusoidal or trapezoidal back-EMF; they tend to have a
waveform somewhere in the middle. These can be either DC
or AC controlled and in fact [17] conducts a study that uses
both strategies in the control.

Figure 2. DC switching. Top - normal motoring switching (two devices

together); center and bottom - regenerative braking (one device on and on
switching continuously to for a step-up chopper from motor to battery).

(a) Mid speed

(b) High speed

Figure 1. Terminal voltage for a DC controller at no load and at mid and
high speed. Upper plot shows extensive PWM switching for voltage
whereas lower plot shows little PWM illustrating DC controller working at
its limit.

On a motorcycle, the regenerative braking is done on the

rear wheel; many racing motorcycles have no rear
mechanical brake. However the main brake is on the front
wheel and this is invariably a mechanical brake. The racing
rules do allow for electric drives on both front and back
wheels and it would be interesting to see if there is a move
use regenerative braking on the front wheel; it need to be
remembered that the braking torque is much higher than the
acceleration torque. The additional weight of a drive on the
front wheel would affect the handling of the motorcycle.
Much information is propriety so news items have to be
relied on to obtain the current state of the technology. Hence
much of the reference material here is gleaned from press
and internet web pages and this can be unreliable. More
lately the Isle of Man has run an independent electric bicycle
race as part of the annual TT races [18] and in 2011 it was
won by a motorcycle fitted with oil-cooled electric motor
that claimed 340 Nm [19][20] with a constant rating of 75
kW, or possibly 100 kW. There is a certain amount of
contradiction to the figures in [19] and [20] and also those

4355

be being quoted for performance for a different design; the

weight appears to be much higher indicating a larger
machine. The values may also be for performance at the road
wheel since the torque is so much higher it suggests that
gearing has already been applied.
900
Torque [Nm]

quoted about the drive unit in [21] (which quotes only a

continuous rating of 60 kW with a peak of 75 kW but a with
much higher torque). However, these are advertising or
popular journal articles, but there is little doubt that the
performance compared to other motorcycles suggested a
large improvement in performance. The machine appears to
have moved from maximum torque control using a DC
controller to having a maximum power range; base speed
and maximum speed are quoted in [21]. It should be borne in
mind that with 240 V, 10 KW-hours of battery storage (or
330 V, 12.5 kW-hours depending on which article is quoted)
then high power operation is only possible for a short time.
In this paper we will address the electromagnetic torque
and thermal aspects of a drive motor using both the simple
maximum torque DC control and then AC control using field
weakening. These aim at maximum power of 60 to 75 kW.
For the simple Kelly controller the maximum may be more
than 60 kW but controlling the motor may be an issue. This
work has already been partially reported in [22].
II.

AC 340 Nm motor torque

Aerodynamic torque

450
300

0
0

18

(1)

where Fd is the force necessary to overcome wind resistance,

is the density of air, v is the velocity in m/s and A is the
area of bicycle in the direction of movement. If a hub motor
is fitted then we need to calculate the wheel speed. The
circumference length of a racing bicycle wheel is about 2 m
which gives a radius of 0.318 m. Therefore if we allow a
maximum speed up to 200 km/hour then the road wheel
speed is 1668 rpm. If we use a motor speed of 6000 rpm,
which is a reasonable speed for a 6 or 8 pole IPM motor, the
gearing is 3.6. The torque required to overcome the wind
resistance is shown in Fig. 3 (a) and this is compared to the
assumed torque speed curve for the Motoczysz E1-PC-10.
This reaches 75 kW of power with a maximum torque of 340
Nm at a road wheel speed of 89 rad/s. With a gearing of 2.5
(this is a low speed running motor and this is quoted as the
lower gear ratio) then the base speed is 2125 rpm. This is
very different from that quoted in [21]. The torque is quoted
is much higher and it appears that the torque and speeds may

50
(a)

MOTORCYCLE LOADING

We now need to consider the mechanical loading. A

racing motorcycle typically has a drag coefficient Cd of 0.6
[23] and an area of 0.36 m2 for a normal bicycle [24], and
[23] quotes about 0.55 m2 for a racing motorcycle. The drag
coefficient is defined by
2 Fd
v2 A

DC 100 Nm motor torque

600

150

Time to speed [s]

The electrical loading calculations were put forward in

[22] but they are also included here for completeness. A
racing duty cycle can be more demanding with hard
acceleration several times a lap, and a high maximum speed;
regenerative braking may also be required. The Isle of Man
SES TT Zero competition race is over a 37.75 mile course
and in 2011 the winning time was 22 min 28 s [18]. This is
an average speed approaching 160 km/hour. If there is only
10 kW-hour of battery energy then the bicycle will be energy
limited.

