DuPlessis F Skripsie2020

UNIVERSITEIT•STELLENBOSCH•UNIVERSITY
jou kennisvennoot • your knowledge partner
Design and comparison of three

surface-mounted PM motors for a light
electric vehicle
Francois Du Plessis
20688830
Report submitted in partial fulfilment of the requirements of the module

Project (E) 448 for the degree Baccalaureus in Engineering in the Department of
Electrical and Electronic Engineering at Stellenbosch University.
Supervisor: Prof RJ Wang
November 2020
Acknowledgements
Aan Jesus Christus, my Verlosser, wat my elke dag krag gee.

Aan Carla, vir jou ondersteuning en liefde.
Aan my ouers, Fanie en Nicolene, vir my grootmaak en bring tot waar ek vandag is.
Aan Kobus, vir al jou hulp en kennis die laaste vier jaar.
To Professor Wang and Stavros Pastellides, for your guidance and support during the
project.
i
UNIVERSITEIT•STELLENBOSCH•UNIVERSITY
jou kennisvennoot • your knowledge partner
Plagiaatverklaring / Plagiarism Declaration
1. Plagiaat is die oorneem en gebruik van die idees, materiaal en ander intellektuele
eiendom van ander persone asof dit jou eie werk is.
Plagiarism is the use of ideas, material and other intellectual property of another’s
work and to present is as my own.
2. Ek erken dat die pleeg van plagiaat ’n strafbare oortreding is aangesien dit ’n vorm
van diefstal is.
I agree that plagiarism is a punishable offence because it constitutes theft.
3. Ek verstaan ook dat direkte vertalings plagiaat is.

I also understand that direct translations are plagiarism.
4. Dienooreenkomstig is alle aanhalings en bydraes vanuit enige bron (ingesluit die

internet) volledig verwys (erken). Ek erken dat die woordelikse aanhaal van teks
sonder aanhalingstekens (selfs al word die bron volledig erken) plagiaat is.
Accordingly all quotations and contributions from any source whatsoever (including
the internet) have been cited fully. I understand that the reproduction of text without
quotation marks (even when the source is cited) is plagiarism
5. Ek verklaar dat die werk in hierdie skryfstuk vervat, behalwe waar anders aangedui,
my eie oorspronklike werk is en dat ek dit nie vantevore in die geheel of gedeeltelik
ingehandig het vir bepunting in hierdie module/werkstuk of ’n ander module/werk-
stuk nie.
I declare that the work contained in this assignment, except where otherwise stated, is
my original work and that I have not previously (in its entirety or in part) submitted
it for grading in this module/assignment or another module/assignment.
Studentenommer / Student number Handtekening / Signature
Voorletters en van / Initials and surname Datum / Date
ii
Abstract
English
This report details the design, analysis and comparison of three surface mounted perma-
nent magnet synchronous motors which feature fractional-slot windings. The concepts
surrounding electric vehicles, traction requirements, permanent magnet technology and
winding topologies are introduced and discussed. Specifically, the concept of fractional-
slot windings, both concentrated and distributed windings, is introduced. Operational
principles of surface mounted permanent magnet synchronous motors are introduced
and discussed along with flux weakening operation of these machines. A design method
is proposed and implemented to obtain initial machine dimensions before the machine
dimensions are optimised to find the lowest cost at which all constraints are satisfied.
Finite-element analysis is used to evaluate machine performance and compare the chosen
winding topologies. The conclusion is drawn that the double-layer concentrated winding
machine provides the best performance w.r.t. cost, with the distributed winding machine
performing marginally worse taking cost into account.
Afrikaans
Hierdie verslag beskryf die ontwerp, analise en vergelyking van drie oppervlak gemonteerde
permanente magneet sinchroon motors met breukdeel gleuf windings (FSCW). Die konsepte
rakende elektriese motors (karre), trekkrag vereistes, permanente magneet tegnologie and
winding uitleg word bekendgestel en bespreek. The konsep van breukdeel gleuf windings,
beide gekonsentreerde en verspreide windings, word bekendgestel en bespreek. Operationele
beginsels van permanente magneet sinchroon motors word bekendgestel en bespreek asook
vloed verswakkende beheer en operasie. ’n Ontwerps prosedure word voorgestel en uitgevoer
om aanvanklike masjien dimensies te vind voor ’n optimiseringsproses uitgevoer word om
die laagste moontlike kostepunt te vind waarby alle beperkings bevredig word. Eindige-
element analise word gebruik om masjien prestasie te evalueer en om die gekose winding
topologië te vergelyk. Daar word tot die gevolgtrekking gekom dat die dubbel laag
gekonsentreerde winding masjien die beste presteer t.o.v. koste, terwyl die verspreide
winding masjien marginaal slegter vaar met koste in ag geneem.
iii
Contents
Declaration ii
Abstract iii
List of Figures vii
List of Tables ix
Nomenclature xi
1. Introduction 1
1.1. Background of the project . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2. Objectives of the project . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.3. Scope of the project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.4. Layout of the report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2. Traction motor drive-train systems 3

2.1. Traction efforts and vehicle physics model . . . . . . . . . . . . . . . . . . 3
2.2. Typical EV drive train systems . . . . . . . . . . . . . . . . . . . . . . . . 4
2.3. Common permanent magnet motor topologies . . . . . . . . . . . . . . . . 5
2.3.1. Surface-mounted PM motors (SPM) . . . . . . . . . . . . . . . . . 6
2.3.2. Interior PM motors (IPM) . . . . . . . . . . . . . . . . . . . . . . . 6
2.4. Surface-mounted PM motor technologies and fractional-slot machines . . . 7
2.5. Choice of topology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3. Characteristics of PM traction motors 10

3.1. Electromagnetic principles of PM motors . . . . . . . . . . . . . . . . . . . 10
3.1.1. Open circuit magnetic field, flux linkage and back-EMF . . . . . . . 10
3.1.2. Phase inductance, winding resistance and iron losses . . . . . . . . 11
3.2. Flux weakening operation of PM traction motors . . . . . . . . . . . . . . 12
3.2.1. CPSR and torque angle charcteristics . . . . . . . . . . . . . . . . . 14
4. Design procedure of PM motor 16

4.1. Design process and specifications for the study . . . . . . . . . . . . . . . . 16
4.2. Pole and slot combinations . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4.3. Winding layouts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
iv
4.3.1. Winding layout: 12-slot 10-pole machine . . . . . . . . . . . . . . . 20
4.3.2. Winding layout: 36-slot 8-pole machine . . . . . . . . . . . . . . . . 20
4.4. Initial sizing of design parameters . . . . . . . . . . . . . . . . . . . . . . . 22
4.4.1. Geometric dimensions . . . . . . . . . . . . . . . . . . . . . . . . . 22
4.4.2. Calculation of number of turns per phase . . . . . . . . . . . . . . . 24
4.5. Additional considerations for final design . . . . . . . . . . . . . . . . . . . 25
4.5.1. Cogging torque and torque ripple . . . . . . . . . . . . . . . . . . . 25
4.5.2. Demagnetization aspects . . . . . . . . . . . . . . . . . . . . . . . . 26
4.6. Optimum design of PM motors . . . . . . . . . . . . . . . . . . . . . . . . 27
4.6.1. Formulation of optimisation problem . . . . . . . . . . . . . . . . . 27
4.6.2. Finite element based optimisation process . . . . . . . . . . . . . . 28
4.6.3. Finite element based simulation methods . . . . . . . . . . . . . . . 29
5. Comparative study of two SPM motors 30

5.1. Dimensions, inductance and cost . . . . . . . . . . . . . . . . . . . . . . . 30
5.2. Main performance characteristics . . . . . . . . . . . . . . . . . . . . . . . 32
5.2.1. Losses and efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . 33
5.2.2. Flux linkage, voltage and current . . . . . . . . . . . . . . . . . . . 35
5.2.3. Output torque, torque ripple and cogging torque . . . . . . . . . . . 38
5.2.4. Demagnetisation and flux densities . . . . . . . . . . . . . . . . . . 38
6. Conclusion and Further work 40

6.1. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
6.2. Recommendations for future work . . . . . . . . . . . . . . . . . . . . . . . 40
Bibliography 41
A. Project Planning Schedule 45
B. Outcomes Compliance 47
B.1. ELO 1: Problem solving . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
B.2. ELO 2: Application of scientific and engineering knowledge . . . . . . . . . 47
B.3. ELO 3: Engineering design . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
B.4. ELO 4: Investigations, experiments and data analysis . . . . . . . . . . . . 48
B.5. ELO 5: Engineering methods, skills and tools, including Information Tech-
nology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
B.6. ELO 6: Professional and technical communication . . . . . . . . . . . . . . 49
B.7. ELO 8: Individual work . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
B.8. ELO 9: Independent Learning Ability . . . . . . . . . . . . . . . . . . . . . 49
v
C. Additional results, equations and derivations 50
C.1. Flux density, thermal(current) aspects and efficiency maps . . . . . . . . . 50
C.2. Results from SEMFEM simulations . . . . . . . . . . . . . . . . . . . . . . 50
C.3. Electrical steel properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
C.4. Open circuit magnetic field calculation . . . . . . . . . . . . . . . . . . . . 55
C.5. Additional inductance and resistance calculations . . . . . . . . . . . . . . 56
D. MATLAB design scripts 57
E. SEMFEM simulation code 61
vi
List of Figures
2.1. Vehicle movement models . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2.2. EV drive trains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.3. Energy conversion system and drive train . . . . . . . . . . . . . . . . . . . 5
2.4. PM motor magnet configurations (4-pole machines) . . . . . . . . . . . . . 7
2.5. Winding topologies for a 4 pole machine: (a) Overlapping distributed
winding (q = 2) (b) Overlapping concentrated winding (q = 1) (c) Non-
overlapping winding with all teeth wound(double-layer) (q = 0.5) (d) Non-
overlapping winding with alternate teeth wound(single-layer) (q = 0.5) . . 8
3.1. Winding coil diagrams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

3.2. Additional dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.3. dq-equivalent circuits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.4. Voltage and current behaviour over speedrange . . . . . . . . . . . . . . . . 15
3.5. Voltage and current behaviour over speed range . . . . . . . . . . . . . . . 15
3.6. Torque angle characteristic . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
4.1. Design process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

4.2. Winding factors for different slot(Ns ) and pole (P) combinations. . . . . . 19
4.3. Winding layouts for 12-slot 10-pole machine . . . . . . . . . . . . . . . . . 21
4.4. Phasor diagrams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.5. Double-layer winding of 36-slot 8-pole machine . . . . . . . . . . . . . . . . 21
4.6. Slot dimensions as used for initial geometric sizing . . . . . . . . . . . . . . 24
4.7. Demagnetization characteristics . . . . . . . . . . . . . . . . . . . . . . . . 27
4.8. Optimisation process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
4.9. Delta-wye connection of motor . . . . . . . . . . . . . . . . . . . . . . . . . 29
5.1. Slot dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

5.2. Preliminary optimised machines . . . . . . . . . . . . . . . . . . . . . . . . 32
5.3. Mesh used in SEMFEM . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
5.4. Mesh used in ANSYS Maxwell . . . . . . . . . . . . . . . . . . . . . . . . . 32
5.5. Magnet losses at 1500 rpm . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
5.6. Magnet losses at 5000 rpm . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
5.7. Flux linkage at 1500 rpm . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
5.8. Induced voltage (back-emf) at 1500 rpm . . . . . . . . . . . . . . . . . . . 35
vii
5.9. Flux linkage at 5000 rpm . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
5.10. Induced voltage (back-emf) at 5000 rpm . . . . . . . . . . . . . . . . . . . 36
5.11. Phase current at 1500 rpm . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
5.12. Phase current at 5000 rpm . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
5.13. Output torque at 1500 rpm . . . . . . . . . . . . . . . . . . . . . . . . . . 37
5.14. Output torque at 5000 rpm . . . . . . . . . . . . . . . . . . . . . . . . . . 37
5.15. Mesh used for cogging torque determination . . . . . . . . . . . . . . . . . 38
5.16. Cogging torque . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
5.17. Flux densities of 12-slot machines at 1500 rpm . . . . . . . . . . . . . . . . 39
5.18. Flux densities of 36-slot machine . . . . . . . . . . . . . . . . . . . . . . . 39
A.1. Initial planning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

A.2. Revised planning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
C.1. Flux densities of 12-slot machines at 5000 rpm . . . . . . . . . . . . . . . . 51

C.2. Efficiency map of 12-slot single-layer machine . . . . . . . . . . . . . . . . 51
C.3. Loss map of 12-slot single-layer machine . . . . . . . . . . . . . . . . . . . 52
C.4. Efficiency map of 12-slot double-layer machine . . . . . . . . . . . . . . . . 52
C.5. Loss map of 12-slot double-layer machine . . . . . . . . . . . . . . . . . . . 53
C.6. Efficiency map of 36-slot machine . . . . . . . . . . . . . . . . . . . . . . . 53
C.7. Loss map of 36-slot machine . . . . . . . . . . . . . . . . . . . . . . . . . . 54
C.8. M19 26G BH-curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
C.9. Machine dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
viii
List of Tables
4.1. Dimensions and specifications of EV drive train . . . . . . . . . . . . . . . 17

4.2. Dimensions and specifications of SPM . . . . . . . . . . . . . . . . . . . . . 18
4.3. Possible winding layouts . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.4. Winding factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
4.5. Current limits for air-cooled PMSMs . . . . . . . . . . . . . . . . . . . . . 22
4.6. Input parameters for design at 1500 rpm . . . . . . . . . . . . . . . . . . . 22
4.7. Initial geometric parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4.8. Cogging torque parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.9. dq-current densities of machines . . . . . . . . . . . . . . . . . . . . . . . . 29
5.1. Inductances of machines . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

5.2. Main machine dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
5.3. Slot parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
5.4. Cost and mass of final machines . . . . . . . . . . . . . . . . . . . . . . . . 31
5.5. Results at 1500 rpm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
5.6. Results at 5000 rpm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
5.7. Phase angle of terminal voltages . . . . . . . . . . . . . . . . . . . . . . . . 35
5.8. Cogging torque of different machines . . . . . . . . . . . . . . . . . . . . . 38
C.1. Current limits for air-cooled PMSMs . . . . . . . . . . . . . . . . . . . . . 50

C.2. Preliminary design results at 1500 rpm (SEMFEM) . . . . . . . . . . . . . 51
C.3. Preliminary design results at 5000 rpm (SEMFEM) . . . . . . . . . . . . . 54
ix
x
Nomenclature
Variables and functions
ACu Copper conductor area.

