QUOTATIONS

The job of a scientist is to listen

carefully to nature, not to tell
nature how to behave.

RICHARD PHILLIPS
FEYNMAN
 DEDICATION

To my beloved son Avaneeth and daughter Late Avni

 LIST OF PUBLICATIONS

I. REFEREED JOURNALS BASED ON THE THESIS

1. Sandeep V. Nair, Kunal Layek, Kamalesh Hatua, “An Unequal Split

Dual Three-Phase PMSM with Extended Torque-Speed Characteristics for
Automotive Application”, IEEE Transactions on Power Electronics, vol. 37,
no. 10, pp. 12437-12449 (2022).
2. S. V. Nair, H. P. and K. Hatua, “Six-Step Operation of a Symmetric Dual
Three-Phase PMSM With Minimal Circulating Currents for Extended Speed
Range in Electric Vehicles,” in IEEE Transactions on Industrial Electronics,
vol. 69, no. 8, pp. 7651-7662 (2022).
3. S. V. Nair, K. Hatua, N. V. P. R. D. Prasad and D. K. Reddy, “A Quick I-f
Starting of PMSM Drive With Pole Slipping Prevention and Reduced Speed
Oscillations,” in IEEE Transactions on Industrial Electronics, vol. 68, no. 8,
pp. 6650-6661, (2021).
4. Sandeep V. Nair, Kamalesh Hatua, N.V.P.R. Durga Prasad, D. Kishore
Reddy, “Quick and seamless transition method for I-f to sensorless vector
control changeover and on-the-fly start of PMSM drives”, IET Electric Power
Applications, 14, (11), p. 2231-2242, (2020).

II. PRESENTATIONS IN CONFERENCES

1. S. V. Nair, V. P, S. Chakraborty, K. Layek and K. Hatua, “Design

Considerations for a Symmetric Dual Three-Phase IPMSM for Battery
Electric Vehicles,” 2022 IEEE International Conference on Power
Electronics, Smart Grid, and Renewable Energy (PESGRE), pp. 1-7, (2022).
2. S. V. Nair, K. Hatua, N. Durga Prasad and D. Kishore Reddy, “A Smooth and
Stable Open-Loop I-F Control for a Surface Mount PMSM Drive by Ensuring
Controlled Starting Torque,” IECON 2018 - 44th Annual Conference of the
IEEE Industrial Electronics Society, pp. 355-360, (2018).
3. S. V. Nair, S. Bhawal, K. Hatua, N. D. Prasad and D. K. Reddy, “Effect of
q-axis Inductance on the Design of an Industrial Inset Permanent Magnet
Synchronous Motor Drive with Restricted DC bus Voltage,” 2018 IEEE
International Conference on Power Electronics, Drives and Energy Systems
(PEDES), pp. 1-5, (2018).
4. S. V. Nair, K. Hatua, N. V. P. R. Durga Prasad and D. K. Reddy, “Pulse turn-
off method for starting of PMSM drive using back-emf position estimation
technique,” 2017 IEEE Transportation Electrification Conference (ITEC-
India), pp. 1-6, (2017).
5. S. Chakraborty, S. V. Nair and K. Hatua, “Effect of V Angle Variation and
Strengthening Rib On Performance Characteristics of IPMSM Motor for
Electric Vehicles,” 2021 1st International Conference on Power Electronics
and Energy (ICPEE), pp. 1-6, (2021).
III. PUBLICATIONS IN CONFERENCE PROCEEDINGS

1. S. V. Nair, V. P, S. Chakraborty, K. Layek and K. Hatua, “Design

Considerations for a Symmetric Dual Three-Phase IPMSM for Battery
Electric Vehicles,” 2022 IEEE International Conference on Power
Electronics, Smart Grid, and Renewable Energy (PESGRE), pp. 1-7, (2022).
2. S. V. Nair, K. Hatua, N. Durga Prasad and D. Kishore Reddy, “A Smooth and
Stable Open-Loop I-F Control for a Surface Mount PMSM Drive by Ensuring
Controlled Starting Torque,” IECON 2018 - 44th Annual Conference of the
IEEE Industrial Electronics Society, pp. 355-360, (2018).
3. S. V. Nair, S. Bhawal, K. Hatua, N. D. Prasad and D. K. Reddy, “Effect of
q-axis Inductance on the Design of an Industrial Inset Permanent Magnet
Synchronous Motor Drive with Restricted DC bus Voltage,” 2018 IEEE
International Conference on Power Electronics, Drives and Energy Systems
(PEDES), pp. 1-5, (2018).
4. S. V. Nair, K. Hatua, N. V. P. R. Durga Prasad and D. K. Reddy, “Pulse turn-
off method for starting of PMSM drive using back-emf position estimation
technique,” 2017 IEEE Transportation Electrification Conference (ITEC-
India), pp. 1-6, (2017).
5. S. Chakraborty, S. V. Nair and K. Hatua, “Effect of V Angle Variation and
Strengthening Rib On Performance Characteristics of IPMSM Motor for
Electric Vehicles,” 2021 1st International Conference on Power Electronics
and Energy (ICPEE), pp. 1-6, (2021).

vi
 ACKNOWLEDGEMENTS

I sincerely thank my supervisor Dr Kamalesh Hatua for nurturing me into a keen

and independent researcher. The intense discussions and brainstorming sessions with
him helped me to take up relevant problems and search for practical solutions. His
enthusiasm for power electronics research and his sincerity to work will always be
the light on my path ahead even after my PhD. His constant support during journal
rejections and challenging periods of my life gave me the strength to have perseverance
and to walk forward.
I feel proud to thank Dr Krishna Vasudevan for his course on machine modelling and
control, which formed the foundation of my research. His meticulous teaching and the
way he listens carefully to students’ questions, and his well-thought answers are skills
I have always wanted to learn.
I am grateful to my Doctoral Committee members Dr Mahesh Kumar, Dr Lakshmi
Narasamma, Dr Shanti Swarup and Dr T. Sundararajan, for helping my research with
thought-provoking questions during my seminars. I immensely thank Dr Srirama
Srinivas, Dr Boby George, Dr Sridharan K, and Dr Jagadeesh Kumar for their
interesting lectures on various courses I attended.
I cordially thank the lab staff Mr Kothandaraman, Mrs Jayasudha Avudhai Tankam
and Mr Veera Balaji, for their support in building my hardware setup and help during
demanding times. I owe a great deal to Mr Murugesan (SCS Engineering works) for
manufacturing the custom-designed PMSM and the testbed. I also thank Mr Dakshina
(Jaya Electricals) and Mr Manoharan for the customised winding of the stator. I am
deeply thankful to Mr NVPR Durga Prasad, Mr Kishore Reddy and Mrs Prasanna of
BHEL R&D, Hyderabad, for their support in executing the project on sensorless vector
control of PMSM drives. I gratefully thank the Department of Science and Technology,
Govt. of India, for providing financial support for the projects I had been involved.
I extend my heartfelt thanks to my labmate Dr Jose Titus for mentoring me in the initial

i
stages of my PhD and the exciting discussions we had on the basics of various subjects.
Words would fail short while thanking my friends Harikrishnan, Arun Chitrabhanu,
Prathibha, Amrutha, Prakash, Akhilesh, Harigopal, Pruthvi, Ganeshan, Surya Prakash
and Prasanna for being together and making my life at IIT Madras a memorable one. I
am also thankful to my neighbours Hari, Suma, Naina, Nikhil, Pavan, Sowmya, Arjun,
Aparna, Sreejith and Amulya for the joyful moments.
I would like to express my gratitude to my juniors in the lab Shivam, Kunal and Veena
for giving me the opportunity to mentor them in the collaborative work on various
industrial projects and for continuing my research work further. I am thankful to my
labmates Vamsi, Himanshu, Vibhav, Shekhar, Surja, Srikrishnan, Saravanan, Mrinmoy,
Shakthi, Shuvankar, Yash, Aloka, Aruni and Pradeep of ESB 101 for helping in creating
a stress-free and harmonious lab atmosphere. The technical discussions with my
colleagues and friends have helped me rethink various concepts from a new perspective
and foster fundamental ideas.
I can’t thank enough my wife, labmate and best friend, Dr Deepthi Sivadas, for her
encouragement and support throughout my PhD journey. She has been a constant
motivation for me to further my research and helped me overcome challenging times in
our life. The technical arguments with her helped me refine my ideas and form a basic
understanding of problems in various other domains.
I am indebted to my parents, Mr Venugopalan Nair and Mrs Geetha Venugopalan and
my grandparents, Mr Krishnankutty Menon (Late) and Mrs Radha A, for raising me up
and allowing me to follow my passion for research. Both my father and grandfather had
inspired me from my childhood with their innovative approach to problem-solving with
limited resources. I also extend my sincere thanks to my mother-in-law Mrs Girijamma
M L, for her encouragement and support during my PhD. I also thank my brother Mr
Sanoop V Nair, for his love and care. Finally, I am thankful to all my teachers, friends
and colleagues who moulded me to be who I am today.

ii
 ABSTRACT

KEYWORDS: Dual three-phase PMSM; Electric vehicles; High-speed PMSM;

I-f control; Starting methods; Winding changeover.

Electric automobiles are gaining increasing popularity due to their impact in reducing
global warming and better performance than conventional Internal Combustion Engine
(ICE) based solutions. Due to the high torque capability and wide operating speed
range, the electric vehicle drive train does not require a transmission gear compared
to ICE based vehicles, reducing the maintenance requirement and vehicle weight. The
purpose of this work is to improve the performance of Permanent Magnet Synchronous
Motors (PMSMs) through modifications in both control as well as machine design for
electric automotive and various industrial drives.
The first part of the thesis proposes a 0◦ displacement dual 3-phase IPMSM (DTP0-
IPMSM) for enhanced output power and operating speed range compared to a 3-
phase PMSM operated at the same DC bus voltage. The available DC bus voltage
is completely utilised by using a six-step operation, extending the operating speed
range to the maximum capability. The issue of circulating current that results in an
increased copper loss in the commonly used split-phase PMSM (30◦ displacement dual
three-phase PMSM) is not present in the proposed configuration. This topology would
eliminate the requirement for DC-DC boost conversion stage used in electric vehicles to
achieve a wide operating speed range. Consequently, the bulky inductor associated with
the boost converter is also eliminated, reducing the vehicle weight, hence improving the
driving range.
The thesis also proposes an unequal split 0◦ winding displacement dual 3-phase IPMSM
(uneq0-IPMSM) that is highly suited for heavy-duty trucks, off-road vehicles and
military trucks. Such applications require a very high initial torque, wide speed range
and high intermittent load capability during high-speed operation. The winding split

iii
ratio of the uneq0-IPMSM can be decided during the design stage to exactly match the
load characteristics, resulting in a lower power rating than an overrated conventional
3-phase PMSM for the same requirement. The uneq0-IPMSM can also operate in the
below base speed region even during high-speed operation, by isolating one set of 3-
phase winding with a higher number of turns from the corresponding inverter using
thyristor switches. This reduces the additional d-axis current requirement compared
to a conventional 3-phase machine operating in field weakening, improving the motor
efficiency in the higher speed range.
For various applications such as compressors, pumps, fans, Heating Ventilation and
Air Conditioning (HVAC) systems, and for critical applications such as electric ship
propulsion and ‘emergency heat and smoke exhaust’, a high precision dynamic control
at low-speed is not necessary. Instead, a simple and reliable control in medium and
high-speed range would be sufficient. A starting method is also proposed in this thesis to
achieve a quick and seamless transition from open-loop I-f control in low-speed region
to sensorless vector control in medium and high-speed range. The critical applications
also mandate a rugged and stable starting method. However, the open-loop I-f control of
PMSM suffers from inherent issues such as mid-frequency instability and pole slipping.
This thesis proposes a dynamic frequency slope control using a torque controller to
achieve quick starting without pole slipping.
Guidelines for designing a DTP0-IPMSM for an electric vehicle are also discussed
in this thesis. A 3 kW 3-phase PMSM that can be reconfigured as a dual 3-phase
or unequal split by changing the terminal connection is designed and manufactured
for testing the concepts presented. All the proposed control methods regarding the
DTP0-IPMSM and uneq0-IPMSM are experimentally demonstrated and validated on
the manufactured 3 kW dual three-phase PMSM setup. Further, the concepts related to
the open-loop starting methods and dynamic slope control are validated experimentally
on a 25 kW 3-phase PMSM drive.

iv
 TABLE OF CONTENTS

Page

ACKNOWLEDGEMENTS. . . . . . . . . . . . . . . . . . . . . . . . . . . i

ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii

LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii

LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xx

GLOSSARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxi

ABBREVIATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxii

NOTATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxv

CHAPTER 1: INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . 1

1.1 Permanent magnet synchronous motors (PMSM) for electric vehicles 2

1.2 Methods for extended constant power region and operating speed range
of IPMSM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.1 Field weakening of IPMSMs . . . . . . . . . . . . . . . . . 4
1.2.2 Re-rated PMSM with high current rated inverter switches . . 6
1.2.3 High DC bus voltage with series connection of cells . . . . 7
1.2.4 Intermediate DC-DC boost converter stage . . . . . . . . . 8
1.2.5 OEW with cascaded inverter and shared DC bus . . . . . . 9
1.2.6 OEW with two isolated voltage sources . . . . . . . . . . . 10
1.2.7 OEW with series compensation using a floating capacitor . . 11
1.2.8 Split-phase PMSM . . . . . . . . . . . . . . . . . . . . . . 13
1.2.9 Star-delta winding switch-over . . . . . . . . . . . . . . . . 14
1.2.10 Series-parallel winding switch-over . . . . . . . . . . . . . 15
1.2.11 Tapped winding changeover . . . . . . . . . . . . . . . . . 17
1.2.12 Five-phase PMSM with winding changeover . . . . . . . . 20

v
 1.2.13 A common feature of conventional winding changeover
methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
1.3 Sensorless vector controlled PMSM for marine propulsion . . . . . 23
1.3.1 Challenges in starting and low-speed control for sensorless
vector controlled PMSM . . . . . . . . . . . . . . . . . . . 24
1.3.2 Starting and changeover process . . . . . . . . . . . . . . . 25
1.4 Methods for changeover from open-loop starting to back-emf based
sensorless control of PMSM . . . . . . . . . . . . . . . . . . . . . 26
1.4.1 Direct transition . . . . . . . . . . . . . . . . . . . . . . . 26
1.4.2 Increased motor acceleration rate based transition . . . . . . 27
1.4.3 Stator current profiling based transition . . . . . . . . . . . 28
1.5 Stability of open-loop starting methods of PMSM . . . . . . . . . . 29
1.6 Motivation and Objectives of the thesis . . . . . . . . . . . . . . . . 30
1.7 Organisation of the thesis . . . . . . . . . . . . . . . . . . . . . . . 34

CHAPTER 2: PERMANENT MAGNET SYNCHRONOUS MOTORS:

BASICS, MODELLING, AND VECTOR CONTROL . . . . . . . . . . . . 36

2.1 Types of PMSM motors . . . . . . . . . . . . . . . . . . . . . . . . 37

2.2 The space phasor concept and rotating magnetic field . . . . . . . . 40
2.2.1 Rotating magnetic field in AC machines . . . . . . . . . . . 44
2.3 Dynamic model of three-phase IPMSM using space phasor . . . . . 46
2.3.1 Equivalent circuit of IPMSM in the rotor reference frame . . 51
2.3.2 Phasor diagram of three-phase IPMSM . . . . . . . . . . . 52
2.4 Controller design for IPMSM . . . . . . . . . . . . . . . . . . . . . 54
2.4.1 Current controller . . . . . . . . . . . . . . . . . . . . . . . 54
2.4.2 Speed controller . . . . . . . . . . . . . . . . . . . . . . . 57
2.5 Space Vector PWM . . . . . . . . . . . . . . . . . . . . . . . . . . 60
2.6 Overmodulation and six-step operation for increased DC bus utilisation 63
2.6.1 Overmodulation region-1 . . . . . . . . . . . . . . . . . . . 63

vi
 2.6.2 Overmodulation region-2 . . . . . . . . . . . . . . . . . . . 65
2.7 Maximum torque per ampere and field weakening in IPMSM . . . . 66
2.8 Optimal current computation of IPMSM considering parameter
variations using lookup table . . . . . . . . . . . . . . . . . . . . . 71
2.9 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

CHAPTER 3: A DUAL 3-PHASE PMSM WITH SIX-STEP OPERATION

FOR ELECTRIC VEHICLES . . . . . . . . . . . . . . . . . . . . . . . . . 77

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
3.2 Proposed power enhancement and extended speed range with dual
three-phase machines . . . . . . . . . . . . . . . . . . . . . . . . . 78
3.2.1 Comparison of proposed dual three-phase PMSM with existing
topologies . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
3.3 Modelling of DTP0-IPMSM . . . . . . . . . . . . . . . . . . . . . 86
3.3.1 Equivalent circuit of DTP0-IPMSM in the rotor reference frame 94
3.3.2 Phasor diagram of DTP0-IPMSM . . . . . . . . . . . . . . 95
3.4 Circulating current in dual three-phase PMSMs during six-step
operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
3.4.1 Circulating current concept . . . . . . . . . . . . . . . . . . 98
3.4.2 Analysis of circulating current in split-phase PMSM . . . . 102
3.5 Proposed circulating current minimisation of high-speed dual-three
phase IPMSM under six-step operation . . . . . . . . . . . . . . . . 103
3.6 Vector control of DTP0-IPMSM with six-step operation . . . . . . . 106
3.6.1 Current vector control (CVC) . . . . . . . . . . . . . . . . 108
3.6.2 Voltage angle control (VAC) for six-step operation . . . . . 109
3.6.3 Transition from CVC to six-step VAC . . . . . . . . . . . . 110
3.6.4 Reverse transition from six-step VAC to CVC . . . . . . . . 112
3.7 Experimental results . . . . . . . . . . . . . . . . . . . . . . . . . 112
3.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123

vii
CHAPTER 4: AN UNEQUAL SPLIT DUAL 3-PHASE PMSM FOR
HIGH STARTING TORQUE AND WIDE OPERATING SPEED RANGE
IN MEDIUM AND HEAVY-DUTY TRUCKS . . . . . . . . . . . . . . . . 125

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125

4.2 Existing solutions for extended constant power operating speed range 126
4.3 Variable torque-speed characteristics using the proposed dual three
phase PMSM with unequal split winding . . . . . . . . . . . . . . . 128
4.3.1 Effect of variation of split ratio on the uneq0-PMSM
characteristics . . . . . . . . . . . . . . . . . . . . . . . . . 132
4.4 Modelling of uneq0-IPMSM . . . . . . . . . . . . . . . . . . . . . 133
4.4.1 Equivalent circuit of the uneq0-IPMSM in the rotor reference
frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
4.4.2 Phasor diagram of the uneq0-IPMSM . . . . . . . . . . . . 137
4.4.3 Vector control of the proposed uneq0-IPMSM . . . . . . . . 138
4.4.4 Transition from two-winding to one-winding operation and vice
versa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
4.5 Parametric plane for PMSM design . . . . . . . . . . . . . . . . . . 145
4.5.1 Overloading performance of various IPMSM designs . . . . 150
4.6 Comparison of the proposed uneq0-PMSM with three-phase PMSM
and DTP0-PMSM for medium/heavy-duty trucks and military vehicles 151
4.7 Comparison of proposed uneq0-PMSM with existing winding
changeover techniques . . . . . . . . . . . . . . . . . . . . . . . . 157
4.8 Experimental results . . . . . . . . . . . . . . . . . . . . . . . . . 162
4.9 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168

CHAPTER 5: MACHINE DESIGN CONSIDERATIONS OF DUAL 3-

PHASE PMSM FOR ELECTRIC AUTOMOTIVE . . . . . . . . . . . . . 170

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170

viii
 5.2 Formation of torque-speed characteristics considering an electric
vehicle specification . . . . . . . . . . . . . . . . . . . . . . . . . . 170
5.3 Design process of a dual three-phase IPMSM for EV application . . 173
5.3.1 Machine ratings and dimensions . . . . . . . . . . . . . . . 173
5.3.2 Number of poles . . . . . . . . . . . . . . . . . . . . . . . 175
5.3.3 DC bus voltage . . . . . . . . . . . . . . . . . . . . . . . . 176
5.3.4 Stator inner diameter, rotor outer diameter, air-gap flux density
and linear current density . . . . . . . . . . . . . . . . . . . 176
5.3.5 Air gap . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
5.3.6 Slot design and winding . . . . . . . . . . . . . . . . . . . 177
5.3.7 Magnet dimensions . . . . . . . . . . . . . . . . . . . . . . 178
5.3.8 No-load air-gap flux density . . . . . . . . . . . . . . . . . 178
5.3.9 Number of turns and wire gauge . . . . . . . . . . . . . . . 179
5.3.10 Slot area and slot width to slot pitch ratio . . . . . . . . . . 180
5.3.11 Design optimisation . . . . . . . . . . . . . . . . . . . . . 181
5.4 Winding configuration of symmetric dual three-phase IPMSM (DTP0-
IPMSM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
5.5 Modelling of an electric vehicle load characteristics for drive cycle
testing of the DTP0-PMSM . . . . . . . . . . . . . . . . . . . . . . 184
5.6 Operation of the 37 kW DTP0-IPMSM with six-step operation through
modified Indian driving cycle . . . . . . . . . . . . . . . . . . . . . 186
5.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189

CHAPTER 6: OPEN-LOOP STARTING AND POWER FAILURE RIDE

THROUGH IN SENSORLESS VECTOR CONTROLLED 3-PHASE
PMSM FOR MARINE PROPULSION . . . . . . . . . . . . . . . . . . . . 191

6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191

6.2 Open-loop I-f starting . . . . . . . . . . . . . . . . . . . . . . . . . 192
6.2.1 Damping of speed oscillations during I-f starting . . . . . . 195

ix
 6.3 Stator voltage integration based sensorless vector control of 3-phase
PMSM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196
6.4 Quick and reliable transition from open-loop I-f to sensorless vector
control for critical applications . . . . . . . . . . . . . . . . . . . . 197
6.5 Analysis of direct transition method . . . . . . . . . . . . . . . . . 198
6.6 Proposed pulse-off method . . . . . . . . . . . . . . . . . . . . . . 203
6.6.1 Computation of the q-axis current demanded by the load . . 206
6.6.2 Simulation of the proposed method . . . . . . . . . . . . . 207
6.6.3 On-the-fly starting for power failure ride through . . . . . . 208
6.7 Determination of pulse-off time . . . . . . . . . . . . . . . . . . . 210
6.7.1 Calculation of interval-1 time duration . . . . . . . . . . . . 211
6.7.2 Calculation of interval-2 time duration . . . . . . . . . . . . 213
6.7.3 Calculation of total stator current decay time . . . . . . . . 214
6.8 Experimental validation . . . . . . . . . . . . . . . . . . . . . . . . 215
6.9 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222

CHAPTER 7: STABILITY OF PMSM IN OPEN-LOOP CONTROL . . 223

7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223

7.2 Instability during open-loop control of PMSM . . . . . . . . . . . . 224
7.2.1 Pole slipping during frequency ramp-up . . . . . . . . . . . 224
7.2.2 Mid-frequency instability and its small-signal stabilisation
using reference frequency modulation . . . . . . . . . . . . 226
7.3 Large-signal dynamic model of PMSM with I-f control . . . . . . . 231
7.4 Proposed dynamic frequency slope control using torque controller . 234
7.4.1 Torque controller . . . . . . . . . . . . . . . . . . . . . . . 234
7.4.2 Total torque demand estimation . . . . . . . . . . . . . . . 239
7.5 Slope attenuator for smooth transition from acceleration interval to
constant speed interval . . . . . . . . . . . . . . . . . . . . . . . . 242
7.6 Experimental results . . . . . . . . . . . . . . . . . . . . . . . . . 244

x
 7.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251

CHAPTER 8: HARDWARE ORGANISATION . . . . . . . . . . . . . . 253

8.1 Power circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253

8.1.1 Custom designed reconfigurable 0 deg displacement 4-split
IPMSM . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
8.1.2 Loading setup . . . . . . . . . . . . . . . . . . . . . . . . . 258
8.1.3 IGBT based 2-level VSI . . . . . . . . . . . . . . . . . . . 259
8.1.4 Thyristor switch box . . . . . . . . . . . . . . . . . . . . . 262
8.1.5 Autotransformer . . . . . . . . . . . . . . . . . . . . . . . 263
8.2 Control circuit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
8.2.1 DSP-FPGA control platform . . . . . . . . . . . . . . . . . 264
8.2.2 Protection board . . . . . . . . . . . . . . . . . . . . . . . 265
8.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267

CHAPTER 9: CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . 268

9.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268

9.2 Summary of present work . . . . . . . . . . . . . . . . . . . . . . . 269
9.3 Future scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270
9.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270

APPENDIX A: Derivation of maximum generated torque per ampere in

the uneq0-IPMSM. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272

REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
 LIST OF TABLES

Table Title Page

1.1 Motor ratings and performance parameters . . . . . . . . . . . . . . 3

3.1 Dual three-phase IPMSM parameters used for simulation . . . . . . 81

3.2 Comparison of proposed method with conventional topologies (Pan
et al., 2014) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
3.3 Stator current harmonics in DTP0-IPMSM and split-phase IPMSM
during six-step operation at 4660 rpm . . . . . . . . . . . . . . . . 106
3.4 Parameters of experimental DTP0-IPMSM prototype . . . . . . . . 113

4.1 Parameters of various three-phase IPMSM designs . . . . . . . . . 153

4.2 Comparison of various IPMSM designs with three-phase, dual three-
phase and uneq0-IPMSM configurations for military truck . . . . . 156
4.3 Comparison of proposed uneq0-PMSM and existing winding
changeover methods . . . . . . . . . . . . . . . . . . . . . . . . . 158
4.4 Parameters of experimental uneq0-IPMSM prototype . . . . . . . . 162

5.1 Specifications and parameters of Mahindra e-Verito (e-Verito; Ehsani

et al., 2018) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
5.2 Design parameters of forced-air cooled 37 kW DTP0-IPMSM . . . 175

6.1 Inset-PMSM parameters . . . . . . . . . . . . . . . . . . . . . . . 199

6.2 Performance comparison of various transition methods . . . . . . . 208

7.1 Motor and controller parameters . . . . . . . . . . . . . . . . . . . 228

xii
 LIST OF FIGURES

Figure Title Page

1.1 Expansion of torque-speed and power-speed characteristics of IPMSM

with field weakening . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2 Power circuit of three-phase PMSM supplied from a 2-level VSI with
high voltage battery . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.3 Power circuit of three-phase PMSM supplied from a 2-level VSI and
DC-DC boost converter . . . . . . . . . . . . . . . . . . . . . . . . 9
1.4 Power circuit of open-end winding IPMSM supplied by cascaded
inverter and shared DC bus . . . . . . . . . . . . . . . . . . . . . . 10
1.5 Power circuit of open-end winding IPMSM supplied by isolated
inverters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.6 Power circuit of open-end winding IPMSM drive with floating capacitor
bridge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.7 Power circuit of split-phase IPMSM . . . . . . . . . . . . . . . . . 12
1.8 Winding arrangement of split-phase IPMSM . . . . . . . . . . . . . 13
1.9 Power circuit of IPMSM with star-delta winding switch-over . . . . 15
1.10 Power circuit of IPMSM with star-delta winding switch-over with
reduced number of switches . . . . . . . . . . . . . . . . . . . . . 15
1.11 Power circuit of IPMSM with series-parallel winding switch-over . 16
1.12 Power circuit of IPMSM with tapped winding switch-over . . . . . 17
1.13 Power circuit of IPMSM with combined star-delta and tapped winding
switch-over configuration . . . . . . . . . . . . . . . . . . . . . . . 18
1.14 Different winding configurations with combined star-delta and tapped
winding switch-over . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.15 Power circuit of IPMSM with tapped winding switch-over
configuration with reduced switches . . . . . . . . . . . . . . . . . 19
1.16 Power circuit of five-phase IPMSM with star, pentagon, pentacle
winding switch-over configuration . . . . . . . . . . . . . . . . . . 21
1.17 Different winding configurations with five-phase IPMSM . . . . . . 21
1.18 General characteristics of winding changeover methods when field
weakening is not used during higher equivalent number of turns
operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.19 General characteristics of winding changeover methods if field
weakening would have been used during higher equivalent number of
turns operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.20 d′ −q ′ reference frame used for I-f control . . . . . . . . . . . . . . 25

2.1 Classification of permanent magnet AC motors . . . . . . . . . . . 36

2.2 Surface-mount PMSM rotor . . . . . . . . . . . . . . . . . . . . . 38
2.3 Rotor structures of permanent magnet AC motors . . . . . . . . . . 38
2.4 Stator winding distribution and rotor of a three-phase IPMSM . . . 40
2.5 mmf distribution in space due to excitation of a-phase with a DC current
of magnitude ‘Im ’ . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

xiii
2.6 Representation of mmf distribution around stator periphery when a-
phase is excited with a DC current of magnitude ‘Im ’ . . . . . . . . 41
2.7 Stator current space phasor in an equivalent two-phase IPMSM . . . 43
2.8 Rotating mmf when the three-phase stator windings are excited with
sinusoidal currents that are phase displaced by 120◦ from each other 45
2.9 Rotor reference frame . . . . . . . . . . . . . . . . . . . . . . . . . 46
2.10 Equivalent circuit of three-phase IPMSM in rotor reference frame . 52
2.11 Phasor diagram of three-phase IPMSM in rotor reference frame . . . 53
2.12 Generalised plant model of IPMSM for current controller design . . 55
2.13 Block diagram of closed-loop current control of IPMSM . . . . . . 55
2.14 Reduced current controller block diagram . . . . . . . . . . . . . . 55
2.15 Simplified mechanical model of IPMSM for speed controller design 58
2.16 Block diagram of speed controller with IPMSM model . . . . . . . 58
2.17 Reduced block diagram of speed controller . . . . . . . . . . . . . 58
2.18 Bode plot of open-loop transfer function of speed control . . . . . . 59
2.19 Space vector diagram of 2-level VSI . . . . . . . . . . . . . . . . . 61
2.20 Reference vector and applied vector during ovm-1 . . . . . . . . . . 64
2.21 Reference vector and applied vector during ovm-2 . . . . . . . . . . 65
2.22 Current limit circle and voltage limit ellipse in ids −iqs plane for IPMSM 68
2.23 MTPA and MTPV lines in ids − iqs plane for IPMSM . . . . . . . . 69
2.24 Flow diagram of iterative process for LUT generation of DTP0-PMSM 73
2.25 Lookup table values of d-axis current reference (id−ref ) . . . . . . . 75
2.26 Lookup table values of q-axis current reference (iq−ref ) . . . . . . . 75