Cd =

750

100
150
Speed [km/hour]

200

Torque requirement

DC 100 Nm motor - acceleration time

AC 340 Nm motor - acceleration time

15
12
9
6
3
0
0

50
(b)

100
150
Speed [km/hour]

200

Acceleration times

Figure 3. Torque and acceleration time comparison.

In this paper an example design is being studied so it is

more important to obtain a realistic characteristic. The
second, and lower constant-torque characteristic, is for a
more straightforward control that can give 100 Nm of torque
which is increased by gearing to 360 Nm of torque at the
road wheel up to 200 km/h. This will give a maximum power
consumption of 60 kW at top speed and it is matched to the
Kelly controller with a 150 V battery supply.
If we assume that the drive motor is torque limited it can
be seen that the aerodynamic force will have an increasing
influence on performance as the speed increases. The
acceleration times are shown are shown in Fig. 3(b) and this
assumes only inertia and wind friction. The weight of the
bicycle and rider is taken as 275 Kg which is artificial since
the Motoczysz is higher in weight, possibly by 30 to 50 Kg.
In Fig. 4 the energy used and distance travelled are given for
a 16 s acceleration up to 200 km/h. The Motoczysz reaches
this speed much sooner and then the speed is constant. At
100 mph (160 km/h) the power being used to overcome the
aerodynamic losses is about 17.3 kW. Therefore, with 10
kW-hours of stored energy there is a possible 34.5 minutes
of constant operation. This illustrates that there is a
restriction in both the top speed and time of a race.

4356

Energy used [kW-hours]

0.25

TABLE I.
DESIGN AND PERFORMANCE PARAMETERS FOR DC
CONTROLLED MACHINE DESIGNS. COMPARISON OF RESULTS LOAD POINT
TAKEN AT 6000 RPM AND 800 A NOMINAL, DC LINK IS VOLTAGE IS 150 V,
DYNAMIC SIMULATION INCLUDING SWITCHING.

DC 100 Nm motor - energy used

AC 340 Nm motor - energy used

0.2
0.15

Machine

0.1

Outer diameter [mm]

Axial core length [mm]
Magnet weight [Kg]
Copper weight [Kg]
Total weight [Kg]
Torque [N]
Copper loss [W]
Iron loss [W]
Efficiency [%]

0.05
0
0

(a)

Distance travelled [m]

800

8
Time [s]

12

16

Energy used during 16 s acceleration

Large 18
slot/6pole
240
150
2.343
5.58
39.02
111.6
1440
1150
96.4

Small 9
slot/6pole
240
85
2.00
4.57
22.33
85.3
915
141
98.1

DC 100Nm motor - distance travelled

AC 340 Nm motor - distance travelled

600
400
200
0
0

4
(b)

8
Time [s]

12

16

Distance travelled

Figure 4. Comparison of energy used and distance.

III.

DRIVE MOTOR DESIGN

In this section the designs for the two motors are put
forward. The designs are highly saturated and require
cooling so the motor design package SPEED PC-BDC is
used in conjunction with the finite element package PC-FEA
and linked to the thermal design package Motor-CAD. This
is the main focus of this paper.
A. DC controlled machine
This is a smaller motor since only 100 Nm of torque is
required. One of the requirements for this machine is to keep
the design as simple as possible and to use coils with one slot
pitch. The specification utilizes a controller with only 150 V
at the DC link but possibly up to 350 A constant and 800 A
for one minute. This means that the voltage constant will be
low but the torque constant is high which are opposing
requirements. It is likely that many parallel paths are needed.
The first design is shown in Fig. 5 and the parameters
and loading simulation results tabulated in Table I. This uses
an 18 slot arrangement with 6 poles and all phase coils were
parallel connected. This had three coils per pole and this
initially looked a reasonable design. The machine was
thermally validated and the maximum power load point was
taken at 6000 rpm (the peak motor speed selected). The
SPEED simulations used a dynamic simulation which
calculated the current using a switching strategy. However,
inspection of coil arrangement shows that currents in coilsides under one pole flow in different directions (two phases
will be conducting at one time) which will restrict the torque.

Figure 5. Large machine with performance predictions: (a) Cross-section

in Motor-Cad showing water jacket (20 L/min of 50% water and 50 %
glycol); (b) Axial-section in Motor-Cad showing compact windings; (c)
Wind layout showing single slot pitch.