Af Frontal area of vehicle.
Aslot Slot area.
Aus Usable slot area.
Bairgap (r, θ) Air-gap flux density considering slotting effects.
Bg Approximate air gap flux density amplitude.
Bg1 Fundamental air gap flux density amplitude.
B̂m Peak air-gap flux density.
BM M Magnetisation margin.
Br Remnant magnet flux density.
Bslotless (r, θ) Air-gap flux density neglecting slotting effects.
Bts Stator tooth magnetic flux density.
Byr Rotor yoke magnetic flux density.
Bys Stator yoke magnetic flux density.
CD Aerodynamic Drag coefficient.
CT Cogging torque factor.
Dag Air gap diameter.
Der External rotor diameter.
Dh1 Diameter after slot enclosure.
Dis Internal stator diameter.
Dout Stator outer diameter.
Dshaf t Shaft diameter.
Em Phase back EMF (RMS).
Erated Open circuit rated line to neutral, RMS voltage.
eRM S Line-to-neutral phase back EMF (RMS).
f, fe Electrical frequency.
fBrn Air-gap flux density coefficient.
xi
ff Slot fill factor.
Fr Rolling resistance.
fr Rolling resistance coefficient.
Ft Tractive effort.
Fw Aerodynamic drag.
g Air gap length.
ggrav Gravitational acceleration (9.81 m/s2 ).
gef f Effective air gap.
hm Magnet thickness.
hys Stator yoke height.
hus Usable slot height.
h1 Slot enclosure depth.
Ich Characteristic current.
id d-axis current.
ig Gear ratio.
Imax Maximum phase current.
io Differential ratio.
iq q-axis current.
IR Rated phase current.
IS Phase current.
J Winding current density.
KBn Air-gap flux density coefficient.
Kc Carter factor.
kj Stacking factor.
Ks Linear current density.
Kwn nth harmonic winding factor.
kwv v th harmonic winding factor.
Kw1 Fundamental winding factor.
KCP SR Constant power speed range ratio.
L, l Length of stator.
lc Approximate coil length.
Ld d-axis inductance.
lew End winding length.
xii
Lew End winding leakage inductance.
Lh Harmonic leakage inductance.
LL Leakage inductance.
Lm Magnetizing inductance.
Lq q-axis inductance.
Ls Synchronous inductance.
l0 Effective stator length.
Lσ Slot leakage inductance.
m Number of phases.
Mv Vehicle mass.
N Cogging torque factor.
Nc Number of turns per coil.
NC Smallest common multiple of Qs and 2p.
Np Rotational speed of electric motor shaft.
Nph , Ns Number of turns per phase.
p Number of pole pairs.
PCore Core losses.
PCu Copper losses.
PF e Iron losses.
Pin Electrical input power.
Ploss Peak total electrical losses.
Pmagnet Magnet losses.
Pout Electric motor output power.
q Number of slots per pole and phase.
Q, Qs Number of stator slots.
rd Wheel radius.
RDC DC winding resistance neglecting skin effect.
Ris Radius of inner stator.
RS Stator winding resistance.
t Time/winding periodicity.
TM G Output torque of magnetic gear.
Tp Output torque from electric motor.
Tr Rated torque at base speed.
xiii
Trip Torque ripple.
Twinding Winding temperature.
V Vehicle speed.
van Line-to-neutral terminal voltage.
vd d-axis terminal voltage.
Vmax Maximum terminal voltage.
vq q-axis terminal voltage.
W Coil pitch in unit of length.
Wew End winding width.
wsb Bottom slot width.
wst Top slot width.
wtt Top tooth width.
w0 Slot opening width.
zQ Number of conductors per slot.
zwire Number of wires per conductor.
αp Magnet pitch to pole pitch ratio.
αu Slot pitch angle.
βT Current angle.
γ Chosen coil pitch in number of slots.
γQ Calculated coil pitch in number of slots.
δRF Torque angle.
η Motor efficiency.
ηt Drive train efficiency.
θ Mechanical angle.
θdeg Phase angle in degrees.
θelec Electrical angle.
θmech Mechanical angle.
λ Flux linkage.
λa Armature flux linkage.
λcoil Coil flux linkage.
λd d-axis flux linkage.
λew End winding permeance factor.
λphase Phase flux linkage.
xiv
λP M Permanent magnet flux linkage.
λq q-axis flux linkage.
λrel (θ) Relative permeance function.
λw End winding permeance factor.
µr Relative permeability.
µ0 Magnetic permeability of free space (4π × 10−7 H/m).
ρair Air density (1.225 kg/m3 ).
ρCu Copper resistivity (0.0171 Ω · mm2 /m).
ρF e Electrical steel resistivity.
σu Slot leakage factor.
τp Pole pitch.
τs Slot pitch angle.
φP Air-gap flux per pole
φrot Flux in rotor.
ψ Length-diameter ratio.
ωe Angular electrical frequency.
ωmech , ωp Rotational speed of electric motor shaft.
xv
Acronyms and abbreviations
CPSR Constant power speed range

CTSR Constant torque speed range
dq direct-quadrature
EMF Electromotive force
EV Electric vehicle
FEA Finite element analysis
FEM Finite element method
FSCW Fractional slot concentrated winding
FSDW Fractional slot distributed winding
GCD Greatest common divisor
IMD Integrated motor drive
IPM Interior permanent magnet motor
LEV Light electric vehicle
MMF Magnetomotive force
PM Permanent magnet
PMSM Permanent magnet synchronous motor
RBPM Radial-bar permanent magnet motor
RMS Root mean square
SBM Surface buried permanent magnet motor
SMC Soft magnetic composite
SPM Surface mounted permanent magnet motor
STPM Spoke-type permanent magnet motor
VPM V-shaped permanent magnet motor
xvi
Chapter 1
Introduction
1.1. Background of the project

Globally, there are increasing concerns surrounding climate change, fossil fuel supply
and energy security [1]. With transportation predicted to contribute 50% of global
greenhouse emissions by 2030 [2], electric vehicles (EVs) offer the potential to reduce
the environmental impact of transportation and increase energy independence, if paired
with renewable energy infrastructure. High-performance electric vehicle drive motors are
one of the critical components needed for the widespread adoption of EVs. Permanent
magnet synchronous motors (PMSM) have been widely adopted in recent years for their
simple structure and high efficiency. However, several aspects require more research [3].
Additional technologies such as integrated motor drives (IMDs) offer significant advantages
when combined with PMSMs [4], [5].
1.2. Objectives of the project

The main objective of the project is to carry out a comparative study on different
configurations of surface-mounted PMSMs through the use of FEA software, with the
intended application in a light electric vehicle (LEV) as described in [6]. This study must
highlight the advantages of the different configurations and evaluate the performance
thereof. Secondary to this, a design procedure must be developed and implemented for
the design of the different machines which is reproducible and takes into account practical
considerations regarding aspects such as manufacturing.
1.3. Scope of the project

The project involves the design and comparison of three surface-mounted PMSMs through
the use of FEA software, designed under the assumption of flux-weakening control. The
machines are not to be built for this study, however practical considerations are to be
taken into account. While additional concepts such as EV drive trains and magnetic gears
are discussed, these are not the focus of this project. In-depth demagnetisation analysis is
1
not to be carried out. While harmonic content is minimised by machine choice, harmonic
analysis of the machines is not carried out. Thermal aspects are considered, but a detailed
analysis of the thermal aspects of the final machines is not carried out.
1.4. Layout of the report

The report will be structured as follows:
• Chapter 2 provides the necessary background information to understand EV traction

requirements, drive train configurations as used in the broader project, PM technolo-
gies and winding configurations focusing on introducing fractional-slot concentrated
winding machines to the reader.
• Chapter 3 gives an overview of the electromagnetic principles of PMSMs. Properties

such as inductance, resistance and losses are discussed with a focus on these quantities
as found in fractional-slot machines. Flux-weakening operation is discussed, the
concept of a constant power speed range in introduced and torque angle characteristics
are discussed.
• Chapter 4 provides detail on the design procedure followed including the selection of
machines w.r.t pole and slot combinations, winding layouts, initial machine sizing
and the optimisation procedure implemented to find the optimal design. Additional
considerations such as cogging torque and demagnetisation are discussed along with
simulation specific parameters.
• Chapter 5 presents the results obtained from both FEA software packages used, one
in-house package and the commercial standard, provides a discussion of these results
and possible reasons for performance charcteristics.
• Chapter 6 provides a summary of the project, a conclusion regarding the comparative

study of the machines and possible improvements and recommendations for future
work.
2
Chapter 2
Traction motor drive-train systems
This chapter describes the vehicle physics model and tractive effort, typical EV drive train
systems and electric motor topologies. SPM technologies and winding topologies are also
introduced and discussed.
2.1. Traction efforts and vehicle physics model

The vehicle movement model with forces acting on the vehicle is shown in Fig. 2.1. The
tractive effort, Ft , is the force that propels the vehicle forward and acts on the vehicle
at the contact area between the tyres and the driving surface. Acting in the opposite
direction to vehicle movement is aerodynamic drag, Fw , and rolling resistance, Fr , given by
Eq. 2.1 and Eq. 2.2 respectively [7]. Applying Newton’s second law under the condition
of zero acceleration, Eq. 2.3 is obtained.
1
Fw = ρair CD Af V 2 (2.1)
2
Fr = Mv ggrav fr (2.2)
Ft = Fw + Fr (2.3)
The torque produced by the electric motor, Tp , is related to Ft by Eq. 2.4, with ig , io ,
ηt the magnetic gear ratio, differential gear ratio and the drive train efficiency respectively.
(a) Vehicle model (b) Model of wheel with rolling resistance indicated
Figure 2.1: Vehicle movement models
3
Tp ig io ηt
Ft = (2.4)
rd
The maximum rotational speed of the motor can be found using Eq. 2.5, with Np(max) ,
Vmax the maximum rotational speed of the motor in revolutions per minute (rpm) and
the maximum speed of the vehicle respectively. The power of the electric motor can be
obtained through Eq. 2.6, with Pout , ωp(max) the output power of the electric motor and
the maximum rotational speed of the motor in rad/s respectively .
πNp(max) rd
Vmax (m/s) = (2.5)
30ig io
Pout = Tp(min) ωp(max) (2.6)
2.2. Typical EV drive train systems

Typical drive train layouts for EV’s are shown in Fig. 2.2 [7]. Note that all layouts are for
rear-wheel drive vehicles. Fig. 2.2a shows the layout typically associated with an internal
combustion engine vehicle, with the engine replaced with an electric motor. Fig. 2.2b
features a similar layout to Fig. 2.2a, however the clutch is removed as the gearbox now
has a fixed transmission ratio. Fig. 2.2c shows a layout where the drive train components
are combined into a single unit, with both axle sides connected to the wheels through a
differential and fixed speed gearbox. Fig. 2.2d shows a layout where each of the rear-wheels
is driven by an independent motor-gearbox combination. Fig. 2.2e is very similar, with
the exception that the fixed speed gearbox is now integrated into the wheel assembly.
Lastly, Fig. 2.2f shows a layout where the electric motor is directly integrated into the
wheel assembly, through the use of an outer rotor motor. An expanded diagram of the
energy conversion system similar to Fig. 2.2b is shown in Fig. 2.3 [7]. A version of this
system is to be implemented in this project, using a magnetic gear [8] with a fixed ratio
and combining the power electronics and motor into a single assembly, called an integrated
motor drive (IMD) [4], [5]. The magnetic gear limits the maximum torque production
of the motor, as shown in Eq. 2.7, with TM G(max) the maximum output torque of the
magnetic gear. With the output power known from Eq. 2.6 and the maximum torque
chosen to be less than Tp(max) , the base speed, ωp(base) , can be obtained from Eq. 2.8.
TM G(max)
Tp(max) = (2.7)
ig
Pout
ωp(base) = (2.8)
Tp(max)
4
(a) (b)
(c) (d)
(e) (f)
Figure 2.2: EV drive trains
Figure 2.3: Energy conversion system and drive train
2.3. Common permanent magnet motor topologies

Substituting conventional field windings with permanent magnets gives the advantage
of reduced copper losses and increased reliability due to the removal of brushes [9].
Neodymium iron boron (NdFeB) magnets are the most powerful magnets available -
providing the strongest magnetic field per unit volume [10]. Accordingly, these magnets
enable a reduction in magnet volume for the same magnetic field strength and reduce
machine size [11]. However, these magnets may vary in price and availability due to
supply-chain concentration and volatility of the rare-earth elements [11]. Various efforts
have been made to find alternatives, including recycling magnets and using ferrite magnets,
but NdFeB magnets remain dominant [11], [12]. There are many different PM rotor
topologies for PM motors, some of which are shown in Fig. 2.4. These can be split into
two main categories, i.e. surface-mounted and interior PM motors.
5
2.3.1. Surface-mounted PM motors (SPM)
The main advantages offered by SPMs are low cost due to simplistic manufacturing
and high performance, as the main field produced by the magnets is practically in the
air gap of the machine and leakage flux is reduced [9], [13]. Since the magnets have a
permeability very close to air and are mounted on a cylindrical rotor, there is very little
difference between the d-and q-axis inductances, known as magnetic saliency [9], [13].
This may have control and operational implications [13] but high quality torque and
EMF is produced. The major disadvantage is the risk of demagnetization, explained
in Section 4.5.2, due to the exposure of the magnets to very strong fields produced by
the stator windings [9]. Another disadvantage is that SPMs with conventional stator
windings typically have low inductance values and are thus considered to be ill suited for
flux weakening operation [9], [14]. Fractional slot concentrated windings (FSCW) offer a
potential solution, explained in more detail in Section 2.4. SPM machines may also require
additional mechanical strengthening to prevent displacement of magnets mounted to the
rotor surface in high speed applications.
2.3.2. Interior PM motors (IPM)

IPMs require more complex rotor design with features such as cutouts in the rotor
laminations to fit the magnets and flux barriers, as shown in Fig. 2.4b. These motors
exhibit a larger magnetic saliency compared to SPMs, resulting in a higher reluctance
torque in addition to the magnetic torque [13]. IPMs may also produce more noise than
SPMs, although this can be mitigated by stator tooth shaping [15]. Several IPM topologies
are shown in Fig. 2.4. Surface buried PM motors (SBM) are similar to SPMs with the
difference being the magnets placed under a thin layer of laminations near the rotor surface.
Using air gap flux barriers, a nearly constant flux distribution on the surface of the rotor
can be obtained, producing high quality torque and EMF [9]. The laminations shield the
PMs from stator fields, reducing the risk of demagnetization. These type of motors are
similar to radial-bar PM motors (RBPM), which also feature a higher flux weakening
capability compared to other IPM types [13]. Spoke-type PM motors (STPM) provide
the highest torque and flux density of PM machines [16]. However, these types of PM
motors offer reduced flux weakening capability, a lower reluctance torque due to reduced
magnetic saliency, and are more susceptible to demagnetisation [16]. V-shaped PM motors
(VPM) provide the ability to adjust the flux concentration and the saliency ratio of the
dq-inductances by adjusting the angle of the V-shaped magnets [13]. These motors exhibit
good performance in torque applications and wide speed range operation [13].
6
(a) SPM (b) IPM with flux barriers (c) SBM
(d) RBPM (e) STPM (f) VPM
Figure 2.4: PM motor magnet configurations (4-pole machines)
2.4. Surface-mounted PM motor technologies and