3.1 DTP0-PMSM obtained from reconfiguration of three-phase PMSM . 79

3.2 Winding distributions for original three-phase PMSM and various
reconfigured dual three-phase PMSMs . . . . . . . . . . . . . . . . 80
3.3 Power circuit of DTP0-PMSM . . . . . . . . . . . . . . . . . . . . 80
3.4 Power-speed characteristics of three-phase and DTP0-IPMSM . . . 82
3.5 Torque-speed characteristics of three-phase and DTP0-IPMSM . . . 82
3.6 Winding distribution and rotor of a DTP0-IPMSM . . . . . . . . . . 87
3.7 Equivalent circuit of DTP0-IPMSM in rotor reference frame . . . . 95
3.8 Phasor diagram for winding-1 of DTP0-IPMSM in rotor reference
frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
3.9 Phasor diagram for winding-2 of DTP0-IPMSM in rotor reference
frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
3.10 Phase current waveforms of split-phase IPMSM during six-step
operation with huge circulating harmonic current components . . . 98
3.11 Harmonic equivalent circuit of dual three-phase SMPMSM in α-β
reference frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
3.12 α − β reference frame for the split-phase PMSM . . . . . . . . . . 102
3.13 Circulating current in split-phase IPMSM during six-step operation . 103
3.14 α − β reference frame for DTP0-IPMSM . . . . . . . . . . . . . . 104
3.15 Circulating current in DTP0-IPMSM during six-step operation . . . 105
3.16 Developed torque in DTP0-IPMSM during six-step operation . . . . 106
3.17 Control system for sensored vector control of DTP0-IPMSM under all
modulation ranges including six-step . . . . . . . . . . . . . . . . . 107

xiv
3.18 Changeover from closed-loop CVC to open-loop six-step VAC . . . 111
3.19 Experimental setup for DTP0-IPMSM drive testing . . . . . . . . . 113
3.20 DTP0-IPMSM efficiency contours . . . . . . . . . . . . . . . . . . 114
3.21 Three-phase IPMSM efficiency contours . . . . . . . . . . . . . . . 115
3.22 DTP0-IPMSM stator current contours . . . . . . . . . . . . . . . . 115
3.23 Three-phase IPMSM stator current contours . . . . . . . . . . . . . 116
3.24 Phase currents & line voltages in six-step mode:- Ch1: ias1 ; Ch2: ias2 ;
Ch3: va1b1 ; Ch4: va2b2 ; Ch-f(t): ia-cir (X-axis:- 7.4 ms/div; Y-axis:-
Ch1,Ch2: 10 A/div; Ch3,Ch4: 200 V/div; Ch-f(t): 20 A/div) . . . . . 117
3.25 α-component of current & voltage in six-step mode:- Ch1: iαs1 ; Ch2:
iαs2 ; Ch3: vα1 ; Ch4: vα2 ; Ch-f(t): iα-cir (X-axis:- 7.4 ms/div; Y-axis:-
Ch1,Ch2: 17 A/div; Ch3,Ch4: 300 V/div; Ch-f(t): 34 A/div) . . . . . 117
3.26 Phase currents and line voltages in linear region:- Ch1: ias1 ; Ch2: ias2 ;
Ch3: va1b1 ; Ch4: va2b2 ; Ch-f(t): ia-cir (X-axis:- 7.6 ms/div; Y-axis:-
Ch1,Ch2: 10 A/div; Ch3,Ch4: 200 V/div; Ch-f(t): 20 A/div) . . . . . 118
3.27 Phase currents and line voltages in ovm1:- Ch1: ias1 ; Ch2: ias2 ; Ch3:
va1b1 ; Ch4: va2b2 ; Ch-f(t): ia-cir (X-axis:- 7.6 ms/div; Y-axis:- Ch1,Ch2:
10 A/div; Ch3,Ch4: 200 V/div; Ch-f(t): 20 A/div) . . . . . . . . . . 119
3.28 Phase currents and line voltages in ovm2:- Ch1: ias1 ; Ch2: ias2 ; Ch3:
va1b1 ; Ch4: va2b2 ; Ch-f(t): ia-cir (X-axis:- 7.6 ms/div; Y-axis:- Ch1,Ch2:
10 A/div; Ch3,Ch4: 200 V/div; Ch-f(t): 20 A/div) . . . . . . . . . . 119
3.29 Transition from CVC to open-loop six-step VAC during speed ramp-
up :- Ch1: ma1 ; Ch2: iqs1 ; Ch3: ids1 ; Ch4: ias1 (X-axis:- 1.32 s/div;
Zoom X-axis:- 60 ms/div; Y-axis:- Ch1: 1 pu/div; Ch2: 4.24 A/div;
Ch3: 10.6 A/div; Ch4: 2.7 A/div) . . . . . . . . . . . . . . . . . . . 120
3.30 Reverse transition from open-loop six-step VAC to CVC during speed
ramp-down:- Ch1: ma1 ; Ch2: iqs1 ; Ch3: ids1 ; Ch4: ias1 (X-axis:-
1.32 s/div; Zoom X-axis:- 60 ms/div; Y-axis:- Ch1: 1 pu/div; Ch2:
4.24 A/div; Ch3: 10.6 A/div; Ch4: 2.7 A/div) . . . . . . . . . . . . . 120
3.31 Performance of proposed method during step torque command at
1462 rpm with τTe_LPF =2 ms:- Ch1: ma1 ; Ch2: Tb e ; Ch3: ωm ; Ch4:
ics1 ; (X-axis:- 200 ms/div; Zoom X-axis:-18.8 ms/div; Y-axis:- Ch1:
1 pu/div; Ch2: 15 Nm/div Ch3: 222 rpm/div; Ch4: 15.6 A/div) . . . 122
3.32 Performance of proposed method during step torque command at
1462 rpm with τTe_LPF = 5 ms:- Ch1: ma1 ; Ch2: Tb e ; Ch3: ωm ; Ch4:
ic1 ; (X-axis:- 200 ms/div; Zoom X-axis:-18.8 ms/div; Y-axis:- Ch1:
1 pu/div; Ch2: 15 Nm/div Ch3: 222 rpm/div; Ch4: 15.6 A/div) . . . 122
3.33 Performance of proposed method during step torque command at
1462 rpm with τTe_LPF =10 ms:- Ch1: ma1 ; Ch2: Tb e ; Ch3: ωm ; Ch4:
ics1 ; (X-axis:- 200 ms/div; Zoom X-axis:-18.8 ms/div; Y-axis:- Ch1:
1 pu/div; Ch2: 15 Nm/div Ch3: 222 rpm/div; Ch4: 15.6 A/div) . . . 122
3.34 Performance of proposed method at 1722 rpm and 21.2 Nm load in
six-step region:- Ch1: ma1 ; Ch2: Tb e ; Ch3: ωm ; Ch4: ics1 ; (X-axis:-
10 ms/div; Y-axis:- Ch1: 2 pu/div; Ch2: 7 Nm/div Ch3: 312 rpm/div;
Ch4: 10 A/div) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123

xv
4.1 Typical torque-speed characteristics requirement for a military truck
(Gao et al., 2003) . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
4.2 Power circuit of the proposed uneq0-PMSM . . . . . . . . . . . . . 129
4.3 uneq0-PMSM obtained from reconfiguration of three-phase PMSM 129
4.4 Operation of the proposed uneq0-PMSM (spec-3 with 1:13 split) . . 130
4.5 Variation of power-speed and torque-speed characteristics with split
ratio in uneq0-PMSM (spec-3) . . . . . . . . . . . . . . . . . . . . 132
4.6 Equivalent circuit of uneq0-IPMSM in rotor reference frame . . . . 137
4.7 Phasor diagram for LS-winding of uneq0-IPMSM in rotor reference
frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
4.8 Phasor diagram for HS winding of uneq0-IPMSM in rotor reference
frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
4.9 Control system of the proposed uneq0-IPMSM with sensored vector
control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
4.10 Transition from two-winding operation to one-winding and vice versa 143
4.11 Flow diagram of winding transition from two-winding operation to one-
winding and vice versa for the proposed uneq0-IPMSM . . . . . . . 144
4.12 Simulation results of uneq0-IPMSM with proposed changeover during
speed ramp-up and ramp-down . . . . . . . . . . . . . . . . . . . . 146
4.12 Simulation results of uneq0-IPMSM with proposed changeover during
speed ramp-up and ramp-down . . . . . . . . . . . . . . . . . . . . 147
4.13 Parameter plane of IPMSM with maximum speed contours . . . . . 148
4.14 Variation of power-speed and torque-speed characteristics for various
design on IPM parameter plane . . . . . . . . . . . . . . . . . . . . 149
4.15 Overloading performance of IPMSMs located on the optimal FW IPM
design line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
4.16 Overloading performance of IPMSM Design-C with twice rated current 152
4.17 Parameter plane for IPMSM design . . . . . . . . . . . . . . . . . 153
4.18 Characteristics of three-phase IPMSM spec-1 with Constant Power
Speed Range (CPSR) 24.7, Vlim = 1 pu . . . . . . . . . . . . . . . . 154
4.19 Characteristics of three-phase IPMSM spec-2 with CPSR 5.87,
Vlim = 3.4 pu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
4.20 Characteristics of three-phase IPMSM spec-3 with CPSR 4.8,
Vlim = 4.47 pu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
4.21 Characteristics of uneq0-IPMSM using reconfigured spec-2 with CPSR
5.87, Vlim = 1 pu . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
4.22 Characteristics of uneq0-IPMSM using reconfigured spec-3 with CPSR
4.8, Vlim = 1 pu . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
4.23 Experimental setup of uneq0-IPMSM fed from two inverters with back-
to-back thyristor switches . . . . . . . . . . . . . . . . . . . . . . . 163
4.24 Experimental results of DTP0-IPMSM with equivalent current contours 164
4.25 Experimental results of proposed uneq0-IPMSM with equivalent
current contours . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164

xvi
4.26 Performance of DTP0-IPMSM during speed increase and decrease
at near full load:- Ch1: ωm ; Ch2: ia1 ; Ch3: Tb e ; Ch4: ia2 ; (X-
axis:- 3.8 s/div; Y-axis:- Ch1: 960 rpm/div; Ch2,Ch4: 10 A/div; Ch3:
7.49 Nm/div) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
4.27 Performance of uneq0-IPMSM during speed increase at near full load
with winding changeover:- Ch1: ωm ; Ch2: ia-LS ; Ch3: Tb e ; Ch4:
ia-HS ; (X-axis:- 1.96 s/div; Zoom X-axis:-20 ms/div; Y-axis:- Ch1:
690 rpm/div; Ch2,Ch4: 12 A/div; Ch3: 6.81 Nm/div) . . . . . . . . 166
4.28 Performance of uneq0-IPMSM during speed increase at 2.27 Nm load
with winding changeover:- Ch1: id-LS ; Ch2: ia-LS ; Ch3: id-HS ; Ch4:
ia-HS ; (X-axis:- 1.84 s/div; Zoom X-axis:-20 ms/div; Y-axis:- Ch1,Ch3:
10.18 A/div; Ch2,Ch4: 10 A/div) . . . . . . . . . . . . . . . . . . . 167
4.29 Performance of uneq0-IPMSM during speed decrease at 2.27 Nm load
with winding changeover:- Ch1: id-LS ; Ch2: ia-LS ; Ch3: id-HS ; Ch4:
ia-HS ; (X-axis:- 1.88 s/div; Zoom X-axis:-20 ms/div; Y-axis:- Ch1,Ch3:
10.18 A/div; Ch2,Ch4: 10 A/div) . . . . . . . . . . . . . . . . . . . 168
4.30 Steady-state operation of uneq0-IPMSM at 990 rpm under full load:-
Ch1: ma-LS ; Ch2: ia-LS ; Ch3: ma-HS ; Ch4: ia-HS ; (X-axis:- 20 ms/div;
Y-axis:- Ch1,Ch3: 1 pu/div; Ch2,Ch4: 5 A/div) . . . . . . . . . . . 169

5.1 Torque-speed characteristics representing specifications of Mahindra

e-Verito BEV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
5.2 Design flow diagram of Dual three-phase IPMSM for EV . . . . . . 174
5.3 Rotor structure with V-shaped PM arrangement and ribs for structural
stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
5.4 Variation of normal component of flux density in air gap along rotor
periphery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
5.5 Plot of developed torque versus stator current space vector angle from
FEM simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
5.6 Flux density distribution during MTPA operation . . . . . . . . . . 182
5.7 Mechanical structure and winding arrangement of the designed 37 kW
DTP0-IPMSM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
5.8 Winding configuration of 37 kW DTP0-IPMSM . . . . . . . . . . . 183
5.9 Vehicle speed profile in MIDC . . . . . . . . . . . . . . . . . . . . 186
5.10 Simulation of the proposed DTP0-IPMSM in MIDC driving cycle . 188
5.11 MIDC operating points of Mahindra e-verito with the proposed DTP0-
IPMSM in torque-speed plane . . . . . . . . . . . . . . . . . . . . 189
5.12 MIDC operating points of Mahindra e-verito with the proposed DTP0-
IPMSM in power-speed plane . . . . . . . . . . . . . . . . . . . . 189

6.1 Synchronously rotating reference frame and rotor reference frame

during I-f open-loop control of PMSM . . . . . . . . . . . . . . . . 193
6.2 Starting frequency profile with I-f control . . . . . . . . . . . . . . 193
6.3 Sensorless position estimation using stator voltage integration . . . 197
6.4 Control system for sensorless vector control of PMSM with I-f starting 198
6.5 Simulation results of direct transition method at the changeover . . . 200

xvii
6.6 Simulation results of direct transition with 15% additional error at the
changeover . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
6.7 Measured line-line back-emf voltage of the 25 kW inset-PMSM: Ch4 -
vab (X-axis:- 50 ms/div; Y-axis:- 5 V/div) . . . . . . . . . . . . . . . 204
6.8 Relative position of back-emf and rotor flux vectors . . . . . . . . . 205
6.9 Simulation results of the proposed pulse-off method . . . . . . . . . 209
6.10 PMSM supplied from inverter . . . . . . . . . . . . . . . . . . . . 210
6.11 Stator current decay during pulse-off interval . . . . . . . . . . . . 211
6.12 Equivalent circuit of PMSM under pulse-off interval . . . . . . . . 211
6.13 Variation of stator currents with current vector angle . . . . . . . . 214
6.14 Stator current decay time variation during pulse-off . . . . . . . . . 215
6.15 Experimental setup of 25 kW inset-PMSM drive . . . . . . . . . . . 216
6.16 Speed and stator current waveforms with pulse-off method under
loaded condition. Ch3: ω b r , Ch4: ias (X-axis:- 0.7 s/div;
Y-axis:- Ch3: 240 mech rpm/div, Ch4: 50 A/div) . . . . . . . . . . 217
6.17 Speed and stator current waveforms with pulse-off method
under no-load. Ch3: ω
b r , Ch4: ias (X-axis:- 0.36 s/div;
Y-axis:- Ch3: 240 mech rpm/div, Ch4: 50 A/div) . . . . . . . . . . 217
6.18 Variation of d-axis and q-axis currents with pulse-off method under
loaded condition. Ch1: iq−ref , Ch2: i′q , Ch3: i′d , Ch4: ias (X-axis:-
0.74 s/div; Y-axis:- Ch1,Ch2,Ch3: 40 A/div, Ch4: 50 A/div) . . . . . 218
6.19 Variation of control system transformation angle with the pulse-off
method at the transition instant. Ch1: θi , Ch3: θc , Ch4: pulse-off signal
(X-axis:- 50 ms/div; Y-axis:- Ch1,Ch3,Ch4: 0.4 unit/div . . . . . . . 219
6.20 Stator current waveforms in the pulse-off duration. Ch2: ias , Ch3: ibs ,
Ch4: ics (X-axis:- 20 µs/div; Y-axis:- Ch2,Ch3,Ch4: 20 A/div . . . . 219
6.21 Speed and stator current waveforms during on-the-fly start with pulse-
off method under loaded condition. Ch3: ω b r , Ch4: ias (X-axis:-
1.04 s/div; Y-axis:- Ch3: 240 mech rpm/div, Ch4: 50 A/div) . . . . . 220
6.22 Speed and stator current waveforms during on-the-fly start with pulse-
off method under no-load. Ch3: ω b r , Ch4: ias (X-axis:- 0.76 s/div; Y-
axis:- Ch3: 240 mech rpm/div, Ch4: 20 A/div) . . . . . . . . . . . . 220
6.23 Variation of d-axis and q-axis currents during on-the-fly
start with pulse-off method under loaded condition. Ch1:
iq−ref , Ch2: i′q , Ch3: i′d , Ch4: ias (X-axis:- 0.92 s/div;
Y-axis:- Ch1,Ch2,Ch3: 40 A/div, Ch4: 50 A/div) . . . . . . . . . . . 221
6.24 Variation of d-axis and q-axis currents during on-the-fly start with
pulse-off method under no-load. Ch1: iq−ref , Ch2: i′q , Ch3: i′d , Ch4: ias
(X-axis:- 0.68 s/div; Y-axis:- Ch1,Ch2,Ch3: 40 A/div, Ch4: 50 A/div) 221

7.1 Variation of the developed torque with the load angle . . . . . . . . 225
7.2 Eigenvalue loci of state transition matrix as a function of ωi with
conventional I-f control . . . . . . . . . . . . . . . . . . . . . . . . 228
7.3 Eigenvalue loci of state transition matrix as a function of ωi with
reference frequency modulation for various CG , where τh is fixed at
0.159 s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230

xviii
7.4 Eigenvalue loci of state transition matrix as a function of ωi with
reference frequency modulation for various τh , where CG is fixed at
0.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
7.5 Large-signal dynamic model of PMSM with open-loop control . . . 233
7.6 Pole slipping during frequency ramp-up with conventional I-f control 233
7.7 Pole slipping with frequency modulated I-f control . . . . . . . . . 233
7.8 Control system with proposed torque controller and slope attenuator 235
7.9 Control system with proposed torque controller . . . . . . . . . . . 237
7.10 Frequency response plot of the closed-loop system transfer function
with proposed integral control . . . . . . . . . . . . . . . . . . . . 238
7.11 Large-signal dynamic model of PMSM using I-f control with torque
controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
7.12 Stable operation during frequency ramp-up using I-f control with torque
controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
7.13 Flux estimation scheme . . . . . . . . . . . . . . . . . . . . . . . . 242
7.14 Speed and torque oscillations at the transition interval using I-f control
with torque controller . . . . . . . . . . . . . . . . . . . . . . . . . 243
7.15 Smooth transition from ramp-up interval to constant speed interval
using slope attenuator . . . . . . . . . . . . . . . . . . . . . . . . . 244
7.16 Hardware setup for 25 kW PMSM drive . . . . . . . . . . . . . . . 245
7.17 Oscillations in speed during low-speed operation: Ch1 - ωi ; Ch2
- Pe ; Ch3 - ω b r ; Ch4 - ias ; (X-axis:- 0.52 s/div; Y-axis:- Ch1,Ch3:
20.1 elec rad/s/div; Ch2: 8 kW/div; Ch4: 40 A/div) . . . . . . . . . 246
7.18 Unstable operation with conventional I-f control: Ch1- ωi ; Ch2 -
Pe ; Ch3 - ω b r ; Ch4 - ias (X-axis:- 1.56 s/div; Y-axis:- Ch1,Ch3:
75.4 elec rad/s/div; Ch2: 8 kW/div; Ch4: 40 A/div) . . . . . . . . . 246
7.19 Stabilization of mid-frequency oscillations in speed with reference
frequency modulation: Ch1 - ωref ; Ch2 - control signal; Ch3 - ω b r ; Ch4
- ias ; (X-axis:- 0.46 s/div; Y-axis:- Ch1,Ch3: 37.7 elec rad/s/div; Ch2:
0.4 unit/div; Ch4: 40 A/div) . . . . . . . . . . . . . . . . . . . . . . 247
7.20 Stabilization of mid-frequency oscillations of currents in d’-q’ frame
with reference frequency modulation: Ch1 - i′q_ref ; Ch2 - i′d ; Ch3 -
i′q ; Ch4 - ias (X-axis:- 0.22 s/div; Y-axis:- Ch1,Ch3: 75 A/div; Ch2:
30 A/div; Ch4: 40 A/div) . . . . . . . . . . . . . . . . . . . . . . . 247
7.21 Stable operation of I-f control with reference frequency modulation
stabilization: Ch1- ωref Ch2 - Pe ; Ch3 - ω b r ; Ch4 - ias (X-axis:-
1.44 s/div; Y-axis:- Ch1,Ch3: 37.7 elec rad/s/div; Ch2: 8 kW/div; Ch4:
40 A/div) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
7.22 Pole slipping during ramp-up: Ch1 - ωref ; Ch3 - ω b r ; Ch4 - ias (X-axis:-
0.66 s/div; Y-axis:- Ch1,Ch3: 50.24 elec rad/s/div; Ch4: 50 A/div) . 249
7.23 Pole slipping during ramp-up: Ch1 - Tmax ; Ch2 - Tb d ; Ch3 - αi (X-axis:-
0.76 s/div; Y-axis:- Ch1,Ch2: 31.83 Nm/div; Ch3: 40 Hz/s/div) . . . 249
7.24 With torque controller during frequency ramp-up: Ch1 - ωref ; Ch2
- αi ; Ch3 - ω b r ; Ch4 - ias ; (X-axis:- 1.36 s/div; Y-axis:- Ch1,Ch3:
136.65 elec rad/s/div; Ch2: 8 Hz/s/div; Ch4: 50 A/div) . . . . . . . 250

xix
7.25 With torque controller during frequency ramp-up: Ch1 - Tmax ; Ch2
- Tb d ; Ch3 - αi ; Ch4 - ks Tmax (X-axis:- 1.48 s/div; Y-axis:- Ch3:
16 Hz/s/div; Ch1,Ch2,Ch4: 15.92 Nm/div) . . . . . . . . . . . . . . 250
7.26 Oscillations in speed at the end of frequency ramp: Ch1 - ωi ; Ch2
- ∆ωi ; Ch3 - ω b r ; Ch4 - ias ; (X-axis:- 1.36 s/div; Y-axis:- Ch1,Ch3:
136.65 elec rad/s/div; Ch2: 25.12 elec rad/s/div; Ch4: 50 A/div) . . 251
7.27 Smooth transition from ramp-up interval to constant speed operation
with slope attenuator: Ch1 - Tmax ; Ch2 - Tb d ; Ch3 - αres ; Ch4 -
ks Tmax (X-axis:- 1.24 s/div; Y-axis:- Ch3: 16 Hz/s/div; Ch1,Ch2,Ch4:
15.92 Nm/div) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
7.28 Smooth transition from ramp-up interval to constant speed operation
with slope attenuator: Ch2 - αres ; Ch3 - ω b r ; Ch4 - ias ; (X-axis:-
1.2 s/div; Y-axis:- Ch2: 32 Hz/s/div; Ch3: 136.65 elec rad/s/div; Ch4:
50 A/div) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252

8.1 Complete power circuit of 1.5 kW uneq0-IPMSM drive test bench . 253
8.2 Custom designed reconfigurable 4-split IPMSM coupled to an
Induction Motor (IM) . . . . . . . . . . . . . . . . . . . . . . . . . 254
8.3 Stator structure of 4-split IPMSM . . . . . . . . . . . . . . . . . . 255
8.4 Winding diagram for the 4-split IPMSM . . . . . . . . . . . . . . . 256
8.5 Insulation at various parts of 4-split IPMSM . . . . . . . . . . . . . 256
8.6 Terminal box of 4-split IPMSM (top-view) . . . . . . . . . . . . . . 257
8.7 Rotor structure of 4-split IPMSM . . . . . . . . . . . . . . . . . . . 257
8.8 N35AH grade NdFeB magnet . . . . . . . . . . . . . . . . . . . . 258
8.9 Accessories for IPMSM motor . . . . . . . . . . . . . . . . . . . . 258
8.10 Loading setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
8.11 Two 2-level VSI with common DC bus . . . . . . . . . . . . . . . . 260
8.12 Inverter components . . . . . . . . . . . . . . . . . . . . . . . . . . 261
8.13 Back-to-back thyristor switch box . . . . . . . . . . . . . . . . . . 262
8.14 10 kVA autotransformer . . . . . . . . . . . . . . . . . . . . . . . . 263
8.15 DSP-FPGA board for drive control . . . . . . . . . . . . . . . . . . 264
8.16 Protection board . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266

A.1 Location of the mmf space vector during MTPA operation of the three-
phase Interior Permanent Magnet Synchronous Motor (IPMSM) . . 273
A.2 Location of mmf space vector during operation of uneq0-IPMSM to
produce equivalent mmf as three-phase IPMSM . . . . . . . . . . . 275

xx
 GLOSSARY

PM10 Particulate Matter with a diameter of 10 µm or less.

d′ −q ′ reference frame Fictitious synchronously rotating reference frame in I-f
control.
d−q reference frame Rotor reference frame.
Sensorless control A control method for any rotating electric machine
(permanent magnet synchronous motor in this thesis)
without using a rotor position/speed sensor.

xxi
 ABBREVIATIONS

AWG American Wire Gauge.

BEV Battery Electric Vehicle.
BLDC Brushless DC.
CPLD Complex Programmable Logic Device.
CPSR Constant Power Speed Range.
CVC Current Vector Control.
CVL Charge Voltage Limit.
DAC Digital to Analog Converter.
DSO Digital Storage Oscilloscope.
DSP Digital Signal Processor.
DTP-PMSM dual 3-phase PMSM.
DTP0-IPMSM 0◦ displacement dual 3-phase IPMSM.
DTP0-PMSM 0◦ displacement dual 3-phase PMSM.
DTP0-SMPMSM 0◦ displacement dual 3-phase SMPMSM.
DVL Discharge Voltage Limit.
EV Electric Vehicle.
FACTS Flexible Alternating Current Transmission System.
FEM Finite Element Method.
FFT Fast Fourier Transform.
FOC Field Oriented Control.
FPGA Field Programmable Gate Array.
FW Field Weakening.
HEV Hybrid Electric Vehicle.
HPF High-Pass Filter.
HS High-Speed.
HVAC Heating Ventilation and Air Conditioning.

xxii
ICE Internal Combustion Engine.
IGBT Insulated-Gate Bipolar Transistor.
IM Induction Motor.
IPM Interior Permanent-Magnet Motor.
IPMSM Interior Permanent Magnet Synchronous Motor.
LED Light-Emitting Diode.
LPF Low-Pass Filter.
LS Low-Speed.
LUT Lookup Table.
MI Modulation Index.
MIDC Modified Indian Driving Cycle.
MTPA Maximum Torque Per Ampere.
MTPV Maximum Torque Per Voltage.
NdFeB Neodymium Iron Boron.
NEDC New European Driving Cycle.
OEW Open-End Winding.
OEW-PMSM Open-End Winding PMSM.
PI Proportional Integral.
PM permanent magnet.
PMSM Permanent Magnet Synchronous Motor.
PWM Pulse Width Modulation.
SMPMSM Surface Mount Permanent Magnet Synchronous Motor.
SMPS Switched Mode Power Supply.
SPWM Sinusoidal Pulse Width Modulation.
SSI Synchronous Serial Interface.
SSSC Static Synchronous Series Compensator.
SVPWM Space Vector Pulse Width Modulation.
SWG Standard Wire Gauge.

xxiii
SynRM Synchronous Reluctance Motor.
THD Total Harmonic Distortion.
uneq0-IPMSM unequal split 0◦ winding displacement dual 3-phase IPMSM.
uneq0-PMSM unequal split 0◦ winding displacement dual 3-phase PMSM.
UV limit Under Voltage trip limit.
VAC Voltage Angle Control.
VOC Volatile Organic Compound.
VSI Voltage Source Inverter.
WLTC Worldwide Harmonised Light Vehicles Test Cycle.

xxiv
 NOTATION

English Symbols
αR Road angle, rad
A State transition matrix
B Input matrix
u Input vector
x State variable vector
mmfnet Resultant mmf, A
Bbm Peak air-gap flux density, Wb/m2
biq Estimated quadrature axis current, A
Tb d Estimated total torque demand, Nm
Tb e Estimated developed torque, Nm
~eb Back-emf vector, V
~is Stator current vector, A
~is−LS ,~is−HS Stator current vector of LS and HS winding in uneq0-PMSM, A
~is1 ,~is2 Stator current vector of winding-1 and winding-2 in DTP-
PMSM, A
I~s-rated Rated stator current vector, A
~v1 , ~v2 , ~v3 , . . . ~v6 Active voltage vectors with various inverter switching states, V
~v7 , ~v8 Zero voltage vectors with various inverter switching states, V
~vp A vector used to realise average value of the reference voltage
vector in SVPWM, V
~vs Stator voltage space vector, V
~vref Reference voltage vector in SVPWM, V
~vs−LS , ~vs−HS Stator voltage space vector of LS and HS winding in uneq0-
PMSM, V
~vs1 , ~vs2 Stator voltage space vector of winding-1 and winding-2 in DTP-
PMSM, V
# »
mmf3φ Resultant mmf vector of the original three-phase PMSM, A
# »
mmfHS Resultant mmf vector of HS winding in uneq0-PMSM, A
# »
mmfLS Resultant mmf vector of LS winding in uneq0-PMSM, A
A RMS linear current density, A/m
Af Vehicle front area, m2
B Viscous friction coefficient, Nms
Bδ1 Peak of fundamental no-load air-gap flux density, Wb/m2
C0 Machine constant, A Wb/m3
CD Aerodynamic drag coefficient
CG Feedback gain constant, (elec rad)2 /(s2 W)
Csnb Snubber capacitance, F
Ds-in Stator inner diameter, m
Dr Rotor outer diameter, m
e α , eβ α and β components of back-emf vector, V
efd-HS , efq-HS d-axis and q-axis control system feed-forward voltages in HS
winding of uneq0-PMSM, V

xxv
efd-LS , efq-LS d-axis and q-axis control system feed-forward voltages in LS
winding of uneq0-PMSM, V
efd1 , efq1 d-axis and q-axis feed-forward voltage in the control system of
DTP-PMSM winding-1, V
efd2 , efq2 d-axis and q-axis feed-forward voltage in the control system of
DTP-PMSM winding-2, V
efd , efq d-axis and q-axis feed-forward voltage in the control system, V
Fg Resistance due to road grade, N
Fr Rolling resistance, N
fr Rolling resistance coefficient
Ft Tractive effort at the driving wheels, N
Fw Aerodynamic drag, N
ffund Fundamental frequency of a waveform, 1/s
fbis Desired bandwidth of current controller, 1/s
fbw Desired corner frequency of the open-loop transfer function of
the speed controller, 1/s
g Acceleration due to gravity, m/s2
gslit Slit gap in V-shaped permanent magnet configuration, m
Ia Magnitude of stator current vector, A
i′d , i′q Direct and quadrature axis currents in synchronously rotating
d′ −q ′ reference frame, A
ix , iy Non-electromechanical energy-conversion-related components
of stator current in DTP-PMSM, A
iα-net , iβ-net α-axis and β-axis current that produces net air-gap flux in DTP-
PMSM, A
iαh-net , iβh-net α-axis and β-axis harmonic current that produces net air-gap flux
in DTP-PMSM, A
iαs1 , iβs1 α-axis and β-axis stator currents of winding-1 in DTP-PMSM,
A
iαs2 , iβs2 α-axis and β-axis stator currents of winding-2 in DTP-PMSM,
A
iαs , iβs α-axis and β-axis stator currents, A
iα1h-min , iβ1h-min Optimal α-axis and β-axis components of total harmonic current
in winding-1 of DTP-PMSM for minimum copper loss, A
iα1h , iβ1h α-axis and β-axis components of total harmonic current in
winding-1 of DTP-PMSM, A
iα2h-min , iβ2h-min Optimal α-axis and β-axis components of total harmonic current
in winding-2 of DTP-PMSM for minimum copper loss, A
iα2h , iβ2h α-axis and β-axis components of total harmonic current in
winding-2 of DTP-PMSM, A
iα-cir , iβ-cir α-axis and β-axis circulating current in DTP-PMSM, A
Ilim_n Normalised rated stator current vector magnitude for IPMSM
parametric plane, pu
Ilim Magnitude of rated stator current vector, A
Irated Rated stator current RMS, A
Iref Reference current, A

xxvi
ismall , ilarge Smallest and largest current at the instant of pulse-off, A
Ia-LS , Ia-HS Magnitude of stator current vector in LS and HS windings of
uneq0-PMSM, A
ias1 , ibs1 , ics1 Instantaneous a, b and c-phase stator currents of winding-1 in
DTP-PMSM, A
ias2 , ibs2 , ics2 Instantaneous a, b and c-phase stator currents of winding-2 in
DTP-PMSM, A
ias , ibs , ics Instantaneous a, b and c-phase stator currents, A
Ich Characteristic current, A
id(0) Direct axis current at the changeover speed producing zero
output torque during two-winding operation of uneq0-PMSM,
A
id−ref , iq−ref Direct and quadrature axis current references, A
id1−ref , iq1−ref Direct and quadrature axis current references of winding-1 in
DTP-PMSM, A
id2−ref , iq2−ref Direct and quadrature axis current references of winding-2 in
DTP-PMSM, A
id-HS-ref , iq-HS-ref Direct and quadrature axis current references of HS winding in
uneq0-PMSM, A
i′d-HS , i′q-HS Direct and quadrature axis currents of HS winding referred to LS
winding in uneq0-PMSM, A
id-HS , iq-HS Direct and quadrature axis currents of HS winding in uneq0-
PMSM, A
id-LS-ref , iq-LS-ref Direct and quadrature axis current references of LS winding in
uneq0-PMSM, A
id-LS , iq-LS Direct and quadrature axis currents of LS winding in uneq0-
PMSM, A
ids1 , iqs1 Direct and quadrature axis currents of winding-1 in DTP-PMSM,
A
ids2 , iqs2 Direct and quadrature axis currents of winding-2 in DTP-PMSM,
A
ids_MTPA , iqs_MTPA Optimal direct and quadrature axis currents for MTPA operation
of IPMSM, A
ids_MTPV , iqs_MTPV Optimal direct and quadrature axis currents for MTPV operation
of infinite-maximum-speed IPMSM, A
Ids , Iqs Steady-state direct and quadrature axis currents, A
ids , iqs Direct and quadrature axis currents in rotor reference frame, A
Id (n, m), Iq (n, m) Direct and quadrature axis current components corresponding to
maximum developed torque for LUT generation, A
Ieq Equivalent current representing conductors with current out of
the total conductors in the machine, A
Im Peak current, A
Iph-RMS Stator phase current RMS, A
iq(max) Maximum allowable quadrature axis current, A
iq_init Quadrature axis current demanded by the load torque for speed
controller initialisation, A

xxvii
iq_load q-axis current demanded by the load, A
Ivar Magnitude of current space vector variable for LUT generation,
A
J Moment of inertia, kg m2
Jeq Equivalent moment of inertia of motor in EV, kg m2
ke Feedback gain, elec rad/(Ws)
kH Offset added to the modulation index in DTP-PMSM control
ki Integral gain of torque controller, (kg s)−1 m−2
kp Proportional gain of torque controller, kg−1 m−2
Ks Safety factor for frequency ramp-up
ks Safety factor for generated torque in dynamic frequency slope
control
kt Torque constant, Nm/A
Kw Slope of linear speed ramp in I-f control, elec rad/s2
Kpω , τω Parameters of speed controller
Kpi , τpi Parameters of current controller
kw1 Fundamental winding factor
lg Air-gap length, m
Ls Stator inductance per-phase, H
ls′ Equivalent active length of machine, m
Lext Inductance of external series inductor, H
Las1−as1 , Lbs1−bs1 a-phase and b-phase stator self inductances of winding-1 in DTP-
PMSM, H
Las1−as2 , Las1−bs2 Mutual inductances between winding-1 and winding-2 in DTP-
PMSM, H
Las1−bs1 , Lbs1−cs1 Mutual inductance between a and b, b and c phases of winding-1
in DTP-PMSM, H
Las2−as2 , Lbs2−bs2 a-phase and b-phase stator self inductances of winding-2 in DTP-
PMSM, H
Las2−bs2 , Lbs2−cs2 Mutual inductance between a and b, b and c phases of winding-2
in DTP-PMSM, H
Las−as , Lbs−bs , Lcs−cs a, b, and c-phase stator self inductances, H
Las−bs , Lbs−cs , Lcs−as Mutual inductance between a and b, b and c, c and a phases, H
Lds-sat , Lqs-sat Saturated direct and quadrature axis self inductances, H
Lds-unsat , Lqs-unsat Unsaturated direct and quadrature axis self inductances, H
Lds-HS-sat , Lqs-HS-sat Saturated direct and quadrature axis self inductances of HS
winding in uneq0-PMSM, H
Lds-HS-unsat , Lqs-HS-unsat Saturated direct and quadrature axis self inductances of HS
winding in uneq0-PMSM, H
Lds-LS-sat , Lqs-LS-sat Saturated direct and quadrature axis self inductances of LS
winding in uneq0-PMSM, H
Lds-LS-unsat , Lqs-LS-unsat Saturated direct and quadrature axis self inductances of LS
winding in uneq0-PMSM, H
Lds , Lqs Direct and quadrature axis self inductances, H
Lll1 Component of winding-1 self leakage inductance corresponding
to flux linking only that winding in DTP-PMSM, H

xxviii
Lll2 Component of winding-2 self leakage inductance corresponding
to flux linking only that winding in DTP-PMSM, H
Lll-HS Component of HS winding self leakage inductance
corresponding to flux linking only that winding in uneq0-
PMSM, H
Lll-LS Component of LS winding self leakage inductance
corresponding to flux linking only that winding in uneq0-
PMSM, H
Lll Component of self leakage inductance corresponding to flux
linking only that winding in DTP-PMSM, H
Llm-HS Mutual leakage inductance between two winding sets referred to
HS winding in uneq0-PMSM, H
Llm-LS Mutual leakage inductance between two winding sets referred to
LS winding in uneq0-PMSM, H
Llm Mutual leakage inductance between two winding sets in DTP-
PMSM and uneq0-PMSM, H
Lls-HS Self-leakage inductance of HS winding in uneq0-PMSM, H
Lls-LS Self-leakage inductance of LS winding in uneq0-PMSM, H
Lls Self-leakage inductance, H
Lmd-HS , Lmq-HS Direct and quadrature axis magnetising inductances of HS
winding in uneq0-PMSM, H
Lmd-LSHS , Lmq-LSHS Direct and quadrature axis mutual inductances between LS and
HS winding in uneq0-PMSM, H
Lmd-LS , Lmq-LS Direct and quadrature axis magnetising inductances of LS
winding in uneq0-PMSM, H
Lmd , Lmq Direct and quadrature axis magnetising inductances, H
Lms Magnetising inductance, H
Lm Per-phase magnetising inductance, H
m Modulation index
mG Gear ratio
m1_small Slope of smallest current at the pulse-off instant in interval-1, A/s
m2_large Slope of the largest current at the pulse-off instant during
interval-2, A/s
mmax1 Limit of linear modulation region
mmax2 Limit of overmodulation region-1
mmax3 Limit of overmodulation region-2
Mveh Vehicle mass, kg
ma1 Modulation index of a-phase winding-1 in DTP0-PMSM
ma-HS Modulation index of a-phase HS winding in uneq0-PMSM
ma-LS Modulation index of a-phase LS winding in uneq0-PMSM
Nc Number of turns per coil in stator winding
ns Mechanical synchronous speed, rps
nLS , nHS Number of turns per phase of LS winding and HS winding in
uneq0-PMSM
Nph Number of turns per-phase
Nse Equivalent number of turns per-phase

