Abstract— There is significant interest in the development of Switched Reluctance Machines for use in automotive traction applications. This has been driven by their low cost compared to Rare Earth Permanent Magnet based motors, coupled with their potential for competitive torque densities. This paper describes the development of a variant of the Switched Reluctance Machine, utilising a Segmental Rotor construction, which has previously been demonstrated to provide up to a 65% improvement in torque densities compared to Switched Reluctance Machines with a conventional toothed rotor construction. The paper describes a strategy which has been developed to optimise the Segmental Rotor SRM in order to achieve performance equivalent to that of the 80kW Interior Permanent Magnet machine utilised in Nissan’s LEAF electric vehicle. Optimisation applies a combination of static and dynamic analyses in order to achieve a full and computationally efficient assessment of motor performance across different regions of the torque speed envelope.

Index Terms— Optimization, Reluctance motors, Traction motors, Automotive components.

I. INTRODUCTION WITCHED Reluctance Machines (SRMs) are experiencing a resurgence in interest. In automotive traction applications

they demonstrate advantages in terms of cost compared to both rare earth permanent magnet brushless motors and also induction machines [1] whilst also offering the potential for high torque densities [2]. However concerns have remained due to their larger size and mass compared to rare earth permanent magnet motors, a consideration in automotive applications, as well as their reputation for torque ripple and acoustic noise [3].

This paper presents a design process which leads to the development of an SRM traction motor aimed at closely matching the specification of the Interior Permanent Magnet (IPM) motor based system employed in Nissan’s Leaf electric vehicle [4].

Recent research [5-10] has highlighted the potential benefits

of the Segmental Rotor SRM. This machine topology has been shown to have the capability to produce up to 65% more torque per unit loss, due to its improved magnetic utilisation, when compared to a conventional SRM. This increased torque density is a result of the characteristic that, with a single phase energised, more teeth carry flux than would be the case in a conventional SRM; this is illustrated in Fig. 1. However this also means that the motor is prone to mutual interactions between neighbouring phases, as two phases may link flux through the same tooth at any time instant.

Fig. 1: The Segmental Rotor SRM (left) can be seen to have better magnetic utilisation with twice as many teeth carrying flux at any one time than with a conventional SRM (right). Both machines are shown with rotors in the peak torque position. A Segmental Rotor SRM with fully pitched windings is shown.

The authors [11] have previously described an approach which enables the optimisation of these machines, utilising static characteristics to judge the best geometry in order to maximise torque per unit loss. However this technique is limited in a number of ways when assessing a motor for use in an automotive traction application. Firstly it takes no account of the mutual interactions between adjacent motor phases and therefore has been found to inaccurately predict dynamic motor torque. Secondly it provides no information as to the iron losses resulting from differing geometries, preventing a full optimisation for all motor losses. Finally this method does not facilitate the assessment of dynamic limitations of the motor in terms of the extents of its torque and speed envelope, clearly critical as torque-speed curves frequently form the basis of automotive traction requirements.

It is also clear that the multi-objective optimisation of SRMs will be complicated and time consuming due to their highly non-linear characteristics. High performance SRMs are generally controlled through the use of control maps, setting current advance angles, conduction angles and peak current

Optimisation of an 80kW Segmental Rotor Switched Reluctance Machine for Automotive

Traction James D. Widmer, Richard Martin, Barrie C. Mecrow

S

This work was supported by EPSRC Platform Grant EP/F067895, “High Efficiency Electrical Energy Conversion”.

James D. Widmer and Richard Martin are with the Newcastle University Centre for Advanced Electrical Drives, School of Electrical and Electronic Engineering, Merz Court, Newcastle Upon Tyne, NE1 7RU – UK. (Tel: +44(0)191 222 3016; e-mail: james.widmer@ncl.ac.uk).

Barrie C. Mecrow is with Newcastle University School of Electrical and Electronic Engineering, Merz Court, Newcastle Upon Tyne, NE1 7RU – UK. (Tel: +44(0)191 222 7329; e-mail: barrie.mecrow@ncl.ac.uk).