fractional-slot machines
Fractional-slot machines are defined as machines where the number of slots per pole and
phase, q, is not an integer. Subsequently, the two categories of fractional-slot machines
are defined as follows for the rest of this report:
• FSCW: A machine with q < 1 and windings on all teeth or on alternate teeth i.e.
windings encircle only a single stator tooth and are non-overlapping.
• FSDW: A machine with q > 1 and q ∈ / N ,where N is the set of natural numbers,
and where windings encircle more than one tooth.
Different winding configurations are shown in Fig. 2.5 with their respective q values [15].
Note that FSCW machines with windings on alternate teeth are referred to as single-
layer windings, while those with windings on all teeth are referred to as double-layer
windings [15]. FSCW PM machines offer several advantages: high power density due to
high slot fill factors, short end turns which reduce copper losses and increase efficiency, low
cogging torque, flux weakening capability and fault tolerance [15]. Distributed overlapping
windings, both FSDW and conventional, are used in PM machines as they generally
provide a more sinusoidal MMF distribution and EMF waveform [15]. FSCW reduce losses
due to shorter end windings which also reduces cost due to copper volume reduction [15].
Simpler manufacturing and increased slot fill factors are possible when segmented stator
structures are used, especially with soft magnetic composite(SMC), plug-in teeth and
7
Figure 2.5: Winding topologies for a 4 pole machine: (a) Overlapping distributed winding
(q = 2) (b) Overlapping concentrated winding (q = 1) (c) Non-overlapping winding with
all teeth wound(double-layer) (q = 0.5) (d) Non-overlapping winding with alternate teeth
wound(single-layer) (q = 0.5)
joint-lapped core structures. By increasing the slot fill factor, ff , as defined in Eq. 2.9 the
power density is increased [15]. Note ACu and Aslot are the copper conductor areas and
slot areas respectively.
ACu
ff = (2.9)
Aslot
FSCWs may have higher inductance values compared to distributed winding SPMs
due to high harmonic leakage inductance [14]. This increases flux-weakening capability
by decreasing the characteristic current, Ich , as defined in Eq. 2.10, and increases the
constant power speed range (CPSR). Optimal flux-weakening occurs when Ich = IR , as the
symmetrical three-phase short-circuit current will equal the rated current, IR . However,
this condition will cause demagnetization, based on Fig. 3.3c and Fig. 3.3 [6].
λP M
Ich = (2.10)
Ld
Fault tolerance requirements include electrical, magnetic, thermal and physical separa-
tion between phases. It is clear that these requirements are best met by the single-layer
FSCW machine, where the windings are physically separate and there is very low mutual
coupling between phases, yet high self-inductance which limits current. Making use of IMD
technology and feeding each phase from a seperate, single-phase inverter further increases
fault tolerance [15]. FSCWs suffer from two potential disadvantages. Rotor losses are
significantly higher due the high harmonic content (both sub and super space harmonics)
and increase as the rotational speed increases. These losses include rotor iron losses, PM
8
losses and losses in conducting retaining sleeves in SPM’s. Single-layer windings further
increase losses due to increased harmonic content. FSCW may also have higher levels of
parasitic effects which include vibration and noise, unbalanced magnetic forced and torque
ripples [15].
Single- and double-layer windings both offer advantages. Since single-layer windings
only have one coil side/slot, manufacturing is simplified and better fault tolerance is
achieved. Single-layer windings also provide higher self-inductance and thus better flux
weakening capability. Double-layer windings provide lower rotor losses, more possible pole
and slot combinations and a more sinusoidal back-EMF waveform due to a lower winding
factor compared to single-layer windings. This lower winding factor can mean lower torque
performance compared to single-layer winding machines [15].
2.5. Choice of topology

The SPM magnet topology is chosen for this design study due to its simplistic manufacturing
and good performance characteristics. It will be paired with both FSCW and FSDW
machines, with the FSDW machine serving as conventional benchmark. A single- and
double-layer FSCW machine with the same slot-pole combination will be studied, to make
a comparison w.r.t. performance.
9
Chapter 3
Characteristics of PM traction motors
This chapter introduces the electromagnetic principles governing the operation of SPMs,
describes characteristic quantities such as inductance, resistance and losses and introduces
flux weakening operation.
3.1. Electromagnetic principles of PM motors

The basic working principle of a three-phase PM synchronous motor can be explained as
follows. Three-phase alternating current is applied to the armature windings in the stator,
producing a rotating magnetic field. The field produced by the magnets on the rotor,
tends to align with this rotating magnetic field and thus rotation is produced. Once the
rotor rotates in synchronism with the rotating magnetic field, a steady electromechanical
torque is produced in addition to an induced voltage in the armature windings [17].
3.1.1. Open circuit magnetic field, flux linkage and back-EMF

Although FEA provides the most accurate results for quantities such as the magnetic
flux density in the air gap, analytical methods can be applied [18], [19]. The air gap flux
density distribution can be calculated from Eq. 3.1, neglecting the effect of slotting. Note
that θ is in mechanical radians and coefficients are calculated in Appendix C.
∞
X
Bslotless (r, θ) = KBn fBrn cos(npθ) (3.1)
n=1,3,5...
From [18], it is known that for FSCW machines, maximum flux linkage occurs when
the center of the magnet pole aligns with the center of the stator tooth around which the
coil is wound, as shown in Fig. 3.1a. From Fig. 3.1a and Eq. 3.1, the coil flux linkage can
be calculated using Eq. 3.2, with γ = 1. This equations can also be applied to FSDW
machines. The flux-linkage per phase is given by Eq. 3.3 [18], [17]. The back-EMF is
given by Eq. 3.4. Note the amplitude of the air-gap flux density is assumed to be constant,
and thus the back-EMF consists only of a speed voltage, induced in the stator windings
by the relative motion of the air-gap flux wave [17].
10
(a) Concentrated winding coil (b) Distributed winding coil
Figure 3.1: Winding coil diagrams
Z γαu
2
λcoil = Nc LRis Bslotless (Ris , θ)dθ (3.2)
− γα2 u
∞
X 1
λphase = 2Nph LRis Kwn KBn fBrn cos(npθ) (3.3)
n=1,3,5... np
dλphase
EM F = −
dt
∞
X (3.4)
= 2Nph LRis ωmech Kwn KBn fBrn sin(npωmech t)
n=1,3,5...
3.1.2. Phase inductance, winding resistance and iron losses

The synchronous inductance mainly consists of the magnetizing inductance, Lm , slot leakage
inductance, Lσ , end-winding inductance, Lew , and the harmonic leakage inductance, Lh .
Analytical expressions are provided in Eq. 3.5 to Eq. 3.9, with no analytical expression
provided for Lh . The end winding permeance factors, λlew and λw , can be found in [19]
while additional information for all equations is provided in Appendix C.5. For FSDW the
magnetizing inductance makes up the largest part of the synchronous inductance while
for FSCW machines, the harmonic leakage inductance contributes the largest part [14].
This is evident when examining the harmonic content of the winding functions of FSCW
and FSDW machines, as in [14]. For both machines, it is also possible for the slot leakage
inductance to make the largest contribution if the slot dimensions are designed in such a
way.
Ls = Lm + Lσ + Lew + Lh (3.5)
2mτp 0
L m = µ0 l (Kw1 Nph )2 (3.6)
π 2 pgef f
11
(b) Simplified end winding di-
(a) Electrical steel (c) Slot dimensions
mensions
Figure 3.2: Additional dimensions
4m 0 2
Lσ = µ0 l Nph σu (3.7)
Q
S3 S1 S7 S6
σu = k1 + k2 ( + ln ) (3.8)
3S6 S5 S6 − S5 S5
4m 2
Lew = µ0 qNph (2lew λlew + Wew λw ) (3.9)
Q
The winding resistance can be calculated using Eq. 3.10, which neglects the skin effect
under the assumption of small diameter wires. Iron losses are mainly caused by induced
voltages in the electrical steel cores due to the alternating flux in the machine. For a sheet
with dimensions as shown in Fig. 3.2a, the eddy current losses can be approximated by
Eq. 3.11. As can be seen, the eddy current losses are heavily dependent on the sheet
thickness, d, frequency and peak air gap flux density. These losses are mitigated by using
thin steel sheets with insulation between, with the ratio of steel to insulation along the
axial length of the machine, defined as the stacking factor, kj . Other iron losses include
hysteresis losses due to the alternating flux, end losses due to leakage flux penetrating the
machine structure and other harmonic losses [19]. Since the PMs consist of solid material,
eddy currents are induced and thus magnet losses must be considered.
Nph
RDC = ρCu lc (3.10)
ACu
whπ 2 f 2 d3 B̂m
2
PF e = (3.11)
6ρF e
3.2. Flux weakening operation of PM traction motors

Flux-weakening operation occurs between the base and maximum speeds, where negative
d-axis flux is produced by the armature current to oppose the PM flux, λP M , and ensure
voltage limits are not exceeded [20]. To further describe flux-weakening operation for an
12
(a) d-axis equivalent circuit (b) q-axis equivalent circuit
(c) dq phasor diagram
Figure 3.3: dq-equivalent circuits
SPM, the dq-reference frame, which rotates synchronously at ωe , is used. The general
dq-equivalent circuits are shown in Fig. 3.3a and Fig. 3.3b along with a dq-axis phasor
diagram in Fig. 3.3c [21]. Note that phasors are not drawn to scale. As can be seen, the
d-axis armature flux, Ld id , opposes λP M , decreasing the d-axis flux and ensuring voltage
limits are adhered to. Neglecting the stator resistance, Rs , and substituting the expressions
for the d-and q-axis flux, from Eq. 3.12 and 3.13, the circuit equations shown in Eq. 3.15
and 3.16 are obtained [6], [21], [22]. Note Ld = Lq = Ls for an SPM. The torque produced
by a SPM is given by Eq. 3.17, with only the q-axis current contributing to the torque as
there is no reluctance torque component, due to no magnetic saliency [20]. The magnet
flux linkage, λP M , is given by Eq. 3.18 with Erated equal to the rated open circuit RMS,
line-to-neutral voltage. The RMS armature flux linkage, λa(RM S) , can be calculated from
λ
Eq. 3.14. Should λa(RM S)
a(base)
> 1, negative d-axis current is required to reduce the flux linkage
to ensure saturation does not occur and that voltage limits are adhered to [17].
λd = λP M + Ld id (3.12)
λq = Lq iq (3.13)
s
λ2d + λ2q
λa(RM S) = (3.14)
2
vd = −Ls iq ωe (3.15)
vq = ωe (λP M + Ls id ) (3.16)
3
Tp = pλP M iq (3.17)
2
13
√
2Erated
λP M = (3.18)
ωe(base)
Circle diagrams within the dq-current plane can be used to analyse flux weakening
operation of SPMs, taking into account voltage and current limits imposed by the inverter
which drives the motor [23], based on Eq. 3.19 and Eq. 3.21, which yields the current limit
circle. Note Vmax in this case is a RMS, line-to-neutral quantity. Substituting Eq. 3.15
and Eq.3.16 into Eq. 3.19, yields Eq. 3.20 for the voltage limit circle with its center at
(− λLP sM , 0). The circle diagram is shown in Fig. 3.5a. Ideal flux-weakening operation for a
finite-speed machine occurs between the blue and red points in Fig. 3.5a. However, since
torque still has to be produced at the maximum speed, from Eq. 3.17, q-axis current must
still be supplied and thus the red point is not reached. The input voltage and current
behaviour over the speed range is shown in Fig. 3.4a and Fig. 3.4b.
2
Vmax ≥ vd2 + vq2 (3.19)
! !2
2
Vmax λP M
≥ i2q + id + (3.20)
ωe Ls Ls
IR2 ≥ i2d + i2q (3.21)
From Fig. 3.5a and Eq. 3.20, the expressions for ωe(base) , in electrical rad/s, can be
determined as shown in Eq. 3.22. Note id = −IR sin βT and iq = IR cos βT , where βT is
the angle by which the armature current leads the q-axis. The electrical frequency, ωe(max) ,
at maximum speed is given by Eq. 3.23 - the point where both circles meet. There is
a single meeting point of the constraint circles for a finite speed motor [6]. As stated
previously, this point is not reached due to torque required at the maximum speed and
thus this speed is slightly higher than the actual maximum speed. It should be noted that
these derivations rely on constant parameters of Ls and λP M , which may vary with flux
levels [21].
Vmax
ωe(base) = q (3.22)
(IR Ls )2 + λ2P M − 2λP M Ls IR sin βT (max)
Vmax
ωe(max) = (3.23)
λP M − IR Ls
3.2.1. CPSR and torque angle charcteristics

The constant power speed range, CPSR, is the ratio between the maximum and base
speeds, between which the output power is constant. Assuming βT (max) = 0, constant
14
(a) Voltage behaviour over speed range (b) Current behaviour over speed range
Figure 3.4: Voltage and current behaviour over speedrange
(a) dq-current plane circle diagram (b) Torque and power characteristic over speed range
Figure 3.5: Voltage and current behaviour over speed range
power output, id = 0 at base speed and iq = 0 at maximum speed, the CPSR can be
defined as shown in Eq. 3.24. This expression is based on the expressions for the terminal
voltage at base and maximum speed as shown in Eq. 3.22 and 3.23 respectively. Note
Ld = Lq = Ls and Imax is the peak current value. As Ls increases, the CPSR increases.
Since Ls = Lm + Ll , either the leakage or magnetizing inductance must increase.
ωe(max)
KCP SR = |P =Constant
ωe(base) out
q (3.24)
(Imax Ls )2 + λ2P M
=
λP M − Imax Ls
The torque angle characteristic of a synchronous motor is shown in Fig. 3.6 with the
torque angle, δRF , defined as the electrical phase angle between the resultant air gap flux
per pole, φR , and MMF of the PMs. For motors, φR leads the rotor MMF wave.
Figure 3.6: Torque angle characteristic
15
Chapter 4
Design procedure of PM motor
This chapter presents the design specifications obtained through equations described in
Chapter 2, described the choice of slot and pole combinations, winding layouts and the
initial sizing procedure of the machine. The finite element based optimisation process is
also described and additional design considerations are discussed.
4.1. Design process and specifications for the study

The design process followed is shown below in Fig. 4.1. Note that optimisation is performed
using an in house FEA package, SEMFEM, while analysis and verification of the designs is
carried out using a commercial package, ANSYS Maxwell. Dimensions and specifications
of the drive train are given in Table 4.1. From Eq. 2.5, the maximum rotational speed of
the motor can be obtained with Eq. 4.1. Applying Eq. 2.3, the output torque of the motor
at maximum speed can be obtained as shown in Eq. 4.2. Note the drive train efficiency is
assumed to be 100 % and the mechanical differential ratio 1:1. Using Eq. 2.6, the motor
power can be determined as 3076.18 W.
30ig Vmax
Np(max) =
πrd (4.1)
= 4879.332 rpm
rd 1 2
Tp(min) = (Mv ggrav fr + ρair CD Af Vmax )
ig 2 (4.2)
= 5.875 Nm
Choosing the output power to be 3 kW, the maximum rotational speed as 5000 rpm
and considering inverter limits, the specifications of the motor are found as shown in Table
4.2. From Eq. 2.7 and Eq. 2.8, the maximum output torque of the motor is limited to
23.09 Nm while the minimum base speed is found to be 1240.7 rpm. Accordingly, the base
speed and maximum torque at base speed are chosen as shown in Table 4.2. Dimensional
constraints are determined by the magnetic gear shaft and casing.
16
Figure 4.1: Design process
Table 4.1: Dimensions and specifications of EV drive train
Parameter Value
Wheel radius [rd ] 0.299 m
Frontal area [Af ] 2.5 m2
Aerodynamic drag coefficient [CD ] 0.3
Rolling resistance coefficient [fr ] 0.013
Grading percentage 0%
Vehicle mass [Mv ] 1000 kg
Maximum vehicle speed [Vmax ] 50 km/h
MG ratio 11:1
Maximum MG torque [TM G(max) ] 254 Nm
4.2. Pole and slot combinations