xxix
P Number of poles
Pe Input power, W
Pmax Maximum output power, W
Prated Rated power, W
Pcu-min Minimum copper loss with harmonic current in DTP-PMSM, W
rd Vehicle wheel radius, m
Rs Stator resistance per-phase, Ω
Rsnb Snubber resistance, Ω
Rs1 Stator resistance per-phase of winding-1 in DTP-PMSM, Ω
Rs2 Stator resistance per-phase of winding-2 in DTP-PMSM, Ω
Rs-HS Stator resistance per-phase of HS winding in uneq0-PMSM, Ω
Rs-LS Stator resistance per-phase of LS winding in uneq0-PMSM, Ω
Si Apparent power, W
t Time, s
T0 Time duration of zero vectors within a sampling time, s
T 1 , T2 Time duration of active vectors within a sampling time in
SVPWM, s
Td Total torque demand, Nm
Te Developed electromagnetic torque, Nm
Tf Frictional torque, Nm
Tl Load torque, Nm
Ts Sampling time, s
Tacc Acceleration torque requirement for EVs, Nm
tdecay_1 , tdecay_2 Time duration of interval-1 and interval-2 during pulse-off
interval, s
Tflat Torque requirement for driving a flat road in EVs, Nm
Tgrad Torque requirement for satisfying gradability specification of
EVs, Nm
Tmax1 , Tmax2 Maximum developed torque with the highest and lowest
equivalent turns configuration in winding changeover methods,
Nm
toff Pulse-off duration, s
trib Rib thickness in V-shaped permanent magnet configuration, m
tc1 Instant of changeover from current vector control to voltage
angle control, s
Te−brk Torque breakpoints for LUT generation, Nm
Te−ref Reference torque to the control system, Nm
Te_n Maximum developed torque for a given stator current magnitude
and motor speed for LUT generation, Nm
Tl(max) Maximum expected load torque during the ramp-up, Nm
Tl-net Net load torque, Nm
Tmax Maximum torque capability, Nm
Tnm-max Maximum developed torque for a given speed and current space
vector magnitude in LUT generation, Nm
ui Output of integral part of torque controller, elec rad/s2
up Output of proportional part of torque controller, elec rad/s2

xxx
Va Magnitude of stator voltage vector, V
vαs1 , vβs1 α-axis and β-axis stator voltages of winding-1 in DTP-PMSM,
V
vαs2 , vβs2 α-axis and β-axis stator voltages of winding-2 in DTP-PMSM,
V
vαs , vβs α-axis and β-axis stator voltages, V
Vlim_n Normalised maximum voltage vector magnitude of the inverter
for IPMSM parametric plane, pu
Vlim Maximum voltage vector magnitude of the inverter, V
Vveh Vehicle speed, m/s
Va-LS , Va-HS Magnitude of stator voltage vectors of LS and HS winding in
uneq0-PMSM, V
vab , vbc Instantaneous line-line voltages between a and b, b and c phases,
V
vao , vbo , vco Instantaneous a-phase, b-phase and c-phase inverter pole
voltages, V
vas1 , vbs1 , vcs1 Instantaneous a-phase, b-phase and c-phase stator voltages of
winding-1 in DTP-PMSM, V
vas2 , vbs2 , vcs2 Instantaneous a-phase, b-phase and c-phase stator voltages of
winding-2 in DTP-PMSM, V
vas , vbs , vcs Instantaneous a-phase, b-phase and c-phase stator voltages, V
Vcir Radius of the circular path during overmodulation region-1 of
SVPWM, V
vd1-CL , vq1-CL d-axis and q-axis voltage references for closed-loop current
vector control of DTP-PMSM winding-1, V
vd2-CL , vq2-CL d-axis and q-axis voltage references for closed-loop current
vector control of DTP-PMSM winding-2, V
vd-OL , vq-OL d-axis and q-axis voltage references for open-loop voltage angle
control, V
′ ′
vd-HS , vq-HS Direct and quadrature axis voltages of HS winding referred to LS
winding in uneq0-PMSM, V
vd-HS , vq-HS Direct and quadrature axis voltages of HS winding in uneq0-
PMSM, V
vd-LS , vq-LS Direct and quadrature axis voltages of LS winding in uneq0-
PMSM, V
Vdc DC bus voltage, V
vds1 , vqs1 Direct and quadrature axis voltages of winding-1 in DTP-
PMSM, V
vds2 , vqs2 Direct and quadrature axis voltages of winding-2 in DTP-
PMSM, V
vds , vqs Direct and quadrature axis voltages, V
Vs Steady-state voltage vector magnitude, V
vs Stator voltage vector magnitude, V
Greek Symbols
α Angular coordinate in stator with respect to a-phase axis, rad
α1 Frequency slope at the end of the ramp-up interval in I-f control,

xxxi
 elec rad/s2
αh Hold angle of ~vp during overmodulation region-2 in SVPWM,
rad
αi Slope of reference speed ramp in I-f control, elec rad/s2
αp Angle of ~vp measured from the base of the sector in SVPWM,
rad
αv V angle of V-shaped permanent magnet configuration, rad
αcir Angle of the circular path during overmodulation region-1 in
SVPWM, rad
αi(ss) Steady-state value of the frequency slope, elec rad/s2
αref Angle of reference voltage vector measured from the base of the
sector in SVPWM, rad
αres Resultant slope of the frequency curve with slope attenuator in
I-f control, elec rad/s2
β Current space vector angle measured from quadrature axis, rad
β1 Optimal current vector angle for MTPA operation of IPMSM,
rad
βA Slope attenuator output
βHS mmf space vector angle of HS winding in uneq0-PMSM, rad
βLS mmf space vector angle of LS winding in uneq0-PMSM, rad
βvar Current vector angle variable for LUT generation, rad
δ Load angle, representing the current space vector angle measured
from direct axis, rad
δ0 Steady-state load angle, rad
ǫ Current vector angle with respect to stationary a-phase axis, rad
ηt Efficiency of vehicle driveline
R Reluctance, 1/H
Rmd , Rmq Direct and quadrature axis reluctances, 1/H
ω Speed of current space vector, elec rad/s
ω0 Steady-state motor speed, elec rad/s
ωb Base speed of operation for finite- and infinite-maximum-speed
IPMSM, elec rad/s
ωm Mechanical rotor speed, rad/s
ωr Electrical rotor speed, elec rad/s
ωr∗ Estimated rotor speed using pulse-off method, elec rad/s
ω12-ch Pre-specified speed for changeover from one-winding to two-
winding operation in uneq0-PMSM, elec rad/s
ω21-ch Pre-specified speed for changeover from two-winding to one-
winding operation in uneq0-PMSM, elec rad/s
ωb_n Normalised maximum operating speed of IPMSM for IPMSM
parametric plane, pu
ωc-v Pre-specified speed for changeover from current vector control
to voltage angle control, elec rad/s
ωchg Changeover speed from I-f to sensorless vector control, elec rad/s
ωdip Speed dip during pulse-off, rad/s
ωfinal Intended speed of steady-state operation, elec rad/s

xxxii
ωHS-b Base speed of HS winding operated alone, elec rad/s
ωHS-max Maximum operating speed of HS winding operated alone in
uneq0-PMSM, elec rad/s
ωLPF Corner frequency of low-pass filter in stator voltage integration,
rad/s
ωLSHS-b Base speed of combined LS-HS operation of uneq0-PMSM, elec
rad/s
ωLSHS-max Maximum operating speed of combined LS-HS operation in
uneq0-PMSM, elec rad/s
ωmax Maximum operating speed of finite-maximum-speed IPMSM,
elec rad/s
ωMTPA Maximum motor speed with MTPA operation, elec rad/s
ωv-c Pre-specified speed for changeover from voltage angle control to
current vector control, elec rad/s
ωb1 , ωb2 Base speed of operation with the highest and lowest equivalent
turns configuration in winding changeover methods, elec rad/s
ωi I-f ramp reference speed, elec rad/s
ωref Reference speed of the motor, elec rad/s
ωvar Motor speed variable for LUT generation, elec rad/s
φ Power factor angle, rad
ΨF Permanent magnet flux linkage per phase, Wb
Ψ∗F Estimated permanent magnet flux linkage using pulse-off
method, Wb
Ψm Magnitude of permanent magnet flux linkage vector, Wb
Ψas1 , Ψbs1 , Ψcs1 a, b and c-phase stator flux linkage of winding-1 in DTP-PMSM,
Wb
Ψas2 , Ψbs2 , Ψcs2 a, b and c-phase stator flux linkage of winding-2 in DTP-PMSM,
Wb
Ψas , Ψbs , Ψcs a, b and c-phase stator flux linkage, Wb
ΨF -HS Permanent magnet flux linkage per phase of HS winding in
uneq0-PMSM, Wb
ΨF -LS Permanent magnet flux linkage per phase of LS winding in
uneq0-PMSM, Wb
ΨF _n Normalised permanent magnet flux linkage per phase for
IPMSM parametric plane, pu
Ψm-LS Magnitude of permanent magnet flux linkage vector of LS
winding in uneq0-PMSM, Wb
ρ Air density, kg/m3
τh High-pass filter time constant, s
τs Stator time constant, s
τfo An equivalent first-order system time constant, s
τTe_LPF Time constant of low-pass filter used for reference torque
filtering, s
τbis Desired time constant of current controller, s
τbw−max Time constant corresponding to maximum system bandwidth
during six-step operation, s

xxxiii
τbw Desired time constant of the open-loop transfer function of the
speed controller, s
θ Initial angle of the a-phase current, rad
θc Control system transformation angle, rad
θr∗ Estimated rotor position using pulse-off method, rad
θemf Position of back-emf vector with respect to the stationary a-phase
axis from line voltage measurement, rad
θerr Error in estimated sensorless position, rad
θi0 Synchronous reference frame transformation angle at the instant
before pulse-off, rad
θi Synchronously rotating reference frame angle (I-f control angle),
rad
∗ ∗
θr0 , θr1 Estimated rotor position at the start and the end of pulse-off
interval, rad
θr Rotor angle measured from reference stator axis, rad
ω
br Estimated rotor speed using stator voltage integration based
sensorless technique, elec rad/s
b αr , Ψ
Ψ b βr α and β components of the estimated rotor flux vector, Wb
b
θr Estimated rotor position, rad
e αr , Ψ
Ψ e βr α and β components of the estimated rotor flux vector using an
independent approach, Wb
~r
Ψ Rotor flux vector, Wb
~s
Ψ Stator flux vector, Wb
~ s1
Ψ Stator flux vector of winding-1 in DTP-PMSM, Wb
~ s2
Ψ Stator flux vector of winding-2 in DTP-PMSM, Wb
ξ Saliency ratio of IPMSM
Miscellaneous
(~x)∗ Conjugate of space vector ‘~x’
∆x Perturbation in x
ẋ Derivative of x
|x| Magnitude of x
~xr Space vector ‘~x’ referred to rotor reference frame
~xs Space vector ‘~x’ referred to stator reference frame
x(n−1) Previous sample of variable ‘x’
x(pu) Variable ‘x’ represented in per-unit, pu
x-fund Fundamental component of variable ‘x’
x−3φ Parameters of original 3-phase PMSM from which DTP-PMSM
is derived

xxxiv
 CHAPTER 1

INTRODUCTION

Road transport is the most commonly used mode of transportation due to its flexibility
and comfort. The total number of cars worldwide is to double by 2040, and the
total number of road vehicles is expected to reach 2-3 billion by 2050 (Agarwal
and Mustafi, 2021). The on-road and off-road vehicles are the major contributors
to various pollutants and greenhouse gases such as CO2 , CO, NOx , Volatile Organic
Compounds (VOCs) and particulate matter (Grambsch, 2001). An increase in the
PM10 (Particulate Matter with a diameter of 10 µm or less) exposure would increase
respiratory and cardiovascular diseases, leading to a reduced life expectancy (WHO
Regional Publications, 2000, 1999).
The global mean surface temperature has risen to nearly 1.5 ◦C over the pre-industrial
levels (Allen et al., 2018). According to Allen et al. (2018), a 1 ◦C rise (likely
between 0.8 ◦C and 1.2 ◦C) above the pre-industrial levels in human-induced warming
is observed in 2017, which is continuing to rise at a rate of 0.2 ◦C (likely between 0.1 ◦C
and 0.3 ◦C) per decade. An animated visualisation of the monthly global temperature
anomaly from 1880-2021 is provided by Goddard Institute of Space Studies (GISS), a
NASA laboratory (SubbaRao, 2022). Hence, more stringent emission standards have
been adopted by various nations to curb global warming and to reduce the associated
health risk due to pollutants. The improved emission standards include Euro-VI in
Europe, BS-VI in India, China-6 in China and Tier-3 & California standard in the US
(Agarwal and Mustafi, 2021). This would also imply that the initial cost and operating
expense of Internal Combustion Engine (ICE) based vehicles would increase further in
the upcoming years due to the stricter emission standards and increased fuel prices.
Electric vehicles are slowly replacing conventional ICE vehicles in most countries.
The support from governments through subsidised prices has also boosted the electric
vehicle boom. Even though Electric Vehicles (EVs) are still expensive compared to ICE
vehicles, the prices are expected to drop with a higher volume of production and through
advancement in battery technology, power electronics and electric motor designs.
Various transportation sectors, including passenger cars, heavy-duty trucks, railway
traction, aerospace application and marine propulsion, have already been electrified to
a great extent. Various nations are also considering using electric vehicles for military
combat vehicles due to their better torque capability, lower noise and thermal signatures.
However, heavy batteries and charging requirements are still hurdles for such special
applications.

1.1 PERMANENT MAGNET SYNCHRONOUS MOTORS (PMSM) FOR

ELECTRIC VEHICLES

Typical requirements of a motor used for an electric vehicle include

• High starting torque

• High torque density

• High power density

• High efficiency throughout its operating region

• High overload capability

• Wide operating speed range

• Reasonable cost

Conventionally, variable transmission using gearboxes is adopted in ICE based vehicles.

The ICEs have a limited torque-speed region, where it can perform satisfactorily.
Hence, a gearbox becomes inevitable for the ICE based propulsion systems, including
passenger cars, heavy-duty trucks, railway traction and ship propulsion. However,
electric motors have a wider range of torque capability compared to ICE, consequently
eliminating the gearbox requirement. Permanent Magnet Synchronous Motors
(PMSMs) are widely adopted for EVs and Hybrid Electric Vehicles (HEVs) due to their
high power density, specific power and high efficiency. The Interior Permanent Magnet
Synchronous Motors (IPMSMs) have magnets embedded inside the rotor compared

2
to Surface Mount Permanent Magnet Synchronous Motors (SMPMSMs), where the
magnets are glued to the rotor surface. The IPMSMs have better demagnetisation
withstand capability and mechanical stability compared to SMPMSMs. With
advancements in technology, electric motors and associated power electronics are
becoming smaller to achieve the same output power and torque levels. The Electrical
and Electronics Technical Team Roadmap of U.S. DRIVE (Driving Research and
Innovation for Vehicle efficiency and Energy sustainability) projects the power density
of electric motors used for EVs to reach 50 kW/L and the cost to be 3.3 $/kW in 2025
(US DRIVE, 2017). An 89% volume reduction of the motor is required to achieve the
updated power density target, considering the 2020 target of 5.7 kW/L. Thus, significant
improvement in motor technology is required to achieve this target. It is to be noted that
an increase in motor speed can increase the output power for the same torque rating.
Hence, there is a trend towards high-speed motor technology in commercial EVs, as
shown in Table. 1.1. Motors of commercial EVs with high peak power density and high
specific power are encircled in the Table. All of them have an operating speed of at
least 10,000 rpm. Even though it is desirable to design higher speed motors for the next
generation EVs, the operating speed range of IPMSMs are restricted due to the limited
DC bus voltage, which is derived from a battery. Various researchers have carried out a
vast amount of work for the extension of constant power region and maximum operating
speed of IPMSM. Each of the existing topologies is discussed in detail in the following
section.

Table 1.1: Motor ratings and performance parameters

With DC-DC converter Without DC-DC converter

Camry LS 600h Prius Accord Sonata Leaf e-Verito
Parameter 2007 2008 2010 2006 2011 2012 2016
Peak Power (kW) 70 110 60 12.4 30 80 31
Peak Torque (Nm) 270 300 207 136 205 280 91
Maximum operating speed (rpm) 14,000 10,230 13,500 6000 6000 10,400 8000
HV battery voltage (V) 244.8 288 201.6 144 270 403.2 72
DC-DC converter output voltage (V) 250 − 650 288 − 650 200 − 650 - - - -
CPSR of the IPMSM 5.65 2.92 4.88 6.89 4.29 3.81 2.66
Motor peak power density (kW/L) 5.9 6.6 4.8 1.5 3.0 4.2 -
Motor peak specific power (kW/kg) 1.7 2.5 1.6 0.5 1.1 1.4 -

3
1.2 METHODS FOR EXTENDED CONSTANT POWER REGION AND
OPERATING SPEED RANGE OF IPMSM

1.2.1 Field weakening of IPMSMs

During low-speed operation, rated developed torque in a three-phase IPMSM is

achieved by operating the motor at rated flux and injecting the rated current at an
appropriate angle with respect to the rotor flux vector (Morimoto et al., 1990). For
IPMSMs, the developed torque consists of both Lorentz torque and reluctance torque.
For different stator current magnitudes, the maximum developed torque (combination
of both Lorentz and reluctance torque) occurs at different stator current angles. The
operation of IPMSM with maximum developed torque for a given stator current
magnitude is termed as Maximum Torque Per Ampere (MTPA) operation. During
MTPA operation, maximum torque is generated with minimum copper loss.
As motor speed increases, the induced back-emf of the motor also increases since the
rotor flux is kept constant. However, beyond a specific operating speed (base speed) at
rated load, the motor terminal voltage exceeds the maximum voltage capability of the
inverter. The motor operation above the base speed is possible only by reducing the
rotor flux to keep the terminal voltage within the inverter limits. This reduction in rotor
flux reduces the developed torque of the motor, even when the rated current is supplied.
In DC motors, the reduction in rotor flux is achieved by reducing the field current.
However, in PMSM, the rotor flux is produced by permanent magnets (PMs) and
hence cannot be controlled directly. Instead, the demagnetising component of the stator
current, termed as the direct axis (d-axis) current is applied such that it opposes the PM
flux. This would reduce the air-gap flux, limiting the terminal voltage for operations
beyond the base speed. The method of air-gap flux reduction for achieving high-speed
operation of the motor is termed as flux weakening or field weakening.
The characteristic current (Ich ) of an IPMSM decides its field weakening capability.
The IPMSM operation is extended to a maximum speed limit using field weakening
as shown in Fig. 1.1, where the limiting speed depends on the machine parameters.

4
 T P
max max

Torque

Power
Normal Field weakening Normal Field weakening
operation operation
0 0
MTPA speed max MTPA speed max

(a) Torque-speed (b) Torque-speed

Fig. 1.1: Expansion of torque-speed and power-speed characteristics of IPMSM with

field weakening

For achieving optimal field weakening, the characteristic current should be equal to the
rated current (Pan et al., 2014; Soong and Ertugrul, 2002; Schiferl and Lipo, 1990),

Ich = Ilim (1.1)

where Ilim is the magnitude of the rated stator current vector. The characteristic current
is defined as
~ r|
|Ψ
Ich = (1.2)
Lds
~ r | is the magnitude of rotor flux vector and Lds is the d-axis self inductance.
where |Ψ
The d-axis self inductance is expressed as

Lds = Lls + Lmd (1.3)

where Lmd is the d-axis magnetising inductance, and Lls is the self-leakage inductance.
In IPMSMs with optimal field weakening, the induced emf due to PM flux can be
completely cancelled by the d-axis self inductance drop by injecting the rated stator
current. Hence, such machines have a theoretical infinite constant power operation and
infinite operating speed range. However, in practical machines, the increase in machine
losses with operating speed reduces the output power, restricting the maximum speed

5
to a finite value.
The class of IPMSMs with Ich > Ilim is termed as finite-maximum-speed IPMSMs
and those with Ich < Ilim are the infinite-maximum-speed IPMSMs (Soong and Miller,
1993b,a). The maximum speed of finite-maximum-speed IPMSMs is limited since the
induced emf due to PM is not fully cancelled by the d-axis inductance drop. Hence, as
the machine speed increases, the machine terminal voltage will exceed the maximum
inverter voltage, thereby limiting the operation. In infinite-maximum-speed IPMSMs,
the large d-axis inductance drop can fully cancel the induced emf due to PM, resulting
in a much higher operating speed capability.
Compared to below base speed operation, where the entire stator current is utilised
for torque production, an additional negative d-axis current is required to maintain the
terminal voltage during field weakening. This would result in additional stator copper
loss and consequently poor operating efficiency in high-speed operation. Further, if the
operating point of the PM in the B-H curve shifts below the knee point during field
weakening, it results in irreversible loss of magnetism. Hence, the stator current should
always be maintained within limits to ensure safe magnet operation.
For IPMSMs with a low self inductance, the operating speed range can be extended by
inserting an external inductor in series (Sebastian and Slemon, 1987). The characteristic
current reduces after inserting the external inductor as given by

~ r|
|Ψ
Ich = (1.4)
Lds + Lext

where Lext is the inductance of the external series inductor. This reduction in
characteristic current increases the field weakening capability of the drive, improving
the operating speed range. However, the additional series inductor increases the system
size and weight and hence discouraged for EVs and HEVs.

6
1.2.2 Re-rated PMSM with high current rated inverter switches

The power rating and the operating speed range of a PMSM drive can be improved by
machine re-rating (Slemon, 1995; Chapman and Krein, 2003). Here, only the stator
winding of a three-phase machine is reconfigured by reducing the number of turns per
phase and increasing the number of parallel paths. Thus, the constant torque operating
region of the machine is extended to a higher speed with the same DC bus voltage,
consequently resulting in higher power capability. The same re-rating concept is applied
when a machine originally in star configuration is reconnected to delta and operated
from an inverter with the same DC bus voltage for achieving an extended operating
√
speed range of 3 times the original speed. However, the current ratings of the inverter
and the thickness of connecting wires need to be increased for injecting a higher current
in the re-rated machine to achieve the same output torque as the original machine. The
re-rating procedure was originally proposed as an alternative to field weakening and to
improve the machine efficiency in high-speed operation. However, using higher current
rated switches or more devices in parallel to achieve the higher current rating of the
inverter is a serious drawback of this configuration.

1.2.3 High DC bus voltage with series connection of cells

The operating speed range of a three-phase PMSM can be improved, and the weight of
copper for wire harness can be reduced by increasing the DC bus voltage with series
connection of Li-ion cells, as shown in Fig. 1.2. However, the performance of the
battery stack during series connection is decided by the weakest cell in the stack. The
charging has to be stopped if any cell reaches its Charge Voltage Limit (CVL), and
the discharging is stopped when any one of them reaches Discharge Voltage Limit
(DVL) to prevent permanent damage of that cell. Variation in cell parameters within
a series connected stack may occur due to limitations in the manufacturing process or
ageing. Thus, cell balancing circuits become necessary for optimal battery stack usage
(Einhorn et al., 2011). Active cell balancing circuits can transfer the energy between

7
 N
HV Vdc S
battery
IPMSM

2-level inverter
Fig. 1.2: Power circuit of three-phase PMSM supplied from a 2-level VSI with high
voltage battery

cells compared to passive methods (using series resistors). However, the complexity of
such active methods increases with the increasing number of series cells. Hence, the
maximum battery voltage in present-day commercial EVs is limited and can be in the
range of 72 V to 403.2 V, as shown in Table. 1.1 (Burress, 2012; Sarlioglu et al., 2017).
From the Table, it is observed that the 2012 Nissan Leaf uses the highest battery voltage
of 403.2 V, whereas Mahindra e-Verito uses the least battery voltage of 72 V. The low
value of DC bus voltage results in high motor current and requires high current rated
devices in the inverter for the same output power, whereas a high battery voltage results
in complex cell balancing requirements.

1.2.4 Intermediate DC-DC boost converter stage

Most present-day commercial EVs use DC-DC boost converters in their power train
(Burress et al., 2011), as shown in Fig. 1.3. The DC-DC boost converters act as an
intermediate stage to couple the low voltage battery to a higher voltage DC bus for the
inverter and motor (Bellur and Kazimierczuk, 2007; Emadi et al., 2006). Thus, the
optimisation of the battery stack and the motor drive can be performed independently.
Further, the DC bus voltage in this configuration can be controlled to improve the
inverter efficiency and to reduce machine noise. At very low operating speed, the
DC bus voltage is regulated at the minimum value (same as battery voltage). In the
above base speed operation, the output voltage of the boost converter is fixed at the

8
 N
Vdc S

IPMSM
Battery

DC-DC boost 2-level inverter

converter
Fig. 1.3: Power circuit of three-phase PMSM supplied from a 2-level VSI and DC-DC
boost converter

maximum allowable limit of the DC bus for achieving the maximum output torque
with rated current at a specific speed. The inverter switching losses are reduced by
optimally varying the DC bus voltage depending on motor speed and load demand in
the below base speed operation. Thus, the inverter operating efficiency is improved in
this configuration compared to the configuration in Section 1.2.3 by using the variable
voltage control. Further, the total waveform distortion of the applied voltage is reduced
during operation at low-speed with a variable DC bus voltage, which reduces the system
noise (Estima and Marques Cardoso, 2012). An overall efficiency improvement is also
observed for light loads at the low-speed region with this configuration (Estima and
Marques Cardoso, 2012). However, the bulky inductors associated with the DC-DC
converter increases the system weight and space requirements of the PMSM drive.
Further, the system reliability also decreases due to this additional power conversion
stage.

1.2.5 OEW with cascaded inverter and shared DC bus

In this topology, two 2-level Voltage Source Inverters (VSIs) are connected to a
common DC link to feed an Open-End Winding PMSM (OEW-PMSM) (Welchko
and Nagashima, 2003), as seen in Fig. 1.4. Since the entire DC bus voltage is
√
directly applied across the phase windings, the motor power can reach 3 times that
of the original star connected three-phase PMSM. The operating speed range is also

9
 Battery Vdc

OEW
IPMSM

Inverter-1 Inverter-2
Fig. 1.4: Power circuit of open-end winding IPMSM supplied by cascaded inverter and
shared DC bus

extended due to the higher voltage availability of this topology. A delta connected
three-phase PMSM obtained from the reconfiguration of a star connected three-phase
PMSM can also achieve the same expansion of power rating and operating speed range.
√
However, the inverter current rating would also increase by 3 times in delta connected
configuration, whereas it remains the same as star connected in the cascaded inverter
topology.
The major drawback of this topology is the existence of a path for common mode
currents to circulate through the inverter and DC bus. Hence, six-step operation of the
inverter would result in huge third harmonic circulation, restricting the drive operation.
It is to be noted that an overmodulation operation with a fundamental voltage of 1.1 pu
is possible with 120◦ conduction mode without any common mode current circulation.

1.2.6 OEW with two isolated voltage sources

The motor output power is doubled, and the operating speed range is extended by using
an OEW-PMSM with an isolated DC bus fed from two different energy sources, as
shown in Fig. 1.5 (Welchko, 2005). If one of the energy sources is selected as a fuel
cell, then the other one can be a battery or a supercapacitor. 3-level voltage output
is obtained by choosing equal DC voltage at both inverters, whereas a 4-level output
voltage is obtained by choosing the DC bus voltages in a 2:1 ratio (say 23 Vdc and 31 Vdc ).
Assuming both the DC bus voltages are equal, the effective DC bus voltage (Vdc-eff ) is

10
 Vdc1 Vdc2

OEW
IPMSM

Inverter-1 Inverter-2

Fig. 1.5: Power circuit of open-end winding IPMSM supplied by isolated inverters

given by

Vdc-eff = Vdc1 + Vdc2 (1.5)

= 2Vdc (1.6)

where Vdc1 and Vdc2 are DC bus voltages of Inverter-1 and Inverter-2, respectively.
Thus, the output power of the OEW-PMSM drive is double that of three-phase PMSM
supplied by a single inverter with DC bus voltage Vdc .
The requirement of two different energy sources with two isolated DC buses limits the
application of this configuration since most of the EVs and HEVs have only a single
energy source.

1.2.7 OEW with series compensation using a floating capacitor

In this topology, a floating capacitor is connected to the DC bus of Inverter-2 to feed an

OEW-PMSM, as shown in Fig. 1.6 (Pan et al., 2014). The Inverter-2 operates similar
to the Static Synchronous Series Compensator (SSSC) used in Flexible Alternating
Current Transmission Systems (FACTSs). Similar to an SSSC, the Inverter-2 provides
controllable reactive power required for the IPMSM during field weakening operation,
whereas the Inverter-1 supplies only the active power. Since the DC bus of Inverter-2
is charged through regeneration, it does not require a pre-charging circuit. The output
voltage of the Inverter-2 can be controlled depending on motor current such that it acts

11
Battery Vdc1 Vdc2 Ccom

OEW
IPMSM

Inverter-1 Inverter-2

Fig. 1.6: Power circuit of open-end winding IPMSM drive with floating capacitor
bridge

as a variable impedance seen by Inverter-1. Thus, according to (1.4), the characteristic

current of IPMSM is reduced as seen from Inverter-1 when Inverter-2 is operated as
a series virtual inductance. Consequently, the operating speed range of the IPMSM is
improved in the field weakening region.
The Inverter-2 operation as an inductor throughout the operating speed range would
reduce the base speed and hence the power rating of the IPMSM. It should be noted
that, since Inverter-2 is a controllable voltage source, it can act as a variable inductor or
capacitor depending on the requirement. Hence, Inverter-2 is operated as a capacitor to
supply the leading power factor requirement of IPMSM in the near base speed region
and as an inductor during deep field weakening. Different control strategies for the
same topology is presented in Lee and Ha (2015); Amerise et al. (2020).
The drawback of this configuration is that the output power of IPMSM cannot exceed
1 pu since the Inverter-2 can only supply reactive power, whereas, in previous OEW-
PMSM configurations, the output power has been enhanced to 1.15 pu and 2 pu by
Welchko and Nagashima (2003) and Welchko (2005), respectively. Further, since the
second inverter in this topology supplies only reactive power to the motor, the drive
operation cannot be continued in the event of a fault in any one phase of the inverter
supplying active power or one phase of the motor.

12
1.2.8 Split-phase PMSM

Multi-phase machines are widely used in various applications like ship propulsion, wind
power systems, and high-power industrial requirements due to their high reliability and
power segmentation compared to three-phase counterparts. Dana TM4 has recently
developed 6-phase and 9-phase PMSM drives for heavy-duty applications like buses,
trucks and tractors (Salem and Narimani, 2019).
The most commonly researched multi-phase topology is the dual three-phase PMSM
with 30◦ winding displacement, widely known as split-phase PMSM (Yang et al., 2019;
Karttunen et al., 2012). The power circuit and winding arrangement of a split-phase
PMSM are shown in Fig. 1.7 and Fig. 1.8, respectively. The split-phase machines
are conventionally used to achieve a reduced DC bus voltage compared to three-
phase machines. However, the split-phase PMSM can also achieve the same power

split−phase
Battery Vdc IPMSM

Inverter-1 Inverter-2
Fig. 1.7: Power circuit of split-phase IPMSM

c2
c1

a1
30◦
b2 a2
b1

Fig. 1.8: Winding arrangement of split-phase IPMSM

13
enhancement and operating speed extension similar to an OEW-PMSM with an isolated
DC bus, as will be explained later in Section 3.2. The split-phase PMSM uses a common
DC bus for both inverters compared to the two isolated DC buses in the Open-End
Winding (OEW) configuration. The split-phase PMSM can be derived using winding
reconfiguration from a three-phase PMSM. In the split-phase configuration with the
same DC bus voltage as the original three-phase drive, double the power rating and
operating speed extension can be achieved. However, the overmodulation and six-step
operation of the split-phase Induction Motor (IM) (Gopakumar et al., 1993) and the
split-phase PMSM (Hu et al., 2017) to fully utilise the available DC bus voltage of
the inverter results in huge harmonic circulating currents. The harmonic circulating
current results in additional copper loss and does not contribute to torque production,
deteriorating the machine efficiency.

1.2.9 Star-delta winding switch-over

Star-delta winding switch-over is one of the most popular method in IMs to achieve a
high starting torque and wide operating speed range (Wang et al., 2010; Kume et al.,
1990). A similar operating speed range extension can be achieved with PMSM also.
It should be noted that circulating harmonic current may flow in PMSM during delta-
connection depending on the harmonics in the machine back-emf. Hence, this method
is limited to sinusoidal back-emf PMSMs.The power circuit of a PMSM for star-delta
winding switch-over is shown in Fig. 1.9. The terminal voltage of the machine in delta
connection is reduced to √13 times that of star connection, extending the operating speed
√
range to 3 times the original speed limit. Conventionally, mechanical contactors are
used for the winding changeover. However, the mechanical contactors have a limited
life span and are associated with a dead-time of nearly 100 ms. There will be output
torque interruptions since the stator currents are reduced to zero to achieve the winding
transition. Such a long duration of torque interruption is not acceptable for most
applications. Hence, power electronic switches can be used to replace the mechanical

14
 Delta

Battery Vdc

IPMSM Star

2-level inverter
Fig. 1.9: Power circuit of IPMSM with star-delta winding switch-over

contactors and achieve minimal dead-time and extended life span. Each bidirectional
switch is realised with two Insulated-Gate Bipolar Transistors (IGBTs) and two diodes.
Hence, 12 IGBTs and 12 diodes are required to realise the 6 bidirectional switches in
Fig. 1.9.
Kume and Sawa (1995) have proposed a configuration for star-delta changeover with a
reduced number of power electronic switches, as shown in Fig. 1.10. Here, only two
inverters (12 IGBTs) and two link switches (2 IGBTs) are used to achieve the power
circuit for star-delta changeover. Both the link switches are disabled and all the Inverter-
2 switches are enabled to achieve the star connection. During delta connection, the
switches in a-phase of Inverter-1 is synchronised with y-phase of Inverter-2, with both
link switches always enabled. Further, b-phase of Inverter-1 with z-phase of Inverter-2

a x
b y
Battery Vdc
c z

IPMSM

Inverter-1 Inverter-2
Fig. 1.10: Power circuit of IPMSM with star-delta winding switch-over with reduced
number of switches

15
and c-phase of Inverter-1 with x-phase of Inverter-2 are also synchronised to achieve
the delta connection. It is to be noted that the motor parameters in the control system
need to be changed after the winding reconfiguration from star to delta. Further, the
line currents are phase shifted by 30◦ in delta configuration compared to the stator
currents during star connection. This configuration proves to be more economical than
the conventional star-delta winding switch-over configuration.