978-1-4673-4974-1/13/$31.00 ©2013 IEEE 427

requirements such that a particular operating pThese are conventionally calculated using largand error runs [12]; an approach which capplied in a time efficient manner to a comple

As a result this paper outlines a new approSegmental Rotor SRMs. This approach comand dynamic analyses to provide a more comof the efficiency and performance characterimotor geometries.

II. OPTIMISATION APPROA

A. Principles of Optimisation The objective of the optimisation described

develop a Segmental Rotor SRM which matching closely that of the Nissan LEAFmachine; in summary a peak torque of 280Nmof 80kW between 2,750rpm and 10,500rpm.

Segmental Rotor SRM optimisation is basecommercially available finite element paevolutionary based solver as detailed in [13allows the comparison of a series of rangeometries, with each geometry assessed throuser defined objective function. Over time, onpoints have been collected, the algorithgeometric search window in order to idensolution.

Three objectives were defined for the opttraction motor. The first objective, illustratedthe motor must achieve the required torque spwithin the constraint of the current and voltagpower converter. In this case of the Nissan Lachieving 280Nm at a speed of approximateDC link voltage of 400V and peak current of

Fig. 2: Showing the key optimisation points for Segment The second objective (again, Fig. 2) is tha

efficient as possible at one or more normal The selection of motor operating points predicted time residency of motor operation

point is achieved. ge number of trial annot be readily ex optimisation. ach to optimising

mbines both static mplete assessment istics of differing

ACH

d in this paper is to has performance F’s IPM traction m and peak power

ed on the use of a ackage using an 3]. This software domly generated ough the use of a nce sufficient data hm narrows the ntify an optimum

timisation of this d in Fig. 2, is that peed characteristic e limits set by the LEAF this means ly 2750rpm for a 500A.

al Rotor SRMs.

at the motor be as operating points. is based on the in different areas

of the torque speed characteristiassessment of the vehicle’s driving data on the baseline operating poinsingle example operating point oassumed. This equates to a constansimulating city driving.

The third objective, providing balathe motor’s mass should be limited iwith its use in an electric vehicleimportant consideration.

B. Objective Function In order to compare the success

optimisation, an objective function being to maximise the function. Inefficiency at each operating point anecessary to develop a method of equefficiency. This has is done by definimotor mass. In this case 10 kg mpercent of motor efficiency was sesomewhat arbitrary but considered rigorous approach would be to dervehicle system modelling activity.

The objective function can then the following equation:

∑ . where n is the number of operation

the weighting factor for each operatat each operating point and kmass a finto an efficiency equivalent.

C. Hybrid Optimisation ProcesThe optimisation process, summa

of static and transient finite element

Fig. 3: Proposed optimisation process whictransient finite element analysis techniques.

ic, as calculated through cycle. In this case, lacking

nts for the Nissan LEAF, a of 3500rpm / 40Nm was nt speed of about 35mph,

ance to the second, was that in order that it is consistent e, where low mass is an

of each step of the motor was defined, with the goal n order to consider motor

as well as motor mass, it is uating motor mass to motor ing a scaling constant to the

motor active mass for each elected; this selection was to be reasonable. A more

rive this value from a full

be calculated according to

. (1)

nal points to be assessed, ki ting point, ηi the efficiency factor which converts mass

ss arised in Fig. 3, is a hybrid analyses.

ch makes use of both static and

428

This process is fully automated through tBasic scripts, coupled to the optimisation tool

In order to confirm the first objective (acfull torque-speed characteristic) a series of stundertaken. For each geometry, a statcharacteristic is generated; this is used to sewinding turns to allow the achievement of pbase speed with the available supply current for full details). The static characteristic is tpredict current waveforms at base speedcharacteristics then form the basis of an assessufficient motoring current can be achieconverter’s voltage limit, in order to achieve that base speed (see Section II.E).

If this static analysis demonstrates that btorque can be achieved then the optimisationnext step. Should it not be possible to achievespeed / peak torque then the geometry is disregeometry generated.

The optimisation then assesses the seobjectives, more detail of which is provided isummary, current waveforms, estimated fromcollated earlier, are generated for each operatintransient with motion 2D simulation undertamotor output power along with both copper abe estimated, allowing calculation of motorefficiency.