The winding factor of a winding is the ratio between the flux linked by the winding
compared to the flux linked by a single-layer full-pitch winding with q = 1 and the same
number of turns per phase, Nph , i.e. the winding factor multiplied with Nph provides the
effective number of turns per phase [24], [17]. Fundamental winding factors can be used
as a guideline for selecting machines as it plays a crucial role in torque production and
machine efficiency [25]. Furthermore, low harmonic winding factors are desired, especially
w.r.t. the fifth and seventh harmonic, to prevent waveform distortion and additional losses.
Fundamental winding factors for different combinations of slots, Ns , and poles, P, are shown
in Fig. 4.2 [25]. The maximum electrical frequency of the motor is usually constrained
to approximately 400 Hz to ensure a good efficiency at high speed. The conditions of
symmetry that govern whether a winding layout is symmetrical or not must also be known.
This ensures windings fed from a symmetrical three-phase supply produce a rotating
magnetic field [19]. While integral slot windings are always symmetrical, fractional-slot
windings must adhere to symmetry conditions. For a given q = nz , as defined in Eq. 4.3,
where z and n are not mutually divisible, the following conditions apply [19]:
Qs
q= (4.3)
2pm
17
Table 4.2: Dimensions and specifications of SPM
Parameter Value
Outer diameter [Dout ] 212 mm
Shaft diameter [Dshaf t ] 30 mm
Output power [Pout ] 3 kW
Maximum voltage [Vmax ] 48 VLL
Phase current [IS = Ilim ] 44 A
Peak losses [Ploss ] 300 W
Base speed [Np(base) ] 1500 rpm
Maximum speed [Np(max) ] 5000 rpm
Torque at base speed [Tp(max) ] 19.1 rpm
Torque at maximum speed [Tp(min) ] 5.73 rpm
Winding temperature[Twinding ] 120 ◦ C
PM material NdFeB N48H
PM remnant flux density [Br ] 1.2899 T
PM recoil permeability [µr ] 1.05
Core material M19 26G
Table 4.3: Possible winding layouts
p n=2 n=4 n=5 n=7 n=8

1 D N N N N
2 B D N N N
3 D N N N N
4 B B N N D
5 D N B N N
1. First condition:
p
• For single-layer windings: n
∈N
2p
• For double-layer windings: n
∈N
n
2. Second condition: 3
∈
/ N for three-phase systems.
Based on these symmetry constraints, the possible windings layouts for different
combinations of p and n are given in Table 4.3. Note single-layer windings are denoted
by S, double-layer windings by D and cases where both single-and double layer windings
are possible by B. N indicates no symmetrical winding layout is possible. Based on
winding factors, frequency constraints and literature, the following machines
are selected for the design study: 12-slot 10-pole single-layer winding machine,
12-slot 10-pole double-layer winding machine and an 36-slot 8-pole double-
layer winding machine. As shown in Table 4.4, all these machines offer high fundamental
winding factors and low harmonic winding factors.
18
Figure 4.2: Winding factors for different slot(Ns ) and pole (P) combinations.
4.3. Winding layouts

To perform winding layouts using coil voltage phasor diagrams (star of slots), some basic
parameters have to be defined for the machine [19]. The slot angle in electrical radians, αu ,
shown in Eq. 4.4, the coil pitch in number of slots, γQ , shown in Eq. 4.5, the number of
phases, m, and the machine periodicity, t, given by Eq. 4.7. For fractional-slot windings,
γQ is not a natural number and thus these will be short-pitched windings, i.e. γ < γQ .
Note m is the number of phases and that all angles are given in electrical radians, defined
in Eq. 4.6, unless specified otherwise. For fractional-slot windings, base-winding phasor
diagrams are used which represent the smallest independent symmetrical section of a
winding, which can be repeated t times in a machine [19].
2πp
αu = (4.4)
Qs
Qs
γQ = (4.5)
2p
θelec = pθmech (4.6)
19
t = GCD{Qs , p} (4.7)
With these quantities known, the phasor diagram can be constructed starting with
Qs
a circle. The circle can be divided into 2m sectors each having 2mt phasors, with two
opposite sectors assigned to a phase- one positive and one negative sector. For three-phase
windings, phases must be seperated by 120 electrical degrees. The phasors are numbered,
1 through Qs , with αu the angle between consecutive slots. For cases where t is an odd
number, natural zone variation must be employed, meaning positive and negative zones
will differ by one phasor. This occurs in some double-layer windings and is compensated
by the reverse happening in the second layer of the winding. For double-layer windings,
after assigning the top layer according to the phasor diagram, the bottom layer is assigned
based on γ [19], [26].
4.3.1. Winding layout: 12-slot 10-pole machine

The phasor diagram of the base winding of both the single-and double-layer winding
machine is shown in Fig. 4.4a. Regarding the double-layer winding, γQ = 1.2, so the coil
pitch is chosen as γ = 1. The winding layouts are shown in Fig. 4.3. The winding factors
for the 12-slot 10-pole machine are given by Eq. 4.8 for the single-layer case and Eq. 4.9
for the double-layer case, with v the harmonic order [18]. Winding factors are summarised
in Table 4.4. Note the third harmonic is neglected as this cancels between phases (m =
3) [19].
vpπ
kwv = sin( ) (4.8)
Qs
vpπ
kwv = sin2 ( ) (4.9)
Qs
4.3.2. Winding layout: 36-slot 8-pole machine

The phasor diagram of the double-layer base winding is shown in Fig. 4.4b. Since γQ = 4.5,
the coil pitch is chosen as γ = 4. The double-layer winding layout is shown in Fig. 4.5.
The fundamental winding factor of the 36-slot 8-pole machine is determined using its base
winding phasor diagram, with negative coil voltage phasors rotated by 180◦ (indicated
with negative signs in Fig. 4.4c). The line of symmetry is found using vector addition,
where after kw1 can be found based on Eq. 4.10 with Z, αρ the number of phasors and the
angle between the ρth phasor and the line of symmetry respectively. kw1 is calculated as
shown in Eq. 4.11.
sin( vπ
2
Z
)X
kwv = cos(αρ ) (4.10)
Z ρ
20
(a) Single-layer 12-slot 10-pole machine (b) Double-layer 12-slot 10-pole machine
Figure 4.3: Winding layouts for 12-slot 10-pole machine
(a) Base winding: 12s10p (b) Base winding: 36s8p (c) Kw1 calculation: 36s8p
Figure 4.4: Phasor diagrams
Figure 4.5: Double-layer winding of 36-slot 8-pole machine
21
Table 4.4: Winding factors
Configuration v=1 v=5 v=7

12s10p(S) 0.966 0.259 0.259
12s10p(D) 0.933 0.067 0.067
36s8p(D) 0.9452 0.1398 -0.0607
Table 4.5: Current limits for air-cooled PMSMs
Ks (kA/m) J (A/mm2 )
Typical value 35-65 3-5
sin( π2 ) π π
kw1 = (4 cos( ) + 2 cos( )) (4.11)
6 18 6
4.4. Initial sizing of design parameters

The design procedure followed is based on work in [27], [28]. This procedure requires
choosing certain parameters, shown in Table 4.6, in addition to those in Table 4.2 before
starting the initial design process. Thermal considerations are included through the choice
of Ks and J, with guidelines shown in Table 4.5. Magnetic saturation is considered through
the choice of B(ysmax) and B(tsmax) and manufacturing constraints through ff . Note the
initial design procedure is performed using parameters at base speed.
4.4.1. Geometric dimensions

Firstly, the air gap flux is calculated. It is approximated as a square wave with amplitude
Bg , given by Eq. 4.12, with a fundamental sinusoidal component Bg1 , given by Eq. 4.13.
Note the Carter factor, Kc , is assumed to be 1.05. A more detailed air-gap flux calculation
Table 4.6: Input parameters for design at 1500 rpm
Parameter 12s10p(S) 12s10p(D) 36s8p(D)

Magnet thickness [hm ] 3 mm 3 mm 2 mm
Slot opening[w0 ] 2 mm 2 mm 2 mm
Slot enclosure depth[h1 ] 4 mm 4 mm 4 mm
Electrical frequency[f] 125 Hz 125 Hz 100 Hz
Air gap size [g] 1 mm 1 mm 1 mm
Pole coverage ratio [αp ] 0.83 0.83 0.83
Bys(max) 1T 1T 1T
Bts(max) 1.6 T 1.6 T 1.6 T
Lamination stacking factor [kj ] 0.95 0.95 0.95
Current density [J] 4 A/mm2 4 A/mm2 4 A/mm2
Linear Current density [Ks ] 35 kA/m 35 kA/m 35 kA/m
Fill factor [ff ] 35% 35% 35%
22
method is shown in Appendix C. The air gap diameter, Dag , measured to the middle of
the air gap, is calculated from Eq. 4.14. Note that the stack length, L, is related to the air
gap diameter as shown in Eq. 4.15, with ψ chosen as 1. With the external rotor diameter,
Der = Dag − 2hm − g, and the internal stator diameter, Dis = Dag + g, the pole pitch and
slot pitch can be calculated from Eq. 4.16 and Eq. 4.17 respectively.
hm
µr
Bg = Br hm (4.12)
µr
+ gKc
4 αp π
Bg1 = Bg sin( ) (4.13)
π 2
√
2π 2
Tr = Bg1 Ks Dag L (4.14)
4
L
ψ= (4.15)
Dag
πDis
τp = (4.16)
2p
πDis
τs = (4.17)
Qs
The stator yoke thickness, hys , and tooth width, wtt , can be calculated from Equations
4.18 and 4.20 respectively. Note that in the case of FSCW machines where the magnet
circumference is comparable to the slot pitch, Equations 4.19 and 4.21 must be used to
account for the flux shunting effect between two adjacent magnets [27].
Bg αp τp
hys = (4.18)
2Bys(max) kj
Bg τs − τp

hys = αp τp − (4.19)
Bys(max) kj 2
Bg τs
wtt = (4.20)
Bts(max) kj
Bg τs − τp

wtt = αp τp − (4.21)
Bts(max) kj 2
To determine further slot dimensions as shown in Fig. 4.6, Dh1 is defined as shown in
Eq. 4.22 with slot enclosure dimensions, h11 and h12 , chosen as 2 mm each to sum to h1 .
The slot width, wst , measured at the same diameter as wtt is calculated as shown in Eq.
4.23.
23
Figure 4.6: Slot dimensions as used for initial geometric sizing
Dh1 = Dis + 2h1 (4.22)
πDh1
wst = − wtt (4.23)
Qs
The required stator slot area is calculated from Eq. 4.24, which excludes the slot
enclosure area. Lastly the stator slot height, hus which excludes the slot enclosure
dimensions, can be calculated from Eq. 4.27. The slot area is approximated as a trapezoid,
as shown in Eq. 4.25, to make initial sizing calculations calculations easier. The outer slot
width, wsb , is given by Eq. 4.26. Note the tooth width is constant due to the trapezoidal
slot shape. The outer stator diameter is given by Eq. 4.28.
πDh1 Ks
Aus = (4.24)
JQs Kw1 ff
wst + wsb

Aus = hus (4.25)
2
!
π(Dh1 + 2hus )
wsb = − wtt (4.26)
Qs
r
2 4πAus
−wst + wst + Qs
hus = 2π (4.27)
Qs
Dout = Dh1 + 2(hus + hys ) (4.28)
4.4.2. Calculation of number of turns per phase

The number of turns per phase can be calculated according to Eq. 4.29, with αp used
for PM machines where the magnet thickness is constant [19]. Note that the back-EMF,
Em , is assumed equal to the line-to-neutral phase voltage value, L0 = L + 2g and ωe is
the electrical frequency in rad/s. The number of turns must be chosen which allows for
an integer number of conductors, zQ , per slot, or per half slot for a double layer winding
24
Table 4.7: Initial geometric parameters

Slot opening depth [h11 ] 2 mm 2 mm 2 mm
Slot height [hslot ] 35.3 mm 36.08 mm 32.95 mm
Slot area [Aus ] 568.87 mm2 588.99 mm2 201.22 mm2
Air gap diameter [Dag ] 74.96 mm 74.96 mm 78.18 mm
Stack Length [L] 74.96 mm 74.96 mm 78.18 mm
Stator yoke thickness [hys ] 21.79 mm 21.79 mm 11.37 mm
Rotor yoke thickness [hyr ] 18.98 mm 18.98 mm 21.59 mm
Tooth width [wtt ] 13.62 mm 13.62 mm 3.81 mm
Slot width [wst ] 8.36 mm 8.36 mm 3.8 mm
Outer diameter[Dout ] 194.15 mm 195.7 mm 171.82 mm
Turns per phase [Nph ] 30 32 36
Turns per coil [Ncoil ] 15 8 3
layout. The condition for a single layer winding is given in Eq. 4.30 and in Eq. 4.31 for a
double layer winding.
√
2Em
Nph = (4.29)
ωe Kw1 αp Bg1 τp L0
Nph
zQ = (4.30)
pq
Nph
zQ(half −slot) = (4.31)
2pq
Flux density in the rotor is calculated to ensure no saturation occurs, based on Eq.
4.32 and Eq. 4.33. Parameters obtained from the initial design process is shown in Table
D −D
4.7. Note hslot = hus + h12 and hyr = er 2 shaf t .
φrot = Bg αp τp L (4.32)
φrot
2
Byr = (Der −Dshaf t )
(4.33)
2
Lkj
4.5. Additional considerations for final design
4.5.1. Cogging torque and torque ripple

Cogging torque is produced by the interaction of MMF harmonics produced by the PM’s
and air gap permeance harmonics due to the effects of stator slotting. It causes the rotor to
tend to align in stable positions which can cause speed ripples and vibration [29]. For the
purposes of this report, three aspects which influence cogging torque are investigated: slot
25
Table 4.8: Cogging torque parameters

NC 60 60 72
CT 2 2 4
αp(opt) 0.8433 0.8433 0.8989
and pole combinations, slot opening width and magnet pole coverage ratio (αp ). From [29],
the factor CT is introduced as a measure of the cogging torque produced by a certain
pole-slot combination, as shown in Eq. 4.34. A larger CT value is related to a larger cogging
torque [29]. NC is the smallest common multiple of Qs and 2p. The optimal magnet pole
coverage ratio is given by Eq. 4.35, taking magnet fringing into account through k2 , which
typically ranges between 0.01 and 0.03. k1 is chosen as 1 to maximise the air gap flux [29].
Note N = N2pC . These quantities are shown in Table 4.8 for the machines investigated, with
k2 = 0.01. Regarding slot openings, the cogging torque must be considered under both
open-circuit and full-load conditions. Under open-circuit conditions, the amplitude of the
cogging torque decreases as the slot opening width decreases. The same phenomenon is
observed under full-load conditions as the stator slotting effect decreases which reduces the
variation of the air gap permeance. However, this is only valid until a certain point where
after the cogging torque amplitude increases again as the tooth-tip leakage flux enhances
local magnetic saturation which causes the air gap permeance variation to increase once
again [30]. Accordingly, an optimal point exists for the slot opening width to decrease
the cogging torque. It should be kept in mind that due to saturation effects, the optimal
design for reducing cogging torque might not result in the minimum torque ripple [30].
2pQs
CT = (4.34)
NC
N − k1
αp(opt) = + k2 (4.35)
N
4.5.2. Demagnetization aspects

A typical demagnetization curve, which refers to the second quadrant of the BH-curve,
is shown in Fig. 4.7a. The magnet originally has a remnant value, BR . As a negative
MMF, i.e. current, is applied to the stator windings, the B-H curve will be traced towards
(a). If the MMF is then increased to zero again, the minor hysteresis loop is traced to
(b). This minor hysteresis loop is approximated by a straight line of which the slope is
the recoil permeability, µR . From Fig. 4.7a, if a small negative current is applied so that
only point (d) is reached, the magnet traces the recoil line back to (c) with almost no
loss of residual magnetization. If a large negative current is applied that goes beyond
(a), also known as the knee point, there is a significant reduction in residual magnetism,
26
(a) Typical demagnetization curve (b) Demagnetization curve of NdFeB N48H
Figure 4.7: Demagnetization characteristics
referred to as demagnetization. The demagnetization curve of NdFeB N48H is shown in