The size of the machine is large as illustrated by the

weight as shown in Table I, so that the thermal stability is
good as shown in Fig. 6(b). The Impregnation Factor is a
coefficient used in the simulation to govern the resin
impregnation in the slot around the coil-turns. This is very
much a manufacturing process and it is an empirical value.
Typically impregnation factors will be 0.6 upwards for a well
manufactured machine. For a bespoke machine aimed at high
temperature operation care would be taken to totally
impregnate the winding and the impregnation factor may be
approaching unity.

4357

The torque ripple in Fig. 6 is high but this is not

important unless it is very high, and this can but reduced or
eliminated using skew as shown later without reducing the
torque significantly. Internal combustion engines have
considerable torque ripple but without a flywheel this can
cause problems as discussed earlier.

120

80
60
40

120

Torque - 3000 rpm half load

Torque - 6000 rpm full load

20
0
0

90
180
270
Rotor rotation [deg elec]

Torque - 6000 rpm full load

100

Torque [Nm]

Torque [Nm]

100

360

80
Torque - 3000 rpm half load

60
40
20
0
0

90
180
270
Rotor rotation [deg elec]

360

Figure 6. Large machine with performance predictions: (a) Instantaneous

torque at full load (800A) and half load (400A); (b) Winding steady-state
temperature variation with 20 L/min flow; variation of impregnation factor.

Figure 8. Reduced size DC machine operation over duty cycle.

Figure 7. Reduced size DC machine topology.

B. Reduced sized DC machine

If the winding is fully pitched than there is a large
increase in torque. However, this complicates the winding
so, in order to reduce the size, then a 9 slot 6 pole
arrangement was designed and this is shown in Fig. 7. This
machine approaches the size of a rear wheel hub motor as
illustrated in [25]. As seen in Table I it is only 22 Kg. The
simulation results at maximum load are also given and these
show that the efficiency is still excellent. A duty cycle
temperature profile is shown in Fig. 8 and this illustrates that
it does get hot when running at full load for longer than one
minute. As with the larger machine, it has a water/glycol
jacket with 20 L/min flow. The thermal simulation for this
machine is 22 minutes (1320 s). There is a an 800 A high
torque acceleration for one minute, the a ten minute 400 A
half maximum load period, then after ten minutes these two
periods are repeated again. This allows the investigation of a
steep temperature rise from cold at full load, then a sustained
period of half load where the motor almost reaches a steady-

4358

state thermal point. The mid period high torque load

illustrates the sharp temperature rise when loaded again for a
short period at maximum current then the return to the halfload temperature state over the remaining period. Since the
machine is forced cooled then the speed of the machine is
not very relevant although it was set to 6000 rpm for the high
torque load and 3000 rpm for half load.
C. Torque ripple reduction
The torque has substantial ripple. This can be easily
removed by skewing the rotor. There are two possible
methods to do this. One simple method, termed step skew in
[26], is to split the rotor in half in the axial direction and then
rotate on half by a certain amount. We can apply this method
to the reduced sized DC machine in the previous section. In
Fig. 9 (a) it can be seen that the torque ripple is reduced
significantly by simple step skewing the rotor by one half
pole. The current phasor is nominally on the q axis that now
lies between these two half-pole sections. There is no
noticeable drop in mean torque.
There is still some torque ripple when using step skew
for this machine. This can be further reduced if continuous
skew is implemented in the rotor manufacture. If the rotor is
skewed over one rotor pole pitch then the results are shown
in Fig. 9 (b). The torque ripple is now almost completely
eliminated and there is still no noticeable drop in mean
torque. The disadvantage with continuous skew is that it is
harder to manufacture and the magnet shape will not be a
straightforward slab shape.

Torque [Nm]

100
80
6000 rpm full load
3000 rpm half load

60

Half pole step skew

Half pole step skew

40
20
0
0

45

90 135 180 225 270

Rotor rotation [deg elec]

315

360

(a) Step skew of half rotor pole

Figure 10. High torque motor: (a) FEA analysis; (b) Motor-Cad cross
section showing slot and stator cooling jacket; (c) Slot representation
showing center cooling duct and 0.3 slot fill.

Torque [Nm]

100
80
6000 rpm full load
3000 rpm half load

60

One pole skew

One pole skew

40
20
0
0

45

90 135 180 225 270

Rotor rotation [deg elec]

315

360

(b) Continuous skew of one rotor pole span

Figure 9. Effect of step and continuous skew on reduced sized DC motor.