The motor mass is then estimated and the ofor each candidate geometry calculated; thesused by the evolutionary algorithm to allow itan optimum solution is identified.

D. Static Prediction of Number of TurnsAs has been described, static finite elemen

used to assess each candidate geometry’s abrequired torque-speed characteristics. Static sas they significantly reduce solution time solution, and also because assumptions cansimplify the complex, non-linear behaviour of

The first stage of this process is to set the nuturns such that the peak motor current from thedeliver the required average motor torque wbase speed, assuming voltage control. For an is generally complex, requiring a number of traorder to set turns and advance angle. This difficult to automate and time consumingoptimisation. A process was therefore developa number of assumptions in order to make estimate of the number of turns required usinof static point solutions.

In Segmental Rotor SRMs, peak motor torproduced at around the mid-point between unaligned rotor positions; this is where the rflux linkage with rotor position is at its hsolution was therefore undertaken for variousin this rotor position and torque values takeelement software. These provide an in

the use of Visual lset. chievement of the tatic analyses are ic flux linkage et the number of eak torque below (see Section II.D then also used to d. These current ssment of whether eved, given the he required power

base speed / peak n proceeds to the

e the required base garded and a new

econd and third in Section II.F. In m the static data ng point and a full aken. This allows and iron losses to r electromagnetic

objective function e results are then t to progress until

s nt calculations are bility to meet the solutions are used

over a transient n be made which f SRMs. umber of winding e supply is able to

whilst operating at SRM this process ansient FE runs in was found to be

g as part of an ped which makes quick, simplified

ng a small number

rque is invariably the aligned and

rate of change of highest. A static s values of MMF

en from the finite ndication of the

instantaneous peak torque value wwhen the motor is running.

An estimation must then be madpeak motor torque scale to average ma representative characteristic, from a Segmental Rotor SRM operating acontrol with optimised phase turnsaverage torque of 278Nm, peak tortorque 203Nm and torque ripple ofripple and its asymmetry from the mSegmental Rotor SRM operating aspeed envelope.

This characteristic can thereforeestimating mean torque from peakgeometries such that:

. . 1

where Tpeak is the required peak motois the desired average torque, krippdecimal and koffset is a constant spemaximum and minimum torque arocase shown in Fig. 4, kripple wouldrepresentative of these machines opspeed.

Fig. 4: Representative torque waveform at bvoltage control, optimised winding turns and

As a target average torque of 280

be calculated that a peak torque orequired. In order to provide a furthebuild effects, a further factor of 1.0resulting in a target peak torque of 4

As the peak converter current is kstraightforward to set the turns suchwhilst still allowing a peak torque ofless than 500A phase current.

E. Static Prediction of CurrentThe objective of the next stage of

whether instantaneous peak torque (4torque (280Nm) and peak power within the available supply VA ratin

which would be achieved

de of how these values of motor torque. Fig. 4 shows finite element analysis, for

at base speed under voltage s and advance angle; here rque of 385Nm, minimum f 65%; this level of torque mean is representative of a at this point in the torque

e be used as the basis for k torque for other motor

1 (2)

or torque assuming that Tave

pple the torque ripple as a ecifying the asymmetry of ound the average. For the d be 0.65 and koffset 1.05, perating at this torque and

base speed, peak torque assuming

advance angle.

Nm is required, it can then of at least 390Nm will be er factor allowing for motor 05 is applied to this value, 410Nm. nown to be 500A, it is then h that turns are minimised f 429Nm to be achieved for

t Rise f the process is to estimate 410Nm) and hence average (80kW) can be achieved

ng at base speed (2750rpm).

429

This can be tested by confirming whether achieve full phase current (500A) by applying(400V) at base speed (2750rpm).

To make this assessment, firstly the staticthe peak torque position is again calculatenumber of winding turns derived in the prevcurrent is increased in steps between zero anflux linkage and torque recorded from thsoftware. In order to allow for end winding in3D machine, a fixed 15% is added to the bopeak torque flux linkages in order to simulate

This data then provides a mapping betweflux linkage and motor instantaneous peakinstantaneous peak torque required to achaverage motoring torque has been estimated ican be used to identify the flux linkage that win order to achieve this average torque.