Fig. 4.7b. When machine design is considered, the concept of a demagnetization margin is
introduced which refers to how far above the knee point the machine operates. The knee
point of NdFeB N48H occurs at 0.2 T.
4.6. Optimum design of PM motors

The aim of any design optimisation can be summarised as follows: find a vector X, that
represents a set of n design variables, each bounded by xil ≤ xi ≤ xih with i = 1, ..., n to
minimise an objective function F(X) and satisfy a set of m constraints, Gk (X)≤ 0 with
k = 1, ..., m [31]. For the design of a fractional-slot machine, the vector X will contain
physical machine dimensions and constraints such as the torque and losses. Traditionally,
this optimisation problem has been solved using gradient based methods [32], [33]. Modern
approaches have also been developed, such as genetic algorithms and particle swarm
optimisation [33], [34], [35] and these have been implemented in design procedures [36], [37].
4.6.1. Formulation of optimisation problem

The optimisation strategy employed will analyse the two critical points at base and
maximum speed, as shown in Fig. 3.5b, using FEA. By analysing the designs at Np(base) ,
a constant torque speed range (CTSR) is ensured. Note Np(init) = 1 rpm. By analysing
the designs at Np(base) and Np(max) , a constant power speed range (CPSR) is ensured.
Adopting the strategy proposed in [13], the optimisation problem is formulated as in [13],
shown below Fig. 4.8. Note that all values are RMS values and the voltage is a line-to-line
quantity. X is a vector containing design variables including the dq-currents, geometric
variables and coil turns [13].
27
Figure 4.8: Optimisation process
Minimise: F(X) = Ctotal

Subject to: T1 ≥ 19.1 Nm T2 ≥ 5.73 Nm
Vmax1 ≤ 48 V Vmax2 ≤ 48 V
IS1 ≤ 42.5 A IS2 ≤ 42.5 A
Ploss1 ≤ 265 W Ploss2 ≤ 265 W
BM M 1 ≥ 0.05 T BM M 2 ≥ 0.05 T
4.6.2. Finite element based optimisation process

The optimisation process was implemented in SEMFEM, an 2D finite element package.
The process followed is shown in Fig. 4.8, starting with the optimisation of initial geometric
variables. The number of turns per phase must then be chosen to be a valid, integer number
as the optimisation may result in non-integer Nph values. Should the choice still satisfy
constraints, a preliminary design is found. Alternatively, the dimensions are optimised
again with Nph constant to check if constraints can be satisfied. If no valid design is found,
the whole process is restarted changing either the choice of Nph or the boundaries. To
ensure, within computational limits, that a global optimum is found, random variable sets
are generated and optimised. Should all constraints be met at a lower price, this design
becomes the final design. Note that optimisations are performed with a coarser mesh
compared to simulations due to time constraints.
28
Table 4.9: dq-current densities of machines
12s10p(S) 12s10p(D) 36s8p(D)

Parameter
1500 rpm 5000 rpm 1500 rpm 5000 rpm 1500 rpm 5000 rpm
2
iq(dens) (A/mm ) 3.845 1.153 3.961 1.188 4.348 1.304
id(dens) (A/mm2 ) -2.392 -4.379 -2.464 -4.511 -2.705 -4.952
Figure 4.9: Delta-wye connection of motor
4.6.3. Finite element based simulation methods

The solution method used in SEMFEM requires dq-current densities to be specified. Using
√
Eq. 3.18 and 3.17, iq can be obtained. The d-axis current is then given by id = 2Is sinβT ,
as explained in Section 3.2. The estimated dq-current densities for the three machines at
base and maximum speeds are shown in Table 4.9, calculated using Eq. 4.36 and Eq. 4.37.
It should be noted that the dq-currents required are the same for all 3 machines under
ideal assumptions. These current densities provide a starting point for the optimisation
process. Analysis of the final machines was carried out in ANSYS Maxwell, where the
machine is driven by the circuit shown in Fig. 4.9, with the apparent power delivered to
the machine given by Eq. 4.38.
iq zQ
iq(dens) = (4.36)
ff Aus
iq zQ
id(dens) = (4.37)
ff Aus
√
S3φ = 3Vt(LL) Is (4.38)
29
Chapter 5
Comparative study of two SPM motors
5.1. Dimensions, inductance and cost

The preliminary machine dimensions found through the optimisation process are shown in
Table 5.2. The final machine dimensions are also shown. Changes can be attributed to
practical considerations, notably the slot fill factor and wire size selection. With reference
to the slot dimensions as shown in Table 5.3 and defined as shown in Fig. 5.1, the
preliminary slot dimensions were adjusted for two reasons. The first is to achieve a slot
fill factor of near 35%, which is achievable if the machines are manufactured in-house.
This slot fill factor must be achieved using standard wire sizes, and thus each conductor
consists of zwire wires. The second reason is to decrease Ls , to ensure a better power factor
as the original machine delivered a poor power factor when analysed in ANSYS Maxwell.
The inductance values of the final machines are shown in Table 5.1. As stated in Chapter
3, it clear that the harmonic leakage inductance contributes significantly towards Ls in
the case of the FSCW machines, while Lm contributes significantly in the FSDW machine
case. This can be attributed to the higher harmonic content of the concentrated windings
due to a less sinusoidal winding distribution. For both FSCW and the FSDW machine,
the slot leakage inductance is the dominant component. This is thought to be the result
of the optimisation process, which seeks to maximise this quantity. The costs and mass of
the final machines are shown in Table 5.4. The 12-slot double-layer machine uses the least
amount of all materials and thus has the lowest cost. This is due to short end windings
which reduce copper usage, a small diameter which reduces steel consumption and thinner
magnets, which greatly reduce costs.
Figure 5.1: Slot dimensions
30
Table 5.1: Inductances of machines
Inductance 12s10p(S) 12s10p(D) 36s8p(D)

Lσ (µH) 586.3 430.67 606.8
Lew (µH) 2.61 1.36 52.74
Lh (µH) 178.34 115.79 7.97
Lm (µH) 68.95 152.6 344.78
Ls (µH) 836.2 700.43 1012.3
RDC (mΩ @ 20 ◦ C) 22.254 21.91 33.678
Table 5.2: Main machine dimensions
12s10p(S) 12s10p(D) 36s8p(D)

Parameter
Preliminary Final Preliminary Final Preliminary Final
Dshaf t (mm) 30 30 30 30 30 30
hyr (mm) 10.984 10.984 18.3 18.3 21.075 20.575
hm (mm) 4.754 4.754 2.32 2.32 2 2.5
αp 0.75 0.75 0.7577 0.7577 0.7258 0.7258
g (mm) 1 1 1 1 1 1
hys (mm) 9.162 10.628 4.555 3.798 5.243 5.188
Dout (mm) 170.142 170.142 156.755 156.755 184.92 184.92
L (mm) 108.673 108.673 95.825 95.825 70.21 70.21
Ncoil 20 20 11 11 5 5
Nph 40 40 44 44 60 60
Table 5.3: Slot parameters
12s10p(S) 12s10p(D) 36s8p(D)

Parameter
hs0 (mm) 3.272 3.272 5.756 4 5.128 5.187
hs1 (mm) 0 0 0 0 0 0
hs2 (mm) 30.639 26 20.205 22 38.243 38.243
bs0 (mm) 4.187 4.187 3.326 3.326 1.8 1.6
bs1 (mm) 5.125 6.2 12.64 14 3.324 5
bs2 (mm) 20.52 27 22.485 24 9.555 9.555
dwire (mm) 3.582 1.628 3.399 1.45 3.61 1.45
zwire 1 4 1 5 1 5
ff (%) 35 38.05 35 38.2 35 37.3
Table 5.4: Cost and mass of final machines
12s10p(S) 12s10p(D) 36s8p(D)

Material
Mass (kg) Cost ($) Mass (kg) Cost ($) Mass (kg) Cost ($)
Copper 2.31 15.41 2.234 14.90 3.434 22.90
Magnets 0.518 25.90 0.273 13.65 0.221 11.05
Steel 9.64 19.28 6.943 13.89 7.002 14.00
Total 12.468 60.59 9.45 42.44 10.657 47.95
31
(a) 12-slot 10-pole (S) (b) 12-slot 10-pole (D)
(c) 36-slot 8-pole machine
Figure 5.2: Preliminary optimised machines
(a) 12s10p(S) (b) 12s10p(D) (c) 36s8p
Figure 5.3: Mesh used in SEMFEM
5.2. Main performance characteristics

The main performance aspects of the preliminary and final machines are shown in Tables
5.5 and 5.6 at 1500 rpm and 5000 rpm respectively. Note that preliminary results were
obtained using the SEMFEM package while final results were obtained through ANSYS
Maxwell. Results compare favourably, with difference attributed to slight dimension
changes and different solution methods. The meshes used for the various machines are
shown in Fig. 5.3 and 5.4 for SEMFEM and Maxwell respectively. Regarding the notation
used in figure legends, the number of slots is indicated followed by a letter indicating a
single (S) or double (D) layer winding. The subscript indicates whether the result relates
to a preliminary (P) or final (F) design.
(a) 12s10p(S) (b) 12s10p(D) (c) 36s8p
Figure 5.4: Mesh used in ANSYS Maxwell
32
Table 5.5: Results at 1500 rpm
12s10p(S) 12s10p(D) 36s8p(D)

Parameter
Tp (Nm) 19.18 18.43 19.14 19.21 19.379 19.43
Trip (%) 3.36 1.94 6.125 4.49 7.184 6.23
Pout (kW) 3.012 2.895 3.043 3.017 3.043 3.053
PCu (W) 198.1 175.1 211.64 166.94 217.37 251.28
PCore (W) 33.06 31.3 39.95 33.02 42.675 30.48
Pmagnet (W) - 6.57 - 7.995 - 0.324
Pin (kW) 3.243 3.076 3.264 3.192 3.303 3.335
η (%) 92.87 93.15 92.29 94.53 92.13 92.4
Power factor 0.9285 0.8522 0.9332 0.8988 0.9439 0.94
Vt(LL) (V) 47.933 48 48.52 48 48.594 48
Is(RM S) (A) 43.006 43.41 42.66 42.71 42.765 42.27
eRM S (V) 26.223 26.95 26.434 26.78 26.413 25.37
λa(RM S) (mWb) 33.388 33.8 33.657 33.8 42.037 40.4
BM M (T) 0.3072 - 0.3052 - 0.1462 -
(a) 12-slot machines (b) 36-slot machine
Figure 5.5: Magnet losses at 1500 rpm
5.2.1. Losses and efficiency

From Tables 5.5 and 5.6, a clear difference between copper and core losses of the preliminary
and final designs can be seen. This is due to different end winding lengths in the different
software packages, changes to dimensions, different wire sizes and a different core loss
calculation method. The 36-slot machine features the highest copper losses at both 1500
and 5000 rpm due to longer end windings, while it has the lowest core losses at 1500 rpm.
The 12-slot double-layer winding machine has the highest core losses at 1500 rpm, while
at 5000 rpm, the 12-slot single-layer winding machine has the highest core losses. Magnet
losses could only be calculated for the final designs using ANSYS Maxwell, and are shown
in Fig. 5.5 and 5.6 at 1500 rpm and 5000 rpm respectively. Taking all losses into account,
the 12-slot double-layer winding machine has the greatest efficiency at 1500 rpm while the
12-slot single-layer winding machine has the greatest efficiency at 5000 rpm.
33
Table 5.6: Results at 5000 rpm
12s10p(S) 12s10p(D) 36s8p(D)

Parameter
Tp (Nm) 5.766 5.779 5.748 5.8255 5.826 5.75
Trip (%) 11.025 5.77 22.2 11.27 15 11.82
Pout (kW) 3.019 3.026 3.015 3.05 3.05 3.01
PCu (W) 179.2 138.67 202.3 151.72 212.55 248.93
PCore (W) 85.4 79.66 61.61 56.24 53.02 57.09
Pmagnet (W) - 51.04 - 55.84 - 2.08
Pin (kW) 3.283 3.208 3.279 3.252 3.315 3.252
η (%) 91.94 94.3 91.95 93.79 91.99 92.57
Power factor 0.9863 0.999 0.9872 0.96 0.9686 0.9298
Vt(LL) (V) 47.157 48 46.647 48 47.843 48
Is(RM S) (A) 43.06 38.62 42.743 40.72 42.763 42.065
eRM S (V) 25.75 27.1 25.286 27.19 25.95 28.69
λa(RM S) (mWb) 9.836 10.17 9.659 10.23 12.39 13.5
BM M (T) 0.2761 - 0.2933 - 0.224 -
Figure 5.6: Magnet losses at 5000 rpm
Figure 5.7: Flux linkage at 1500 rpm
34
Table 5.7: Phase angle of terminal voltages
12s10p(S) 12s10p(D) 36s8p(D)

1500 rpm 5000 rpm 1500 rpm 5000 rpm 1500 rpm 5000 rpm
◦
θdeg ( ) 62 71 52 58 53.2505 55
fe (Hz) 125 416.67 125 416.67 100 333.33
Figure 5.8: Induced voltage (back-emf) at 1500 rpm
5.2.2. Flux linkage, voltage and current

The flux linkage and induced voltages at 1500 rpm are shown in Fig. 5.7 and 5.8. The
36-slot machine produces the most sinusoidal waveform, resulting in a sinusoidal back-emf.
Due to flattening at the top of the flux linkage waveforms of the 12-slot machines, distortion
occurs around the zero-crossings of the back-emf waveforms. The flux linkage and induced
voltages at 5000 rpm are shown in Fig. 5.9 and 5.10. Significant flattening now occurs for
all machines, and distorted back-emf waveforms are obtained. In general, higher induced
voltages (RMS) are obtained for the final designs. While sinusoidal currents are specified
for the simulation of preliminary designs, sinusoidal currents which correlate well are
drawn when the simulation is performed using voltage sources, as shown in Fig. 5.11 and
5.12. The power factor, calculated using Eq. 4.38, of the final designs was found to be
lower than that of the preliminary designs, with the exception of the 12-slot single-layer
machine at 5000 rpm. The input voltages are all of the form shown in Eq. 5.1, with phase
angles as shown in Table 5.7 and ωe as defined in Eq. 5.2. Note these are line-to-neutral
quantities and that the phase angle is the torque angle, as described in Chapter 3.
π
van = 39.1918 sin(ωe t + θdeg ∗ ) (5.1)
180
π
ωe = pNp (5.2)
30
35
Figure 5.9: Flux linkage at 5000 rpm
Figure 5.10: Induced voltage (back-emf) at 5000 rpm
Figure 5.11: Phase current at 1500 rpm
36
Figure 5.12: Phase current at 5000 rpm
Figure 5.13: Output torque at 1500 rpm
Figure 5.14: Output torque at 5000 rpm
37
Table 5.8: Cogging torque of different machines
Cogging torque 12s10p(S) 12s10p(D) 36s8p(D)

Peak-to-peak (Nm) 0.1672 0.8448 0.7976
(a) 12s10p(S) (b) 12s10p(D) (c) 36s8p
Figure 5.15: Mesh used for cogging torque determination
5.2.3. Output torque, torque ripple and cogging torque