1.2.10 Series-parallel winding switch-over

The series-parallel winding switch-over is used to obtain better performance by

circulating current elimination and speed range extension compared to star-delta
winding changeover (Tsai et al., 2001; Huang and Chang, 1999; Zhitkova and Hameyer,
2016). The series connection is used to obtain a higher torque in the low-speed
operation, and the parallel connection is used to maintain terminal voltage within
limits during high-speed operation. The power circuit of a PMSM drive with series-
parallel winding switch-over provision is shown in Fig. 1.11. In this configuration,
the winding turns are split equally. The SSR-S bidirectional switches are turned on
and SSR-P switches are turned off to achieve the series connection of stator windings.
The SSR-P switches are turned on and SSR-S switches are turned off to realise the
parallel connection. The terminal voltage is reduced to half in parallel connection
compared to the series connection, extending the maximum operating speed. The

SSR-P

SSR-S

SSR-P
SSR-P
SSR-S
Battery Vdc SSR-P
SSR-P
SSR-S

SSR-P

IPMSM
2-level inverter
Fig. 1.11: Power circuit of IPMSM with series-parallel winding switch-over

16
control parameters of the machine need to be modified when the winding configuration
is changed from series to parallel and vice versa. However, the stator current phase
angle remains the same during changeover between series and parallel connection,
compared to 30◦ phase shift during star-delta changeover. Further, there is no circulating
current since the machine is always star-connected during series as well as parallel
configurations. 18 IGBTs and 18 diodes are required in this configuration to realise the
9 bidirectional switches in Fig. 1.11, compared to 6 bidirectional switches in star-delta
winding reconfiguration.

1.2.11 Tapped winding changeover

The concept of tapped winding configuration in PMSM for extended speed range
(Kume et al., 1991) is similar to that of the autotransformer. Selecting more number
of turns results in higher output voltage, and a lower number of turns reduces the output
voltage in an autotransformer. Similarly, in the tapped winding motor, high output
torque is obtained in the low-speed region by selecting the tap corresponding to more
turns. As the motor speed increases, the number of winding turns is reduced by selecting
the corresponding tap to limit the motor terminal voltage. Thus, different torque-speed
characteristics are obtained from the same motor extending its operating speed range.
12 IGBTs and 12 diodes are required to realise 6 bidirectional switches for a two-speed
PMSM with tapped winding as shown in Fig. 1.12. Thus, the number of switches in
tapped winding PMSM drive is equal to that of the star-delta switch-over. However,
the complete winding is not utilised in this configuration during high-speed operation.
Hence, the resulting copper loss is higher in the tapped winding compared to series-
parallel and star-delta configuration for achieving the same speed range. The advantage
of this configuration is that, the ratio in which the windings are divided can be chosen
depending on the torque-speed requirement of the application. However, this design
flexibility is not present in star-delta and series-parallel winding changeover methods.
A method to combine the benefits of star-delta and tapped winding is discussed by

17
Battery Vdc

bidirectional
2-level inverter IPMSM
switch box

Fig. 1.12: Power circuit of IPMSM with tapped winding switch-over

Chen et al. (2005); Chen and Cheng (2006). The power circuit of the combined star-
delta and tapped winding reconfigurable PMSM drive is shown in Fig. 1.13. Here, four
different winding arrangements are possible: long-star, short-star, long-delta, and short-
delta, as shown in Fig. 1.14a, Fig. 1.14b, Fig. 1.14c and Fig. 1.14d, respectively. Each
winding configuration results in different torque-speed characteristics, with long-star
having the maximum developed torque in the low-speed region and short-delta having
the maximum operating speed limit. However, 12 bidirectional switches are required to
realise this configuration.
A method to implement the tapped winding configuration with a single inverter and a
minimal number of additional switches is proposed by Swamy et al. (2006), as shown

N2 turns N1 turns
Battery Vdc

N2 turns N1 turns

2-level inverter N2 turns N1 turns

IPMSM

Fig. 1.13: Power circuit of IPMSM with combined star-delta and tapped winding
switch-over configuration

18
 (a) Long-star (b) Short-star (c) Long-delta (d) Short-delta

Fig. 1.14: Different winding configurations with combined star-delta and tapped
winding switch-over

in Fig. 1.15. Here, only two IGBTs and two diode bridge rectifiers are required to
obtain a center-tapped configuration. The common star point can be obtained either at
the winding end taps or at the center taps. During low-speed operation, a higher number
of turns is obtained by keeping the center taps open and by making the common star

2-level inverter
IPMSM

a a1 a2

b b1 b2
Battery Vdc
c c1 c2

D1 DB1
R
SW1
Csnb R
R D2
D3 DB2

SW2
D4

SW box
Fig. 1.15: Power circuit of IPMSM with tapped winding switch-over configuration with
reduced switches

19
point at the winding end taps. Thus, the maximum output torque is obtained with the
given inverter current rating in the low-speed operation. As the motor speed increases,
the stator winding turns are reduced by shifting the star point from winding ends to
nHS
the center taps. The back-emf reduces by a factor of nLS
after the winding changeover
from the higher number of turns (nLS ) to the lower number of turns (nHS ), expanding
the operating speed range of the motor. A snubber capacitor (Csnb ) is used to reduce the
voltage spike across outgoing IGBT.
Compared to the conventional tapped winding motor, the configuration in Fig. 1.15 has
a lower terminal voltage at the open taps of the winding during high-speed operation.
For a case where the center tap divides the number of turns equally, the terminal voltage
of the open end will not exceed the rated motor terminal voltage since the neutral point is
located at the winding center. It is to be noted that, in the conventional tapped winding,
a high terminal voltage appears across the open taps during high-speed operation.
This reduces the reverse voltage blocking capability requirement of the IGBTs used
for winding disconnection in this configuration compared to the conventional tapped
winding motors.

1.2.12 Five-phase PMSM with winding changeover

Similar to star-delta configuration in three-phase PMSM, 5-phase PMSMs are also

reconnected in star, pentagon and pentacle configuration to achieve a wide operating
speed range (Sadeghi et al., 2012). The power circuit configuration of the five-phase
PMSM with bidirectional switches for winding changeover is shown in Fig. 1.16, and
the different winding configurations are shown in Fig. 1.17. 15 bidirectional switches
and a 5-phase inverter are required to realise the winding changeover. For the same
1 1
operating speed, the terminal voltages of the windings are in the ratio 1: √1.382 : √3.618
with star, pentagon and pentacle, respectively. Even though a wider operating speed
range and more reconfiguration options are possible in the five-phase PMSM compared
to three-phase PMSM, the requirement for a large number of bidirectional switches is a

20
 IPMSM

Battery Vdc

pentagon star

5-phase inverter

pentacle

Fig. 1.16: Power circuit of five-phase IPMSM with star, pentagon, pentacle winding
switch-over configuration

(a) Star (b) Pentagon (c) Pentacle

Fig. 1.17: Different winding configurations with five-phase IPMSM

drawback of this configuration.

1.2.13 A common feature of conventional winding changeover methods

The general torque-speed and power-speed characteristics of conventional winding

changeover techniques can be represented as shown in Fig. 1.18. Here, Tmax1 and Tmax2
represent the maximum developed torque with the highest and lowest equivalent turns

21
configuration, respectively. Also, ωb1 and ωb2 represent the corresponding base speeds
of torque characteristics corresponding to Tmax1 and Tmax2 , respectively. For winding
changeover methods with more than two splits, there will be additional extensions to
torque-speed characteristics of Fig. 1.18. For simplicity, the characteristics of winding
changeover methods with only two different winding configurations are considered
here. Fig. 1.18 represents the torque-speed and power-speed characteristics considering
the operation of high equivalent turns winding in below base speed region alone. It is
observed that a significant power dip occurs in the power-speed characteristics in the
vicinity of transition from the high equivalent turns to low equivalent turns operation, as
shown in Fig. 1.18b. The power dip occurs since field weakening is not utilised during
high equivalent turns operation in order to prevent uncontrolled regeneration during
winding transitions. Fig. 1.19 represents the corresponding characteristics where the
high equivalent turns winding operates with both below base speed and field weakening
control. The dip in power could be significantly reduced if the field weakening of high
equivalent turns had been possible, as seen from Fig. 1.19b.
A reverse transition from low equivalent turns winding to high equivalent turns winding
would result in uncontrolled regeneration in the conventional winding changeover
methods due to loss of control during the changeover. At the instant of reverse
transition, the induced emf due to PMs of high equivalent turns winding will be much
higher than the maximum voltage capability of the inverter. During the transition, there
will be a short interval when the high equivalent turns winding is connected to the
inverter, and the negative d-axis current is not sufficiently high to maintain the field
weakening. The uncontrolled regeneration would charge the DC bus capacitor to a
high voltage and may also result in an overcurrent trip. Thus, it is a common practice
to use the high equivalent turns only in the below base speed region, so the terminal
voltage is always within the maximum inverter limit even during the winding transition.
Consequently, a significant power dip exists in the power-speed characteristics of most
of the conventional winding changeover methods.

22
 T High High
max1 P
equivalent max equivalent turns
Torque
turns Low Power

Power
equivalent dip
turns Low
equivalent
T
max2 turns
0 b1 speed b2 max 0 b1
speed b2 max
(a) Torque-speed (b) Power-speed

Fig. 1.18: General characteristics of winding changeover methods when field

weakening is not used during higher equivalent number of turns operation

T High High equivalent Reduced power dip

max1 P
equivalent max turns
Low Region of
Torque

Power
turns equivalent additional
turns power
Low
T equivalent
max2
turns
0 b1
speed b2 max 0 b1 speed b2 max
(a) Torque-speed (b) Power-speed

Fig. 1.19: General characteristics of winding changeover methods if field weakening

would have been used during higher equivalent number of turns operation

In this thesis, two dual 3-phase PMSM (DTP-PMSM) configurations to achieve a wide
operating speed range are proposed. The first method achieves both enhanced output
power as well as an extension of the operating speed range, which is highly suitable for
EV applications. The second method achieves a wide constant power region and high
overloading capability for medium and heavy-duty trucks and military vehicles. It also
achieves field weakening during the winding transition, reducing the power dip present
in the power-speed characteristics of conventional winding changeover methods. A
brief overview of the proposed methods is provided in Section 1.6 and a detailed
explanation is given in the subsequent chapters.

1.3 SENSORLESS VECTOR CONTROLLED PMSM FOR MARINE

PROPULSION

Together with EVs and HEVs, PMSM drives are also used for ship propulsion and
submarine applications. The rotor position information is critical for the control of

23
a PMSM drive. However, position sensors increase cost and reduce reliability of the
drive. Position sensors can be avoided by implementing sensorless vector control for
the PMSM drive.
Sensorless vector control techniques for PMSMs (Vas, 1998) can be classified into
(a) Open-loop estimators using stator model together with open-loop starting (Fatu
et al., 2008)
(b) Open-loop estimators with an improved stator model for low-speed performance
(Ghaderi and Hanamoto, 2011)
(c) Observer-based (extended Kalman filter (Bolognani et al., 1999), Luenberger
observer (Baratieri and Pinheiro, 2014; Poulain et al., 2008), and sliding mode
observer (Kim et al., 2011; Qiao et al., 2013)) speed and position estimators
(d) Estimators based on inductance variation using high-frequency injection
techniques (Consoli et al., 2001; Corley and Lorenz, 1998; Kim et al., 2004)
(e) Estimators using artificial intelligence (neural network and fuzzy-logic based
(Kung and Tsai, 2007))

Sensorless vector control methods (b), (c), (d) and (e) can be used for PMSM drives
(Quang et al., 2014) requiring high dynamic performance in all speed ranges and
continuous low-speed control. The back-emf based sensorless technique, which belongs
to category (a) provides a simple and computationally less intensive solution for
applications involving medium- and high-speed range control. The method (b) is similar
to (a), but includes compensation for stator resistance variation, inverter nonlinearity,
and dead-time effect, thus achieving improved low-speed performance.
For a variety of applications such as compressors, pumps, fans, and Heating Ventilation
and Air Conditioning (HVAC) systems, and also for critical applications such as
electric ship propulsion, and ‘emergency heat and smoke exhaust’, a high precision
dynamic control at low-speed is not necessary. Instead, these applications require
simple and rugged control in medium- and high-speed range. Hence, the sensorless
vector control methods (b), (c), (d) and (e) would be an overkill considering the
computational complexity and cost. Thus, back-emf based sensorless vector control
would be appropriate for such applications requiring only medium- and high-speed
control.

24
1.3.1 Challenges in starting and low-speed control for sensorless vector
controlled PMSM

The back-emf of PMSM is zero during standstill and is very low during the low-
speed operation. The low back-emf during low-speed makes the position estimation
highly sensitive to motor parameters and measurement inaccuracies. Hence, the
conventional back-emf based sensorless methods may fail during starting and very
low-speed operation. Therefore, an additional starting method is commonly used in
conjunction with back-emf based sensorless vector control for medium- and high-speed
control. Here, an open-loop method is used to start the motor from standstill, and once
sufficient speed is attained, a change over to sensorless control can be performed.
The open-loop control methods for PMSM includes V/f control and I-f control. The V/f
control does not require any current or speed feedback for the control. Thus, the current
sensor outputs are used only for protection and not for control. However, since there is
no current feedback in V/f control, there may be overcurrents during sudden loading,
and the motor may lose synchronism. The I-f open-loop control (Fatu et al., 2008) has
an inner closed-loop current control and is devoid of outer speed loop. Thus, the motor
currents are always maintained at the rated value due to the closed-loop current control,
even during transients. Hence, the I-f open-loop control has better stability and ensures
that the stator currents are always within limits.

1.3.2 Starting and changeover process

During I-f control, the current reference is kept constant at the rated value, and the
reference speed is ramped-up. The control system transformation angle is obtained
by integrating the reference speed. Hence, the currents are controlled in a fictitious
synchronously rotating reference frame (d′ −q ′ ), as shown in Fig. 1.20. The d′ −q ′
reference frame is not the same as the rotor reference (d−q) frame. The current vector

25
 q ′-axis
q-axis
~is ωr d-axis
iqs ~r
Ψ
δ d′-axis
N ωi
θi θr

S stator axis

Fig. 1.20: d′ −q ′ reference frame used for I-f control

(~is ) is aligned along the q ′ -axis, since the reference currents are selected as

id−ref = 0 (1.7)

iq−ref = Ilim (1.8)

Under no-load, the error angle (θr −θi ) between the d−q frame and d′ −q ′ reference
frame is nearly 90◦ , so that ids ≈ Ilim and iqs ≈ 0. Similarly, during full load, the error
angle is nearly 0◦ such that ids = 0 and iqs = Ilim .
As the PMSM picks up speed with I-f open-loop control, the accuracy of the sensorless
vector control improves. Once sufficient accuracy is attained, a control transition to the
sensorless vector control is performed using a suitable transition algorithm. A smooth
changeover can be achieved either using the direct transition method or methods based
on increased motor acceleration rate and current profiling.

1.4 METHODS FOR CHANGEOVER FROM OPEN-LOOP STARTING TO

BACK-EMF BASED SENSORLESS CONTROL OF PMSM

1.4.1 Direct transition

The direct transition methods are used to achieve a quick changeover from the I-
f control to the sensorless vector control and do not require any intermediate stage
for control transition. In direct transition (Baratieri and Pinheiro, 2013), a quick
changeover is performed by changing the control system transformation angle from

26
the synchronously rotating reference frame angle to the already converged sensorless
estimated angle. Also, the q-axis current reference and the current controller states
are redefined according to the load torque requirement of the drive. The load
torque requirement is computed using the sensorless position information before
the changeover instant. However, accurate rotor position information is critical for
achieving a smooth transition. Hence, offset in measurement, the effect of inverter
dead-time, and stator resistance variation should be properly compensated. Further, a
positive d-axis current during I-f control causes motor saturation and results in d-axis
inductance variation (Lu et al., 2010). The effect of saturation in the inductances is
difficult to predict (Zhu et al., 2007; Guglielmi et al., 2006), and would necessitate
additional Lookup Tables (LUTs) for storing the motor parameters to achieve better
accuracy. Such high precision dynamic control, which includes compensation for motor
parameter variation would be unnecessary for applications requiring simple and rugged
control in the medium- and high-speed range. Further, speed and current oscillations
during I-f ramp-up also result in position estimation error with the sensorless algorithm.
Consequently, a transition interval is necessary to align the d′ −q ′ reference frame with
the d-q frame, ensuring a smooth variation in stator current and preventing machine
saturation during the changeover.

1.4.2 Increased motor acceleration rate based transition

A smooth transition from I-f control to sensorless vector control can be achieved if
both d′ −q ′ reference frame and d−q frame are aligned at the transition instant. Under
such a condition, negligible change in the reference d-q currents and control system
transformation angle occurs, resulting in negligible stator current transients at the
changeover. For SMPMSM, both reference frames are aligned when the current space
vector (controlled along the q ′ − axis) gets aligned along the q-axis. This would occur
when the injected stator current is just sufficient to meet the load demand.

27
The total developed torque of a machine (Te ) at steady-state is given by

dωm
Te = Tl + J + Bωm + Tf (1.9)
dt

where Tl is the load torque, J is the moment of inertia, B is the viscous friction
coefficient, Tf is the frictional torque, and ωm is the motor speed.
During light load and low acceleration, the entire current space vector would be along
the d-axis, with only a small component along the q-axis, resulting in a large error
between the two reference frames. By increasing the motor acceleration rate, the two
reference frames would start aligning due to the increased J dωdtm component of the total
developed torque. Thus, a smooth transition is achieved if the changeover is performed
at the instant when both reference frames gets aligned due to the increased acceleration
rate.
A first-order compensator that results in increased motor acceleration rate is used by
Fatu et al. (2008); Li et al. (2010), such that the synchronously rotating reference
frame aligns to the sensorless estimated rotor reference frame. Similarly, a speed-
dependent gain is used by Stirban et al. (2010) for aligning both the reference frames.
Since the magnitude of the current vector is kept constant in both methods, the
increased motor acceleration rate aligns both reference frames. Thus, variation in stator
current and effect of machine saturation is reduced during the changeover, achieving a
smooth transition. However, the successful convergence of these methods depends on
compensator parameters.

1.4.3 Stator current profiling based transition

Current profiling based methods are used in literature to slowly reduce the current
vector magnitude so that the synchronously rotating reference frame aligns with the
sensorless estimated angle. Most of the authors use a linearly decreasing current profile
(Wang et al., 2012; Kung et al., 2016; Kung and Risfendra, 2016; Ortombina et al.,
2017). Current profiles which are inversely proportional to speed (Wang et al., 2014),

28
exponentially decreasing with time (Wang et al., 2017), and with a negative slope
proportional to the square of error angle (Zhang et al., 2019) are also used. Further,
reference current reduction using a closed-loop integral control is proposed by Niu et al.
(2016). Even though both increased acceleration rate methods and current profiling
methods provide a smooth changeover, a finite time is required for the transition
interval.
The methods based on increased motor acceleration and current profiling are well suited
for applications such as compressors, pumps, fans, and HVAC systems. However, they
cannot be used for critical applications like electric ship propulsion and ‘emergency
heat and smoke exhausts’ where a quick startup is necessary, due to their finite transition
time.
In this thesis, a quick and smooth changeover from open-loop I-f starting to back-emf
based sensorless vector control is achieved using a proposed ‘pulse-off’ method. A brief
overview of the proposed method is provided in Section 1.6, and a detailed explanation
is given in Chapter 6. The stability of the PMSM drive during open-loop I-f starting also
needs to be considered to make it more suitable for critical applications like electric ship
propulsion and ‘emergency heat and smoke exhausts’.

1.5 STABILITY OF OPEN-LOOP STARTING METHODS OF PMSM

As discussed in Section 1.3.2, the reference frequency has to be ramped up in the I-f
control before the control transition to the sensorless vector control. The selection of
frequency ramp slope plays a critical role in the stability of both I-f and V/f control
of PMSMs. Different profiles are used in the literature for ramping up the reference
frequency. The most common method is to use a linear frequency profile as adopted
by Yang et al. (2015); Zhan et al. (2017); Baratieri and Pinheiro (2013); Wang et al.
(2012, 2017). Besides, frequency profiles such as f ∝ t2 (Fatu et al., 2008), f ∝
(1− cos(ωf t)) (Stirban et al., 2010), ‘S’-shaped profile (Xu and Wang, 1998), and linear
frequency profile with a low pass filter (Wang et al., 2014; Iepure et al., 2012) are also

29
found. According to Krishnan and Ghosh (1989), the frequency reference should be
modified independently for each case depending on the rotor inertia and initial position
for successful starting. Hence, it is usually decided by trial and error. However, an
improper choice of frequency slope leads to pole slipping during open-loop control (Xu
and Wang, 1998).
Pole slipping is usually prevented by restricting the slope of frequency to a small value.
Wang et al. (2012) use a speed ramp-up rate, which is limited by the maximum load
torque permissible at starting. Hence, for drive systems with high expected load torque
during starting and high inertia, the ramp-up rate is selected to be a small value with
this method. Similarly, in the control method used by Baratieri and Pinheiro (2013), the
expression for transition time (from ramp-up region to constant speed region) tends to
infinity for large values of maximum expected load torque, consequently deteriorating
the starting performance. Closed-loop I-f control is proposed by Yang et al. (2015) to
improve the stator current utilisation of the drive. It uses a torque estimator to vary the
current space vector angle to an optimal value, reducing the stator current for achieving
the same output torque. However, the dynamic performance of the drive in the ramp-up
interval is not discussed. Yu et al. (2018) propose a closed-loop I-f control where the
torque angle is maintained at 90◦ to meet the load torque with minimum stator current.
Even though this method reduces the initial reverse motion and improves the system
response to load variations, the frequency ramp slope is selected to be a fixed value.
In all these cases, the machine has a slow ramp-up even during no-load starting, which
increases the starting time requirement for the drive.
In addition to pole slipping during frequency ramp-up, an inherent instability called
“mid-frequency instability” exists during I-f control (alike V/f control) of PMSM in
the absence of damper windings (Kan and Tzou, 2012). The system becomes unstable
for operation beyond a certain frequency. The stability of PMSM can be assessed by
analysing the eigenvalues of the linearized small signal model. The eigenvalues include

30
stator poles1 , which represent fast electrical transients and rotor poles1 , which represent
slow dynamics of the mechanical subsystem. As a result of loose coupling between
stator and rotor poles (Perera et al., 2003), the rotor poles shift to the right half of the
s-plane when operated above a certain frequency, causing system instability.
The system damping is improved by modulating the reference speed in accordance with
the perturbations in input power (Borisavljevic et al., 2010; Haichao et al., 2017). A
first order high-pass filter is used to extract the perturbations in input power. Further,
the perturbation in input power is added to the reference speed after multiplying with
a gain, which is inversely proportional to the machine speed. The eigenvalues of the
resulting system after adding the stabilisation loop are confined to the left half of the
s-plane with variation in speed, consequently achieving stable operation.
Even though mid-frequency stabilisation is well addressed in the literature, the pole
slipping issue during I-f starting is not researched in detail. A comprehensive
consideration of mid-frequency instability as well as pole slipping is required to achieve
a quick and reliable starting while using open-loop I-f control for critical applications.

1.6 MOTIVATION AND OBJECTIVES OF THE THESIS

A wide variety of methods exist to extend the operating speed range of PMSMs for
EV and HEV applications. Field weakening is commonly used in PMSM to fully
utilise the inverter current and voltage limits to extend the operating speed range.
Also, the machine design plays a critical role in deciding the shape of torque-speed
characteristics, field weakening capability, and hence the maximum operating speed
range. Further, multi-phase machines are becoming more popular in EVs and HEVs due
to their better reliability and enhancement in power capability as well as the operating
speed range. It is also observed that different types of winding changeover methods
exist that are used to extend the constant power operating range and the maximum
speed of operation. The motivation for this thesis was to formulate a PMSM drive to
1
here poles represent the denominator of the system transfer function

31
achieve a wide operating speed range suitable for EVs and HEVs and overcome the
limitations of existing configurations to the extent possible.
0◦ displacement dual 3-phase PMSM (DTP0-PMSM) (Barcaro et al., 2010; Liang
et al., 2020; Vaseghi et al., 2011) is commonly used to improve the fault tolerance
in motor drive systems. Here, the focus was mainly on reducing the short circuit
current, magnetic decoupling of the two windings and smooth changeover to one
winding operation during a fault. In this thesis, six-step operation in a DTP0-PMSM
for an electric vehicle application is proposed to achieve double the power rating of the
original three-phase PMSM and extended speed range. The six-step operation ensures
complete utilisation of the available DC bus voltage. Compared to the split-phase
PMSM discussed earlier, the DTP0-PMSM has inherent circulating current elimination
and is thus suitable for six-step operation. The performance of the proposed method is
experimentally validated on a 3 kW DTP0-PMSM setup.
Certain applications like heavy-duty trucks and off-road electric vehicles are
characterised by very high tractive effort in the low-speed region for obstacle
negotiation and hill climbing, and have wide operating speed range requirements. An
unequal split 0◦ winding displacement dual 3-phase PMSM (uneq0-PMSM) is also
proposed in this thesis to achieve both extension of speed range as well as reduced field
weakening requirement for such applications. The uneq0-PMSM achieves improved
overloading capacity and efficiency in the high-speed operating region, compared to
a three-phase PMSM with a wide Constant Power Speed Range (CPSR). In uneq0-
PMSM, the winding turns of the two three-phase winding sets are not equal and wound
with 0◦ winding displacement between them. Hence, the uneq0-PMSM is the general
case of the conventional dual three-phase PMSM (equal turns). When the operating
speed of uneq0-PMSM exceeds a specific speed, the three-phase winding set with the
higher number of turns (Low-Speed (LS) winding) is cut-off using thyristor switches.
As a result, the machine operates with only the three-phase winding set with a lower
number of turns (High-Speed (HS) winding), achieving below base speed control even

32
in the high-speed operating region of the machine. The uneq0-PMSM has a different
shape of torque-speed and power-speed characteristics compared to the conventional
three-phase PMSM, enabling the high efficiency operation in the high-speed region.
The split ratio of winding turns can be chosen appropriately to suite the torque-speed
characteristic for a given requirement. Thus, various torque-speed characteristics can
be realised from a given IPMSM stator and rotor by only changing the winding split
ratio using the proposed uneq0-PMSM configuration. As a result, the designer has an
additional degree of freedom to achieve the torque-speed characteristics for a given
application. The proposed concepts are experimentally validated on a 1.5 kW unequal
split 0◦ winding displacement dual 3-phase IPMSM (uneq0-IPMSM) with 1:3 winding
split ratio.
This thesis also deals with a sensorless starting method of PMSM drives that is highly
suitable for critical applications like marine propulsion and emergency heat and smoke
exhaust. Here, the objective is to develop a quick and reliable starting method that is
simple to implement. The conventional transition methods based on current profiling
and increased acceleration rate require a finite time for the changeover, hence is not
suitable for critical applications. A novel pulse-off method is proposed in this thesis
to achieve a quick and smooth changeover from I-f to sensorless vector control by
accurately estimating the rotor angle. In the proposed method, the inverter pulses are
turned off for a short duration after the motor reaches enough speed for sensorless
vector control to work. In the pulse-off duration, the rotor position is estimated
using the measured back-emf after the current decreases to zero. Hence, the proposed
method accurately determines the rotor position independent of machine parameters
such as stator resistance and inductance compared to the conventional direct transition
methods. Simple hall effect based voltage transducers are used to implement the
proposed method. Even though the additional isolated voltage sensors increase cost
and number of hardware components, it will not be significant compared to the total
system cost in high-power applications, and also of less concern for applications where

33
a quick and reliable starting is the priority. Furthermore, the line voltage sensors can
also be used for accurate PM flux estimation during the self-commissioning of PMSM
drives.
The conventional open-loop I-f control of PMSM has inherent instabilities: mid-
frequency instability and pole slipping. Since an arbitrary choice of the frequency
ramp slope in I-f starting leads to pole slipping, the frequency ramp slope is fixed to
a minimum value, which deteriorates the dynamic performance during starting. Critical
applications like ship propulsion and ‘emergency heat and smoke exhaust’ demand
quick and reliable starting without any instability. In this thesis, an approach to vary
the frequency ramp slope dynamically depending on the moment of inertia and load
torque is proposed. The proposed method overcomes the pole slipping instability and
achieves a quick and stable ramp-up using I-f open-loop control. The reasons for pole
slipping and mid-frequency instability during frequency ramp-up are analysed in detail.
The total torque demand is maintained well within the machine capability using the
proposed torque controller, thus preventing pole slipping. Since the small-signal model
of PMSM is insufficient to capture the dynamics corresponding to pole slipping during
frequency ramp-up, a nonlinear large-signal dynamic model of PMSM with I-f control
is used for analysis.
The performance of both pulse-off starting method and dynamic frequency slope control
for stability is experimentally verified on a 25 kW PMSM drive, which is started using
open-loop I-f control.

1.7 ORGANISATION OF THE THESIS

Chapter 2 of the thesis gives an overview of different types of PMSMs. The dynamic
model of IPMSM is derived using space phasor modelling. The speed and current
controller design for implementing the sensored vector control are explored. The
inverter operation using Space Vector Pulse Width Modulation (SVPWM) in linear,
overmodulation, and six-step regions are explained. A detailed study of maximum

34
torque per ampere control in the below base speed region of operation and field
weakening for the above base speed control of IPMSM is also provided.
Chapter 3 proposes the six-step operation of 0◦ displacement dual 3-phase IPMSM
(DTP0-IPMSM) suitable for EV applications. Expansion of power rating and operating
speed of DTP0-IPMSM compared to a conventional three-phase IPMSM is explored.
The modelling of DTP0-IPMSM using the space phasor concept is discussed, which
is used to explain the circulating current phenomenon. The control architecture of the
DTP0-IPMSM with MTPA operation in below base speed region of operation and field
weakening for high-speed operation is provided. The operation in linear modulation
region is achieved using Current Vector Control (CVC), whereas the overmodulation
and six-step operation are performed using an open-loop Voltage Angle Control (VAC).
An algorithm for smooth back-and-forth transition from CVC to VAC in the six-step
operation is also given. The performance of the proposed method is experimentally
validated on a 3 kW DTP0-IPMSM setup, and the results are provided.
Chapter 4 proposes an unequal split 0◦ winding displacement dual 3-phase PMSM
(uneq0-PMSM) for achieving high starting torque and wide operating speed range.
The effect of variation of winding split ratio on the torque-speed characteristics is
analysed. Further, the control system of uneq0-PMSM fed from two inverters is also
provided. A transition algorithm is also proposed for the smooth winding changeover
from two winding operation in low-medium speed to one winding operation for high-
speed range. The modelling of uneq0-IPMSM is also discussed. The proposed concepts
are experimentally validated on a 1.5 kW uneq0-IPMSM with a 1:3 winding split ratio,
and the results are given.
Chapter 5 discusses machine design considerations for the proposed dual three-phase
IPMSM. The power rating of the DTP0-IPMSM is decided by considering various
EV requirements. The flux density levels and developed electromagnetic torque are
computed using FEMM. This chapter also provides a brief outline on the general
machine design steps, which are common to both three-phase IPMSM and DTP0-

35
IPMSM.
Chapter 6 proposes a pulse-off method for achieving smooth control changeover from
open-loop I-f control for starting to sensorless vector control for continuous operation.
The concepts of open-loop I-f control and stator voltage integration based sensorless
vector control are discussed in detail. The control architecture of a three-phase PMSM
with the proposed pulse-off starting method is elaborated. The proposed control is
experimentally validated on a 25 kW inset-PMSM drive and the results are provided.
Chapter 7 proposes a dynamic frequency slope control to overcome pole slipping
during I-f control of PMSM. The causes of pole slipping are discussed in detail using
the large signal model of PMSM. The causes and existing solutions for mid-frequency
instability in PMSM with open-loop control is also provided. A torque controller is
designed for implementing the dynamic frequency slope control such that it does not
interfere with the existing control to overcome mid-frequency instability. A slope
attenuator is also proposed to reduce the speed and current oscillations during the
transition from ramp-up interval to constant speed operation with I-f control. The
proposed methods are experimentally validated on a 25 kW inset-PMSM drive.
Chapter 8 describes the hardware organisation of the drive setup. An IPMSM with four
winding sets is used, which can be connected as either three-phase IPMSM, DTP0-
IPMSM or uneq0-IPMSM by changing the motor terminal connections. Each power
circuit component constituting the drive is explained in detail. The details of the control
board, encoder interface board, sensors and protection board are also provided.
Chapter 9 concludes the thesis and also discuss the future work.

36
 CHAPTER 2

PERMANENT MAGNET SYNCHRONOUS MOTORS:

BASICS, MODELLING, AND VECTOR CONTROL

A review of the three-phase PMSM motor drive is provided in this chapter. A space
phasor based method is used for modelling and control of both the three-phase and dual
three-phase IPMSM throughout this thesis. Hence, the concept of space phasor and
modelling of the three-phase IPMSM using space phasor is explained in detail in this
chapter. A brief recapitulation of types of PMSM, rotating magnetic field, controller
design and optimal operation of PMSM is also provided for a better understanding of
multi-phase PMSM operation discussed in later chapters.
Three-phase AC motors with PMs are classified as shown in Fig. 2.1. The Brushless DC
(BLDC) motor and PMSM differ in terms of the shape of induced emf and the control
technique used. The BLDC motors have a trapezoidal back-emf, whereas the PMSMs

Permanaent magnet
AC motors

PMSM BLDC

Axial flux Radial flux

SMPMSM IPMSM

Inset I−shaped V−shaped double delta−shaped

V−shaped

Fig. 2.1: Classification of permanent magnet AC motors

have a sinusoidal back-emf. The control of PMSM is performed with sinusoidal current
control, whereas, in BLDC motors, a quasi-square wave current control is used with
only two phases conducting at a given time. BLDC motors are preferred in low power
applications like electric scooters, washing machines, compressors, and pumps, which
requires high power density but not very stringent torque ripple specification. PMSMs
are commonly used in both low and high power applications like machine tooling,
industrial drives, passenger cars, electric traction, ship propulsion, and aerospace
applications requiring high torque and power density, high efficiency, and smooth
operation.