In using this data to calculate whether tlinkage is achievable at a given speed, an advelectrical degrees is assumed, meaning that vobe applied with the rotor positioned 180deg ipeak torque position and therefore with the profile. Equally a conduction angle of 180meaning that full supply volts (in this case applied across this electrical angle.

By using the simple relationship: . (where φ is the flux linkage, v the voltag

phase and t the period over which that voltagtrivial to calculate whether at a given motosufficient time for the flux linkage, and therefoto climb to the level required.

As a result it is possible to use this anawhether the motor could achieve base speedascertaining whether it would be possible to pflux linkage, given available DC link volts, inthe required torque.

For each optimisation run this current riseused to either select, if peak torque could be aa particular geometry prior to moving onto optimisation.

F. Dynamic Assessment of Motor LossesOnce it is established that a motor geo

achieve peak torque at base speed, the noptimisation process is to undertake a traelement simulation in order to calculate lossoperating points.

For SRMs, a high fidelity assessment worequiring that, for each operating point, an oangle be selected and that a current contrimplemented. An alternative is to use a sicurrent source to approximate such a currentHowever it is important that a reasonable acurrent rise and fall is made so that iron loss ca

it is possible to g full supply volts

c characteristic in d. Assuming the vious step, phase nd 500 amps and

he finite element nductance in a real oth unaligned and this effect. en phase current, k torque. As the hieve the desired in Section D, this

would be required

this level of flux vance angle of 90 oltage would first in advance of the same inductance

0deg is assumed, 400V) would be

(3)

ge applied to the ge is applied) it is or speed there is

fore phase current,

alysis to confirm d, peak torque by produce sufficient n order to deliver

e test is therefore achieved, or reject the next stage of

s metry is able to next step in the ansient 2D finite es at the selected

ould be onerous, optimum advance ol scheme to be imple trapezoidal t control scheme. approximation of alculations remain

realistic. In order to simplify this process

approach, similar to that applied prea representative current waveform a

The required current to achieve thestimated from the static torque assmore the likely relationship between

The current waveform assumes torque position, with zero degreeconduction angle is 180deg. Statcharacteristics can again be used in oof current rise time. A static flux linkis calculated in the unaligned pcharacteristic for the peak torque would perhaps be better to also cacurrent characteristic several otherpeak torque position, this is not founthe fidelity of the operation (see Secharacteristics are instead estimaextrapolation of the other flux liunaligned and peak torque positions

These static characteristics are thephase inductance around the unalignwill rise and around the aligned pAgain equation (3) is used as the allowing the calculation of current ri

Fig. 5: Comparing the estimated current waveof motor losses and a full, PI current control w

These waveforms are used as the

motion 2D finite element analysis ofrequired operating points. Copper anand motor efficiency calculated. Finis calculated and each geometry com

III. OPTIMISATI

A single topology of Segmental comprising a machine featuring 12 and alternately wound teeth (see topology was considered, as with number of allowable stator to

s, a current rise estimation eviously, is used to develop s shown in Fig. 5.

he necessary torque is again sessment, considering once n peak and average torque. that it is centred on peak

e advance angle and that ic flux linkage / current

order to provide an estimate kage / current characteristic position and the existing position reused. Whilst it alculate the flux linkage / r electrical angles beyond nd to significantly improve ection V) and instead these ated as a simple linear nkage data points in the . en used to estimate average ned position where current

position where it will fall. basis of this calculation,

ise and fall times.

eform for the dynamic assessment waveform.

e basis of a transient with f the motor operating at the nd Iron losses are estimated nally the objective function mpared.

ION SETUP Rotor SRM is considered, stator slots, 10 stator poles Fig. 6). Only this single this class of machine the oth and rotor segment

430

combinations is limited. For a 3 phase single tooth wound machine combinations of either 6.x-5.x (indicating 6.x Stator Slots and 5.x Rotor Segments) or 6.x-7.x series, with x being a positive, even, integer number, are possible [11] due to geometrical limitations on the rotor geometry and other factors such as unbalanced magnetic pull. In the case of a 10,500 rpm machine, rotor segment numbers are further limited by electrical frequency. The ten segment rotor, considered the practical minimum, already has an electrical frequency of 1.75kHz, close to the limit for a power converter of this size.