The output torque waveforms are shown in Fig. 5.13 and 5.14 at 1500 rpm and 5000 rpm
respectively, with preliminary and final results comparing favourably. The average torque
obtained, shown in Tables 5.5 and 5.6, meets traction requirements with the exception of
the 12-slot single-layer machine at 1500 rpm. The cogging torque wave forms are shown in
Fig. 5.16 for the three preliminary designs, with the maximum and minimum values shown
in Table 5.8. The 12-slot single-layer machine has the lowest cogging torque, correlating
well with the discussion from Section 4.5.1. The discussion does not correlate well with
the obtained result for the other two machines, with the 36-slot machine providing a lower
cogging torque than the double-layer 12-slot machine. This shows that while the factor
CT gives an estimation of cogging torque performance, it is not the only aspect to be
considered. It is known to accurately determine the cogging torque, a very fine mesh is
needed if FEA is used [29]. The mesh used for the three machines is shown in Fig. 5.15.
The torque ripple, calculated as a percentage of the average torque, is shown in Tables 5.5
and 5.6. The final design results feature lower torque ripple values, attributed to different
solvers and mesh discretisation, with the 12-slot single-layer machine performing best.
5.2.4. Demagnetisation and flux densities

The demagnetisation margins of the preliminary designs are shown in Tables 5.5 and
5.6. The 12-slot machines feature bigger margins due to weaker fields produced by stator
windings. In-depth demagnetisation analysis was not performed on the final designs. Flux
densities of the 12-slot machines at 1500 rpm are shown in Fig. 5.17 while flux densities of
the 36-slot machine at 1500 and 5000 rpm are shown in Fig. 5.18, where flux weakening
can be observed. Flux densities of the 12-slot machines are shown in Appendix C, along
with efficiency and loss maps generated under the assumption that the motor is always
operated at maximum torque per ampere (MTPA).
38
(a) 12-slot 10-pole machine (b) 36-slot 8-pole machine
Figure 5.16: Cogging torque
(b) 12s10p(D)
(a) 12s10p(S)
Figure 5.17: Flux densities of 12-slot machines at 1500 rpm
(a) Flux densities at 1500 rpm (b) Flux densities at 5000 rpm
Figure 5.18: Flux densities of 36-slot machine
39
Chapter 6
Conclusion and Further work
6.1. Conclusion
This report detailed the concepts surrounding fractional slot SPMs in the context of EVs,
described the design of three such machines and the comparison of their performance.
Comparison of the machines, subject to the given constraints, show the 12-slot double-layer
machine to deliver the required performance at the lowest cost with the added benefit of
ease of manufacturing. The 36-slot machine provides the greatest flux weakening ability
and CPSR due its high synchronous inductance, while featuring a slightly smaller cogging
torque compared to the 12-slot double-layer machine due to smaller slot openings. The
design process used was found to be sufficient and is reproducible should future machines
need to be designed.
6.2. Recommendations for future work

While the motors have been evaluated using FEA software, which is considered to be
very accurate, the motors need to be built to truly evaluate and compare the advantages
of the different configurations. A very important aspect would be the evaluation of the
demagnetisation aspects of the machines, which may change some design aspects as shown
in [6]. Additional losses, such as friction and windage losses, can be evaluated in this way.
Furthermore, the performance of the motor when used in conjunction with a flux-weakening
control system must be investigated. Fault tolerance can also then be evaluated. With
practicality in mind, performance of the motors can be evaluated under the constraint of
a fixed outer diameter.
Possible improvements to the motors include investigating the effect of tooth shape,
tooth tip shaping effects due to chamfers and reduction of cogging torque and torque
ripple through the use of closed stator slots and skewing [38], [39]. The possible negative
effects of these design choices, such as voltage distortion, must also be considered [40].
Furthermore, the full advantage of FSCW machines could be fully exploited by using
modular stator manufacturing techniques to increase slot fill factor and thereby power
density [15].
40
Bibliography
[1] J. Wang, X. Yuan, and K. Atallah, “Design Optimization of a Surface-Mounted

Permanent-Magnet Motor With Concentrated Windings for Electric Vehicle Applica-
tions,” IEEE Transactions on Vehicular Technology, vol. 62, no. 3, pp. 1053–1064,
2013.
[2] S. Hosseinpour, H. Chen, and H. Tang, “Barriers to the wide adoption of electric vehi-
cles: A literature review based discussion,” in 2015 Portland International Conference
on Management of Engineering and Technology (PICMET), 2015, pp. 2329–2336.
[3] B. Qu, Q. Yang, Y. Li, M. Sotelo, S. Ma, and Z. Li, “A Novel Surface Inset Permanent
Magnet Synchronous Motor for Electric Vehicles,” Symmetry, vol. 12, p. 179, 2020.
[4] W. Lee, S. Li, D. Han, B. Sarlioglu, T. A. Minav, and M. Pietola, “A Review of

Integrated Motor Drive and Wide-Bandgap Power Electronics for High-Performance
Electro-Hydrostatic Actuators,” IEEE Transactions on Transportation Electrification,
vol. 4, no. 3, pp. 684–693, 2018.
[5] J. Wang, Y. Li, and Y. Han, “Integrated Modular Motor Drive Design With GaN
Power FETs,” IEEE Transactions on Industry Applications, vol. 51, no. 4, pp. 3198–
3207, 2015.
[6] S. Pastellides, S. Gerber, R. J. Wang, and M. J. Kamper, “Design of a Surface-

Mounted PM Motor for Improved Flux Weakening Performance,” ICEM, vol. 31,
no. 1, pp. 1772–1778, 2020.
[7] M. Ehsani, Y. Gao, S. Gay, and A. Emadi, Modern Electric, Hybrid Electric, and
Fuel Cell Vehicles. Boca Raton, Florida: CRC Press, 2004.
[8] P. M. Tlali, R. Wang, and S. Gerber, “Magnetic gear technologies: A review,” in 2014
International Conference on Electrical Machines (ICEM), 2014, pp. 544–550.
[9] I. D. Kakoudakis, E. K. Karamanis, I. I. Kikidis, and A. G. Kladas, “Comparison of

Permanent Magnet Motor Configurations for Electric Vehicle Applications,” in 2018
XIII International Conference on Electrical Machines (ICEM), 2018, pp. 2058–2064.
[10] P. Shewane, A. Singh, M. Gite, and A. Narkhede, “An Overview of Neodymium

Magnets over Normal Magnets for the Generation of Energy,” International Journal
41
on Recent and Innovation Trends in Computing and Communication, vol. 2, no. 12,
pp. 4056–4059, 2014.
[11] D. Prosperi, A. Bevan, G. Ugalde, C. Tudor, G. Furlan, S. Dove, P. Lucia, and

M. Zakotnik, “Performance comparison of motors fitted with magnet-to-magnet
recycled or conventionally manufactured sintered NdFeB,” Journal of Magnetism and
Magnetic Materials, vol. 460, pp. 448–453, 2018.
[12] Y. Kim, S. Lee, E. Lee, B. Cho, and S. Kwon, “Comparison of IPM and SPM motors
using ferrite magnets for low-voltage traction systems,” in 2015: EVS28 International
Electric Vehicle Symposium and Exhibition, 2015, pp. 1–7.
[13] S. Pastellides, S. Gerber, and R. J. Wang, “Design Strategy and Comparison of

Four PM Motor Topologies for a 2kW Traction Application,” in 2019 Southern
African Universities Power Engineering Conference/Robotics and Mechatronics/Pat-
tern Recognition Association of South Africa (SAUPEC/RobMech/PRASA), 2019, pp.
358–363.
[14] T. J. A.M. EL-Refaie, “Optimal flux weakening in surface pm machines using fractional-
slot concentrated windings,” IEEE Transactions on Industry Applications, vol. 41,
no. 3, pp. 790–800, May/June 2005.
[15] A. M. EL-Refaie, “Fractional-Slot Concentrated-Windings Synchronous Permanent

Magnet Machines: Opportunities and Challenges,” IEEE Transactions on Industrial
Electronics, vol. 57, no. 1, pp. 107–121, 2010.
[16] E. Carraro, N. Bianchi, S. Zhang, and M. Koch, “Design and Performance Comparison
of Fractional Slot Concentrated Winding Spoke Type Synchronous Motors With
Different Slot-Pole Combinations,” IEEE Transactions on Industry Applications,
vol. 54, no. 3, pp. 2276–2284, 2018.
[17] S. Umans, Fitzgerald & Kingsley’s Electric Machinery. McGraw-Hill Education,

2013. [Online]. Available: https://books.google.co.za/books?id=REI9XwAACAAJ
[18] D. Ishak, Z. Q. Zhu, and D. Howe, “Comparison of PM brushless motors, having

either all teeth or alternate teeth wound,” IEEE Transactions on Energy Conversion,
vol. 21, no. 1, pp. 95–103, 2006.
[19] J. P. H. Hrabovcova, T. Jokinen, Design of Rotating Electrical Machines. John Wiley

& Sons, Ltd, 2008.
[20] T. M. Jahns, “Flux-Weakening Regime Operation of an Interior Permanent-Magnet

Synchronous Motor Drive,” IEEE Transactions on Industry Applications, vol. IA-23,
no. 4, pp. 681–689, 1987.
42
[21] L. Jolly, M. A. Jabbar, and Liu Qinghua, “Optimization of the constant power speed
range of a saturated permanent-magnet synchronous motor,” IEEE Transactions on
Industry Applications, vol. 42, no. 4, pp. 1024–1030, 2006.
[22] H. Liu, Z. Q. Zhu, E. Mohamed, Y. Fu, and X. Qi, “Flux-Weakening Control of

Nonsalient Pole PMSM Having Large Winding Inductance, Accounting for Resistive
Voltage Drop and Inverter Nonlinearities,” IEEE Transactions on Power Electronics,
vol. 27, no. 2, pp. 942–952, 2012.
[23] W. L. Soong and T. J. E. Miller, “Field-weakening performance of brushless syn-

chronous AC motor drives,” IEE Proceedings - Electric Power Applications, vol. 141,
no. 6, pp. 331–340, 1994.
[24] EMETOR. (2020) Winding factor. [Online]. Available: https://www.emetor.com/

glossary/winding-factor/
[25] e. a. Y.Tang, “Investigation of winding topologies for permanent magnet in-wheel

motors,” The International Journal for Computation and Mathematics in Electrical
and Electronic Engineering, vol. 31, no. 1, pp. 88–107, 2012.
[26] L. Alberti and N. Bianchi, “Theory and design of fractional-slot multilayer windings,”
in 2011 IEEE Energy Conversion Congress and Exposition, 2011, pp. 3112–3119.
[27] S. Vaschetto, A. Tenconi, and G. Bramerdorfer, “Sizing procedure of surface mounted

PM machines for fast analytical evaluations,” in 2017 IEEE International Electric
Machines and Drives Conference (IEMDC), 2017, pp. 1–8.
[28] A. Sorgdrager, “Development of a line-start permanent-magnet synchronous machine,”

Master’s thesis, North-West University, North-West University, November 2014.
[29] Z. Q. Zhu and D. Howe, “Influence of design parameters on cogging torque in

permanent magnet machines,” IEEE Transactions on Energy Conversion, vol. 15,
no. 4, pp. 407–412, 2000.
[30] D. Wu and Z. Q. Zhu, “Design Tradeoff Between Cogging Torque and Torque Ripple
in Fractional Slot Surface-Mounted Permanent Magnet Machines,” IEEE Transactions
on Magnetics, vol. 51, no. 11, pp. 1–4, 2015.
[31] N. Bianchi and S. Bolognani, “Design optimisation of electric motors by genetic

algorithms,” IEE Proceedings - Electric Power Applications, vol. 145, no. 5, pp.
475–483, 1998.
[32] R. Ramarathnam, B. G. Desai, and V. S. Rao, “A Comparative Study of Minimization

Techniques for Optimization of Induction Motor Design,” IEEE Transactions on Power
Apparatus and Systems, vol. PAS-92, no. 5, pp. 1448–1454, 1973.
43
[33] S. Rao, Engineering Optimisation: Theory and Practice. John Wiley & Sons, Ltd,
2009.
[34] Y. Duan, R. G. Harley, and T. G. Habetler, “Method for multi-objective optimized

designs of Surface Mount Permanent Magnet motors with concentrated or distributed
stator windings,” in 2009 IEEE International Electric Machines and Drives Confer-
ence, 2009, pp. 323–328.
[35] M. Mutluer and O. Bilgin, “Design optimization of PMSM by particle swarm opti-
mization and genetic algorithm,” in 2012 International Symposium on Innovations in
Intelligent Systems and Applications, 2012, pp. 1–4.
[36] W. Gu, X. Zhu, L. Quan, and Y. Du, “Design and Optimization of Permanent Magnet
Brushless Machines for Electric Vehicle Applications,” Energies, vol. 8, no. 12, pp.
13 996–14 008, 2015.
[37] Y. Duan, “Method for Design and Optimization of Surface Mount Permanent Magnet
Machines and Induction Machine,” Ph.D. dissertation, School of Elect. and Computer
Eng., Georgia Institute of Tech., Atlanta, GA, USA, 2010.
[38] N. Soda and M. Enokizono, “Relation Between Stator Core Shape and Torque Ripple
for SPM Motor,” in 2018 XIII International Conference on Electrical Machines
(ICEM), 2018, pp. 955–960.
[39] K. Seo, Y. Kim, and S. Jung, “Stator teeth shape design for torque ripple reduction in
surface-mounted permanent magnet synchronous motor,” in 2014 17th International
Conference on Electrical Machines and Systems (ICEMS), 2014, pp. 387–390.
[40] D. Wu, Z. Q. Zhu, and X. Ge, “Effectiveness of Terminal Voltage Distortion Mini-
mization Methods in Fractional Slot Surface-Mounted Permanent Magnet Machines
Considering Local Magnetic Saturation,” IEEE Transactions on Energy Conversion,
vol. 31, no. 3, pp. 1090–1099, 2016.
44
Appendix A
Project Planning Schedule
Figure A.1: Initial planning
45
Figure A.2: Revised planning
46
Appendix B
Outcomes Compliance
B.1. ELO 1: Problem solving

The initial problem is stated to be the comparison and analysis of different SPM synchronous
motors with unconventional winding topologies, in the context of use in electric vehicles.
Firstly, this problem statement requires understanding these technologies, as discussed in
Chapter 2 of this report where electric vehicles, PM technology and winding topologies
are discussed. Chapter 3 provides some of the characteristics in terms of the machine
attributes and the operation of these machines under flux-weakening control - a very
popular control method.
A secondary problem is the design process of these machines. This requires a design
process to be developed, starting from initial machine specifications (externally determined
in this case) and narrowing it down to obtain the exact physical dimensions of the machine.
As described in Chapter 4, the process consists of 2 parts. Firstly, an analytical method is
used to determine initial geometric parameters of the machine. While providing a starting
point, this method alone cannot ensure all specifications are met in the most effective
way. Thus, the second part of the process involves a design optimisation through the use
of software to ensure specifications are adhered to in the entire operational range of the
machine and that it is done in the most cost-effective way, a very practical consideration.
With the design stage completed, the machines must be analysed and conclusions
drawn from results obtained. This is done in Chapter 5, where the results of the different
machine designs are shown and discussed. Possible reasons for differences and performance
are discussed, based on theory from literature also discussed in earlier chapters and from
additional analysis relying on knowledge from undergraduate modules.
B.2. ELO 2: Application of scientific and engineering

knowledge
This outcome is achieved through the use of scientific and engineering knowledge throughout
the project. Firstly, the basic working principles of synchronous motors must be understood,
relying on modelling of these machines, based on electromagnetic theory, through circuit
47
diagrams, phasor diagrams and equations as shown in Chapter 3. To design physical
dimensions of the machines, these electromagnetic principles must now be combined with
practical sizing considerations to obtain dimensions as shwon in Chapter 4. To analyse
the machines, the underlying principles must be understood so that conclusions can be
drawn from results, as shown in Chapter 5.
B.3. ELO 3: Engineering design