2.1 TYPES OF PMSM MOTORS

PMSMs are further classified based on the orientation of field flux as axial flux PMSMs
and radial flux PMSMs. In the axial flux PMSM, the magnetic flux is along the
axial direction (parallel to the axis of rotation), whereas, in the radial flux PMSM,
the magnetic flux is radially outwards relative to the axis of rotation. The axial flux
motors are preferred in applications like in-wheel hub motor, which requires short
length and wide diameter. The axial flux motors are gaining popularity due to their
high torque and power density. However, the conventional radial flux motors are
still dominant in the EV market due to their simple construction and easy to analyse
2D electromagnetic characteristics. The radial flux motors can be further classified
into Surface Mount Permanent Magnet Synchronous Motor (SMPMSM) and Interior
Permanent Magnet Synchronous Motor (IPMSM). In SMPMSMs, the magnets are
glued to the rotor surface, as shown in Fig. 2.2 and secured with Kevlar tapes or
glass banding. Hence, SMPMSMs are commonly used for low-speed applications
due to their weak mechanical stability. IPMSMs have magnets embedded inside the
rotor, achieving better mechanical stability than the SMPMSM. Various IPMSM rotor
structures are shown in Fig. 2.3. In inset PMSMs, the magnets are placed in grooves
on the rotor periphery, as shown in Fig. 2.3a. In other IPMSM configurations, the

38
 S q-axis

N N
d-axis

S S

N
Fig. 2.2: Surface-mount PMSM rotor

magnets are placed in a spoke shape, I-shape, V-shape, double V-shape and delta-
shaped configuration as shown in Fig. 2.3b, Fig. 2.3c, Fig. 2.3d, Fig. 2.3e, and
Fig. 2.3f, respectively. The magnets are embedded well inside the rotor in all these
configurations, achieving high mechanical stability and a smooth rotor surface. Hence,
IPMSMs are most suitable for high power, high-speed EV and HEV applications. Most
of the commercial electric vehicles in the present-day market uses IPMSM. It is to be

S q-axis
S
q-axis q-axis
N N d-axis NN S S

S N d-axis
S N N
N d-axis
N S
S S N S
S S N N
N S
(a) Inset PMSM (b) Spoke type PMSM (c) I-shaped PMSM
S S S
q-axis q-axis q-axis

N N N
N d-axis N d-axis N d-axis

S S S
(d) V-shaped PMSM (e) double V-shaped PMSM (f) delta-shaped PMSM

Fig. 2.3: Rotor structures of permanent magnet AC motors

39
noted that all the commercial EVs in Table. 1.1 uses IPMSM except Mahindra e-Verito,
which uses a three-phase Induction Motor (IM).
The permeability of PMs is comparable to that of air. Hence, in SMPMSM, as the
magnets are embedded outside the rotor, a uniform reluctance is observed from the
stator independent of the rotor position. The per-phase magnetising inductance of a
machine (Lm ) is inversely proportional to the reluctance ( R) as given by

N2
Lm = (2.1)
R

where Nph is the number of turns per-phase of the stator winding. Thus, SMPMSM has
a constant stator inductance due to the uniform reluctance seen by the stator flux.
In IPMSMs, there exist a low reluctance path from N-pole to S-pole of the rotor and
vice-versa that prevents the rotor flux from linking the stator coils, deteriorating the
machine torque capability. Hence, IPMSMs require flux barriers to prevent the flux
circulation within the rotor and channel the flux to the stator. As a result, there is a
variation in the reluctance of the IPMSM depending on the rotor position and stator
excitation angle. The rotor magnet axis of PMSM is termed as the direct axis (d-axis),
and the interpolar axis is termed as the quadrature axis (q-axis). When the stator is
excited along the d-axis, maximum reluctance is observed since both slots for magnets
and flux barriers create resistance to the flux path. A minimum reluctance is observed
when the stator is excited along the q-axis since the stator flux can pass parallel to the
flux barriers and the PM slot.
Let Rmd and Rmq represent the reluctance along rotor d-axis and q-axis, respectively.
The magnetising inductances of an IPMSM along the d-axis (Lmd ) and q-axis (Lmq )
are inversely proportional to Rmd and Rmq , respectively. Since Rmd is greater than
Rmq , Lmd is less than Lmq . As a consequence of the inductance variation due to rotor
saliency, an additional reluctance torque can be extracted from IPMSMs compared to
SMPMSMs. The reluctance torque of IPMSMs is in addition to the Lorentz torque
extracted from permanent magnets.

40
The space phasor concept and generation of rotating magnetic field in a three-phase
machine is discussed in the next section. The principle of rotating magnetic field
explained here applies to all AC machines having a sinusoidal winding distribution, like
IMs, synchronous machines, PMSMs, and Synchronous Reluctance Motor (SynRM).

2.2 THE SPACE PHASOR CONCEPT AND ROTATING MAGNETIC FIELD

Consider a two-pole three-phase IPMSM with 3 sets of sinusoidally distributed

windings that are placed 120◦ apart from each other, as shown in Fig. 2.4. Assume
that the coils have Nse equivalent number of turns and are excited with currents ias , ibs
and ics in a-phase, b-phase and c-phase, respectively. The equivalent number of turns is
given by
Nse = Ns kw1 (2.2)

where Ns is the number of turns per phase and kw1 is the fundamental winding factor.
Let the a-phase axis be selected as the reference axis, and ‘α’ represent the angular
coordinate in stator with respect to the a-phase axis. Thus, the b-phase axis is located
at α = 120◦ , and the c-phase axis is at α = 240◦ . Since the windings are sinusoidally

Current into the plane

bs axis Current out of the plane
as’
rotor axis

cs N bs

θr α
as axis

bs’ S cs’

as
cs axis
Fig. 2.4: Stator winding distribution and rotor of a three-phase IPMSM

41
  2π 4π
i
+ejα ias + ibs e−j 3 + ics e−j 3 (2.5)

The current space phasor (Leonhard, 2001; Ranganathan, 2004) is defined as

~is = ias + ibs ej 2π3 + ics ej 4π3 (2.6)

The corresponding conjugate current space phasor is given by

~i∗s = ias + ibs e−j 2π3 + ics e−j 4π3 (2.7)

Hence, the resultant mmf is expressed in terms of the current space phasor as

Nse  −jα~ 
mmfnet = e is + ejα~i∗s (2.8)
2

The resultant mmf also has a sinusoidal spacial distribution. However, the instantaneous
magnitude and angular velocity of the resultant mmf depend on the instantaneous stator
currents.
The space phasor is different from the constant complex phasors used for representing
the steady-state ac quantities. In general, the machine variables like mmf, flux, or
voltage can be expressed as a space phasor, as given by

2π 4π
f~ = fa + fb ej 3 + fc ej 3 (2.9)

Only the mmf and flux space phasor have a physical meaning, whereas the current
and voltage space phasors are fictitious quantities used for mathematical modelling
of machines. The space phasor definition does not necessitate that the quantities be
sinusoidal. It can be any signal, and the space phasor gives the magnitude and spatial
orientation of the complex vector when it is used to excite the windings of a machine.
The current space phasor can be resolved into two components—one along the a-phase

43
axis and the other along a perpendicular axis.

~is = iαs + jiβs (2.10)

Here, iαs and iβs represent the stator currents of an equivalent two-phase machine that
can produce the same current space phasor as the three-phase machine. The equivalent
two-phase machine is shown in Fig. 2.7. Here, the α-axis is along the a-phase axis and
the β-axis is perpendicular to it.

β-axis

~is

ωt + θ
3−φ stator
α-axis axis
Fig. 2.7: Stator current space phasor in an equivalent two-phase IPMSM

The currents of the two-phase machine are expressed in terms of three-phase machine
currents using (2.6) and (2.10) as

ibs ics
iαs = ias − − (2.11)
√ 2 2
3
iβs = (ibs − ics ) (2.12)
2

Thus, it can be concluded that the same resultant mmf as the original three-phase
machine can be obtained in a two-phase machine, by injecting iαs in one winding which
is along the reference axis and iβs in another winding perpendicular to it.
In general, the variables in the three-phase domain can be converted to two-phase to
produce the same complex vector using the following transformation derived from

44
(2.11) and (2.12) as  
    f
1 1  a
fα  1 − 2 − 2  



 = √ √  f
 b (2.13)
3 3  
fβ 0 2 −2
fc

2.2.1 Rotating magnetic field in AC machines

Let a-phase, b-phase and c-phase coils of the three-phase machine be excited with
sinusoidal currents, which are phase displaced by 120◦ from each other as given by

ias = Im cos(ωt + θ) (2.14)

2π
ibs = Im cos(ωt + θ − ) (2.15)
3
2π
ics = Im cos(ωt + θ + ) (2.16)
3

where ω is the angular frequency, Im is the peak current and θ is the initial angle of the
a-phase current.
The current space phasor is obtained by substituting (2.14), (2.15) and (2.16) in (2.6) as

n  h i
~is = Im ej(ωt+θ) + e−j(ωt+θ) + ej(ωt+θ− 2π3 ) + e−(jωt+θ− 2π3 ) ej 2π3
2h i o
2π 2π 4π
+ ej(ωt+θ+ 3 ) + e−j(ωt+θ+ 3 ) ej 3 (2.17)
3
= Im ej(ωt+θ) (2.18)
2

The location of the current space phasor in the equivalent two-phase machine is shown
in Fig. 2.7. The current phasor has a magnitude ‘ 23 Im ’, and is displaced from the a-phase
axis by an angle ‘ωt + θ’. The corresponding conjugate current space phasor is given
by substituting (2.14), (2.15) and (2.16) in (2.7) as

~i∗s = 3 Im e−j(ωt+θ) (2.19)

2

45
 β-axis

PM axis
q−axis d−axis

θr

N
3 − φ stator axis
α-axis

S
Fig. 2.9: Rotor reference frame

axis) is perpendicular to it. Let the rotor axis be at an angle θr from the stator axis.
The transformation of space phasor from the stationary reference frame to the rotor
reference frame is obtained as
f~r = f~s e−jθr (2.22)

where θr is the angle of rotor measured in anticlockwise direction from the stator axis.
Throughout the thesis, the superscript ‘s’ denotes that the corresponding space phasor
is referred to the stator reference frame and ‘r’ denotes the rotor reference frame.
The space phasor in the rotor reference frame can be resolved into two components,
where one is along the rotor axis (fd ) and the other perpendicular to it (fq ) as

f~r = fd + jfq (2.23)

The machine variables are expressed in the rotor reference frame (d-q frame) for
implementing the vector control. By using the rotor reference frame transformation, an
AC machine (say PMSM or IM) can be controlled similar to a DC motor with decoupled
control of torque and flux. The model of the PMSM is also greatly simplified using this
transformation, which is discussed in the next section.

2.3 DYNAMIC MODEL OF THREE-PHASE IPMSM USING SPACE PHASOR

The IPMSM have a salient rotor, as discussed in Section 2.1. Hence, the self inductance
of each phase has a variation with the rotor position. To simplify the analysis, this

47
variation is approximated as a double angle variation about a mean value. Hence, the
a-phase self inductance for Fig. 2.4 is expressed as (Krause et al., 2013),

Las−as = Lls + LA − LB cos(2θr ) (2.24)

Lmax +Lmin Lmax −Lmin

where Lls is the leakage inductance, LA = 2
, LB = 2
. Here, Lmax
represents the maximum inductance and Lmin represents the minimum inductance with
variation in rotor position. With this approximation of sinusoidal inductance variation,
the corresponding self and mutual inductances of other phases can be expressed as

LA 2π
Las−bs = Lbs−as = − − LB cos(2θr − ) (2.25)
2 3
LA 4π
Las−cs = Lcs−as = − − LB cos(2θr − ) (2.26)
2 3
4π
Lbs−bs = Lls + LA − LB cos(2θr − ) (2.27)
3
LA
Lbs−cs = Lcs−bs = − − LB cos(2θr ) (2.28)
2
2π
Lcs−cs = Lls + LA − LB cos(2θr − ) (2.29)
3

The flux linkage equation of the three-phase IPMSM is given by

Ψas = Las−as ias + Las−bs ibs + Las−cs ics + ΨF cos θr (2.30)

 
2π
Ψbs = Lbs−as ias + Lbs−bs ibs + Lbs−cs ics + ΨF cos θr − (2.31)
3
 
4π
Ψcs = Lcs−as ias + Lcs−bs ibs + Lcs−cs ics + ΨF cos θr − (2.32)
3

The phase voltage equations of a three-phase IPMSM is given by

dΨas
vas = Rs ias + (2.33)
dt
dΨbs
vbs = Rs ibs + (2.34)
dt
dΨcs
vcs = Rs ics + (2.35)
dt

48
where Rs is the per-phase stator resistance.
The voltage space phasor in the stationary reference frame is obtained in terms of
~ s ) using (2.33), (2.34) and (2.35)
current space phasor (~iss ) and flux space phasor (Ψ s

as

v~ss = vas + avbs + a2 vcs (2.36)

~s
dΨ
= Rs~iss + s
(2.37)
dt

2π
~ ss ) is
where a = ej 3 . Here, the flux space phasor in the stationary reference frame (Ψ
given by
~ s = Ψas + aΨbs + a2 Ψcs
Ψ (2.38)
s

The flux space phasor in the stationary reference frame is solved by substituting (2.30),
(2.31), (2.32), and (2.6) in (2.38) as

 
~ ss 3 3 3
Ψ = Lls + LA ~iss − LB ej2θr (~iss )∗ + ΨF ejθr (2.39)
2 2 2

The voltage space phasor equation of IPMSM in the stationary reference frame is solved
by substituting (2.39) in (2.37) as

  s
~ 3 d~is 3 d(~is )∗ 3
~s s
vs = Rs is + Lls + LA − LB ej2θr s − j3LB ωr ej2θr (~iss )∗ + j ωr ΨF ejθr
2 dt 2 dt 2
(2.40)
where ωr is the electrical speed of the rotor.

49
The voltage space phasor is resolved into α − β components and is expressed as

   
3 diαs 3 diαs diβs
vαs = Rs iαs + Lls + LA − LB cos 2θr + sin 2θr
2 dt 2 dt dt
3
− 3LB ωr (−iαs sin 2θr + iβs cos 2θr ) − ωr ΨF sin θr (2.41)
  2 
3 diβs 3 diαs diβs
vβs = Rs iβs + Lls + LA − LB sin 2θr − cos 2θr
2 dt 2 dt dt
3
− 3LB ωr (iαs cos 2θr + iβs sin 2θr ) + ωr ΨF cos θr (2.42)
2

It is to be noted that the voltage equations are dependent on the rotor position, making it
complex for analysis and control implementation. The voltage equations are simplified
by changing the frame of reference to the rotor reference frame using coordinate
transformations. The general coordinate transformation from the rotor reference frame
to the stator reference frame is given by,

f~s = f~r ejθr (2.43)

The voltage equation in the rotor reference frame is derived by substituting all the stator
reference frame variables in terms of the rotor reference frame variables in (2.42) as

   3
 
d ~r jθr  3 d
v~sr ejθr ~r
=Rs is e jθ r
+ Lls + LA is e − LB ej2θr (~irs ejθr )∗
2 dt 2 dt
(2.44)
3
− j3LB ωr ej2θr (~irs ejθr )∗ + j ωr ΨF ejθr
2

Simplifying (2.44) gives the voltage equation in the rotor reference frame as

   
~ 3 d ~r 3 d ~r ∗ 3
~r r
vs =Rs is + Lls + LA i − LB (i ) + jωr Lls + LA ~irs
2 dt s 2 dt s 2
3 3
− jωr LB (~irs )∗ + j ωr ΨF (2.45)
2 2

The voltage space phasor and current space phasor are resolved into d−q components

50
as

v~sr = vds + jvqs (2.46)

i~rs = ids + jiqs (2.47)

Finally, the voltage equations of IPMSM are expressed in d-q components in the rotor
reference frame as

dids
vds = Rs ids + (Lls + Lmd ) − ωr (Lls + Lmq )iqs (2.48)
dt
diqs 3
vqs = Rs iqs + (Lls + Lmq ) + ωr (Lls + Lmd )ids + ωr ΨF (2.49)
dt 2

where Lmd = 23 (LA + LB ) = 32 Lmax , Lmq = 23 (LA − LB ) = 32 Lmin . Lmd and Lmq
represent the magnetising inductances of IPMSM in the d-q frame.
The self inductances of IPMSM in the d-q frame are given by

Lds = Lls + Lmd (2.50)

Lqs = Lls + Lmq (2.51)

The developed electromagnetic torque of IPMSM can be derived from the instantaneous
power balance equation.

2
Pe = Re{~vs~i∗s } (2.52)
3
2
= Re{(vds + jvqs )(ids − jiqs )} (2.53)
3
2
= (vds ids + vqs iqs ) (2.54)
3
(2.55)

By substituting (2.48) and (2.49) in (2.55), and neglecting the resistive drop and

51
transient terms, the active power output of the machine is obtained as

 
2 3
Pe = −ωr (Lls + Lmq )iqs ids + ωr (Lls + Lmd )ids iqs + ωr ΨF iqs (2.56)
3 2
 
2 3
= ωr (Lmd − Lmq )iqs ids + ωr ΨF iqs (2.57)
3 2

The developed electromagnetic torque is given by

Pe
Te = (2.58)
ωm

2
where ωm is the mechanical rotor speed (ωm = ω ).
P r

Thus, the developed electromagnetic torque of IPMSM in the rotor reference frame
(d-q) variables is derived as,

2P
Te = ((Lmd − Lmq )iqs ids + Ψm iqs ) (2.59)
32

where Ψm = 23 ΨF .
The dynamic equations for an SMPMSM are obtained by substituting Lm = Lmd =
Lmq in the IPMSM voltage and torque equations.

2.3.1 Equivalent circuit of IPMSM in the rotor reference frame

The equivalent circuit of IPMSM in the rotor reference frame is derived from the
dynamic model equations (2.48) and (2.49), as shown in Fig. 2.10. A mutual coupling
voltage exists between the d-axis and q-axis equivalent circuits. Hence, the ids current in
the d-axis equivalent circuit affects the control of q-axis current in the q-axis equivalent
circuit and vice versa. These cross-coupling terms should be appropriately included as
feed-forward to the control system to achieve good dynamic performance during the
vector control of IPMSM.

52
 Lls + Lmd ωr (Lls + Lmq )iqs
ids Rs

vds

(a) d-axis equivalent circuit

Lls + Lmq ωr (Lls + Lmd )ids

iqs Rs

vqs 3
ω Ψ
2 r F

(b) q -axis equivalent circuit

Fig. 2.10: Equivalent circuit of three-phase IPMSM in rotor reference frame

2.3.2 Phasor diagram of three-phase IPMSM

The phasor diagram of IPMSM in the rotor reference frame is derived considering the
steady-state model. The steady-state voltage equations are obtained by neglecting the
transient terms in (2.48) and (2.49) as

vds = Rs ids − ωr (Lls + Lmq )iqs (2.60)

3
vqs = Rs iqs + ωr (Lls + Lmd )ids + ωr ΨF (2.61)
2

The phasor diagram of IPMSM in the d-q frame is constructed using (2.60) and (2.61),
as shown in Fig. 2.11.
In IPMSM, Lmd is less than Lmq . Hence, to produce the maximum output torque,
a positive q-axis current and a negative d-axis current must be injected according to
(2.59). Hence, the entire operation of IPMSM with MTPA during below base speed

53
 q−axis

Rsids ωr Lqsiqs

jRsiqs
jωr Ldsids
j 32 ωr ΨF
vsr
~

~ir
s
jiqs

δ
~r
Ψ
d−axis
ids θr

Stator reference
axis
Fig. 2.11: Phasor diagram of three-phase IPMSM in rotor reference frame

region of operation and field weakening for above the base speed of operation is
achieved with the current space vector controlled in the second quadrant of ids −iqs
plane. It is clearly seen from the phasor diagram that jωr Lds ids term directly opposes
the machine back-emf (j 23 ωr ΨF ). Hence, during above base speed operation, the
magnitude of the terminal voltage (|~vsr |) is controlled to be within inverter capability
by injecting a negative ids , achieving field weakening. Fig. 2.11 shows below base
speed operation, where the IPMSM operates with a lagging power factor. As the
IPMSM enters field weakening, the current space phasor (~irs ) gradually aligns with
the voltage space phasor (~vsr ), achieving unity power factor operation. As the motor
speed increases, the ~irs moves further away from ~vsr due to the negative d-axis current

54
injection, causing a leading power factor operation. The IPMSM can extract maximum
output power at unity power factor operation in the field weakening region.

2.4 CONTROLLER DESIGN FOR IPMSM

Hierarchical control is used to implement speed control of the IPMSM. The control
system consists of an inner current loop and an outer speed loop. The inner loop is
designed to be much faster than the outer loop. Hence, each control loop is designed
independently without affecting the dynamics of the other loop. The control loop design
of the IPMSM is similar to that of the DC motor since the rotor reference model of the
IPMSM resembles the DC machine model. The inner current controller is designed
initially, followed by the outer speed control.

2.4.1 Current controller

The voltage equations of the IPMSM in the d-q frame can be generalised as

dis
vs = Rs is + Ls + eb (2.62)
dt

For the d-axis model of the IPMSM, vs =vds , is =ids , Ls =Lds , eb = − ωr Lqs iqs ,whereas,
for the q-axis model of the IPMSM, vs =vqs , is =iqs , Ls =Lqs , eb =ωr Lds ids + 32 ωr ΨF
Taking Laplace transform of (2.62) gives

Vs (s) − Eb (s) = Is (s)(sLs + Rs ) (2.63)

Eq. 2.63 can be rearranged as

Vs (s) − Eb (s)
Is (s) = (2.64)
sLs + Rs

The generalised model of the IPMSM obtained from (2.64) for which the controller
needs to be designed is shown in Fig. 2.12. Here, the back-emf acts as a disturbance
to the control system of the plant when the plant is modelled as a first-order system.

55
 Eb(s)
1
Vs (s) Is(s)
sLs+Rs

Fig. 2.12: Generalised plant model of IPMSM for current controller design

Further, there is cross-coupling between the d-axis and q-axis equivalent circuit due
to the components of back-emf as discussed in Section 2.3.1. Hence, the back-emf
components are fed forward to the controller to cancel its effect during control system
design, as shown in Fig. 2.13.

2
Motor model
Vdc Eb (s)
(Feed forward) Eb(s)

Vdc 1
Iref (s)
PI 2 sLs+Rs Is (s)

Fig. 2.13: Block diagram of closed-loop current control of IPMSM

The resulting model of the closed-loop current controlled system after the cancellation
of back-emf is shown in Fig. 2.14. A 2-level VSI is used for the control of the PMSM.
Since the switching frequency of the inverter is large compared to the bandwidth of
the current controller, the switching delay and the delay due to digital implementation
can be neglected for the inverter model. Hence, the inverter model used for the
Vdc
current controller design can be represented as a simple gain 2
(where Vdc is the
DC bus voltage). This assumption will not significantly affect the performance of the

Kpi(1+sτpi ) Vdc Vs (s)

Iref (s) 1 Is (s)
sτpi 2 sLs+Rs
PI controller Inverter gain Motor model

Fig. 2.14: Reduced current controller block diagram

56
Proportional Integral (PI) controller when the switching frequency is greater than nearly
10 times the current controller bandwidth.
A pole-zero cancellation method is used to achieve a first-order response for the current
controller. The open-loop transfer function of the system is given by

Kpi (1 + sτpi ) Vdc 1

G(s)H(s) = (2.65)
sτpi 2 Rs (sτs + 1)

Ls
where τs is the stator time constant (τs = R s
).
Assume that the bandwidth of the resulting first-order system be fbis , then the open-loop
transfer function is obtained as
1
G1 (s) = (2.66)
sτbis
1
where τbis is the time constant (τbis = 2πfbis
).
The current controller gain τpi in (2.65) is chosen as τs to achieve pole-zero cancellation,
achieving a first-order response for the system.

τpi = τs (2.67)

The current controller parameters are obtained by equating (2.65) and (2.66).

Kpi Vdc 1 1
=
sτpi 2 Rs sτbis

The current controller gain Kpi is obtained as

L
Kpi = Vdc
(2.68)
2
× τbis

It should be noted that, for the form of PI controller used in Fig. 2.14, the proportional
Kpi
gain is Kpi and the integral gain is τpi
.
Most traction applications do not require speed control; instead, a simple torque control
would be sufficient. For such applications, the person driving the vehicle acts as the

57
outer speed loop. He/She controls the reference torque input depending on the intended
speed of the vehicle that he/she desires to maintain. In such cases, the inner current
control loop itself is sufficient to achieve the torque control. Here, the reference current
is obtained from the motor model to achieve the required torque. The reference currents
for IPMSMs are usually obtained from a Lookup Table (LUT) due to the nonlinear
variation of machine parameters due to core saturation. Detailed analysis of LUT
implementation for current control of the IPMSM is given in Section 2.8. Speed control
is required for applications like fan and pump, where the operator would directly set the
reference speed of operation. The speed controller design is discussed in the following
section.

2.4.2 Speed controller

The speed controller acts as an outer loop for the speed controlled system. The speed
controller is designed to have a lower bandwidth (nearly 1/10 times) compared to
the current control so that the dynamics of the current controller will not affect its
performance. The mechanical model of the motor is derived as follows.
The developed torque (Te ) equation of IPMSM is given by

dωm
Te = Tl + J + Bωm + Tf (2.69)
dt
dωm
Te − (Tl + Bωm + Tf ) = J (2.70)
dt

where Tl is the load torque, J is the moment of inertia, B is the viscous friction
coefficient, and Tf is the frictional torque.
The load torque, viscous friction and frictional torque are jointly represented as net load
torque (Tl-net ) since it is difficult to obtain these parameters, which also simplifies the
machine model.
Te (s) − Tl-net = Jsωm (s) (2.71)

Te (s) − Tl-net
ωm (s) = (2.72)
Js

58
The resulting mechanical model of IPMSM is shown in Fig. 2.15.

Tl-net

1
Te (s) ωm (s)
Js

Fig. 2.15: Simplified mechanical model of IPMSM for speed controller design

The complete model of IPMSM with the speed controller and the current controller
is shown in Fig. 2.16. The load torque acts as a disturbance input to the speed control
loop. During controller design, the disturbance input is fixed as zero. The reduced block
diagram after neglecting load torque is shown in Fig. 2.17. The speed PI controller is
designed using the symmetric optimum pole-placement method to achieve maximum
phase margin for ensuring stability. The bode plot of the open-loop transfer function of
the speed control system (Fig. 2.17) is shown in Fig. 2.18. Here, the two poles at the
origin of the open-loop transfer function result in an initial slope of −40 dB/dec.
As the frequency increases, the zero due to the controller causes the slope to change
1
from −40 dB/dec to −20 dB/dec at τw
. Further, another pole due to the current controller
1 1
causes the slope to again change back to −40 dB/dec at τbis
. Let fbw = τbw
, be the
desired corner frequency of the open-loop transfer function of the speed controller.

Tl-net

1 1
ωref (s) PI kt ωm (s)
1+sτbis Te (s)
Js
Speed Current Motor model
controller controller

Fig. 2.16: Block diagram of speed controller with IPMSM model

Kpw (1+sτw ) 1 1
ωref (s) kt ωm (s)
sτw 1+sτbis Te(s) Js
Speed controller Current controller Motor model

Fig. 2.17: Reduced block diagram of speed controller

59
 −40 dB/decade

|GH|(dB)
−20 dB/decade

1
τbis
1 1 ω(rad/s)
τw τbw

−40 dB/decade

6 GH(rad)
Maximum
phase margin

ω(rad/s)

Fig. 2.18: Bode plot of open-loop transfer function of speed control

According to the symmetric optimum technique, the maximum phase margin for this
1
system can be achieved by ensuring that the crossover frequency τbw
occurs at the
1 1
geometric mean of τw
and τbis
. It should also be ensured that the crossover frequency
occurs when the slope of the bode plot is −20 dB/dec. From the phase plot of the speed
control system shown in Fig. 2.18, it is observed that the peak of phase occurs at the
crossover frequency, achieving maximum phase margin.
1 1
The crossover frequency is obtained as the geometric mean of τw
and τbis
.

r
1 1 1
= (2.73)
τbw τw τbis

60
The controller gain (τw ) is obtained as

2
τbw
τw = (2.74)
τbis

At the gain crossover frequency, the magnitude of the transfer function is unity. Hence,

Kpw (1 + sτw ) 1 1
kt =1 (2.75)
sτw 1 + sτbis sJ

where s = 2πfbw .
Let τw ≫ τbw and τbw ≫ τbis . Thus, 1 + sτw ≈ sτw and 1 + sτbis ≈ 1. Hence, (2.75)
reduces to

Kpw τw s kt
= 1 (2.76)
τw s sJ
Kpw kt
= 1 (2.77)
2πfbw J

The controller gain (Kpw ) is given by

2πfbw J
Kpw = (2.78)
kt
J
Kpw = (2.79)
kt τbw

2.5 SPACE VECTOR PWM

Space Vector Pulse Width Modulation (SVPWM) can achieve a 15% higher output
voltage than Sinusoidal Pulse Width Modulation (SPWM) for the same DC bus voltage.
Further, SVPWM also enables easier implementation of overmodulation operation till
six-step for improved DC bus voltage utilisation. Hence, SVPWM is used as the
modulation technique for the inverter throughout this thesis.
A 2-level VSI has a total of 8 valid switching combinations. This consist of 6 active
voltage vectors (~v1 , ~v2 , . . . ~v6 ) and 2 zero voltage vectors (~v7 ,~v8 ). The zero vectors

61
correspond to the inverter state when either all top switches are on or all bottom
switches are on (with the other side switches in the complementary state). The other
6 combinations would generate a space phasor (called as space vector from here on)
as discussed in Section 2.2, which forms the vertices of a regular hexagon in α − β
reference frame as shown in Fig. 2.19. The following transformation is used to obtain
the space vector from the pole voltages.

vbo vco
vαs = vao − − (2.80)
√ 2 2
3
vβs = (vbo − vco ) (2.81)
2

~v3 ~v2 Inscribed

circle

~vref
~v0
~v4 ~v1
~v7
(|~v1|=Vdc)

~v5 ~v6
Fig. 2.19: Space vector diagram of 2-level VSI

It is to be noted that the magnitude of each active vector is Vdc while using the above
definition. A rotating voltage space vector is obtained during the control by appropriate
switching of the inverter. A circular trajectory of the space vector is desirable for motor
control since it generates sinusoidal output voltages. Reference space vector at each
instant is realised by switching between the active vectors and zero vectors. During each
sampling instant, the reference voltage is realised in an average sense by distributing the
switching duration of active and zero vectors by ensuring volt-sec balance.
Let the reference voltage vector (~vref ) be located in sector-1 at a particular instant of

62
time and Ts be the sampling time. The switching harmonics are minimum when active
vectors corresponding to the vertices of the sector in which the reference vector is lying
is used for its realisation. ~v1 and ~v2 are the active vectors used together with zero vectors
(~v7 , ~v8 ) when ~vref lies in sector-1. Assume that the total sampling time is divided into
T1 , T2 and T0 , where ~v1 is applied for T1 time duration, ~v2 is applied for T2 duration and
zero vector for T0 duration.
Applying volt-sec balance and resolving ~vref along ~v1 and perpendicular component,

|~v1 |T1 + |~v2 |T2 cos 60◦ = |~vref |Ts cos αref (2.82)

|~v2 |T2 sin 60◦ = |~vref |Ts sin αref (2.83)

Here, |~v1 | = |~v2 | = Vdc , and αref is the angle of ~vref measured from the base of the
sector in which ~vref is located at that instant.
|~vref |
Let the Modulation Index (MI) be defined as m = Vdc
. Solving (2.82) and (2.83),

sin(60◦ − αref )
T1 = mTs (2.84)
sin 60◦
sin(αref )
T2 = mTs (2.85)
sin 60◦
T0 = Ts − T1 − T2 (2.86)

The time duration of the active and zero vectors when the reference vector is in a
different sector is also obtained using (2.84), (2.85) and (2.86). To realise a reference
vector in a particular sector, the active vectors that lie on the vertices of that sector is
used.
The active vectors can be used to realise the reference voltage vector only when it is
within the hexagonal boundary. A sinusoidal output voltage without any lower order
harmonics is produced when the system operates in the linear modulation range. The
linear modulation range is defined when ~vref lies within the inscribed circle of the space
vector hexagon as marked in Fig. 2.19. Hence, the limit of the linear region occurs when

63
~vref traces the inscribed circle.
For battery-operated electric vehicles, the available DC bus voltage is limited. Hence,
to maximise the utilisation of DC bus voltage, the IPMSM needs to be operated outside
the linear modulation region, referred as the overmodulation region of operation. The
maximum fundamental output voltage is produced when the inverter operates in six-
step mode, where there are only six switching per fundamental cycle. This also reduces
the switching losses of the inverter compared to the linear modulation region. However,
the lower order harmonics in voltage as well as motor current increases as the operation
moves nearer to the six-step mode. The implementation of the overmodulation and
six-step operation is discussed in the next section.

2.6 OVERMODULATION AND SIX-STEP OPERATION FOR INCREASED

DC BUS UTILISATION

To achieve a smooth transition from the linear region to six-step operation, the
reference vectors tracing circular trajectory outside the inscribed circle must also
be realised. However, a proper voltage compensation algorithm is necessary to
maintain linearity between the reference vector to be realised and the actual applied
vector in the overmodulation region. In this thesis, the overmodulation algorithm
given by Venugopal and Narayanan (2006) is implemented to compensate for voltage
nonlinearity during the transition from the linear modulation region to the six-
step operation. In this scheme, the entire overmodulation region is divided into
overmodulation region-1 (ovm-1) and overmodulation region-2 (ovm-2). The six-step
mode represents the limit of ovm-2 operation.

2.6.1 Overmodulation region-1

In the overmodulation region-1, the space vector to be realised (~vref ) lies just outside
the voltage limit hexagon, as shown in Fig. 2.20. Hence, the average value of the
reference vector is realised by using another vector (~vp ) by ensuring volt-sec balance.
~vp follows a circular path while moving near the edges of the hexagon and then moves

64
 ~v2

~vref

~vp
αref
αp vcir
~v7 αcir
~v0 |~v1|=Vdc ~v1

Fig. 2.20: Reference vector and applied vector during ovm-1

along the hexagon in each sector shown by the blue line in Fig. 2.20. Here, Vcir denotes
the radius of the circular path. The volt-sec lost while moving along the hexagonal
boundary is compensated by the additional gain while moving in the circular region
with Vcir > |~vref |. αcir in Fig. 2.20 marks the angle of the circular trajectory. Even
though the magnitude of ~vref and ~vp are different, they are aligned to each other. Hence,
the angle of ~vp is given by
αp = αref (2.87)

The magnitude of ~vp is given by



 0.866Vdc
 , if αcir < αref < π
− αcir
sin( π3 +αp ) 3
|~vp | = (2.88)


Vcir , otherwise

The relation between the magnitude of reference vector (|~vref |), Vcir and αcir is given
by,

6  π π αcir 
|~vref | = Vcir αcir + sin( + αcir ) ln | cot( + )| (2.89)
π
√ 3 6 2
3 π
Vcir = Vdc cosec( + αcir ) (2.90)
2 3

A LUT is used to generate Vcir and αcir for a given |~vref |. The LUT is generated by

65
solving (2.89) and (2.90) simultaneously. The limiting value of ovm-1 occurs when ~vp
moves only along the hexagon without tracing any circular region (αcir =0).

2.6.2 Overmodulation region-2

Overmodulation region-1 cannot maintain linearity between the reference voltage and
the applied voltage vector when the reference space vector magnitude is more than
0.9085Vdc . Here, volt-sec of the reference voltage vector cannot be compensated using
the additional voltage margin with the circular trajectory in ovm-1. In ovm-2, the
reference vector is realised by holding ~vp at the corners of the hexagon for a calculated
time and then traces the path along the edges of the hexagon for the remaining time,
as shown in Fig. 2.21. The hold angle (αh ) is calculated to achieve volt-sec balance
between the reference voltage and the applied voltage vector.