Fig. 6: A 12-10, single tooth would Segmental Rotor SRM.

A parameterised, half model of the 12-10 motor is generated

such that all motor dimensions may be varied continuously. Basic design rules are applied in accordance with [5, 11]. These include that, for example, the gap between stator tooth tips be equal to the gap between rotor segments and that the rotor segment must have the same span as each stator tooth tip. No limitations are placed on motor outer diameter or length, with it being expected that the mass penalty function would limit total motor volume. Limitations are however placed on minor motor dimensions to avoid the creation of geometries where, for example, stator slots become closed or which otherwise are impractical.

IV. OPTIMISATION RESULTS Table 1 presents the results of a series of five optimisation

runs and contrasts them with the performance of the IPM Motor from Nissan’s LEAF electric vehicle. Despite the simplifications applied to the optimisation approach, optimisation runs took between 5 and 15 days to converge, running on a modern server.

In one case, SRSRM2, a motor that is both volumetrically smaller and lighter than the Nissan LEAF is predicted. However, as will be seen in Section VI, the development of this design into a practical prototype results in its performance becoming marginal when compared to the requirement to match the performance of the Nissan Leaf. In all other cases larger and heavier motors result from the optimisation and in fact it is the largest of the resulting, optimised machines (SRSRM4) which is selected for prototype manufacture (see section VI).

It is interesting to note that whilst each optimisation run produces a similar objective function, dimensions differ

significantly due to the randomness of the optimisation process. This suggests that the optimisation ‘surface’ is relatively flat.

TABLE 1

COMPARISON OF SEGMENTAL ROTOR SRM DESIGNS RESULTING FROM A SERIES OF FIVE OPTIMISATION RUNS AND THE NISSAN LEAF’S IPM MACHINE

Stat

or O

D

(mm

)

Stac

k L

engt

h (m

m)

Vol

ume

(l)

Act

ive

Mas

s (k

g)

Eff

icie

ncy

(%)

Obj

ectiv

e Fu

nctio

n

SRSRM1

241.4 151.5 6.9 35.8 97.5 93.9

SRSRM2

229.6 136.4 5.6 30.2 97.3 94.3

SRSRM3

191.0 241.9 6.9 37.1 97.2 93.5

SRSRM4

246.2 159.1 7.6 42.1 97.3 93.1

SRSRM5

250.0 155.2 7.6 38.6 97.5 93.6

Nissan LEAF IPM

200.4 151.5 5.7 32.0 Not Known

-

V. FINITE ELEMENT VALIDATION OF OPTIMISATION Despite its limitations, the SRSRM2 design still forms a

valid basis upon which to validate the assumptions which underpinned the optimisation process; even with this marginal design the optimisation process must still be shown to provide valid results, based on the approach and assumptions described in Section II.

Two operating points are considered; firstly base speed, peak torque, validating the current rise modelling. For the solution to be valid the current rise would need to be sufficient to allow the optimised machine to produce the required output power / torque. The FE analysis assumed that the motor was running at an advance angle of 90degrees, with 180degree conduction period and under current control.. The results of this analysis are summarised in Table 2. Current and Torque graphs are shown in Fig. 7 and Fig. 8 respectively with a flux density plot of the motor being shown in Fig. 9.

TABLE 2

COMPARING THE RESULTS OF BASE SPEED PEAK TORQUE (2750RPM, 280NM) ANALYSIS FROM ESTIMATES DURING OPTIMISATION AND FULL 2D FE.