This outcome is achieved through the implementation of procedural and non-procedural
design methods. The main design aspect of the project is the design of three different
machines, starting with an initial set of specifications and ending with three machines which
conform to these specifications, based on simulation and analysis. The machine design
process is formalised as much as possible, to ensure it is reproducible with as much detail
as possible provided in Chapter 4. However, it consists of less formally structured parts,
such as the machine optimisation, also described in Chapter 4, and practical considerations
implemented in Chapter 5.
B.4. ELO 4: Investigations, experiments and data

analysis
Various motor topologies featuring overlapping and non-overlapping windings were in-
vestigated based on tractive effort required and other specifications. The performance
characteristics of each identified candidate design were calculated using coupled-field cir-
cuits as explained in Chapter 4. The machines were then analysed under various conditions
to ensure specifications were met using two different FEA software packages, SEMFEM
and ANSYS MAxwell. The results obtained were then processed to be analysed and
discussed, as shown in Chapter 5, with possible reasons provided for performance. Further
analysis over the entire speed range of the motors is shown in Appendix C.
B.5. ELO 5: Engineering methods, skills and tools,

including Information Technology
This outcome is demonstrated through the use of various software tools. MATLAB was used
to plot data and to effectively perform calculations, especially iterative calculations such as
the calculation of geometric parameters described in Section 4.4. Analysis, simulation and
optimisation of the machines was performed in two software packages, ANSYS Maxwell
and SEMFEM. The SEMFEM package required the use of the Python programming
48
language. Report writing was performed using a LATex compiler while diagrams were
drawn using both Microsoft Excel and draw.io, an online diagram editor.
B.6. ELO 6: Professional and technical communication

This outcome is demonstrated through this report, the oral presentation and accompanying
slides as well as the project poster. It is further achieved through the effective communi-
cation of technical information through the use of equations and diagrams. Results are
presented effectively through figures and tables. Weekly meetings with my supervisor
served as verbal professional communication, with additional correspondence via email.
B.7. ELO 8: Individual work

This outcome is firstly demonstrated by completion of this project and the report which
required large amounts of individual work. Furthermore, the design method, choice of
machines to investigate and conclusions drawn from analysis are my own based on literature
and guidance from my study leader. Various aspects had to be collected from different
sources and combined to create this report as shown in Chapter 2, 3 and 4, demonstrating
this outcome.
B.8. ELO 9: Independent Learning Ability

This outcome is demonstrated by obtaining the necessary knowledge to design, optimise
and analyse SPM synchronous motors. This required knowledge of the following aspects
which were not obtained through undergraduate modules: SPM synchronous motors,
winding topologies and the layout thereof, flux-weakening operation, design of the physical
dimensions of machines, machine optimisation, use of FEA software and use of the
programming language Python. Knowledge was obtained through conference papers,
journal articles, textbooks and software training events.
49
Appendix C
Additional results, equations and

derivations
C.1. Flux density, thermal(current) aspects and efficiency

maps
The flux densities in the 12-slot machines at 5000 rpm are shown in Fig. C.1, illustrating
the weakening of the flux in the machines when compared to the flux densities shown in
Fig. 5.17 at 1500 rpm. The linear current densities, Ks and current densities, J, of the
final machines are shown in Table C.1.
The efficiency and loss maps of the 12-slot single-layer machine are shown in Fig. C.2
and C.3 respectively. The efficiency and loss maps of the 12-slot double-layer machine
are shown in Fig. C.4 and C.5 respectively. The efficiency and loss maps of the 36-slot
machine are shown in Fig. C.6 and C.7 respectively. All three machines show wide ranges
of high efficiency. The 12-slot single-layer machines shows the lowest losses at low speeds
relative to its losses at high speed, while the 36-slot machine shows high losses over the
entire speed range when operated at the maximum torque value at all respective speeds.
C.2. Results from SEMFEM simulations

Additional results obtained from preliminary simulations is shown in Tables C.2 and C.3
below.
Table C.1: Current limits for air-cooled PMSMs
Final value 12s10p(S) 12s10p(D) 36s8p(D)

Ks (kA/m) 51.36 47.387 60.31
J (A/mm2 ) 5.12 5.002 4.981
50
(a) 12s10p(S) (b) 12s10p(D)
Figure C.1: Flux densities of 12-slot machines at 5000 rpm
Figure C.2: Efficiency map of 12-slot single-layer machine
Table C.2: Preliminary design results at 1500 rpm (SEMFEM)

Current density [J] 4.152 A/mm2 4.55 A/mm2 4.038 A/mm2
d-axis current [id ] −38.433 A −36.78 A −36.403 A
q-axis current [iq ] 47.138 A 47.818 A 48.297 A
d-axis flux linkage [λd ] 21.938 mWb 23.791 mWb 31.854 mWb
q-axis flux linkage [λq ] 41.812 mWb 41.23 mWb 50.196 mWb
51
Figure C.3: Loss map of 12-slot single-layer machine
Figure C.4: Efficiency map of 12-slot double-layer machine
52
Figure C.5: Loss map of 12-slot double-layer machine
Figure C.6: Efficiency map of 36-slot machine
53
Figure C.7: Loss map of 36-slot machine
Table C.3: Preliminary design results at 5000 rpm (SEMFEM)

2 2
Current density [J] 3.949 A/mm 4.445 A/mm 3.993 A/mm2
d-axis current [id ] −58.77 A −58.453 A −58.447 A
q-axis current [iq ] 15.94 A 15.397 A 15.534 A
d-axis flux linkage [λd ] 5.934 mWb 5.684 mWb 8.839 mWb
q-axis flux linkage [λq ] 12.58 mWb 12.42 mWb 15.13 mWb
Figure C.8: M19 26G BH-curve
54
C.3. Electrical steel properties
The BH-curve of the M19 26G electrical steel as used in ANSYS Maxwell is shown in Fig.
C.8.
C.4. Open circuit magnetic field calculation

The coefficients used for the calculation of the open-circuit magnetic field are given below,
with dimensions defined as shown in Fig. C.9. Note Rm = Rr + hm and r is the radius at
which the air-gap flux density is calculated.
π
sin((np + 1)αp 2p )
A1n = π (C.1)
(np + 1)αp 2p
π
sin((np − 1)αp 2p )
A2n = π (C.2)
(np − 1)αp 2p
Br
Mrn = αp (A1n + A2n ) (C.3)
µ0
Br
Mθn = αp (A1n − A2n ) (C.4)
µ0
Mn = Mrn + npMθn (C.5)
1 Mrn 1
A3n = (np − ) + (C.6)
np Mn np
Rr np+1 Rr 2np
KB1n = (A3n − 1) + 2( ) − (A3n + 1)( ) (C.7)
Rm Rm
µr + 1 Rr 2np µr − 1 Rm 2np Rr 2np

KB2n = (1 − ( ) )− (( ) −( ) ) (C.8)
µr Ris µr Ris Rm
np
KB3n = (C.9)
(np)2 − 1
µ0 Mn KB1n
KBn = KB3n (C.10)
µr KB2n
r np−1 Rm np+1 Rm np+1

fBrn = ( ) ( ) ( ) (C.11)
Ris Ris r
55
Figure C.9: Machine dimensions
C.5. Additional inductance and resistance calculations

For Eq. 3.6, τp is defined as shown in Eq. 4.16, gef f as shown in Eq. C.12 and l’ as shown
in Eq. C.13.
hm
gef f = +g (C.12)
µr
l0 = l + 2g (C.13)
Additional slot leakage parameters are given below in Eq. C.14 to Eq. C.16.
9
k1 = 1 − (C.14)
16
3
k2 = 1 − (C.15)
4
W
=1− (C.16)
τp
γ
The coil length, as used in Eq. 3.10, is given in Eq. C.17. Note W = τ.
γQ p
lc = 2l + 2.4W + 0.1 (C.17)
56
Appendix D
MATLAB design scripts
The design script as used in MATLAB to obtain initial geometric parameters is shown
below. Note only the script for the 12-slot 10-pole single-layer winding machine is shown
as the same script was used for the other machines, with appropriate inputs parameters
changed. Also note that to use this script for distributed winding machines, Eq. 4.20 and
Eq. 4.18 must be used.
1 clear all ;
2 % Rated data
3 Tr = 19.1;
4 nr = 1500;
5 Vr = 48;
6
7 % Material
8 Br = 1.2899;
9 ur = 1.05;
10 Ki = 0.95;
11 u0 = 4* pi *10ˆ( -7) ;
12 % Material stress loadings
13 % maximum stator yoke flux
14 Bysmax = 1;
15 % maximum stator teeth flux
16 Btsmax = 1.6;
17 % maximum rotor yoke flux
18 Byrmax = 1;
19 % current density in A / mm ˆ2
20 J = 4;
21
22 % Geometrical and winding specs

23 lambda = 1;
24 % number of poles , q , chosen short pitch ratio and number of coils per
phase
25 poles = 10;
26 q = 0.4;
27 Qs = 12;
28 yQ = 5/6;
29 numCoils = 2;
30 % shaft diameter in mm
57
31 Dmin = 30;
32 Dir = Dmin ;
33 % air gap thickness in mm
34 hag = 1;
35 % magnet thickness in mm
36 lm = 3;
37 % magnet to pole coverage ratio
38 alpham = 0.83;
39 % slot opening in mm
40 w0 = 2;
41 % slot enclosure dimensions in mm
42 h11 = 2;
43 h12 = 2;
44 % fill factor as percentage
45 kfill = 35;
46 % fundamental winding factor
47 Kw1 = 0.966;
48
49 % Calculate Bg
50 lmacc = lm / ur ;
51 % assume Kc = 1.05
52 hagacc = hag *1.05;
53 Bg = Br * ( lmacc /( lmacc + hagacc ) ) ;
54 Bg1 = (4/ pi ) * Bg * sin ( alpham *( pi /2) ) ;
55
56 % Calculate new parameters

57 C1 = (( sqrt (2) * pi ) /4) ;
58 % linear current density
59 KsAperm = 350*100;
60 % air gap diameter in mm
61 Dag2 = ( nthroot (( Tr /( C1 * Bg1 * KsAperm * lambda ) ) ,3) ) *1000;
62 % axial length in mm
63 Lcore2 = lambda * Dag2 ;
64 % External rotor diameter - measured at outside of rotor yoke in mm
65 Der2 = Dag2 - 2* lm - 2*( hag /2) ;
66 % internal stator diameter in mm
67 Dis2 = Dag2 + 2*( hag /2)
68 taup = ( pi * Dis2 ) / poles ;
69 taus = ( pi * Dis2 ) /( Qs ) ;
70 % in mm
71 hys2 = ( Bg /( Bysmax * Ki ) ) *( alpham * taup -(( taus - taup ) /2) ) ;
72 wtt2 = ( Bg /( Btsmax * Ki ) ) *( alpham * taup -(( taus - taup ) /2) ) ;
73 h1 = h11 + h12 ;
74 Dh12 = Dis2 + 2* h1 ;
75 wst2 = (( pi * Dh12 ) / Qs ) - wtt2 ;
76 % usable slot area in m ˆ2
77 Aus2 = ( KsAperm * pi *( Dh12 /1000) ) /(( J *10ˆ6) * Qs * Kw1 *( kfill /100) ) ;
58
78 % in mm ˆ2
79 Aus2 = Aus2 *10ˆ6;
80 hus2 = (( -1* wst2 ) + sqrt ( wst2 ˆ2+(4* pi * Aus2 / Qs ) ) ) /(2* pi / Qs ) ;
81 % external stator diameter in mm
82 Des2 = Dh12 +2*( hus2 + hys2 ) ;
83 wsb2 = ( pi *( Dh12 +2* hus2 ) / Qs ) - wtt2 ;
84 hts2 = hus2 + h1 ;
85 refspeed = 1500;
86 fr = refspeed * poles /120;
87 omegar = 2* pi * fr ;
88 lacc = Lcore2 + 2* hag ;
89 taup = ( pi * Dis2 ) / poles ;
90 % number of truns per phase
91 Nph = ( sqrt (2) *( Vr / sqrt (3) ) ) /( omegar * Kw1 * alpham * Bg1 *( taup /1000) *( lacc
/1000) ) ;
92 % choose number of turns per phase - integer value
93 Nphchoice = 30;
94 % number of conductors per slot
95 zQ = (6/ Qs ) * Nphchoice ;
96 % recalculate new Bg1 to check it does not vary too much
97 Bg1new = ( sqrt (2) *( Vr / sqrt (3) ) ) /( omegar * Kw1 * alpham * Nphchoice *( taup /1000)
*( lacc /1000) ) ;
98 % magnet width
99 wm = alpham * taup ;
100 % use m units
101 fluxp = Bg *( wm /1000) .*( Lcore2 /1000) ;
102 fluxp1 = (2/ pi ) * Bg1new *( taup /1000) *( Lcore2 /1000) ;
103 % flux in rotor yoke
104 Byr = ( fluxp /2) /(((( Der2 - Dir ) /2) /1000) *( Lcore2 /1000) * Ki ) ;
105 % calculate number of turns per coil
106 Ncoil = Nphchoice / numCoils ;
107 % write data to text file
108 fileID = fopen ( ’ initialParams . txt ’ , ’w ’) ;
109 fprintf ( fileID , ’ % -25 s % -22 s \ n ’ , ’ Dimensions ’ , ’ Value ( all units in mm ) ’) ;
110 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ Air gap ( g ) ’ , hag ) ;
111 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ Magnet thickness ( hpm ) ’ , lm ) ;
112 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ Magnet coverage ’ , alpham ) ;
113 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ Dag2 ’ , Dag2 ) ;
114 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ Lcore2 ’ , Lcore2 ) ;
115 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ Inner stat diam ( Dis ) ’ , Dis2 ) ;
116 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ Rotor ext diam ( Der ) ’ , Der2 ) ;
117 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ Stator yoke height ( hys ) ’ , hys2 ) ;
118 fprintf ( fileID , ’ % -12 s %22.2 f \ n \ n ’ , ’ External stator diam ( Des ) ’ , Des2 ) ;
119 fprintf ( fileID , ’ % -25 s % -22 s \ n ’ , ’ Slot Dimensions ’ , ’ Value ( all units in
mm ) ’) ;
120 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ Tooth width ( wtt ) ’ , wtt2 ) ;
121 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ Slot opening ( w0 ) ’ , w0 ) ;
59
122 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ h11 ’ , h11 ) ;
123 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ h12 ’ , h12 ) ;
124 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ wst2 ’ , wst2 ) ;
125 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ wsb2 ’ , wsb2 ) ;
126 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ Usable slot height ( hus ) ’ , hus2 ) ;
127 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ Usable slot area ( Aus ) ’ , Aus2 ) ;
128 fprintf ( fileID , ’ % -25 s %22.2 f \ n \ n ’ , ’ Fill factor ( kfill ) ’ , kfill ) ;
129 fprintf ( fileID , ’ % -25 s % -22 s \ n ’ , ’ Electrical aspects ’ , ’ Values ’) ;
130 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ Nph ’ , Nph ) ;
131 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ Nphchoice ’ , Nphchoice ) ;
132 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ Conductors per slot ( zQ ) ’ , zQ ) ;
133 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ Turns per coil ( Ncoil ) ’ , Ncoil ) ;
134 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ Linear curr dens ( A / cm ) ’ , KsAperm /100) ;
135 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’J ( A / mm ˆ2) ’ ,J ) ;
136 fprintf ( fileID , ’ % -25 s %22.2 f \ n \ n ’ , ’ Rated voltage LL ( Vr ) ) ’ , Vr ) ;
137 fprintf ( fileID , ’ % -25 s % -22 s \ n ’ , ’ Magnetic aspects ’ , ’ Values ’) ;
138 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ Bg1new ’ , Bg1new ) ;
139 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ Bg1old ’ , Bg1 ) ;
140 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ Byr ’ , Byr ) ;
141 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ Bys ’ , Bysmax ) ;
142 fprintf ( fileID , ’ % -25 s %22.2 f \ n ’ , ’ Bts ’ , Btsmax ) ;
143 fclose ( fileID ) ;
Listing D.1: MATLAB script
60
Appendix E
SEMFEM simulation code
1 # !/ usr / bin / env python3