~v2

~vref

αref
~vp
~v7 αh
~v0 |~v1|=Vdc ~v1

Fig. 2.21: Reference vector and applied vector during ovm-2

The applied voltage vector magnitude (|~vp |) and the applied voltage vector angle (αp )
are given by




 Vdc , if 0 ≤ αref < αh



|~vp | = 0.866Vdc
, if αh ≤ αref < ( π3 − αh ) (2.91)

 sin( π3 +αp )




Vdc , if ( π3 − αh ) ≤ αref < π
3

66
 



 0, if 0 ≤ αref < αh



αp = αref , if αh ≤ αref < ( π3 − αh ) (2.92)






π, if ( π3 − αh ) ≤ αref < π
3 3

The relation between the reference vector (~vref ) and the hold angle (αh ) is given by,

√ !
6 3 π αh
|~vref | = Vdc sin(αh ) + ln | cot( + )| (2.93)
π 2 6 2

Another LUT is used to generate αh for a given |~vref | in ovm-2 by solving (2.93). As the
reference voltage magnitude increases in ovm-2, the hold angle increases, decreasing
the length of space vector trajectory along the edges of the hexagon. Finally, the limiting
value of ovm-2 occurs when the hold angle (αh ) becomes 30◦ . Then, each active vector
is applied for 60◦ duration resulting in six-step operation. The maximum utilisation of
the inverter occurs during the six-step mode.

2.7 MAXIMUM TORQUE PER AMPERE AND FIELD WEAKENING IN

IPMSM

The operating region of an IPMSM is divided into below base speed and above base
speed regions. The motor can produce rated torque with optimal current injection only
in the below base speed region of operation. However, in the above base speed region,
the optimal current cannot be maintained due to voltage limit constraints of the inverter.
The MTPA operation in the below base speed region and field weakening in the above
base speed region is discussed (Morimoto et al., 1990) as follows.
The combined operating limits of the motor and the inverter are expressed as,

Ia ≤ Ilim (2.94)

Va ≤ Vlim (2.95)

67
where Ia is the magnitude of stator current space vector (~is ) and Va is the magnitude of
the stator voltage space vector (~vs ). The rated current vector magnitude (Ilim ) is decided
by the continuous current rating of the stator and the inverter switches. The maximum
voltage vector magnitude (Vlim ) for the linear modulation region of operation is given
by
Vlim = 0.866Vdc (2.96)

The operation with higher voltage limits can be obtained with overmodulation and six-
step operation as discussed in Section 2.6.
An ids −iqs plane is used to analyse the operating region of the IPMSM. The locus
of stator current limit can be obtained as (2.97), which corresponds to a circle in the
ids −iqs plane.
2
i2ds + i2qs = Ilim (2.97)

The stator voltage equation by neglecting the stator resistance drop is expressed as
(2.98), which corresponds to an ellipse in the ids −iqs plane.

 2
Vlim
= (Lqs iqs )2 + (Lds ids + Ψm )2 (2.98)
ωr

The stator current circle and the voltage limit ellipse in the ids −iqs plane for the
finite-maximum-speed IPMSM and the infinite-maximum-speed IPMSM are shown in
Fig. 2.22a and Fig. 2.22b, respectively. The basic concepts of the finite- and infinite-
maximum-speed IPMSM are already explained in Section 1.2.1. The feasible operating
region of the IPMSM is the common intersecting area inside the current limit circle and
the voltage limit ellipse. It is observed that the voltage limit ellipse shrinks as the motor
speed increases in both the finite- and infinite-maximum-speed IPMSM, as shown in
Fig. 2.22. For the finite-maximum-speed IPMSM, the voltage limit ellipse shrinks to
a point outside the current limit circle as the motor speed increases. Hence, there is
no common intersection of the current limit circle and the voltage limit ellipse beyond
a certain speed of operation (ωmax in Fig. 2.22a). However, for the infinite-maximum-

68
 Voltage limit Current limit
ellipses ωr =ωx iqs circle

ωr =ωy

ωr =ωz
ids

ωx<ωy <ωz
(a) Finite-maximum-speed IPMSM
Voltage limit Current limit
ellipses iqs circle

ωr =ωz
ids
ωr =ωy

ωr =ωx
ωx<ωy <ωz
(b) Infinite-maximum-speed IPMSM

Fig. 2.22: Current limit circle and voltage limit ellipse in ids − iqs plane for IPMSM

speed IPMSM, the voltage limit ellipse shrinks to a point inside the current limit circle,
extending the theoretical maximum operating speed to infinity.
The developed torque of the IPMSM is given by (2.99), neglecting the effect of stator
leakage.
2P
Te = [(Lds − Lqs )iqs ids + Ψm iqs ] (2.99)
32

The d−q components of stator current for achieving the maximum torque is obtained by
differentiating (2.99) with respect to one of the current components (either ids or iqs ) for
a constant current magnitude by rewriting (2.99) in one variable form. Let ‘β’ denote
the angle of current space vector measured from q-axis. The condition for MTPA is

69
obtained as (Morimoto et al., 1990; Jahns et al., 1986),

ids_MTPA = −Ia sin β1 (2.100)

iqs_MTPA = Ia cos β1 (2.101)

Here, the optimal current vector angle (β1 ) is given by

( p )
−Ψm + (Ψm )2 + 8(ξ − 1)2 L2ds Ia2
β1 = sin−1 (2.102)
4(ξ − 1)Lds Ia

Lqs
where ξ is the saliency ratio (ξ = Lds
).
The locus of all points that satisfies MTPA for the finite-maximum-speed IPMSM
and the infinite-maximum-speed IPMSM is shown in Fig. 2.23a and Fig. 2.23b,

MTPA line
Current limit
ωr =ωb iqs circle

MTPV line A1

A4 A3

ωr =ωmax ids

(a) Finite-maximum-speed IPMSM

MTPA line
MTPV line Current limit
iqs circle

A1
ω b ω2
r= = A2
ω ω r
A4
ids

(b) Infinite-maximum-speed IPMSM

Fig. 2.23: MTPA and MTPV lines in ids − iqs plane for IPMSM

70
respectively. The maximum torque is generated with the rated stator current at A1
in Fig. 2.23. The motor is operated along the MTPA line during the below base speed
operation to achieve minimum copper loss for the specified torque. Similar to MTPA,
Maximum Torque Per Voltage (MTPV) is the locus of maximum output torque for
a given inverter voltage, as shown in Fig. 2.23. For MTPV, the current limit is not
considered. Point ‘A2’ (corresponding to speed ‘ω2 ’) in Fig. 2.23b represents the
minimum operating speed at which an infinite-maximum-speed IPMSM can operate
at MTPV. Below this speed, the current limit of the machine would be violated with
MTPV. The condition for MTPV are,

−Ψm
ids_MTPV = − ∆id (2.103)
Lds
r 
2
Vlim
ωr
− (Lds ∆id )2
iqs_MTPV = (2.104)
ξLds




0, ξ=1
∆id = r
2 (2.105)

 −ξΨm + (ξΨm )2 +8(ξ−1)2 (
Vlim
)
 ωr
, ξ 6= 1
4(ξ−1)Lds

As the motor speed increases from zero, the voltage limit ellipse shrinks such that it
crosses the point A1 when speed reaches ‘ωb ’ in Fig. 2.23a and Fig. 2.23b. This speed
of operation beyond which the rated torque cannot be extracted is termed as the base
speed of the motor. The expression for the base speed (ωb ) is obtained by substituting
the MTPA condition (2.100) and (2.101) in the IPMSM voltage equation (2.98),

Vlim
ωb = p (2.106)
(Lqs iqs_MTPA ) + (Lds ids_MTPA + Ψm )2
2

In infinite-maximum-speed IPMSMs, for operating speed ωb <ω<ω2 , the maximum

motor output torque is limited by both voltage and current constraints. Here, the
maximum output torque at a given speed during ωb <ω<ω2 is achieved by the
intersection points of the voltage limit ellipse and the current limit circle. Thus, the

71
operating point moves from A1 to A2 along the circle in Fig. 2.23b, when the maximum
torque needs to be extracted. For all operations above ωb , if the torque requirement is
less than the maximum torque, then the optimal operating point is obtained from the
intersection of the voltage limit ellipse and the constant torque curve (corresponding to
the load requirement). When the operating speed exceeds ω2 , the maximum output
torque is obtained by following the MTPV curve. A theoretical infinite speed can
be achieved by the infinite-maximum-speed IPMSM by following the MTPV curve
as the speed increases beyond ω2 . However, the actual speed of both finite- and infinite-
maximum-speed IPMSMs would be limited due to the increased core and mechanical
losses with the operating speed and due to the minimum load present in the system.
For finite-maximum-speed IPMSMs, for operating speed ωb <ω<ωmax , the operating
point moves from A1 to A3 along the circle in Fig. 2.23a, when maximum torque needs
to be extracted. The operation along MTPV is not feasible in the finite-maximum-
speed IPMSM since the MTPV curve exists outside the current limit circle, as shown
in Fig. 2.23a. Hence, the operating speed is limited to ωmax at point A3, where the
voltage limit ellipse touches the current limit circle only at one point. Above this speed,
there is no common intersection of the voltage limit ellipse and the current limit circle,
restricting the IPMSM operation. The maximum operating speed of a finite-maximum-
speed IPMSM is given by
Vlim
ωmax = (2.107)
(Ψm − Lds Ilim )

2.8 OPTIMAL CURRENT COMPUTATION OF IPMSM CONSIDERING

PARAMETER VARIATIONS USING LOOKUP TABLE

As discussed in Section 2.4.1, electric traction applications require a torque control to be

implemented, which is achieved using current control. Here, a reference current is given
to the control system depending on the torque demand. For IPMSMs, the optimal values
of ids and iqs varies depending on the operating point in the torque-speed characteristics.
The optimal reference currents can be solved using the equations derived in Section 2.7

72
for both the finite-maximum-speed IPMSM and the infinite-maximum-speed IPMSM.
However, the effect of stator resistance and inductance variation due to saturation is
not considered in this equation based solutions. Hence, off-line solutions of optimal
currents of the IPMSM model using recursive (Cheng and Tesch, 2010) or iterative
solutions are preferred for better drive performance. The solution results are stored in
an LUT and used directly as the current reference to the control system depending on
the torque requirement and operating speed of the motor.
In this thesis, an iterative process is proposed for the LUT generation of the IPMSM,
as shown in Fig. 2.24. The machine parameters, DC bus voltage (Vdc ), magnitude of
rated current vector (Ilim ) are defined initially. LUTs with d-axis and q-axis inductance
variation with d-axis current (ids ) and q-axis current (iqs ) are also provided as input
for the LUT generation algorithm. Here, the space vector voltage limit (Vlim ) for LUT
generation is limited to the linear modulation region.

Vlim = 0.95 × 0.866Vdc (2.108)

The basic concept of LUT generation is that for every values of current space vector
magnitude (Ivar ) and motor speed (ωvar ), an optimal current space vector angle exists
that results in the maximum torque generation at that speed. Hence, the current space
vector angle with respect to the q-axis (βvar ) is varied in steps for each Ivar and ωvar .
The developed torque at each possible βvar is computed and stored in an array, provided
the machine voltages are within the inverter limits. Finally, the maximum developed
torque for a given speed and current space vector magnitude (Tnm-max ) is computed
from the array, and the corresponding βvar is obtained. The βvar corresponding to the
maximum torque output is used to calculate the d-q current components Id (n,m) and
Iq (n,m), which is then stored in a 2D LUT. Here, speed (ωvar ) and current (Ivar ) are
varied in steps in two nested loops to compute the complete torque-speed characteristics
of the IPMSM. The speed loop increments the speed from zero, whereas the current loop
decrements the current from the rated value.

73
 Set the machine
parameters, Vdc and Ilim

Initialize ωvar =0, n=0

Flag = 0

Initialize Ivar =Ilim

Initialize β=0, k=0

Calculate Id (k), Iq (k)

Calculate |~vs |

No If
|~vs | < Vlim
Yes
Calculate Te (k)

Flag = 1, k++
Increment βvar
If
βvar < 90◦ Yes
No
Tnm−max = max(T e(k))
Save Id and Iq Decrement Ivar
corresponding to max(Te(k))
Decrement Te−brk
No If m−−
Ivar =Ilim
Yes Calculate Id (n,m) Iq (n,m)
by interpolation of saved Id , Iq
Te−brk =Tnm−max
  to Te−brk and add to LUT
m=f loor TTe−step
e−brk

Yes
If No
If No Tnm−max ≤ Te−brk
ωvar =0

Yes

Te0 = Tnm−max

If
n++ flag = 0 or No
Ivar ≤ 0
Increment ωvar
Yes
If
No Ivar = Ilim
Yes
ωmax =ωvar

End

Fig. 2.24: Flow diagram of iterative process for LUT generation of DTP0-PMSM

74
The purpose of the LUT is to generate id−ref and iq−ref , while the motor speed (ωm ) and
reference torque (Te−ref ) are provided as inputs. Hence, the LUT should have evenly
spaced ωm and Te−ref breakpoints. During each iteration of the current loop, whenever
the computed maximum torque for a given operating point (Tnm−max ) crosses the
breakpoint values of reference torque (Te−brk ), then Id (n,m) and Iq (n,m) are interpolated
to the breakpoints and saved to the LUT. In Fig. 2.24, ‘n’ corresponds to the speed row
and ‘m’ corresponds to the torque column of the resulting LUT. The ‘flag’ denotes
whether at least one iteration of βvar has satisfied the voltage limit constraint for a
given Ivar and ωm . If the flag remains ‘0’ even after varying βvar from 0◦ to 90◦ , then it
means that there is no feasible operation if the current is further reduced. Hence, further
current decrement is stopped and the program continues with the next iteration of the
speed loop. However, if the flag remains zero even when Ivar = Ilim , then the operation
at that speed is infeasible, marking the limiting speed of the drive (ωmax ). During the
iterative process, the maximum torque developed during below base speed operation is
saved as Tmax .
The LUTs of id−ref and iq−ref obtained using the proposed algorithm given in the flow
diagram (Fig. 2.24) for a 3 kW, 34 Nm, 1875 rpm IPMSM is plotted as shown in
Fig. 2.25 and Fig. 2.26, respectively.

2.9 CONCLUSION

This chapter provided a brief outline of the types, modelling, and control of PMSM
motors. The modelling of the PMSM was performed using the space phasor approach,
which gives a good insight into the rotating magnetic field concept and the control
required for PMSM motors. The development of the speed and current controllers
were also discussed in detail. The SVPWM technique for inverter switching was
also presented. The SVPWM is extended to incorporate the overmodulation and six-
step operation of the inverter for achieving improved DC bus voltage utilisation. The
PMSM drives are usually controlled in MTPA in the below base speed region and Field

75
 Fig. 2.25: Lookup table values of d-axis current reference (id−ref )

Fig. 2.26: Lookup table values of q-axis current reference (iq−ref )

Weakening (FW) for above the base speed operation. The concepts of MTPA and FW
of the IPMSM were explained using the current limit circle and the voltage limit ellipse
in the ids − iqs plane. Finally, the solution for optimal current references considering
parameter variation was derived using an iterative flow diagram. The optimal current
references are stored in an LUT to reduce the computational complexity of the Digital
Signal Processor (DSP).

76
The next chapter proposes a dual three-phase PMSM for achieving an expansion of
power rating and operating speed range compared to the conventional three-phase
PMSM. The dual three-phase PMSM is highly suited for EV and HEV applications.
The dual three-phase PMSM reduces the vehicle weight and improve efficiency since
it eliminates the requirement for a DC-DC boost converter used in conjunction with
conventional three-phase PMSM drives.

77
 CHAPTER 3

A DUAL 3-PHASE PMSM WITH SIX-STEP OPERATION

FOR ELECTRIC VEHICLES

3.1 INTRODUCTION

A wide operating speed range and enhanced output power capability are necessary
to meet the ever-increasing demand for high efficient and high power-dense motors
for EV and HEV applications, as discussed in Section 1.1. Present-day commercial
EVs use either a high voltage DC bus by connecting more cells in series or a DC-DC
boost converter to attain the specific targets. However, the requirement of complex
cell balancing circuits while stacking more cells and the additional weight and space
requirement for bulky inductors associated with DC-DC converters are still drawbacks
of the existing system. Alternate solutions include the use of OEW configurations and
winding changeover techniques.
Multi-phase and multi-star machines are becoming popular in EV applications due to
their power segmentation capability, reduction in power electronic switch ratings and
improved fault tolerance. Multi-phase machines have been conventionally used in high
power applications like ship propulsion and large industrial drives. Despite the large
amount of published work on multi-phase PMSMs for EVs and HEVs, commercial
realisations are limited. Hence, more research on multi-phase machines and further
exploration of their advantages is required if it has to replace the conventional 3-phase
solutions.
Conventionally, multi-phase machines are realised with either a parallel splitting of
the conductors in the winding to reduce the current rating of devices (He et al., 2010;
Parsa, 2005) or a serial splitting accompanied with a reduction of DC bus voltage to
reduce the device voltage ratings (Gopakumar et al., 1993; Parsa, 2005). In this thesis,
a new approach for doubling the rated power of an existing three-phase machine by
reconfiguring its winding and using it as a dual three-phase machine (2×3 phase) is
presented. The increase in rated power is achieved in the dual three-phase machine with
the same insulation level and per phase current as of the original three-phase machine.
Here, coils corresponding to each phase winding are split serially, reducing the number
of turns per phase to half and fed from two different converters with the same DC bus
voltage and switch ratings as the original three-phase drive. Thus, it would double the
rated power as well as extend the maximum operating speed of the drive as detailed in
Section 3.2. Hence, multi-phase PMSMs would be ideal for traction applications that
require a wide operating speed range.
Dual three-phase machines are classified based on displacement angle between the two
winding sets as
(a) Symmetric dual three-phase machine (either 0◦ or 60◦ displacement) (Luo et al.,
2015; Barcaro et al., 2010; Che and Hew, 2015; Oleschuk et al., 2007; Liang
et al., 2020)
(b) Asymmetric dual three-phase machine or split-phase machine (30◦ displacement)
(Zhu et al., 2020; Zhou et al., 2016; Wang et al., 2013; Yazdani et al., 2009;
Marouani et al., 2008; Hadiouche et al., 2006, 2004; Gopakumar et al., 1993)

Even though the proposed method of power enhancement and extended speed range is
explained for a 0◦ displacement dual 3-phase IPMSM (DTP0-IPMSM) in Section 3.2,
it is applicable to all dual three-phase machines, including 60◦ displacement dual three-
phase IPMSMs and split-phase machines. In this thesis, six-step operation in the DTP0-
IPMSM for an electric vehicle application is proposed to achieve double the power
rating and extended speed range compared to the original three-phase IPMSM, ensuring
minimal circulating current. The concept of circulating current is explained in detail in
Section 3.4.

3.2 PROPOSED POWER ENHANCEMENT AND EXTENDED SPEED

RANGE WITH DUAL THREE-PHASE MACHINES

Consider a three-phase PMSM with rated power Prated , base speed ωb , and nc turns per
coil connected to an inverter with a DC bus voltage Vdc . The 0◦ displacement dual 3-

79
phase PMSM (DTP0-PMSM) is realised by splitting the coils in each phase of the three-
phase machine into equal halves. Thus, the number of turns per coil of the resulting dual
nc
three-phase machine is 2
. The process of reconfiguration of a three-phase PMSM to
form a DTP0-PMSM is shown in Fig. 3.1.

nc
nc 2

+ nc
2
nc axis-1 nc
turns/coil nc 2
3-φ 2 axis-2
axis winding−1

winding−2

Original 3-phase PMSM DTP0-PMSM

Fig. 3.1: DTP0-PMSM obtained from reconfiguration of three-phase PMSM

The dual three-phase PMSM can be wound as either with 0◦ displacement (DTP0-
PMSM) or 30◦ displacement (split-phase PMSM) between the two three-phase winding
sets. The winding configurations of the three-phase PMSM, DTP0-PMSM and split-
phase PMSM are shown in Fig. 3.2a Fig. 3.2b, and Fig. 3.2c, respectively. Here, a,b,c
correspond to a-phase, b-phase and c-phase of the three-phase PMSM in Fig. 3.2a.
Further, in Fig. 3.2b and Fig. 3.2c, a1, b1 and c1 correspond to winding-1 and a2, b2,
c2 correspond to winding-2 in the DTP-PMSM. It is to be noted that, in the DTP0-
PMSM, two three-phase winding sets are wound in the same slot to achieve 0◦ spacial
displacement.
The power circuit of the DTP0-PMSM is shown in Fig. 3.3. Conventionally, DC
Vdc
bus voltage is reduced to 2
to reduce the device voltage ratings (Gopakumar et al.,
1993; Parsa, 2005), when the number of turns per coil is reduced to half for operating
the reconfigured dual three-phase PMSM. However, in the proposed method, the DC
bus voltage is maintained at Vdc since the winding insulation level of the original
inverter fed three-phase machine is already rated for Vdc . Thus, the base speed of the

80
 nc turns/coil

a a −c −c b b −a −a c c −b −b

30◦
360◦
(a) Single layer winding for three-phase PMSM
nc turns/coil
2
a1 a1 −c2 −c2 b1 b1 −a2 −a2 c1 c1 −b2 −b2
a2 a2 −c1 −c1 b2 b2 −a1 −a1 c2 c2 −b1 −b1
(b) Double layer winding for DTP0-PMSM
nc turns/coil
2
−b2 a1 a2 −c1 −c2 b1 b2 −a1 −a2 c1 c2 −b1
a1 a2 −c1 −c2 b1 b2 −a1 −a2 c1 c2 −b1 −b2
(c) Double layer winding for split-phase PMSM

Fig. 3.2: Winding distributions for original three-phase PMSM and various
reconfigured dual three-phase PMSMs

resulting machine becomes 2ωb , doubling the output power of the proposed DTP0-
PMSM compared to the original three-phase PMSM. This method of doubling the
output power using winding reconfiguration of the three-phase PMSM as a DTP0-
PMSM is termed as proposed method-A (Prop-A) throughout the thesis. The operating
speed range is further extended using the six-step operation of the converter. It should
be noted that the thermal rating and cooling requirements of the resulting DTP0-PMSM
need to be improved to compensate for higher core loss and mechanical losses due to
the extraction of higher power and high-speed operation.
By using the proposed concept, the windings of a 15.5 kW original three-phase IPMSM
are split serially and the resulting three-phase sets are connected to two three-phase

battery Vdc

DTP0−
Inverter-1 IPMSM Inverter-2

Fig. 3.3: Power circuit of DTP0-PMSM

81
VSIs. The resulting configuration operates as a 31 kW DTP0-IPMSM, thus doubling
the power level without violating the DC bus voltage limit and the machine current
ratings. The 15.5 kW three-phase IPMSM and the corresponding reconfigured 31 kW
DTP0-IPMSM with parameters given in Table. 3.1 are simulated in MATLAB/Simulink
to obtain the power-speed characteristics and torque-speed characteristics as shown in
Fig. 3.4 and Fig. 3.5, respectively. MTPA control is used in the below base speed
region, and field weakening control is employed for the above base speed operation.
The torque-speed characteristics requirement of a 15.5 kW HEV and a 31 kW Battery
Electric Vehicle (BEV) requirement is superimposed on Fig. 3.4 and Fig. 3.5.
Constant Power Speed Range (CPSR) is conventionally used to measure the
performance of the drives used for electric vehicle applications. It is defined as the
speed range for which the drive can maintain constant power operation with respect to
the base speed.

Maximum speed with rated power capability

CPSR = (3.1)
Base speed

The same winding reconfiguration method with six-step operation in Prop-A can also
be used to extend the CPSR without utilising the power enhancement feature of DTP0-
PMSM, as observed in Fig. 3.4. Such extension of CPSR of the conventional three-
phase PMSM would be ideal for Hybrid Electric Vehicle (HEV) applications with a
low hybridisation ratio (Rahman et al., 2000). Here, the power rating of the motor
is kept same as the original motor by operating only in the region marked by HEV

Table 3.1: Dual three-phase IPMSM parameters used for simulation

Parameter Value Parameter Value

Prated 31 kW Rs 13 mΩ
Vdc 200 V ΨF 0.0832 Wb
ωb 1820 rpm Lds 0.247 mH
Irated 90 A Lqs 0.593 mH
ωmax 13,000 rpm Poles 8
Tmax 163 Nm

82
ranging from 3 to 7, whereas heavy-duty EVs like electric trucks require CPSR up to
25 (Mohammadi, 2018; Giulii Capponi et al., 2015) to have a single gear transmission.
Thus, the wide CPSR requirement of electric vehicles can be achieved using Prop-B
and enhancement of power rating as well as operating speed is achieved using Prop-
A with the reconfigured DTP0-PMSM obtained from an original three-phase PMSM.
Thus, the proposed DTP0-IPMSM with six-step operation eliminates the DC-DC boost
converters and the associated bulky inductor used together with a three-phase PMSM
having a limited CPSR in EV and HEV applications.

3.2.1 Comparison of proposed dual three-phase PMSM with existing topologies

A comparison of the proposed solution with some of the conventional topologies

discussed in Section 1.2 is provided in Table. 3.2, which is an expansion of the table
given by Pan et al. (2014). It is observed from the table that the proposed DTP0-
PMSM is a better solution compared to other topologies in terms of maximum device
voltage, inverter KVA rating, elimination of additional inductor, output voltage slope,
circulating current, fault-tolerant operation during winding open-circuit, and motor
insulation requirement. The proposed method can either double the power rating at
a higher speed (Prop-A) or enhance CPSR by operating it with the same power rating
as the original three-phase PMSM (Prop-B) as explained in Section 3.2. It is to be noted
that the advantages of more levels of output voltage and reduction in switching losses
under linear modulation with OEW-PMSM are not present in the proposed topology.
However, switching loss reduction is achieved in the six-step mode of operation using
the proposed method. The table entries where the proposed DTP0-PMSM gives an
advantage compared to any other topology is highlighted in grey for easy understanding.
The advantages of the proposed method are explained by considering each table entry
in the following section. In Table. 3.2, 1 pu refers to the rating of a three-phase PMSM
from which various topologies are derived.

(a) Maximum motor terminal voltage

The proposed DTP0-PMSM configuration is obtained by splitting the windings

84
 Table 3.2: Comparison of proposed method with conventional topologies (Pan et al., 2014)

Parameters Drv-1 Drv-2 Drv-3 Drv-4 Drv-5 Prop-A Prop-B

Max motor voltage 2 pu 2 pu 1.15 pu 2 pu 2 pu 1 pu 1 pu
Total source voltage 1 pu 2 pu 1 pu 1 pu 2 pu 1 pu 1 pu
Max device voltage 1 pu 1 pu 1 pu 2 pu 2 pu 1 pu 1 pu
No. of switches 12 12 12 8 6 12 12
Max motor power 1 pu 2 pu 1.15 pu 2 pu 2 pu 2 pu 1 pu
Inverter KVA rating 2 pu 2 pu 2 pu 4 pu 2 pu 2 pu 2 pu
Extra inductor no no no yes no no no
Stored energy in cap 2 pu 2 pu 1 pu 4 pu 4 pu 1 pu 1 pu
CPSR expanded same same same same same expanded
Onset of FW < 2 pu 2 pu 1.15 pu 2 pu 2 pu 2 pu 2 pu
Output voltage waveform 3-level 3-level 3-level on phase 2-level 2-level 2-level 2-level
Voltage slope dV
dt
1 pu 1 pu 1 pu 2 pu 2 pu 1 pu 1 pu
Switching loss reduction yes no no yes no six-step mode six-step mode
Circulating current no no yes no no no no
Fault tolerant with OC
no no no no no yes yes
winding faults
Motor insulation req same same same high high same same
Drv-1: OEW with series compensation using a floating capacitor
Drv-2: OEW with two isolated voltage sources
Drv-3: OEW with shared DC bus
Drv-4: 2-level inverter with DC/DC boost converter
Drv-5: 2-level inverter with doubled supply voltage
Prop-A: Proposed DTP0-PMSM with six-step operation for output power enhancement
Prop-B: Proposed DTP0-PMSM with six-step operation with same power rating as original 3-phase PMSM for
CPSR enhancement

of an original three-phase PMSM and feeding from two inverters with the same
DC bus voltage Vdc . Hence, the terminal voltage remains at 1 pu throughout the
operation. However, in all other configurations except Drv-3 in Table. 3.2, the
terminal voltage becomes 2 pu when the machine is operated at 2ωb . In Drv-3,
the motor voltage reaches 15% more since the DC voltage is directly applied to
the phase winding.

(b) Total source voltage

The total source voltage refers to the total voltage applied from different sources
to the motor. Since the proposed DTP0-PMSM has only one DC bus, the total
source voltage remains the same as the original three-phase PMSM (1 pu). In
Drv-2, the total source voltage is 2 pu since it requires two isolated voltage
sources to supply each inverter. Drv-5 also requires a total source voltage of
2 pu, since the voltage at DC bus is doubled by stacking more batteries in series.

(c) Maximum device voltage

The proposed DTP0-PMSM requires switches of 1 pu voltage rating since the DC
bus voltage is the same as the original three-phase PMSM. Whereas, Drv-4 and
Drv-5 require 2 pu device rating because of the doubled voltage at the DC bus.

(d) Number of switches

The proposed DTP0-PMSM requires 12 switches compared to 8 and 6 switches in
Drv-4 and Drv-5. In Drv-4, 6 switches correspond to the inverter, and 2 switches
correspond to the boost converter. The requirement of additional switches and
their associated gate drive circuitry is a disadvantage of the proposed method.
However, the additional switches improve reliability compared to the original

85
 three-phase PMSM, as explained in the “Fault tolerance with open circuit winding
fault” section given below.

(e) Maximum motor power

The proposed DTP0-PMSM can be either operated to extract double the power of
the original three-phase PMSM (Prop-A) or operated at the same power level to
extend the CPSR (Prop-B). Hence, Prop-A has 2 pu power rating, whereas Prop-
B has only 1 pu power rating. Both Drv-4 and Drv-5 can achieve 2 pu power
rating due to the availability of double DC bus voltage. Drv-2 can also achieve
2 pu power rating by the use of two isolated voltage sources. However, Drv-1 and
Drv-3 can achieve a power rating of only 1 pu and 1.15 pu, respectively.

(f) Inverter kVA rating

The proposed DTP0-PMSM requires an inverter kVA rating of 2 pu, since two
inverters with each having the same kVA rating as the original three-phase drive
are required to feed the drive. All other topologies except Drv-4 require 2 pu
inverter kVA rating. In Drv-4, the inverter as well as the DC/DC boost converter
are rated for 2 pu each. Hence, the total converter kVA rating becomes 4 pu.

(g) Extra inductor

No additional inductor is required for the proposed DTP0-PMSM. Also, no other
configuration except Drv-4 requires an additional inductor. In Drv-4, a bulky
inductor is required for the DC/DC boost converter used in the drive.

(h) Stored energy in capacitor

Since the proposed DTP0-PMSM uses the same DC bus as the original three-
phase PMSM, the energy stored in the capacitor remains at 1 pu. However, the
stored energy in the capacitor of Drv-1 and Drv-2 is 2 pu since they use two
DC buses. Moreover, Drv-4 and Drv-5 require 4 pu energy to be stored in the
capacitor due to the doubling of capacitor voltage.

(i) CPSR
The proposed DTP0-PMSM operated with the same power rating of a three-phase
PMSM (Prop-B) achieves an extended CPSR. Further, the same CPSR extension
is obtained in Drv-1 also. Even though Drv-2, Drv-3, Drv-4, Drv-5 and Prop-
A improve the motor power rating, the CPSR remain the same as the original
three-phase PMSM (1 pu).

(j) Onset of field weakening

Field weakening is initiated in the proposed DTP0-PMSM control only at 2 pu
speed compared to the original three-phase PMSM. Hence, the efficiency is
improved in the low and medium speed region due to the reduction in d-axis
current required for field weakening. Drv-2, Drv-4 and Drv-5 also start field
weakening at 2 pu speed. However, the field weakening for Drv-3 starts at 1.15 pu
speed since the DC bus voltage is directly applied across phase windings in this
configuration.

(k) Output voltage waveform

The proposed DTP0-PMSM produces the same 2-level output voltage as the

86
 original three-phase PMSM. However, the OEW configurations (Drv-1, Drv-2,
Drv-3) can produce a 3-level output voltage.

(l) Voltage slope dV

dt
Voltage slope refers to the slope of PWM pulses applied at the motor terminal.
Here, the comparison is made assuming the same switching duration of the
devices. The voltage slope of the proposed DTP0-PMSM and the original three-
phase PMSM are equal (1 pu) since the DC bus voltage remains the same. For
OEW configurations (Drv-1, Drv-2, Drv-3) also, the voltage slope remains at
1 pu. However, the voltage slope is 2 pu for Drv-4 and Drv-5 since the DC bus
voltage is doubled.

(m) Switching loss reduction

The proposed method reduces switching loss in the six-step mode but not in the
linear modulation region. In Drv-1, the switching loss is reduced even in the
linear region by reducing the DC bus voltage of the floating capacitor. In Drv-4,
the DC bus voltage is reduced in the low-speed region using the boost converter.

(n) Circulating current

Circulating currents are not present in the proposed DTP0-PMSM due to the
symmetrical winding configuration. No other configuration except Drv-3 has the
circulating current issue. In Drv-3, a path exists for the common-mode current
since the two inverters connected to the two winding ends share the same DC bus.

(o) Fault tolerance with open circuit winding fault

In case of an open circuit fault in any phase or inverter leg, the proposed DTP0-
PMSM can continue the operation with the other set of inverter and three-phase
winding. No other configuration compared here has the open circuit winding
fault tolerance. Thus, the proposed method improves the reliability of the original
three-phase PMSM drive, making it more suitable for EV application.

(p) Motor insulation requirement

The motor insulation is majorly decided by dV dt
for inverter fed motors. In the
proposed DTP0-PMSM, the DC bus voltage as well as the motor terminal voltage
remain the same as the original three-phase PMSM. Hence, the same motor
insulation grade as the original three-phase PMSM can be used for the DTP0-
PMSM since the insulation is already rated for PWM operation. However, an
additional winding spacer needs to be placed in the stator slot of the DTP0-
PMSM to separate the two three-phase winding sets for double layer winding
distribution. The motor insulation has to be increased for Drv-4 and Drv-5 due to
high dV
dt
with the doubled DC bus voltage. Motor insulation for Drv-1, Drv-2 and
Drv-3 remain nearly the same since dV dt
is 1 pu, even though the motor terminal
voltage is 2 pu.