D. AC controlled machine
This section addresses the design of a higher
specification machine. This machine assumes AC control
and field weakening so that constant power operation is
possible. Some basic parameters are given in Table II and an
FEA solution together with the Motor-CAD model is shown
in Fig. 10. The winding layout is similar to the 9/6 machine
above. The round slots allow a smooth flow of flux around
the winding and shorten the flux path. This machine weighs
less than the large machine above but there is a higher level
of cooling. There is a very good impregnation level in the
winding and the thermal conductivity is 1W/C/m. There is

4359

an interface gap between the housing and the stator

lamination of 0.005 mm. The water jacket and slot water
jacket settings are 12 L/min using Paratherm LR with an
inlet temperature of 65C. The winding in the slot is for
illustrative purposes with give an indication to the slot fill. In
reality it is a random-wound coil structure with heavy
impregnation.
It was decided to use a spoke type magnet arrangement
because this allowed flux concentration and the phase control
maintains the current on the rotor q axis.
The simulation results are given in Fig. 11. What is
apparent is that it is possible to operate the machine at these
high torque levels for some time providing that there is
excellent cooling; the current density in the winding at full
load is 25 A/mm2. Again, the same 22 minute simulation was
used to examine the thermal performance of the machine as
shown in Fig. 11.
This spoke design, with round slots, is an interesting
design. It appears to aid the flux in the machine core. It may
be necessary to address the issue of slot cooling and whether
it would be possible to put cooling ductwork directly down
the center of the slot. This would help address issues related
to the cooling of the winding hot spot as shown in Fig. 11.
The temperature rise is at the limit of F type insulation and
any inconsistencies in the winding impregnation will drive
this temperature higher.
TABLE II.
DESIGN AND PERFORMANCE PARAMETERS FOR AC
CONTROLLED MACHINE DESIGN FULL LOAD POINT TAKEN AT 2118 RPM
AND 424 A RMS NOMINAL FOR LINE CURRENT (339 A DC LINK); DYNAMIC
SIMULATION INCLUDING SWITCHING.
Outer diameter [mm]
Axial core length [mm]
Magnet weight [Kg]
Copper weight [Kg]
Total weight [Kg]
Torque [N]
Copper loss [W]
Iron loss [W]
Efficiency [%]

240
120
4.76
2.36
35.17
330.0
3315
436
95.1

Winding hot spot

168

Winding
mean
Winding cool spot
Housing
Magnet
24
0

Time [s]

1980

IV.

CONCLUSIONS

This paper outlines the load requirements for an electric

motorcycle and briefly describes different motor designs for
both a brushless DC and a brushless AC PM designs. The
machines are aimed at mounting in the frame of the bicycle.
There are several hub motors available (for electric scooters
but also for some racing bicycles) but to get the high power
required for racing, and low wheel mass, it is more
appropriate to mount the motor in the frame.
These are high performance machines subjected to
extreme demands. Forced cooling is required and this paper
has reviewed cooling and duty cycling in two different
designs.
The control is important; there is limited voltage and
energy so high efficiency is necessary. Simulations using
thermal design software illustrated that these machines
operate on the thermal limit but they are electromagnetically
very efficient, reaching efficiencies well in excess of 90 %.
REFERENCES
[1]