Estimate from Optimisation

Results from 2D FE

Peak Current

499.2A 496.7A

Time for Current to rise to Peak

996.6μs 945.4μs

Peak Torque

410.0Nm 356.5Nm

Average Torque

280.0Nm 311.1Nm

431

Fig. 7: Phase current waveforms for motor at base speed

Fig. 8: Torque characteristic for motor operating at base s

Fig. 9: Segmental Rotor SRM R3-2, resulting from theprocess. Finite element plot shows motor operating at bas

TABLE 3 COMPARING THE RESULTS OF ANALYSIS OF MOTOR PERFO

/ 40NM FROM THE OPTIMISATION PROCESS AND FULL

Estimate from Optimisation

R

Average Torque 38.9Nm Mechanical Power 14.3kW RMS Current 39.9A Copper Loss(1) 168.3W Iron Loss 375.6W Total Loss 543.8W Efficiency 96.3%

(1) Copper loss assumes winding temperature o

/ peak torque.

speed / peak torque.

e hybrid optimisation se speed peak torque.

ORMANCE AT 3500RPM 2D FE ANALYSIS. Results from 2D

FE 36.0Nm 13.2kW 38.2A

154.0W 289.5W 443.5W 96.7%

f 150degC

A higher average torque was predelement model than that predictedprocess. This is consistent with thwhich is to ensure that the optimisperformance when modelled in 2D, build effects.

However peak torque is found to is because the combination of 90 degdegree conduction angle led tinstantaneous negative torque, as shof advance and conduction angles w

The second test was to compare operating point (3500rpm, 40Nmprocess with a full PWM based curreFE. The results of this analysis are s

The results showed close correlatdifference between the level of torquFE modelled efficiencies matched cl

VI. CONSIDERATIONS FOR A

Whilst the optimisation process iscorrelation to a more realistic, curanalysis, such an analysis is in its ow

A number of other factors therefaccount: these include 3D effects, cgeometry in order to make it manthermal considerations.

As has been discussed earlier, significant impact on the base speedthe motor. A static, 3D FE analysis hshowed that 3D effects would increaligned and unaligned positions validating the assumptions made dthis effect is allowed for in the motowould not adversely impact low speremains a risk that in a practical macwould be reduced sufficiently to lim

More significantly, it was necessathe rotor segment root in order sufficiently robust to operate at overspeed allowance. The revised resulted in an increase in unaligimpacting both absolute motor torqand also increasing the current ristorque production at base speed.

Finally, thermal considerations continuous and short term opermachines. Modified as discussed abdesign would produce losses in the oand peak torque. This has been fouthermal modelling, with windings etemperature rise of greater than 150dwould have to potentially be de-rateprovide performance comparable toLEAF.

As a result the SRSRM2 motor d

dicted by the detailed finite d during the optimisation he optimisation objective, sed motor has a margin of

allowing for 3D and other

be less than predicted; this gree advance angle and 180 to significant levels of hown in Fig. 8. Adjustment would reduce this effect.

the results for the selected m) from the optimisation ent control simulation in 2D hown in Table 3.

tion, whilst there was some ue produced, predicted and losely.

A PRACTICAL MACHINE s shown to result in a good rrent control based 2D FE wn right very idealised. fore need to be taken into changes made to the motor nufacturable and resulting

3D effects will have a d / peak torque capability of has been undertaken which ease inductance in both the

by approximately 14%; uring optimisation. Whilst

or optimisation process and eed torque capability, there chine the rate of current rise

mit base speed torque. ary to modify the design of to allow the rotor to be

12,000 rpm plus 20% rotor segment root design

gned inductance of 10%, que capability per unit loss se time constant, limiting

are expected to limit the ration of these electrical bove, the SRSRM2 motor

order of 14kW at base speed und to be too high through experiencing a steady state degC. As a result the motor ed and therefore would not o that fitted to the Nissan

esign was considered to be

432

too marginal for construction as a prototype. Instead the SRSRM4 was selected in order to provide more comparable performance. At time of writing, a prototype motor is under construction, based on this this design (see Table 1 for details). Fig. 10 shows the motor frame when compared to IPM traction motor from Nissan’s LEAF electric vehicle. This prototype motor will be used as the basis for a full series of static and dynamic tests across the required torque, speed range.

Fig. 10: Showing the resulting Segmental Rotor SRM frame alongside the Nissan’s IPM Motor used in the Nissan LEAF.