87
3.3 MODELLING OF DTP0-IPMSM

The detailed space phasor (alternatively called as space vector) based dynamic
modelling of three-phase IPMSM was discussed in Section 2.3. Similar to the three-
phase IPMSM, the dynamic model of the DTP0-IPMSM is also developed using the
space phasor methodology. In the DTP0-IPMSM, both the three-phase winding sets
have an equal number of turns and are wound one above the other. It is to be noted that
both the windings have strong mutual coupling, as they share the same slots. Hence,
current in one winding set causes an induced emf in the other winding set and vice
versa. Thus, the control of the DTP0-IPMSM has to be performed with appropriate
feedforward terms to achieve decoupled control and ensure good dynamic performance.
This is the major difference between the modelling as well as control of the three-phase
IPMSM and the DTP0-IPMSM.
Fig. 3.6 shows the winding distribution and the rotor of a DTP0-IPMSM. Let as1, bs1,
and cs1 correspond to the phases of winding-1 and as2, bs2, and cs2 correspond to the
phases of winding-2. It is assumed that all the windings are sinusoidally distributed in
space.

bs2 axis Current into the plane

Current out of the plane
as2’
bs1 axis
as1’ rotor axis
cs2 N bs1 bs2

cs1
θr
as1 axis as2 axis

bs2’ cs1’

bs1’ S
cs2’
as1
cs1 axis as2

cs2 axis

Fig. 3.6: Winding distribution and rotor of a DTP0-IPMSM

88
The flux linkage equation of the DTP0-IPMSM is given by

Ψas1 = Las1−as1 ias1 + Las1−bs1 ibs1 + Las1−cs1 ics1 + Las1−as2 ias2 + Las1−bs2 ibs2

+Las1−cs2 ics2 + ΨF cos θr (3.2)

Ψbs1 = Lbs1−as1 ias1 + Lbs1−bs1 ibs1 + Lbs1−cs1 ics1 + Lbs1−as2 ias2 + Lbs1−bs2 ibs2
 
2π
+Lbs1−cs2 ics2 + ΨF cos θr − (3.3)
3
Ψcs1 = Lcs1−as1 ias1 + Lcs1−bs1 ibs1 + Lcs1−cs1 ics1 + Lcs1−as2 ias2 + Lcs1−bs2 ibs2
 
4π
+Lcs1−cs2 ics2 + ΨF cos θr − (3.4)
3
Ψas2 = Las2−as1 ias1 + Las2−bs1 ibs1 + Las2−cs1 ics1 + Las2−as2 ias2 + Las2−bs2 ibs2

+Las2−cs2 ics2 + ΨF cos θr (3.5)

Ψbs2 = Lbs2−as1 ias1 + Lbs2−bs1 ibs1 + Lbs2−cs1 ics1 + Lbs2−as2 ias2 + Lbs2−bs2 ibs2
 
2π
+Lbs2−cs2 ics2 + ΨF cos θr − (3.6)
3
Ψcs2 = Lcs2−as1 ias1 + Lcs2−bs1 ibs1 + Lcs2−cs1 ics1 + Lcs2−as2 ias2 + Lcs2−bs2 ibs2
 
4π
+Lcs2−cs2 ics2 + ΨF cos θr − (3.7)
3

Here, the self and mutual inductances of the DTP0-IPMSM can be expressed as,

Las1−as1 = Las2−as2 = Lls + LA − LB cos(2θr ) (3.8)

Las1−bs1 = Lbs1−as1 = Las2−bs2 = Lbs2−as2 = Las1−bs2 = Lbs2−as1 = Las2−bs1 = Lbs1−as2

LA 2π
=− − LB cos(2θr − ) (3.9)
2 3
Las1−cs1 = Lcs1−as1 = Las2−cs2 = Lcs2−as2 = Las1−cs2 = Lcs2−as1 = Las2−cs1 = Lcs1−as2
LA 4π
=− − LB cos(2θr − ) (3.10)
2 3
4π
Lbs1−bs1 = Lbs2−bs2 = Lls + LA − LB cos(2θr − ) (3.11)
3
Lbs1−cs1 = Lcs1−bs1 = Lbs2−cs2 = Lcs2−bs2 = Lbs1−cs2 = Lcs2−bs1 = Lbs2−cs1 = Lcs1−bs2
LA
=− − LB cos(2θr ) (3.12)
2
2π
Lcs1−cs1 = Lcs2−cs2 = Lls + LA − LB cos(2θr − ) (3.13)
3

89
Las1−as2 = Las2−as1 = Llm + LA − LB cos(2θr ) (3.14)
4π
Lbs1−bs2 = Lbs2−bs1 = Llm + LA − LB cos(2θr − ) (3.15)
3
2π
Lcs1−cs2 = Lcs2−cs1 = Llm + LA − LB cos(2θr − ) (3.16)
3

The phase voltage equations of the DTP0-IPMSM is given by

dΨas1
vas1 = Rs ias1 + (3.17)
dt
dΨbs1
vbs1 = Rs ibs1 + (3.18)
dt
dΨcs1
vcs1 = Rs ics1 + (3.19)
dt
dΨas2
vas2 = Rs ias2 + (3.20)
dt
dΨbs2
vbs2 = Rs ibs2 + (3.21)
dt
dΨcs2
vcs2 = Rs ics2 + (3.22)
dt

The voltage space vector for winding-1 in the stationary reference frame is obtained
~ ss1 ) using (3.17),
in terms of the current space vector (~iss1 ) and the flux space vector (Ψ
(3.18) and (3.19) as,

s
~vs1 = vas1 + avbs1 + a2 vcs1 (3.23)
~s
dΨ
= Rs~iss1 + s1
(3.24)
dt

2π
where a = ej 3 . Similarly, the voltage space vector for winding-2 in the stationary
reference frame is obtained in terms of the current space vector (~iss2 ) and the flux space
~ s ) using (3.20), (3.21) and (3.22) as,
vector (Ψ s2

s
~vs2 = vas2 + avbs2 + a2 vcs2 (3.25)
~ ss2
dΨ
= Rs~iss2 + (3.26)
dt

~ s and Ψ
Here, the flux space vectors in the stationary reference frame (Ψ ~ s ) are given
s1 s2

90
by

~ s = Ψas1 + aΨbs1 + a2 Ψcs1

Ψ (3.27)
s1

~ ss2 = Ψas2 + aΨbs2 + a2 Ψcs2

Ψ (3.28)

The flux space vector for winding-1 in the stationary reference frame is solved by
substituting (3.2), (3.3), (3.4), and (2.6) in (3.27) as

   
~ ss1 = 3 3 3 3
Ψ Lls + LA is1 + Llm + LA ~iss2 − LB ej2θr (~iss1 +~iss2 )∗ + ΨF ejθr (3.29)
~s
2 2 2 2

where Llm is the mutual leakage between winding-1 and winding-2. Similarly, the flux
space vector for winding-2 is obtained as,

   
~ ss2 = 3 3 3 3
Ψ Lls + LA is2 + Llm + LA ~iss1 − LB ej2θr (~iss1 +~iss2 )∗ + ΨF ejθr (3.30)
~s
2 2 2 2

The voltage space vector equation for winding-1 of the DTP0-IPMSM in the stationary
reference frame is solved by substituting (3.29) in (3.24) as,

    s
3 d~iss1 3 d~is2 3 d
s
~vs1 =Rs~iss1 + Lls + LA + Llm + LA − LB ej2θr (~iss1 + ~iss2 )∗
2 dt 2 dt 2 dt
3
− j3LB ωr ej2θr (~iss1 + ~iss2 )∗ + j ωr ΨF ejθr
2
(3.31)

Similarly, voltage space vector equation for winding-2 is given by

    s
3 d~iss2 3 d~is1 3 d
s
~vs2 =Rs~iss2 + Lls + LA + Llm + LA − LB ej2θr (~iss1 + ~iss2 )∗
2 dt 2 dt 2 dt
3
− j3LB ωr ej2θr (~iss1 + ~iss2 )∗ + j ωr ΨF ejθr
2
(3.32)

The voltage space vector for winding-1 is resolved in α−β components and is expressed

91
as,

   
3 diαs1 3 diαs2
vαs1 = Rs iαs1 + Lls + LA + Llm + LA
2 dt 2 dt
 
3 d(iαs1 + iαs2 ) d(iβs1 + iβs2 )
− LB cos 2θr + sin 2θr
2 dt dt
− 3LB ωr (−(iαs1 + iαs2 ) sin 2θr + (iβs1 + iβs2 ) cos 2θr )
3
− ωr ΨF sin θr (3.33)
2    
3 diβs1 3 diβs2
vβs1 = Rs iβs1 + Lls + LA + Llm + LA
2 dt 2 dt
 
3 d(iαs1 + iαs2 ) d(iβs1 + iβs2 )
− LB sin 2θr − cos 2θr
2 dt dt
3
− 3LB ωr ((iαs1 + iαs2 ) cos 2θr + (iβs1 + iβs2 ) sin 2θr ) + ωr ΨF cos θr (3.34)
2

The voltage space vector for winding-2 is resolved in α−β components and is expressed
as,

   
3 diαs2 3 diαs1
vαs2 = Rs iαs2 + Lls + LA + Llm + LA
2 dt 2 dt
 
3 d(iαs1 + iαs2 ) d(iβs1 + iβs2 )
− LB cos 2θr + sin 2θr
2 dt dt
− 3LB ωr (−(iαs1 + iαs2 ) sin 2θr + (iβs1 + iβs2 ) cos 2θr )
3
− ωr ΨF sin θr (3.35)
2    
3 diβs2 3 diβs1
vβs2 = Rs iβs2 + Lls + LA + Llm + LA
2 dt 2 dt
 
3 d(iαs1 + iαs2 ) d(iβs1 + iβs2 )
− LB sin 2θr − cos 2θr
2 dt dt
3
− 3LB ωr ((iαs1 + iαs2 ) cos 2θr + (iβs1 + iβs2 ) sin 2θr ) + ωr ΨF cos θr (3.36)
2

The voltage equation in the rotor reference frame is derived by substituting all stator
reference frame variables in terms of rotor reference frame variables as done in
Section 2.3. The simplified voltage equations in the rotor reference frame are given

92
by
  r   r
3 d~is1 3 d~is2
r
~vs1 =Rs~irs1+ Lls + LA + Llm + LA
2 dt 2 dt
 
3 d 3
− LB (~irs1 + ~irs2 )∗ + jωr Lls + LA (~irs1 + ~irs2 )
2 dt 2
3 3
− jωr LB (~irs1 + ~irs2 )∗ + j ωr ΨF (3.37)
2 2
  r   r
3 d~i 3 d~is1
r
~vs2 =Rs~is2 + Lls + LA
r s2
+ Llm + LA
2 dt 2 dt
 
3 d 3
− LB (~irs1 + ~irs2 )∗ + jωr Lls + LA (~irs1 + ~irs2 )
2 dt 2
3 3
− jωr LB (~irs1 + ~irs2 )∗ + j ωr ΨF (3.38)
2 2
Finally, the voltage equations for winding-1 and winding-2 of the DTP0-IPMSM in the
rotor reference are expressed in d-q components of the voltage space vector and the
current space vector as,

dids1 dids2
vds1 = Rs ids1 + (Lls + Lmd ) + (Llm + Lmd ) − ωr (Lls + Lmq )iqs1
dt dt
− ωr (Llm + Lmq )iqs2 (3.39)
diqs1 diqs2
vqs1 = Rs iqs1 + (Lls + Lmq ) + (Llm + Lmq ) + ωr (Lls + Lmd )ids1
dt dt
3
+ ωr (Llm + Lmd )ids2 + ωr ΨF (3.40)
2
dids2 dids1
vds2 = Rs ids2 + (Lls + Lmd ) + (Llm + Lmd ) − ωr (Lls + Lmq )iqs2
dt dt
− ωr (Llm + Lmq )iqs1 (3.41)
diqs2 diqs1
vqs2 = Rs iqs2 + (Lls + Lmq ) + (Llm + Lmq ) + ωr (Lls + Lmd )ids2
dt dt
3
+ ωr (Llm + Lmd )ids1 + ωr ΨF (3.42)
2

where Lmd = 23 (LA + LB ) = 32 Lmax , Lmq = 23 (LA − LB ) = 32 Lmin . Lmd and Lmq
represent the magnetising inductances of the DTP0-IPMSM in the d-q frame.
The parameters of the DTP0-IPMSM can be expressed in terms of the original three-
phase IPMSM. Each winding set of the DTP0-IPMSM has half the number of turns per

93
phase of the original three-phase IPMSM. Hence,

Rs−3φ
Rs = (3.43)
2
Lls−3φ
Lls = (3.44)
4
Lmd−3φ
Lmd = (3.45)
4
Lmq−3φ
Lmq = (3.46)
4

where subscript 3φ corresponds to parameters of the original three-phase IPMSM.

Assume that the self-leakage (Lls ) consists of a component of mutual leakage with other
winding (Llm ) and another component corresponding to leakage flux linking only that
winding (Lll ). Then, the DTP0-IPMSM voltage expression can be rewritten as,

dids1 d(ids1 + ids2 )

vds1 = Rs ids1 + Lll − ωr Lll iqs1 + (Llm + Lmd )
dt dt
− ωr (Llm + Lmq )(iqs1 + iqs2 ) (3.47)
diqs1 d(iqs1 + iqs2 )
vqs1 = Rs iqs1 + Lll + ωr Lll ids1 + (Llm + Lmq )
dt dt
3
+ ωr (Llm + Lmd )(ids1 + ids2 ) + ωr ΨF (3.48)
2
dids2 d(ids1 + ids2 )
vds2 = Rs ids2 + Lll − ωr Lll iqs2 + (Llm + Lmd )
dt dt
− ωr (Llm + Lmq )(iqs1 + iqs2 ) (3.49)
diqs2 d(iqs1 + iqs2 )
vqs2 = Rs iqs2 + Lll + ωr Lll ids2 + (Llm + Lmq )
dt dt
3
+ ωr (Llm + Lmd )(ids1 + ids2 ) + ωr ΨF (3.50)
2

The developed electromagnetic torque of the DTP0-IPMSM can be derived from the
instantaneous power balance equation.

2
Pe = Re{~vs1~i∗s1 + ~vs2~i∗s2 } (3.51)
3
2
= Re{(vds1 + jvqs1 )(ids1 − jiqs1 ) + (vds2 + jvqs2 )(ids2 − jiqs2 )} (3.52)
3
2
= (vds1 ids1 + vqs1 iqs1 + vds2 ids2 + vqs2 iqs2 ) (3.53)
3

94
By substituting (3.39) - (3.42) in (3.53) and neglecting the resistive drop and transient
terms, the active power output of the machine is obtained as

 
2 3
Pe = ωr (Lmd − Lmq )(iqs1 + iqs2 )(ids1 + ids2 ) + ωr ΨF (iqs1 + iqs2 ) (3.54)
3 2

Thus, the developed electromagnetic torque of the DTP0-IPMSM in the rotor reference
Pe
frame (d-q) variables is derived using Te = ωr
as,

2P
Te = [(Lmd − Lmq )(iqs1 + iqs2 )(ids1 + ids2 ) + Ψm (iqs1 + iqs2 )] (3.55)
32

where Ψm = 23 ΨF .
The dynamic equations for a 0◦ displacement dual 3-phase SMPMSM (DTP0-
SMPMSM) is obtained by substituting Lms = Lmd = Lmq in the voltage and torque
equations of DTP0-IPMSM.

3.3.1 Equivalent circuit of DTP0-IPMSM in the rotor reference frame

The equivalent circuit of the DTP0-IPMSM in the rotor reference frame is derived from
the dynamic model equations (3.47), (3.48), (3.49), and (3.50), as shown in Fig. 3.7.
It is to be noted that except first three terms, all other terms are the same in the d-
axis equations of winding-1 (3.47) and winding-2 (3.49). A similar property exists for
the q-axis equations of winding-1 (3.48) and winding-2 (3.50) also. Thus, a common
section exists for the d-axis equivalent circuit of winding-1 and winding-2, as shown in
Fig. 3.7a. Similarly, a common section also exists for the q-axis equivalent circuit of
winding-1 and winding-2, as seen in Fig. 3.7b. These common sections represent the
equivalent circuit corresponding to the air gap flux produced by the machine. Hence,
the electromagnetic torque of the machine and the air gap flux are produced only due to
the current flowing in the common section in the q-axis as well as the d-axis equivalent
circuit. The individual branches of the winding sets shown before the common section
represents the stator resistance drop and the stator self leakage flux linking only that

95
 Lll ωr Lll iqs1
ids1 Rs
Llm+Lmd ωr (Llm+Lmq )(iqs1+iqs2 )

vds1 ids2 Rs Lll ωr Lll iqs2 (ids1+ids2 )

vds2
(a) d-axis equivalent circuit

Rs Lll ωr Lll ids1

iqs1
Llm+Lmq ωr (Llm+Lmd )(ids1+ids2 )

vqs1 iqs2 Rs Lll ωr Lll ids2 (iqs1+iqs2 )

3
ωΨ
2 r F
vqs2
(b) q -axis equivalent circuit

Fig. 3.7: Equivalent circuit of DTP0-IPMSM in rotor reference frame

winding (Lll ).
Further, a mutual coupling voltage exists between the d-axis and the q-axis equivalent
circuit similar to the three-phase IPMSM discussed in Section 2.3.1. These cross-
coupling terms should be appropriately fed forward to the control system to achieve
good dynamic performance during the vector control of the DTP0-IPMSM.

3.3.2 Phasor diagram of DTP0-IPMSM

The phasor diagram of the DTP0-IPMSM in the rotor reference frame is derived
considering the steady-state model. The steady-state voltage equations are obtained
by neglecting the transient terms in (3.39), (3.40), (3.41), and (3.42) as

96
 vds1 = Rs ids1 − ωr (Lls + Lmq )iqs1 − ωr (Llm + Lmq )iqs2 (3.56)
3
vqs1 = Rs iqs1 + ωr (Lls + Lmd )ids1 + ωr (Llm + Lmd )ids2 + ωr ΨF (3.57)
2
vds2 = Rs ids2 − ωr (Lls + Lmq )iqs2 − ωr (Llm + Lmq )iqs1 (3.58)
3
vqs2 = Rs iqs2 + ωr (Lls + Lmd )ids2 + ωr (Llm + Lmd )ids1 + ωr ΨF (3.59)
2
The phasor diagram for winding-1 and winding-2 of the DTP0-IPMSM in the d-q frame
is constructed using (3.56)- (3.59) as shown in Fig. 3.8 and Fig. 3.9, respectively.
The current space vector is located in the second quadrant to maximise the developed
torque according to (3.55), similar to a three-phase IPMSM discussed in Section 2.3.2.
It should be noted that the developed torque in (3.55) is proportional to the algebraic
addition of d-q components of currents in winding-1 and winding-2. Thus, maximum
developed torque is obtained with the minimal current when the current space vectors
of winding-1 and winding-2 are aligned. Further, since both the windings have the same
winding thickness, the magnitude of the currents in both winding sets should also be

q-axis
Rsids1 ωr (Llm+Lmq )iqs2
jRsiqs1
ωr (Lls+Lmq )iqs1
jωr (Lls+Lmd)ids1
j 23 ωr ΨF

jωr (Llm+Lmd)ids2
~ir jiqs1
s1
~ r
vs1 ~ir jiqs2
s2

δ
~r
Ψ d-axis
ids1 θr
ids2
Stator reference axis
Fig. 3.8: Phasor diagram for winding-1 of DTP0-IPMSM in rotor reference frame

97
 q-axis
Rsids2 ωr (Llm+Lmq )iqs1
jRsiqs2
ωr (Lls+Lmq )iqs2
jωr (Lls+Lmd)ids2
j 32 ωr ΨF

jωr (Llm+Lmd)ids1 ~ir

s2 jiqs1
~ir jiqs2
s1
r
~
vs2

δ
~r
Ψ d-axis
ids1 θr
ids2
Stator reference axis
Fig. 3.9: Phasor diagram for winding-2 of DTP0-IPMSM in rotor reference frame

the same for achieving a minimum copper loss. As a result of these two observations, to
achieve optimal current control of the DTP0-IPMSM, the current space vectors of both
windings should be equal in magnitude as well as in phase. Hence, during the control,
the d-axis currents of winding-1 and winding-2 are controlled to be equal, together with
q-axis currents of both winding sets also controlled to be equal. The phasor diagrams
shown in Fig. 3.8 and Fig. 3.9 are constructed with equal current space vectors for both
the winding sets.

3.4 CIRCULATING CURRENT IN DUAL THREE-PHASE PMSMS DURING

SIX-STEP OPERATION

The split-phase machine has been the most researched among the dual three-phase
machines due to its inherent 6th harmonic torque ripple elimination (Singh et al., 2003).
In order to fully utilise the available DC bus voltage of the inverter, operation in the
overmodulation region up to six-step needs to be implemented in the dual three-phase
machines. However, the overmodulation and six-step operation of split-phase induction

98
3.4.1 Circulating current concept

The circulating current is defined as the harmonic current, which does not contribute
to the air gap flux production but results in additional stator copper loss (Xu and Ye,
1995; Hadiouche et al., 2004). Here, the concept of circulating current is visualised
in a simple manner from the perspective of additional stator copper loss without
using any complex transformations. Finally, the resulting expression is correlated
with the non-electromagnetic energy component of stator current obtained from matrix
transformation (Hadiouche et al., 2004; Hadiouche et al., 2000). To explain the concept
of circulating current, the harmonic model of a DTP0-SMPMSM (where Lds =Lqs ) in
the α-β reference frame is used, as shown in Fig. 3.11. However, the concept is also
applicable to both the split-phase IPMSM and the DTP0-IPMSM. In the harmonic
equivalent circuit, the voltage sources that were present in the fundamental model
are short-circuited. The subscripts 1 and 2 represent the variables corresponding to
winding-1 and winding-2, respectively. Rs , Lll , Llm and Lms represent the stator
resistance, self leakage inductance, mutual leakage inductance and mutual inductance,
respectively. Let iα1h and iα2h be the total harmonic current of winding-1 and winding-
2, respectively, in the α-reference frame. The total harmonic current in winding-1 (iα1h )
is obtained as the actual current (iαs1 ) less the fundamental current (iα1-fund ).

iα1h−min
1
0
Rs1 Lll−1
iαh−net
iα1h 0
1 1
0 11
00
iα−cir √
iα2h 0
1 1
0 3Llm 23 Lms
Rs2 Lll−2
1
0
iα2h−min
Fig. 3.11: Harmonic equivalent circuit of dual three-phase SMPMSM in α-β reference
frame

100
There exist an optimal value for both iα1h and iα2h to produce the net air-gap flux
component iαh-net , which result in a minimum total copper loss while flowing in
winding-1 and winding-2, respectively.

iαh-net = iα1h +iα2h = iα1h-min +iα2h-min (3.60)

The current divider rule of parallel circuits is used to obtain the optimal currents that
result in a minimum copper loss. Here, the self leakage inductance is neglected while
applying the current divider rule.

iαh-net Rs2 iαh-net Rs1

iα1h-min = ; iα2h-min = (3.61)
Rs1 + Rs2 Rs1 + Rs2

The minimum copper loss with harmonic current is obtained as

Pcu-min = i2α1h-min Rs1 + i2α2h-min Rs2 (3.62)

The total α-component of stator current in winding-1 (iα1h ) is decomposed into two, as
shown in Fig. 3.11.

(a) Component that contributes to air-gap flux (iα1h-min )

(b) Component that circulates only in the stator winding and does not link with the
air-gap flux (iα-cir )

The circulating current is obtained as the actual current less the optimal current.

iα1h Rs1 − iα2h Rs2

iα-cir = iα1h − iα1h-min = (3.63)
Rs1 + Rs2
iβ1h Rs1 − iβ2h Rs2
iβ-cir = iβ1h − iβ1h-min = (3.64)
Rs1 + Rs2

Due to the symmetry in the two winding sets, Rs1 is assumed to be equal to Rs2 . Further,
considering MTPA operation, the fundamental components of both windings are
controlled to be the same (iα1-fund =iα2-fund and iβ1-fund =iβ2-fund ). Hence, the circulating

101
currents in the α-β reference frame (3.63) and (3.64) reduces to

iαs1 − iαs2 iβs1 − iβs2

iα-cir = ; iβ-cir = (3.65)
2 2

Thus, the total winding currents (consisting of fundamental and harmonic) of winding-
1 and winding-2 in the α − β reference frame should be the same for dual-three phase
machines to eliminate the circulating current. The condition for circulating current
elimination is given by

iα−net iβ−net
iαs1 = iαs2 = ; iβs1 = iβs2 = (3.66)
2 2

where iα−net and iβ−net are the net current flowing through the common branch of the
equivalent circuit in the α and β reference frame, respectively.
The optimal harmonic stator currents of the DTP0-PMSM with equal parameters are
given by

iαh-net
iα1h-min = iα2h-min = (3.67)
2
iβh-net
iβ1h-min = iβ2h-min = (3.68)
2

Hadiouche et al. (2004); Hadiouche et al. (2000) defines a transformation for a general
dual three-phase machine with an arbitrary shift angle ‘α’ between the windings
to obtain the non-electromechanical energy-conversion-related components of stator
variables. The transformation used for the DTP0-PMSM is obtained by substituting
α=0 and is given by
    
 ix  1 −0.5 −0.5 −1 0.5 0.5  iabc1 
 =K √ √ √ √   (3.69)
iy 0 − 23 2
3
0 2
3
− 2
3
i abc2

where K= √13 , iabc1 = [ ias1 ibs1 ics1 ]T and iabc2 = [ ias2 ibs2 ics2 ]T .

102
Equation (3.69) is reduced to
     
ix   iαs1 −iαs2   iα-cir 
 =K  = 2K   (3.70)
iy −iβs1 +iβs2 −iβ-cir

√ √
where iαs1 = [ 1 −0.5 −0.5 ][ iabc1 ], iβs1 = [0 2
3
− 2
3
][ iabc1 ] iαs2 = [ 1 −0.5 −0.5 ][ iabc2 ],
√ √
iβs2 = [0 2
3
− 2
3
][ iabc2 ] are the three-phase to two-phase stationary transformations
used.
Hence, ix and iy components corresponding to non-electromechanical energy-
conversion-related variables in the existing literature is equivalent to the circulating
current defined in this thesis, when the two three-phase winding sets are exactly
symmetrical. It should be noted that (3.63) and (3.64) gives a generalised expression for
the circulating current, which can be used even when the parameters in the two winding
sets of the dual three-phase PMSM are different.

3.4.2 Analysis of circulating current in split-phase PMSM

In the split-phase PMSM, two sets of three-phase windings are displaced from each
other by 30◦ . The neutrals of the two windings are isolated to eliminate zero sequence
currents. The split-phase PMSM model with the same parameters as the DTP0-PMSM
in Table. 3.1 is simulated in MATLAB/Simulink during the six-step operation. The
reference frame transformations used for the analysis of split-phase PMSM is shown
in Fig. 3.12. The voltages and currents in the α-β reference frame are extracted to

q-axis a2-axis
β-axis ωr ~
30◦ Ψr
d-axis
θr
α-axis
a1-axis
Fig. 3.12: α − β reference frame for the split-phase PMSM

103
rating and reduces the inverter switching losses, the large circulating current limits its
practical usage.

3.5 PROPOSED CIRCULATING CURRENT MINIMISATION OF HIGH-

SPEED DUAL-THREE PHASE IPMSM UNDER SIX-STEP OPERATION

In this thesis, the high-speed operation of the DTP0-IPMSM is achieved by

implementing the six-step operation. The use of DTP0-IPMSM provides a simple
solution to ensure minimal circulating current even during six-step operation, without
using any complex control algorithms.
From the power circuit of the DTP0-PMSM shown in Fig. 3.1, it is observed that the two
sets of three-phase windings are aligned to each other in contrast to 30◦ displacement in
the split-phase PMSM. The reference frame transformations used for the analysis of the
DTP0-IPMSM is shown in Fig. 3.14. A 15.5 kW, three-phase IPMSM is reconfigured
as a 31 kW DTP0-IPMSM and simulated in MATLAB/Simulink. The parameters of the
DTP0-IPMSM is given in Table. 3.1. Since there is no winding displacement between
the two three-phase sets, the applied voltages are also in-phase and aligned to each
other to produce a maximum resultant voltage vector. Thus, the voltages of the two
winding sets in the α-β reference frame are also equal instantaneously (vαs1 = vαs2
and vβs1 = vβs2 ), as observed in Fig. 3.15. Since the two windings are symmetric,
equal currents are generated in the α-β reference frame to produce a total effective
current iα-net . Thus, the condition iαs1 =iαs2 = iα-net
2
is satisfied at all operating regions
as discussed in Section 3.4.1. Hence, the circulating currents are zero in both the α-

q-axis
β-axis ωr
~r
Ψ
d-axis
θr
α-axis
a1-axis a2-axis
Fig. 3.14: α − β reference frame for DTP0-IPMSM

105
Table 3.3: Stator current harmonics in DTP0-IPMSM and split-phase IPMSM during
six-step operation at 4660 rpm

Harmonic order 5th 7th 11th 13th THD

DTP0-IPMSM 2.5% 1.3% 0.5% 0.35% 3.53%
Split-phase IPMSM 87.2% 44.6% 0.53% 0.34% 98.5%

et al., 2012). During the six-step operation of the 31 kW DTP0-IPMSM under

consideration at 4660 rpm, the magnitude of 5th and 7th harmonic components in the
phase current is only 2.5% and 1.3% with respect to the fundamental compared to
87.2% and 44.6% for the split-phase IPMSM, as given in Table. 3.3. Further, the
Total Harmonic Distortion (THD) of phase current for the DTP0-IPMSM is only 3.53%
compared to 98.5% for the split-phase IPMSM. The torque ripple of the DTP0-IPMSM
at a constant speed of 9620 rpm during the six-step mode is obtained as 13.2%, as
shown in Fig. 3.16. This corresponds to 0.029 pu when normalised with the maximum
developed torque at MTPA.

50
Speed = 9620 rpm
Torque ripple = 4.62 Nm (0.029pu)
Developed torque (Nm)

40

30

20

10

0
0 1 2 3 4 5
time (s) -3
×10

Fig. 3.16: Developed torque in DTP0-IPMSM during six-step operation

3.6 VECTOR CONTROL OF DTP0-IPMSM WITH SIX-STEP OPERATION

The control system of the DTP0-IPMSM is shown in Fig. 3.17. The control is designed
using the dynamic model of the DTP0-IPMSM given in Section 3.3. In the control
of the dual three-phase IPMSM, each set of three-phase winding is controlled in the

107
 stage-2 vq-OL
change efq1
PI
iq−ref vq1-CL stage-1 vq1-out vαs1 θv1
dq αβ
iqs1 change SVPWM Vdc
Te-ref efd1 vd-OL
id−iq id−ref PI αβ v V6 θ Module
vd1-CL βs1 MI1 ibs1
LUT stage-1 ias1
efq2 change vd1-out
ωr ids1 PI θr
stage-1 vq2-out kH ias2
change MI2 ibs2
vq2-CL v αs2 αβ SVPWM
iqs2 efd2 vq-OL dq
PI
stage-1 Module
αβ vβs2 V θ θv26
θr
change vd2-out
ids2 vd2-CL speed
vd-OL calc
stage-1 change (vd1) ω r
ωm
P
vd1-CL 0 ids1 efd1 2
ωr
vd1-out iqs1 Feed
vd-OL 1 efq1
ids2 forward efd2 iαs1
changeover calculation ias1 abc αβ ids1
signal iqs2 efq2 iβs1
xin reset ibs1 αβ dq iqs1
0 yLPF θr
id-ref open-loop vd-OL iαs2
vd1-CL(n−1) ias2 abc αβ ids2
LPF iq-ref VAC i
vd-OL(n−1) yinit calculation vq-OL ibs2 αβ βs2 dq iqs2
ωr

Fig. 3.17: Control system for sensored vector control of DTP0-IPMSM under all
modulation ranges including six-step

synchronous rotor reference frame (d-q frame) similar to a conventional three-phase

IPMSM. ids1 and iqs1 in the control system correspond to currents in the d-q frame of
winding-1, whereas ids2 and iqs2 correspond to that of winding-2.
In the equivalent three-phase IPMSM, there exists an optimal d-axis current and q-axis
current for a given speed which produces the required torque with minimum current
magnitude, and is termed as Maximum Torque Per Ampere (MTPA) operation. An
LUT is commonly used to generate the optimal q-axis current reference (iq-ref ) and the
d-axis current reference (id-ref ). In this thesis, an iterative method is used to form the
LUT, which includes the effect of stator resistance and the variation of d-axis and q-axis
inductances due to saturation as discussed in Section 2.8. Thus, MTPA control in the
below base speed operation and field weakening during the above base speed operation
are achieved without using any additional algorithm for the three-phase IPMSM.
In the DTP0-IPMSM, the two sets of three-phase windings are aligned to each
other with half the number of turns compared to the equivalent three-phase machine.
Hence, the same mmf vector that produces MTPA in the three-phase machine is
realised in the DTP0-IPMSM when both the windings are operated with equal current

108
references. Thus, the optimal currents for the maximum output torque with the
DTP0-IPMSM occurs when the d-axis current reference of winding-1 and winding-
2 are the same (id1-ref =id2-ref =id-ref ), and the q-axis current references are also the
same (iq1-ref =iq2-ref =iq-ref ), as discussed in Section 3.3.2. The LUT for the optimal
current generation of the DTP0-IPMSM is obtained by modifying the voltage and
torque equation of the three-phase IPMSM and by using the same procedure described
in Section 2.8. The DTP0-IPMSM is operated in torque control mode for traction
applications. During low and medium speed operation, a closed loop Current Vector
Control (CVC) is used.

3.6.1 Current vector control (CVC)

In CVC, PI controllers are used to control the d-q currents of winding-1 and winding-2.
During CVC, the switches in the ‘stage-1 change’ block in Fig. 3.17 are connected to
‘0’ for passing the closed-loop d-q voltages, and the offset kH is kept inactive. The
rotor position (θr ) is sensed using an absolute encoder and is used for back and forth
transformation between the rotor reference frame (d-q) and the stationary reference
frame (α-β).
In the DTP0-IPMSM, additional cross-coupling voltages between the two winding sets
are present, as seen from (3.56)-(3.59). Since winding-1 and winding-2 share the same
slots with a double layer distribution, the self leakage inductance and mutual leakage
inductance are nearly the same (Lls ≈L′lm ). Hence, the cross-coupling voltages using
the machine model reduces to

efd1 = efd2 = −ωr Lqs (iqs1 + iqs2 ) (3.71)

3
efq1 = efq2 = ωr Lds (ids1 + ids2 )+ ωr ΨF (3.72)
2

where ΨF is the PM flux. Here, d-axis self inductance (Lds ) and q-axis self inductance
(Lqs ) are given by
Lds = Lls + Lmd ; Lqs = Lls + Lmq (3.73)

109
where Lls is the self leakage inductance, Lmd is the d-axis magnetising inductance
and Lmq is the q-axis magnetising inductance. These cross-coupling voltages are fed
forward to improve the system dynamics, as shown in Fig. 3.17. The drive operation
in the below base speed operation and above base speed operation are performed using
CVC in the linear modulation region. However, a six-step operation with CVC would
result in issues like current controller saturation and reduced torque dynamics. Hence,
Voltage Angle Control (VAC) methods (Lee et al., 2014; Stojan et al., 2012; Morimoto
et al., 2007; Monajemy and Krishnan, 1999; Kim and Seok, 2013) are commonly used
for the six-step operation to overcome the limitation of conventional Field Oriented
Control (FOC).

3.6.2 Voltage angle control (VAC) for six-step operation

The VAC methods are generally classified into open-loop VAC and closed-loop VAC
with PI controllers. Even though closed-loop VAC methods achieve steady-state voltage
control even in the presence of machine parameter variations (Bolognani et al., 2011)
compared to open-loop methods, they have instability issues. Hence, the controller
parameters should be adjusted using stability analysis techniques and additional control
methods like d-axis current feedback have to be included to ensure stability (Lee et al.,
2014; Stojan et al., 2012). Open-loop VAC methods are discussed by Morimoto et al.
(2007); Monajemy and Krishnan (1999); Kim and Seok (2013). A two-dimensional
LUT derived from the steady-state motor model is used by Monajemy and Krishnan
(1999); Kim and Seok (2013) to obtain the voltage angle for a given torque and speed
for a three-phase IPMSM during six-step control.
In this thesis, an LUT based open-loop VAC is adapted for the DTP0-IPMSM due to its
simple implementation and stability during high-speed operation with six-step control.
Since the magnitude of space vector voltage is fixed during the six-step mode, only
one degree of control freedom is available. Hence, the d-q currents of winding-1 and
winding-2 are controlled by varying only the voltage space vector angle.