D. G. Dorrell, M.-F. Hsieh, M. Popescu, L. Evans, D. A Staton, and

V. Grout, "A Review of the Design Issues and Techniques for RadialFlux Brushless Surface and Internal Rare-Earth Permanent Magnet
Motors," IEEE Trans. on Ind. Electr, vol. 58, no. 9, 2011, pp 3741
3757.
[2] D. G. Dorrell, M.-F. Hsieh and A. M. Knight, Alternative Rotor
Designs for High Performance Brushless Permanent Magnet
Machines for Hybrid Electric Vehicles, IEEE Trans. on Magn., vol.
48, no. 2, 2012, pp 835 838.
[3] Mitch Olszewski, Evaluation of the 2007 Toyota Camry Hybrid
Synergy Drive System, Oak Ridge National Laboratory, U.S.
Department of Energy, USA, 2009.
[4] L. D. Marlino, Report on Toyota Prius Motor Thermal Management,
Oak Ridge National Laboratory, U.S. Dept. of Energy, USA, 2005.
[5] M. Olszewski, T. A. Burress, S. L. Campbell, C. L. Coomer, C. W.
Ayers, A. A. Wereszczak, J. P. Cunningham, L. D. Marlino, L. E.
Seiber, H. T. Lin, Evaluation of the 2010 Toyota Prius Hybrid
Synergy Hybrid Drive System, Oak Ridge National Laboratory, U.S.
D. of E., USA, March 2011.
[6] Y. Takano, M. Takeno, T. Imakawa, A. Chiba, N. Hoshi, M.
Takemoto and S. Ogasawara, "Torque density and efficiency
improvements of a Switched Reluctance Motor without rare earth
material for hybrid vehicles," IEEE Energy Conversion Congress and
exposition (ECCE), Atlanta, Sept 2010 , pp 2653 2659.
[7] Y. Takano, M. Takeno, N. Hoshi, A. Chiba, M. Takemoto, S.
Ogasawara and M. A. Rahman, "Design and analysis of a switched
reluctance motor for next generation hybrid vehicle without PM
materials", International Power Electronics Conference, Sapporo,
Japan, 21-24 June 2010, pp 1801 1806.
[8] X. D. Xue, K.W. E. Cheng, T.W. Ng and N. C. Cheung, "MultiObjective Optimization Design of In-Wheel Switched Reluctance
Motors in Electric Vehicles", IEEE Trans. on Ind. Elec., vol. 57, no.
9, Sept. 2010, pp 2980 2987.
[9] D. G. Dorrell, M. Popescu, L. Evans, D. A. Staton and A. M. Knight,
"Comparison of Permanent Magnet Drive Motor with a Cage
Induction Motor Design for a Hybrid Electric Vehicle", IEEE IPEC
Conference, Sapporo, Japan, June 2010.
[10] J. Weinert, J. Ogden, D. Sperling and A. Burke, The future of
electric two-wheelers and electric vehicles in China, Energy Policy,
vol. 36, 2008, pp 2544 2555.

Figure 11. High torque motor: Temperature variation over duty cycle with
transient of 3 cycles of 1minute full torque and 10mins 0.5 torque (full
torque is 330 Nm as given in Table II).

4360

[11] J. Weinert, C. Ma and C. Cherry, The transition to electric bikes in

China: history and key reasons for rapid growth, Transportation,
vol. 34, 2007, pp 301 208.
[12] Formula Student Germany, Formula Student Electric Rules 2011, V.
1.1.0, Release: 26. April. 2011.
[13] TTXGP, Technical Rules, 2012 Season, http://www.egrandprix.com.
[14] http://kellycontroller.com/khb1480124-144v800aopto-bldccontrollerwith-regen-p-822.html
[15] M.-J. Yang, H.-L. Jhou, B.-Y. Ma, and K.-K. Shyu, A CostEffective Method of Electric Brake With Energy Regeneration for
Electric Vehicles, IEEE Trans. on Ind. Elec., vol. 56, no. 6, June
2009, pp 2203 2212.
[16] J. W. Kolar, T. Friedli, J. Rodriguez and P. W. Wheeler, Review of
Three-Phase PWM ACAC Converter Topologies, IEEE Trans. on
Ind. Elec., vol. 58, no. 11, Nov. 2011, pp 4988 5006.
[17] D. Lu, J. Li, M. Ouyang and J. Gu, Research on hub motor control
of four-wheel drive electric vehicle, IEEE Vehicle Power and
Propulsion Conference (VPPC), 2011, pp 1 5.

[18] www.iomtt.com/TT-2012/TT-Zero.aspx.
[19] http://www.motorcyclenews.com/MCN/News/newsresults/Generalne
ws/2010/Februaryfeb2710-motoczysz-launches-plugnplaypowertrain/
[20] http://motoczysz.com/motorcycles/e1pc_2010
[21] http://www.d1g1taldr1ve.com/d1-tech-specs/
[22] D. G. Dorrell, Design of Brushless Permanent-Magnet DC Motors
for Racing Motorcycles, IEEE International Symposium on
Industrial Electronics (ISIE), Hangzhou, May 2012.
[23] http://www.schultzengineering.us/aero.htm
[24] D. G. Wilson, "HPV science", Technical Journal of the IHPVA, No.
54, Spring 2003, pp 4 13.
[25] http://www.ev-propulsion.com/motorcycle-hub-motors.html
[26] H. S. Chen, D. G. Dorrell and M. C. Tsai, "Design and Operation of
Interior Permanent Magnet Motors with Two Axial Segments and
High Rotor Saliency", IEEE Trans. on Magnetics, vol 46, no. 9, Sept.
2010, pp 3664 3675.

4361