2 import semfem4 as sf4
3 import matplotlib . pyplot as plt
4 from math import pi
5 from math import sqrt
6 import pprint
7 import numpy as np
8 import os
9 import json
10 import sys
11 from scipy . interpolate import *
12
13 class desig n_simu latio n ( sf4 . simulation ) :

14 def draw_rotor_pole ( self , v , i ) :
15 rotor_mesh = [ v [ ’ mesh1 ’] , v [ ’ mesh2 ’] , v [ ’ mesh3 ’ ]]
16 if i % 2 == 0:
17 ds = sf4 . slots . i n n e r _ s u r f a c e _ p m _ p o l e 2 ( sf4 , v [ ’p ’] , v [ ’ mpp ’] , v [
’ mt ’] , v [ ’ rot_yt ’] , v [ ’ r_iy ’] , v [ ’ r_iy ’] , v [ ’ Br ’] , v [ ’ ur ’] , rotor_mesh )
18 else :
19 ds = sf4 . slots . i n n e r _ s u r f a c e _ p m _ p o l e 2 ( sf4 , v [ ’p ’] , v [ ’ mpp ’] , v [
’ mt ’] , v [ ’ rot_yt ’] , v [ ’ r_iy ’] , v [ ’ r_iy ’] , - v [ ’ Br ’] , v [ ’ ur ’] , rotor_mesh )
20 return ds
21 def draw_stator_slot ( self , v , i ) :
22 stator_dvars = [ v [ ’Q ’] , v [ ’ ro ’] , v [ ’ ro ’] , v [ ’ st_yt ’] , v [ ’ st_tt ’] , v [ ’
sat ’] , v [ ’ sot ’] , v [ ’ spp ’] , v [ ’ sop ’ ] ,250]
23 stator_mesh = [ v [ ’ mesh1 ’] , v [ ’ mesh2 ’] , v [ ’ mesh3 ’ ]]
24 ds , ew = sf4 . slots . o u t e r _ p a r a l l e l _ t e e t h _ r o u n d _ s l o t ( sf4 , self .
winding [ i ] , stator_dvars , stator_mesh )
25 return ds
26 def simulate ( self , m , v ) :
27 with open ( ’v . json ’ , ’w ’) as file :
28 file . write ( json . dumps (v , indent =4) )
29 # Initialize empty results dictionary
30 y = {}
31
32 # Point 1 simulation
61
33 self . s t a n d a r d _ c u r r e n t _ d e n s i t y _ s i m (m , v [ ’ steps ’] , v [ ’ nrpm1 ’] , v [ ’
idens_d1 ’] , v [ ’ idens_q1 ’] ,
34 coil_ff = v [ ’ coil_ff ’] ,
coil_turns = v [ ’ coil_turns ’] ,
35 fpl_output =1)
36
37 y1 = self . standard_post (m , winding_temp = v [ ’ winding_temp ’] ,

print_results = False )
38
39 # Point 1 results
40 y [ ’ T1 ’] = y1 [ ’ T_av ’]
41 # y [ ’ T_rip1 ’] = (( max ( self . ts . torq_vec [: , 0]) ) - min ( self . ts .
torq_vec [: , 0]) ) / y1 [ ’ T_av ’]
42 y [ ’ eff1 ’] = y1 [ ’ eff ’]
43 y [ ’ Is1 ’] = ( max ( self . ts . i_vec [: ,0]) ) /( sqrt (2) )
44 y [ ’ i_d1 ’] = y1 [ ’ i_d ’]
45 y [ ’ i_q1 ’] = y1 [ ’ i_q ’]
46 y [ ’ i1_rms1 ’] = y1 [ ’ i1_rms ’]
47 y [ ’ i_rms1 ’] = y1 [ ’ i_rms ’]
48 y [ ’ idens_rms1 ’] = y1 [ ’ idens_rms ’]
49 y [ ’ e_rms1 ’] = y1 [ ’ e_rms ’]
50 y [ ’ u_rms1 ’] = y1 [ ’ u_rms ’]
51 y [ ’ u_ll1 ’] = y1 [ ’ u_ll ’]
52 y [ ’ P_out1 ’] = y1 [ ’ P_out ’]
53 y [ ’ P_in1 ’] = y1 [ ’ P_in ’]
54 y [ ’ P_copper1 ’] = y1 [ ’ P_c ’]
55 y [ ’ P_core1 ’] = y1 [ ’ P_core ’]
56 y [ ’ P_loss1 ’] = y1 [ ’ P_loss ’]
57 y [ ’ PF1 ’] = y1 [ ’ PF ’]
58 y [ ’ demag_margin1 ’] = y1 [ ’ demag_margin ’]
59 y [ ’ fl_d11 ’] = y1 [ ’ fl_d1 ’]
60 y [ ’ fl_q11 ’] = y1 [ ’ fl_q1 ’]
61 y [ ’ fl_rms1 ’] = y1 [ ’ fl_rms ’]
62 y [ ’ tcog1_max ’] = max ( self . ts . torq_vec [: ,0])
63 y [ ’ tcog1_min ’] = min ( self . ts . torq_vec [: ,0])
64
65 np . savetxt ( ’ torque1500 . txt ’ , self . ts . torq_vec [: ,0])

66 np . savetxt ( ’ curr1500 . txt ’ , self . ts . i_vec [: ,0])
67 np . savetxt ( ’ flink1500 . txt ’ , self . ts . flink_vec [: ,0])
68 f_flink = splrep ( self . ts . time_vec , self . ts . flink_vec [: ,0] , k =3)
69 voltage = splev ( self . ts . time_vec , f_flink , der =1)
70 np . savetxt ( ’ voltage1500 . txt ’ , voltage )
71
72 # Point 2 simulation
73 self . s t a n d a r d _ c u r r e n t _ d e n s i t y _ s i m (m , v [ ’ steps ’] , v [ ’ nrpm2 ’] , v [ ’
idens_d2 ’] , v [ ’ idens_q2 ’] ,
74 coil_ff = v [ ’ coil_ff ’] ,
62
coil_turns = v [ ’ coil_turns ’] ,
75 fpl_output =1)
76
77 y2 = self . standard_post (m , winding_temp = v [ ’ winding_temp ’] ,

print_results = False )
78
79 # Point 2 results
80 y [ ’ T2 ’] = y2 [ ’ T_av ’]
81 # y [ ’ T_rip2 ’] = (( max ( self . ts . torq_vec [: , 0]) ) - min ( self . ts .
torq_vec [: , 0]) ) / y2 [ ’ T_av ’]
82 y [ ’ eff2 ’] = y2 [ ’ eff ’]
83 y [ ’ Is2 ’] = ( max ( self . ts . i_vec [: ,0]) ) /( sqrt (2) )
84 y [ ’ i_d2 ’] = y2 [ ’ i_d ’]
85 y [ ’ i_q2 ’] = y2 [ ’ i_q ’]
86 y [ ’ i1_rms2 ’] = y2 [ ’ i1_rms ’]
87 y [ ’ i_rms2 ’] = y2 [ ’ i_rms ’]
88 y [ ’ idens_rms2 ’] = y2 [ ’ idens_rms ’]
89 y [ ’ e_rms2 ’] = y2 [ ’ e_rms ’]
90 y [ ’ u_rms2 ’] = y2 [ ’ u_rms ’]
91 y [ ’ u_ll2 ’] = y2 [ ’ u_ll ’]
92 y [ ’ P_out2 ’] = y2 [ ’ P_out ’]
93 y [ ’ P_in2 ’] = y2 [ ’ P_in ’]
94 y [ ’ P_copper2 ’] = y2 [ ’ P_c ’]
95 y [ ’ P_core2 ’] = y2 [ ’ P_core ’]
96 y [ ’ P_loss2 ’] = y2 [ ’ P_loss ’]
97 y [ ’ PF2 ’] = y2 [ ’ PF ’]
98 y [ ’ demag_margin2 ’] = y2 [ ’ demag_margin ’]
99 y [ ’ fl_d12 ’] = y2 [ ’ fl_d1 ’]
100 y [ ’ fl_q12 ’] = y2 [ ’ fl_q1 ’]
101 y [ ’ fl_rms2 ’] = y2 [ ’ fl_rms ’]
102 y [ ’ tcog2_max ’] = max ( self . ts . torq_vec [: ,0])
103 y [ ’ tcog2_min ’] = min ( self . ts . torq_vec [: ,0])
104
105 np . savetxt ( ’ torque5000 . txt ’ , self . ts . torq_vec [: ,0])

106 np . savetxt ( ’ curr5000 . txt ’ , self . ts . i_vec [: ,0])
107 np . savetxt ( ’ flink5000 . txt ’ , self . ts . flink_vec [: ,0])
108 f_flink = splrep ( self . ts . time_vec , self . ts . flink_vec [: ,0] , k =3)
109 voltage = splev ( self . ts . time_vec , f_flink , der =1)
110 np . savetxt ( ’ voltage5000 . txt ’ , voltage )
111
112 y [ ’ C_total ’] = y1 [ ’ C_total ’]

113 y [ ’ C_steel ’] = y1 [ ’ C_steel ’]
114 y [ ’ C_copper ’] = y1 [ ’ C_copper ’]
115 y [ ’ C_magnet ’] = y1 [ ’ C_magnet ’]
116 y [ ’ M_total ’] = y1 [ ’ M_total ’]
117 y [ ’ M_steel ’] = y1 [ ’ M_steel ’]
118 y [ ’ M_copper ’] = y1 [ ’ M_copper ’]
63
119 y [ ’ M_magnet ’] = y1 [ ’ M_magnet ’]
120 y [ ’ d_wire ’] = y1 [ ’ d_wire ’]
121 with open ( ’ output . json ’ , ’w ’) as file :
122 file . write ( json . dumps (y , indent =4) )
123
124 return y
125
126 def setup_simulation ( v ) :

127 v [ ’ spp ’] = v [ ’ wst ’ ]/( v [ ’ wst ’] + v [ ’ wtt ’ ])
128 v [ ’q ’] = v [ ’Q ’ ]/(2* v [ ’p ’ ]*3)
129 v [ ’R ’] = v [ ’ Dout ’ ]/2
130 v [ ’ ro ’] = v [ ’R ’]
131 v [ ’ st_yt ’] = v [ ’ thickSyoke ’]
132 v [ ’ rot_yt ’] = v [ ’ thickRyoke ’]
133 v [ ’ sot ’] = v [ ’ h1 ’]
134 v [ ’ Dis ’] = v [ ’ Dir ’] + 2* v [ ’ thickRyoke ’] + 2* v [ ’ mag_thick ’] + 2* v [ ’ ag
’]
135 v [ ’ st_tt ’] = v [ ’R ’] - v [ ’ thickSyoke ’] - v [ ’ h1 ’] - v [ ’ Dis ’ ]/2
136 distperSlotgap = ( pi * v [ ’ Dis ’ ]) /( v [ ’Q ’ ])
137 v [ ’ sop ’] = v [ ’ w0 ’ ]/( distperSlotgap )
138 v [ ’ mpp ’] = v [ ’ alphamag ’]
139 v [ ’ mt ’] = v [ ’ mag_thick ’]
140 v [ ’ r_iy ’] = v [ ’ Dir ’ ]/2
141 v [ ’ mesh1 ’] = v [ ’ mesh_scale ’ ]*1 e -3
144 sim = d esign _simul ation (2* v [ ’p ’] , 2* v [ ’p ’] , v [ ’Q ’] , v [ ’L ’] , v [ ’ Dout ’
] , v [ ’ N_layers ’] , v [ ’ coil_span ’] ,
145 project = v [ ’ project ’] , verbosity = v [ ’ verbosity
’] , p er i o di c i ty _ f ac t o rs =[])
146 return sim
147
148 if __name__ == ’ __main__ ’:

149 with open ( ’ params . json ’ , ’r ’) as jsonfile :
150 v = json . load ( jsonfile )
151 sim = setup_simulation ( v )
152 ds , m = sim . make_mesh ( v )
153 y = sim . simulate (m , v )
Listing E.1: Python script
64

DuPlessis F Skripsie2020

Uploaded by

Copyright:

Available Formats

DuPlessis F Skripsie2020

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

DuPlessis F Skripsie2020

Uploaded by

Copyright:

Available Formats

UNIVERSITEIT•STELLENBOSCH•UNIVERSITY

jou kennisvennoot • your knowledge partner

Design and comparison of three

Report submitted in partial fulfilment of the requirements of the module

Supervisor: Prof RJ Wang

Aan Jesus Christus, my Verlosser, wat my elke dag krag gee.

Plagiaatverklaring / Plagiarism Declaration

3. Ek verstaan ook dat direkte vertalings plagiaat is.

4. Dienooreenkomstig is alle aanhalings en bydraes vanuit enige bron (ingesluit die

Studentenommer / Student number Handtekening / Signature

Voorletters en van / Initials and surname Datum / Date

List of Figures vii

2. Traction motor drive-train systems 3

3. Characteristics of PM traction motors 10

4. Design procedure of PM motor 16

5. Comparative study of two SPM motors 30

6. Conclusion and Further work 40

A. Project Planning Schedule 45

D. MATLAB design scripts 57

E. SEMFEM simulation code 61

2.1. Vehicle movement models . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

3.1. Winding coil diagrams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

4.1. Design process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

5.1. Slot dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

A.1. Initial planning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

C.1. Flux densities of 12-slot machines at 5000 rpm . . . . . . . . . . . . . . . . 51

4.1. Dimensions and specifications of EV drive train . . . . . . . . . . . . . . . 17

5.1. Inductances of machines . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

C.1. Current limits for air-cooled PMSMs . . . . . . . . . . . . . . . . . . . . . 50

Variables and functions

ACu Copper conductor area.

CPSR Constant power speed range

1.1. Background of the project

1.2. Objectives of the project

1.3. Scope of the project

1.4. Layout of the report

• Chapter 2 provides the necessary background information to understand EV traction

• Chapter 3 gives an overview of the electromagnetic principles of PMSMs. Properties

• Chapter 6 provides a summary of the project, a conclusion regarding the comparative

Traction motor drive-train systems

2.1. Traction efforts and vehicle physics model

Figure 2.1: Vehicle movement models

Pout = Tp(min) ωp(max) (2.6)

2.2. Typical EV drive train systems

Figure 2.2: EV drive trains

Figure 2.3: Energy conversion system and drive train

2.3. Common permanent magnet motor topologies

2.3.2. Interior PM motors (IPM)

(d) RBPM (e) STPM (f) VPM

Figure 2.4: PM motor magnet configurations (4-pole machines)

2.4. Surface-mounted PM motor technologies and

2.5. Choice of topology

Characteristics of PM traction motors

3.1. Electromagnetic principles of PM motors

3.1.1. Open circuit magnetic field, flux linkage and back-EMF

Figure 3.1: Winding coil diagrams

3.1.2. Phase inductance, winding resistance and iron losses

Figure 3.2: Additional dimensions

3.2. Flux weakening operation of PM traction motors