VII. CONCLUSIONS Segmental Rotor SRMs are challenging to optimise because

of their non-linear behaviour. The methodology presented in the paper has combined static and transient finite element model runs in order to allow a comprehensive optimisation of these machines.

Static runs have been used to estimate the maximum performance of the motor, confirming that full power can be achieved at base speed. This approach has been shown to allow the development of a valid pass/fail criteria based on the required peak performance of the motor.

Static characteristics have also been used to develop estimated current waveforms which have then in turn be used to support a transient analysis of motor losses and therefore efficiency; these have also shown a good correlation with simulated PWM current control approach applied in 2D FE.

Whilst this optimisation approach has been found to take considerable computer times (between 10 and 20 days per run) it has also nevertheless be shown to be significantly faster and more reliable than methods based on fully dynamic analysis. Whilst slower than techniques based solely on static FE calculations, it has also shown to be more robust as it allows an assessment of both copper and iron losses as well as taking full account of magnetic interactions between phases.

Optimisation has shown that Segmental Rotor SRMs are likely to be able to produce performance equivalent to a class leading IPM, however practical machines are expected to be 30% larger and larger and have similarly higher mass. However they will also be significantly cheaper, requiring no rare earth magnets and operating with a power converter of the same VA

rating as that used with a permanent magnet machine.

REFERENCES

[1] D. G. Dorrell, et al., "Comparison of different motor design drives for hybrid electric vehicles," in Energy Conversion Congress and Exposition (ECCE), 2010 IEEE, 2010, pp. 3352-3359.

[2] H. Neudorfer, et al., "Comparison of three different electric powertrains for the use in hybrid electric vehicles," in Power Electronics, Machines and Drives, 2008. PEMD 2008. 4th IET Conference on, 2008, pp. 510-514.

[3] Z. Q. Zhu and C. C. Chan, "Electrical machine topologies and technologies for electric, hybrid, and fuel cell vehicles," in Vehicle Power and Propulsion Conference, 2008. VPPC '08. IEEE, 2008, pp. 1-6.

[4] Ishikawa S. Sato Y., Okubo T., Abe M. and Tamai K., "Development of High Response Motor and Inverter System for the Nissan LEAF Electric Vehicle," presented at the SAE 2011 World Congress & Exhibition, Detroit, Michigan, United States, 2011.

[5] B. C. Mecrow, et al., "Segmental Rotor switched reluctance motors with single-tooth windings," Electric Power Applications, IEE Proceedings -, vol. 150, pp. 591-599, 2003.

[6] B. C. Mecrow, et al., "Switched reluctance motors with Segmental Rotors," Electric Power Applications, IEE Proceedings -, vol. 149, pp. 245-254, 2002.

[7] Xiaoyuan Chen, et al., "New designs of switched reluctance motors with Segmental Rotors," presented at the PEDM 2010, Brighton, UK, 2010.

[8] J. Oyama, et al., "The fundamental characteristics of novel switched reluctance motor with segment core embedded in aluminum rotor block," in Electrical Machines and Systems, 2005. ICEMS 2005. Proceedings of the Eighth International Conference on, 2005, pp. 515-519 Vol. 1.

[9] R. Vandana, et al., "A Novel High Power Density Segmented Switched Reluctance Machine," in Industry Applications Society Annual Meeting, 2008. IAS '08. IEEE, 2008, pp. 1-7.

[10] N. Vattikuti, et al., "A novel high torque and low weight segmented switched reluctance motor," in Power Electronics Specialists Conference, 2008. PESC 2008. IEEE, 2008, pp. 1223-1228.

[11] J. D. Widmer and B. C. Mecrow, "Optimised Segmental Rotor Switched Reluctance Machines with a greater number of rotor segments than stator slots," in Electric Machines & Drives Conference (IEMDC), 2011 IEEE International, 2011, pp. 1183-1188.

[12] T. J. E Miller, Switched Reluctance Motors and their Control. Oxford: Magna Physics Publications - Oxford University Press, 1993.

[13] P. Neittaanmaki, et al., Inverse Problem and Optimal Design in Electricity and Magnetism: Oxford University Press, 1996.

433