110
To achieve six-step operation, the modulation region has to change from the linear
region and transit through the overmodulation region. In the overmodulation region,
the number of inverter switching per fundamental cycle reduces compared to the
linear region. The overmodulation algorithm discussed in Section 2.6 (Venugopal
and Narayanan, 2006) is used in this thesis, where the entire region is divided into
overmodulation region-1 (ovm-1) and overmodulation region-2 (ovm-2). In ovm-1,
the trajectory of the average voltage space vector is partially circular and partially
hexagonal. As the required fundamental output voltage increases in ovm-1, the
trajectory approaches a complete hexagon. A further increase in fundamental voltage is
achieved using ovm-2. In ovm-2, the output voltage is realised by holding the voltage
vector at the active vectors (tip of the hexagon) for a hold time and then traversing
through the hexagon for the remaining time. Finally, at the end of ovm-2, the output
voltage is realised with only six switching per fundamental cycle, achieving six-step
operation. Let Modulation Index (MI) be defined as the ratio of the magnitude of
reference space vector voltage to the DC bus voltage. Hence, MI at the limit of linear
region (mmax1 ), ovm-1 (mmax2 ), and ovm-2 (mmax3 ) are 0.866, 0.9085, and 0.9549,
respectively.
An intermediate open-loop VAC mode is also used in between CVC and six-step
operation to smoothen the changeover process. In this intermediate stage MI is in the
range mmax1 <MI<mmax3 , whereas in six-step mode MI is fixed at mmax3 . Hence, in
the intermediate stage, the modulation is changed from linear to ovm-1 and ovm-2, and
finally to six-step operation.
During both the intermediate stage and the six-step mode, the reference voltages in the
d-q frame is obtained using the machine model and with the same current references
obtained from LUT during CVC. The d-axis and q-axis voltage references for open-
loop VAC are obtained as,

vd-OL = Rs id-ref −2ωr Lqs iq-ref (3.74)

111
 3
vq-OL = Rs iq-ref + 2ωr Lds id-ref + ωr ΨF (3.75)
2

where Rs is the stator resistance. Here, the same reference voltages (vd-OL and vq-OL )
are given to both windings assuming that the windings are symmetric. Hence, the α-β
voltages of winding-1 and winding-2 are the same during open-loop VAC operation.
Consequently, the circulating currents are minimised in linear, overmodulation and six-
step operation as discussed in Section 3.5. A sudden change of control from closed-loop
CVC to open-loop VAC based six-step mode can result in a large over current. Hence,
a transition algorithm is required for achieving a smooth changeover.

3.6.3 Transition from CVC to six-step VAC

The transition from closed-loop CVC to open-loop VAC based six-step is implemented
in two stages, as shown in Fig. 3.18. In stage-1, the control is transferred from CVC
to open-loop VAC with modulation restricted to linear region (MI<mmax1 ). Further, in
stage-2, an offset (kH ) is continuously added to MI to gradually change the modulation
region of open-loop VAC from linear through ovm-1 and ovm-2, to finally reach the
six-step mode. The stage-1 transition is realised using a Low-Pass Filter (LPF) with
zero input and a changeover switch, as shown in the enlarged ‘stage-1 change’ block in

vd1-out vd1-CL vd-OL

transition
CVC transition 0.955 to six-step
vd1-out to VAC
vd1-CL MI
m0
VAC
0 tc1 time 0 time
vd1-CL(n−1)−vd-OL(n−1) 0.955 − m0

yLPF kH

0 tc1 time 0 time

(a) stage-1 transition (for vd1 ) (b) stage-2 transition

Fig. 3.18: Changeover from closed-loop CVC to open-loop six-step VAC

112
Fig. 3.17. The stage-1 changeover process for vd1 is shown in Fig. 3.18a. Whenever the
machine speed crosses a pre-specified speed for changeover (ωc-v ) marked by instant
tc1 in Fig. 3.18a, stage-1 is initiated by the ‘changeover signal’ shown in Fig. 3.17.
Here, the changeover switch is changed to ‘1’, which passes vd-OL to vd1-out . Also, the
integrator in the LPF is initialised to vd1-CL(n−1) −vd-OL(n−1) as observed in Fig. 3.18a.
Here, the subscript (n−1) denotes the values at the sampling instant previous to tc1 . As
a result, vd1-out will not have a sudden transition but continues to have vd1-CL(n−1) even
after the changeover. As time progresses, the output of the LPF (yLPF ) slowly decays to
zero from its initial value since the input is zero. As a result, vd1-out transits smoothly
from vd1-CL to vd−OL as the LPF output changes from its initial value to zero, as shown
in Fig. 3.18a.
The stage-2 transition represents the intermediate open-loop VAC mode and its
implementation is highlighted with ‘green’ blocks in Fig. 3.17. Let ‘m0 ’ denote the
initial modulation index when stage-2 is initiated. In this stage, an increasing offset
(kH ) which starts from zero, is added to the space vector magnitude to change the
operation from linear region (MI=m0 ) through ovm-1 and ovm-2 to finally reach the
six-step mode (MI=0.955). Whenever MI reaches mmax3 , kH is clamped to that value,
latching the operation to the six-step mode. The plot of both MI and kH during stage-2
transition is shown in Fig. 3.18b. During further operation in the six-step mode, only the
angle of the voltage space vector is controlled depending on the d-q current references.
A similar transition strategy is required to change the control from six-step VAC to CVC
when the speed reduces from the high-speed operation.

3.6.4 Reverse transition from six-step VAC to CVC

The control transition from open-loop six-step VAC to CVC is implemented in one
stage. Hysteresis is provided during the reverse transition to prevent multiple control
transitions when the machine is operated near the changeover speed (ωc-v ). Hence, the
reverse changeover speed (ωv-c ) is selected at a lower value compared to ωc-v . When

113
the operating speed goes below ωv-c , the reverse changeover process is initiated by
decreasing kH linearly from its present value back to zero. When kH is decreased, the
space vector modulation changes from the six-step mode through the overmodulation
region and finally reaching the linear region. Once kH reaches zero, all the switch
positions in Fig. 3.17 are changed from ‘1’ to ‘0’, and the integrators in the d-axis and
q-axis current PI controllers of CVC are initialised to vd−OL and vq-OL , respectively.
Thus, a seamless changeover from open-loop VAC to closed-loop CVC is achieved.

3.7 EXPERIMENTAL RESULTS

The performance of the proposed DTP0-IPMSM with six-step operation is

experimentally validated on a 3 kW low power, low-speed prototype with the
parameters given in Table. 3.4. The DTP0-IPMSM is operated using torque control
whose shaft is connected to an 11 kW IM operating with speed control. A resistive brake
chopper is provided at the DC bus terminal of the IM drive to achieve regeneration when
the DTP0-IPMSM operates in the motoring mode. The experimental setup consisting
of the DTP0-IPMSM, IM and two inverters is shown in Fig. 3.19. The proposed drive
control is implemented on a TMS320F28335 DSP with a sampling time of 50 µs, and
the switching frequency is chosen as 10 kHz. The proposed square-wave operation is
tested at high-speed operating region by increasing the reference speed of the IM drive.
The control algorithm is changed from closed-loop CVC to open-loop VAC based six-
step operation using the proposed transition algorithm when the machine speed exceeds
1500 rpm.

Table 3.4: Parameters of experimental DTP0-IPMSM prototype

Parameter Value Parameter Value

Prated 3 kW Rs 3.25 Ω
Vdc 300 V ΨF 0.579 Wb
ωb 900 rpm Lds-unsat 24 mH
Irated 6.5 A Lqs-unsat 45.13 mH
ωmax 2360 rpm Lds-sat 11.6 mH
Tmax 34 Nm Lqs-sat 42 mH
Poles 4

114
 Fig. 3.19: Experimental setup for DTP0-IPMSM drive testing

Fig. 3.20 and Fig. 3.21 show the efficiency contours obtained from test data of the
3 kW DTP0-IPMSM and the corresponding three-phase IPMSM, respectively. The
maximum torque limit curves during operation in linear region and six-step region are
obtained from the simulation of the 3 kW DTP0-IPMSM and the 1.5 kW three-phase
IPMSM modelled in MATLAB. It is observed that the maximum torque points obtained
using simulation and experiments are nearly matching, validating the model used for
simulation. The wider range of operation while using the DTP0-IPMSM compared to
the three-phase IPMSM is evident from the torque-speed characteristics. For a given
torque requirement, the optimal d-q currents of the DTP0-IPMSM at ‘2ωm ’ will be
nearly equal to that of the three-phase IPMSM at ‘ωm ’ speed. Due to the similarity
in operating conditions, the efficiencies of both the machines are nearly the same at
most operating points. Hence, as seen from Fig. 3.20 and Fig. 3.21, the efficiency
contour of the DTP0-IPMSM appears to be a horizontally stretched version of the three-
phase IPMSM. The maximum efficiency for both the machines occurs in the light load
condition at the starting of field weakening operation (just after their own base speed).
Further, an increase in efficiency is observed for the DTP0-IPMSM during high-speed

115
 50
Efficiency 0.9
45 Max torque linear region
Max torque six-step region
40 six-step 0.85
Linear region
35 operation operation
0.6
0.63
0.64 650.61
660.62

Torque (Nm)

687
0. 0.8

00.72

Efficiency
0.

.76
0.6
30

0.
00.75
.0007.
69
0.74
0.73
0.61

0.

0
0..0
71
0.71
0.

6
.
.6
06
62

.6
0.72

9
0.7

0.
00.

87
6.6654
25

0 00.0.0.06.06.0.
.6 6766 5 463
.662

63
7 3
1 0.75
.
0 .74 0.
20 0 .756 000..0886.58
45
6 463
8621
00..66.767

0.
6

0
0.
0
0.77

0.8873 0.
0 7 0 219

62616798
..888
0000.7
0..77 000..0.6 00.

.0.70.7.7.
767754737721 9 8

0..7.66219

87
34
0 .789 070..0636.70

56
6.50 0.7

000.0
15 64.

68700.0.780
47. 000.0.

0.7
0.00..0.00...70.6 0 .7 8 365621 0.0

.777 .7
0 0. 0 7.70
..
0620
.1606. 6
0.81 07
0080.
90
. 76 0
6.6
37 7506.
.... 4

6
00. 77
4 .36
567
87798 621

00.7

00.82
4
10

835
0.86
0.87
.882

7
6
.7 4
.75
8

.81
79
82
81
0.8
7

.8
.777988881234 5 3 0.65

0.7
0.
0.
0.
0 .

0
88 8 6
00.00.0.0.0..00.0.0.80.8.8788.99

0
9
5 0

0 0.6
500 1000 1500 2000
speed (rpm)

Fig. 3.20: DTP0-IPMSM efficiency contours

50
efficiency 0.9
45 Max torque linear region
0.85
40
35 0.8
Torque (Nm)

0.64

30 0.75 Efficiency
0.5
0
0.
6 54
.6
5
52

0.6
25
8
6
4
0 0.0.50.

6
0.

0. 0.7
46.26 58

64 0.70..0

68
20
0.

0.65
0.
05
..06.

.6

15
7472 866

62
524
6
8676
00.0.00..7060

0.6
02.07..476
.0

10
6468
88 6

824
00...77

00.0..886 0.55
5
0 0.5
500 1000 1500 2000
speed (rpm)
Fig. 3.21: Three-phase IPMSM efficiency contours

range (near 1700 rpm) with the six-step operation.

The plot of stator current contours of the 3 kW DTP0-IPMSM and the three-phase
IPMSM are shown in Fig. 3.22 and Fig. 3.23. It is observed that the maximum
achievable torque is limited by the rated stator current of 6.5 A in both below base
speed and field weakening region. The torque capability of the DTP0-IPMSM above
1500 rpm is enhanced by using six-step operation, as seen in Fig. 3.22.

116
 50
Phase current RMS
45 Max torque linear region 6
40 Max torque six-step region
Linear region six-step
operation operation 5
35
66.5
Torque (Nm)

Current (A)
30
4

5.5
25 6.

5
4.5 5

6
20 4
3
4. 5

6.56
3.5 4 5 5.5 5.5
15 3 6.5
6
5
2.5 5
10 2 2. 3. 4. 2
25 3 5 4

6.56
5. 55
1.5

3. 3
1.

4. 4
2.5

5
5 5

5
1 1 2
1
0
500 1000 1500 2000
speed (rpm)
Fig. 3.22: DTP0-IPMSM stator current contours

50
Phase current RMS
45 Max torque linear region 6
40
5
35
Torque (Nm)

Current (A)
30
6

4
25
5.555

20
6.56

3
4.

3.5 4
15 3

10 2
5..5
6.55
4

1.
2.
3.53

52
56

5
4
5

1
1
0
500 1000 1500 2000
speed (rpm)
Fig. 3.23: Three-phase IPMSM stator current contours

The steady-state six-step operation of the DTP0-IPMSM under 15 Nm load at 2000 rpm
is shown in Fig. 3.24 and Fig. 3.25. Fig. 3.24 shows the phase currents and line
voltages for winding-1 and winding-2. The a-phase current is equivalent to scaled α-
axis current (ias1 = 23 iαs1 and ias2 = 23 iαs2 ) when there is no neutral connection, according
to the transformation used for conversion from the abc to stationary α-β reference
frame. Hence, the circulating current is obtained by subtracting the corresponding

117
pulsation, leading to a higher torque ripple in the six-step operation compared to
the linear operation with CVC. However, an increase in the frequency of harmonic
components during high-speed operation results in higher attenuation by the self
inductance of the IPMSM, resulting in lower torque ripple at high-speed operation.
Further, the speed oscillation is negligible due to the high machine inertia, as seen in
Fig. 3.34.

3.8 CONCLUSION

In this chapter, a method of reconfiguring a three-phase PMSM and operating it as

a DTP0-PMSM for doubling the power rating and extending the maximum operating
speed is proposed. Here, the resulting machine has the same insulation level and DC
bus voltage as the original machine. Further, implementation of the six-step operation in
the DTP0-IPMSM with minimal circulating current is proposed to further improve the
constant power speed region of electric vehicles. The proposed DTP0-IPMSM also
achieves better reliability compared to existing solutions used for power rating and
CPSR enhancement. This approach can be used to eliminate the additional DC-DC
converter from the EV powertrain that uses a PMSM with limited CPSR. Finally, a
transition method for a smooth changeover from closed-loop CVC to open-loop VAC
based six-step operation and vice-versa for the DTP0-IPMSM was proposed. The
proposed methods were experimentally validated on a 3 kW laboratory prototype.
The next chapter proposes an unequal split 0◦ winding displacement dual 3-phase
PMSM (uneq0-PMSM). Here, the number of turns of winding-1 and winding-2 are
different. The uneq0-PMSM has a different torque-speed characteristic shape compared
to the conventional three-phase and dual three-phase PMSM. Further, the designer has
an additional degree of freedom to vary the winding split ratio to suit the torque-speed
characteristic requirement for the given application. The DTP0-PMSM forms a subset
of the uneq0-PMSM with a 1:1 split ratio. The modelling, control, and applicability of
uneq0-PMSM are discussed in detail in the following chapter.

125
 CHAPTER 4

AN UNEQUAL SPLIT DUAL 3-PHASE PMSM FOR HIGH

STARTING TORQUE AND WIDE OPERATING SPEED
RANGE IN MEDIUM AND HEAVY-DUTY TRUCKS

4.1 INTRODUCTION

A dual three-phase PMSM with six-step operation was proposed in the previous chapter
(Chapter 3) to improve the performance of electric vehicles and hybrid electric vehicles.
However, the requirements for certain EV applications like heavy-duty trucks, off-road
EVs and military vehicles are different from that of passenger cars. Such vehicles are
characterised by very high tractive effort requirements in the low-speed region and a
wide operating speed range compared to the base speed. The constant power speed
range of heavy-duty trucks and tractors can be up to 25 compared to passenger cars
having a CPSR in the range 3 to 7. Hence, compared to increasing the power rating of
the motor, an increase in the constant power region of operation is more desirable for
such applications. The use of wide Constant Power Speed Range (CPSR) IPMSMs
reduces the power rating of the inverter switches as well as satisfy the heavy-duty
vehicle requirement.
The typical torque-speed characteristics of a military truck are shown in Fig. 4.1 (Gao
et al., 2003). Most medium/heavy-duty trucks and off-road vehicles also have similar
characteristics. It is to be noted from Fig. 4.1 that a very high tractive effort is required in
the low-speed region for obstacle negotiation and hill climbing. However, the tractive
effort is low for high-speed operations like traffic cruising and high-speed highway
driving. Since the operating speed is low during high torque demand and the torque
demand is low during high-speed operation, a motor power requirement of 18.6 kW/ton
and a CPSR of 24.7 can encompass all continuous operating requirements, as seen in
Fig. 4.1. The load requirements like hard acceleration and emergency brake are only
 1.6 g mph time
Intermittent

Tractive effort ratio (force/weight)

Continuous ------- ----- ------
1.4 Heavy Traffic cruising 0.02 40 cont.
towing Highway driving 0.03 50 cont.
Level cross country 0.03 25 cont.
1.2 Obstacle High speed highway 0.04 70 cont.
negotiation Cross country sprint 0.107 40 min.
1 Traffic acceleration 0.2 20 sec.
Hard Acceleration 0.5 25 sec.
0.8 Hard Emergency Emergency brake 0.5 50 sec.
acceleration brake Hill climbing 0.7 6 cont.
Heavy towing 1.0 4 min.
0.6 Hill
Obstacle negotiation 1.0 5 sec.
climbing High speed highway hp/ton kW/ton
0.4 Cross country sprint
150 112
Traffic 125 93.2
100 74.6
0.2 acceleration Traffic cruising Highway driving 75 55.9
50 37.3
Level cross country 25 18.6
0
0 10 20 30 40 50 60 70 80 90
speed (mph)

Fig. 4.1: Typical torque-speed characteristics requirement for a military truck (Gao
et al., 2003)

intermittent duty and hence can be met by overloading the 18.6 kW/ton machine for
a short time. It is to be noted that the motor used for such application should have
a characteristic with a wide constant power operating region, as shown in Fig. 4.1.
However, if the constant power operating range of the motor used is limited, then it will
result in an overdesign. The motors with a CPSR of 12.3, 8.2, 6.1, 4.9 and 4.1 are also
able to satisfy the load requirement but require a higher power rating of 37.3 kW/ton,
55.9 kW/ton, 74.6 kW/ton, 93.2 kW/ton, and 112 kW/ton, respectively, as seen from
Fig. 4.1. Hence, a motor with a wide constant power region is most suited for heavy-
duty truck, military and off-road applications. The existing solutions used to achieve a
wide constant power region in PMSMs are discussed in the next section.

4.2 EXISTING SOLUTIONS FOR EXTENDED CONSTANT POWER

OPERATING SPEED RANGE

A detailed review of existing methods for the extended operating speed range of
PMSM is given in Section 1.2. The most intuitive method to achieve a wide constant
power operating region is to design the motor with the characteristic current (Ich )
approximately equal to the motor rated current (|I~s-rated |). Such motors would have

127
a wide constant power operating region with field weakening. The motor operating
speed can theoretically extend to infinity when Ich < |I~s-rated | (infinite-maximum-
speed IPMSM). However, such motors with wide CPSR have limited overloading
capability in the high-speed operation when the saliency ratio of the motor is low (Soong
and Ertugrul, 2002). A high saliency ratio design results in increased manufacturing
complexity and lower mechanical stability; hence is not suitable for heavy-duty electric
vehicles.
The overloading capability is high for PMSMs with low CPSR since the field weakening
requirement is low. However, the low CPSR machines have a low operating speed
range. The operating speed range of a low CPSR PMSM can be improved by re-rating
(Chapman and Krein, 2003). The re-rated machine has fewer turns with more parallel
paths than the original machine, which would increase the base speed of the machine,
delaying the start of field weakening. However, this results in an increase in the inverter
switch current rating.
Winding changeover methods are more suitable compared to multi-phase and OEW
solutions when the requirement is to extend the constant power region and not to
enhance the output power rating. The winding changeover techniques do not require
field weakening in the high-speed operation compared to multi-phase and OEW
based solutions. This would reduce copper losses due to the reduced d-axis current
requirement, achieving improved efficiency. However, a large number of bidirectional
switches are required for most of the winding changeover methods, as discussed in
Section 1.2. Further, a significant dip in the combined steady-state torque-speed
characteristics with different winding interconnections is inevitable for all winding
changeover methods, as discussed in Section 1.2.13.
In this thesis, an unequal split 0◦ winding displacement dual 3-phase PMSM (uneq0-
PMSM) is proposed to achieve both extension of speed range as well as reduced field
weakening requirement for improved overloading capacity and efficiency in the high-
speed operating region. In the uneq0-PMSM, the winding turns of the two three-

128
phase winding sets are not equal and is wound with 0◦ winding displacement between
them. Hence, the uneq0-PMSM is the general case of the conventional dual 3-phase
PMSM (DTP-PMSM) having equal turns ratio. In the proposed uneq0-PMSM, when
the machine speed exceeds a specific speed, the three-phase winding set with higher
number of turns (Low-Speed (LS) winding) is cut-off using thyristor switches. As a
result, the machine operates with only the three-phase winding set with lower number
of turns (High-Speed (HS) winding), achieving below base speed control even in the
high-speed operating region of the machine. The uneq0-PMSM has a different shape of
torque-speed and power-speed characteristics compared to the conventional three-phase
PMSM, enabling high efficiency operation in the high-speed region. The split ratio of
winding turns can be chosen appropriately to suit the torque-speed characteristic for
a given requirement. Thus, various torque-speed characteristics can be realised from
a given IPMSM stator and rotor by only changing the winding split ratio using the
proposed uneq0-PMSM configuration. As a result, the designer has an additional degree
of freedom to achieve the torque-speed characteristics for a given application.

4.3 VARIABLE TORQUE-SPEED CHARACTERISTICS USING THE

PROPOSED DUAL THREE PHASE PMSM WITH UNEQUAL SPLIT
WINDING

The power circuit diagram of the proposed unequal split 0◦ winding displacement dual
3-phase PMSM (uneq0-PMSM) is shown in Fig. 4.2. The uneq0-PMSM is derived by
reconfiguring only the stator winding of a conventional three-phase PMSM, as shown in
Fig. 4.3. The winding coils in each phase of the three-phase PMSM (‘nLS + nHS ’ turns
per coil) are split serially in an unequal proportion to derive two three-phase winding
sets. The winding with a higher number of turns (nLS turns per coil) is denoted as the
Low-Speed (LS) winding, and the winding with a lower number of turns (nLS turns per
coil) is represented as the High-Speed (HS) winding. The wire gauges and insulation
levels of both LS and HS windings are equal to that of the original three-phase PMSM.

129
 bidirectional
ac switches

Inverter−1
LS
winding Vdc

HS Inverter−2
winding

uneq0−PMSM

Fig. 4.2: Power circuit of the proposed uneq0-PMSM

nLS +nHS nLS

+ nLS
nHS
nLS+nHS nHS
turns/coil 3-φ axis HS-axis LS-axis

High−speed Low−speed
winding winding
Original 3-phase PMSM uneq0-PMSM

Fig. 4.3: uneq0-PMSM obtained from reconfiguration of three-phase PMSM

The two winding sets have isolated neutrals and are fed from two similar rated inverters
with a common DC bus, as shown in Fig. 4.2. Here, the two three-phase winding sets
have 0◦ spacial displacement with respect to each other. Two back-to-back connected
thyristors are also connected between the inverter-1 and the LS winding to disconnect
that winding during high-speed operation.
The operation of the proposed uneq0-PMSM is demonstrated in Fig. 4.4, showing the
variation of torque, power, d-q currents and power factor of both LS and HS windings
with motor speed. The uneq0-PMSM with parameters corresponding to that of spec-3 in
Table. 4.1 having a split ratio of 1:13 is used here. The optimal values of the reference
d-axis and q-axis currents are obtained from a pre-defined LUT. The LUT is derived
by solving the machine model in MATLAB such that the maximum torque output is

130
as equal q-axis current is seen in Fig. 4.4. As the operating speed increases and crosses
the base speed of combined LS-HS operation (ωLSHS-b ), field weakening is initiated in
both windings by injecting more negative d-axis current to maintain the magnitude of
LS winding stator voltage vector (Va-LS ) within the maximum limit. The maximum
voltage limit is set as the output voltage limit of linear modulation using SVPWM.
However, beyond a specific speed (ωLSHS-max ) the LS winding cannot be operated since it
exceeds the current or voltage limit even with field weakening. Hence, the pulses to the
back-to-back connected thyristor switches are disabled to isolate the LS winding from
the inverter-1 at a pre-specified speed (ω21-ch ). Consequently, the power circuit changes
to a conventional three-phase configuration where only the HS winding is operated
from the inverter-2. Thus, the current in the LS winding reduces to zero, as observed
in Fig. 4.4. In the proposed uneq0-PMSM, the winding split ratio of the machine
is decided such that the base speed of HS winding operated alone (ωHS-b ) is greater
than ωLSHS-max . Hence, the HS winding can operate in the below base speed region
after transition even though the machine operates in the medium/high-speed region.
This below base speed operation even during medium/high-speed using the proposed
uneq0-PMSM, improves its overload capability. It also reduces the additional d-axis
current requirement compared to the conventional three-phase PMSM with high CPSR
to satisfy the similar load torque requirement. As the speed increases further, field
weakening is used in the HS winding by injecting more −id-HS to extract its full torque
capability and to achieve a wide operating speed range, as seen from Fig. 4.4. The
maximum operating speed with the HS winding operated alone is denoted as ωHS-max .
The power factor of the machine is near unity in both low-speed (LS-HS combined
operation) and medium speed region (HS alone operation), whenever the operation is
with below base speed control, as observed from Fig. 4.4. Hence, the machine efficiency
and overload capability are improved compared to the conventional three-phase or DTP-
PMSM with high CPSR, whose major operating region is in field weakening.

132
4.3.1 Effect of variation of split ratio on the uneq0-PMSM characteristics

Different shapes of torque-speed and power-speed characteristics are obtained by

varying the winding split ratio of the uneq0-PMSM. Fig. 4.5 shows various
characteristics when a three-phase IPMSM is connected as three-phase, DTP, and
uneq0-PMSM with different winding split ratios. The three-phase PMSM with
parameters corresponding to that of spec-3 in Table. 4.1 is used here. The total
number of turns per phase of the three-phase PMSM is assumed to be 14. Here, the
DTP-PMSM forms a subset of the uneq0-PMSM when the winding split ratio is 7:7
(equal split). The feasible winding split ratios for the uneq0-PMSM are 1:13, 2:12,
3:11, 4:10, 5:9, and 6:8 since there are 14 turns per phase. The resulting uneq0-

0.8
uneq0-PMSM (1:13)
uneq0-PMSM (2:12)
0.6 uneq0-PMSM (3:11)
Torque (pu)

uneq0-PMSM (4:10)
uneq0-PMSM (5:9)
0.4
uneq0-PMSM (6:8)
DTP-PMSM (7:7)
0.2 3 -PMSM

0
0 5 10 15 20 25 30
speed (pu)
(a) Torque-speed characteristics
2.5
DTP-PMSM (7:7) uneq0-PMSM (1:13)
3 -PMSM uneq0-PMSM (2:12)
2
uneq0-PMSM (3:11)
Power (pu)

uneq0-PMSM (4:10)
1.5 uneq0-PMSM (5:9)
uneq0-PMSM (6:8)
1

0.5

0
0 5 10 15 20 25 30
speed (pu)
(b) Power-speed characteristics

Fig. 4.5: Variation of power-speed and torque-speed characteristics with split ratio in
uneq0-PMSM (spec-3)

133
PMSMs with 6:8, 5:9 and 4:10 split ratios have a reduced operating area in the torque-
speed and power-speed characteristics compared to the DTP-PMSM. However, with a
3:11 split ratio, the shape of torque-speed characteristics starts to deviate from that of
conventional three-phase PMSM. It consists of a low-speed region where two windings
are operated and a high-speed region with one-winding operation. During one-winding
operation (HS winding alone), the machine is operated in the below base speed region
by disconnecting the LS winding from the inverter, extending its operating speed range
further to that of the two-winding configuration. This extended operation with one
winding gets more prominent with winding split ratios of 2:12 and 1:13. The resulting
shapes of torque-speed characteristics with such extreme split ratios fit the heavy-duty
automotive requirements in Fig. 4.1.

4.4 MODELLING OF UNEQ0-IPMSM

The dynamic model of the uneq0-IPMSM is similar to the DTP0-IPMSM derived in

Section 3.3. However, since the turns ratio of two three-phase winding sets is different
in the uneq0-IPMSM, the parameters of each winding have to be scaled appropriately.
For all parameters, subscript ‘LS’ denotes that it relates to LS winding, ‘HS’
denotes HS winding and ‘LSHS’ denotes the mutual terms. Assume that the
uneq0-IPMSM has a winding split ratio of nHS :nLS , where nLS and nHS represent
the number of turns per phase of LS winding and HS winding, respectively. The
parameters of the uneq0-IPMSM can be represented in terms of the three-phase PMSM
parameters as Rs-LS =d1 Rs−3φ ; Lls-LS =d21 Lls−3φ ; ΨF -LS =d1 ΨF ; Lmd-LS =d21 Lmd−3φ ;
Lmq-LS =d21 Lmq−3φ ; Rs-HS =d2 Rs−3φ ; Lls-HS =d22 Lls−3φ ; ΨF -HS =d2 ΨF ; Lmd-HS =d22 Lmd−3φ ;
Lmq-HS =d22 Lmq−3φ ; Lmd-LSHS =d1 d2 Lmd−3φ ; Lmq-LSHS =d1 d2 Lmq−3φ where Rs−3φ is the
per phase stator resistance, ΨF is the PM flux, Lmd−3φ is the d-axis magnetising
inductance, Lmq−3φ is the q-axis magnetising inductance and Lls−3φ is the self leakage
inductance of the original three-phase PMSM from which the uneq0-IPMSM was
derived. Here d1 = nLSn+n
LS
HS
and d2 = nLSn+n
HS
HS
.

134
The voltage equations for the LS and HS windings of the uneq0-IPMSM expressed in
the rotor reference frame are given by

d d
vd-LS =Rs-LS id-LS + (Lls-LS +Lmd-LS ) id-LS +(Llm +Lmd-LSHS ) id-HS
dt dt
−ωr (Lls-LS +Lmq-LS )iq-LS −ωr (Llm +Lmq-LSHS )iq-HS (4.1)
d d
vq-LS =Rs-LS iq-LS + (Lls-LS +Lmq-LS ) iq-LS +(Llm +Lmq-LSHS ) iq-HS
dt dt
3
+ωr (Lls-LS +Lmd-LS )id-LS +ωr (Llm +Lmd-LSHS )id-HS + ωr ΨF -LS (4.2)
2
d d
vd-HS =Rs-HS id-HS + (Lls-HS +Lmd-HS ) id-HS +(Llm +Lmd-LSHS ) id-LS
dt dt
−ωr (Lls-HS +Lmq-HS )iq-HS −ωr (Llm +Lmq-LSHS )iq-LS (4.3)
d d
vq-HS =Rs-HS iq-HS + (Lls-HS +Lmq-HS ) iq-HS +(Llm +Lmq-LSHS ) iq-LS
dt dt
3
+ωr (Lls-HS +Lmd-HS )id-HS +ωr (Llm +Lmd-LSHS )id-LS + ωr ΨF -HS (4.4)
2

where ωr is the electrical rotor speed, Lmd-LSHS is the mutual inductance between LS
and HS winding and Llm is the mutual leakage inductance between LS and HS winding.
Assume that the self leakage of LS winding (Lls-LS ) consist of a component of mutual
leakage with other winding (Llm-LS ) and another component corresponding to leakage
flux linking only that winding (Lll-LS ). Similarly, the self leakage of HS winding (Lls-LS )
consists of a component of mutual leakage with other winding (Llm-HS ) and another
component corresponding to leakage flux linking only that winding (Lll-HS ). Then, the
uneq0-IPMSM voltage expression can be rewritten as,

d d
vd-LS =Rs-LS id-LS + Lll-LS id-LS +(Llm-LS +Lmd-LS ) id-LS
dt dt
d
+(Llm +Lmd-LSHS ) id-HS −ωr Lll-LS iq-LS −ωr (Llm-LS +Lmq-LS )iq-LS
dt
−ωr (Llm +Lmq-LSHS )iq-HS (4.5)
d d
vq-LS =Rs-LS iq-LS + Lll-LS iq-LS + (Llm-LS +Lmq-LS ) iq-LS
dt dt
d
+(Llm +Lmq-LSHS ) iq-HS +ωr Lll-LS id-LS +ωr (Llm-LS +Lmd-LS )id-LS
dt

135
 3
+ωr (Llm +Lmd-LSHS )id-HS + ωr ΨF -LS (4.6)
2
d d
vd-HS =Rs-HS id-HS + Lll-HS id-HS + (Llm-HS +Lmd-HS ) id-HS
dt dt
d
+(Llm +Lmd-LSHS ) id-LS −ωr Lll-HS iq-HS −ωr (Llm-HS +Lmq-HS )iq-HS
dt
−ωr (Llm +Lmq-LSHS )iq-LS (4.7)
d d
vq-HS =Rs-HS iq-HS + Lll-HS iq-HS + (Llm-HS +Lmq-HS ) iq-HS
dt dt
d
+(Llm +Lmq-LSHS ) iq-LS +ωr Lll-HS id-HS +ωr (Llm-HS +Lmd-HS )id-HS
dt
3
+ωr (Llm +Lmd-LSHS )id-LS + ωr ΨF -HS (4.8)
2

The voltage equations of the uneq0-IPMSM can be referred to either LS winding or

HS winding side to obtain a common equivalent circuit. Transformation of variables in
uneq0-IPMSM is performed similar to that of transformers. In this thesis, the variables
of HS winding are referred to LS winding side to represent the machine in a common
equivalent circuit in the rotor reference frame.
The resulting voltage equations for the LS and HS winding of the uneq0-IPMSM in the
d-q frame (rotor reference frame) after referring all the variables to the LS winding are
given by

d
vd-LS =Rs-LS id-LS + Lll-LS id-LS −ωr Lll-LS iq-LS
dt
d
+(Llm-LS +Lmd-LS ) (id-LS +i′d-HS )−ωr (Llm-LS +Lmq-LS )(iq-LS +i′q-HS ) (4.9)
dt
d
vq-LS =Rs-LS iq-LS + Lll-LS iq-LS +ωr Lll-LS id-LS
dt
d
+(Llm-LS +Lmq-LS ) (iq-LS + i′q-HS )+ωr (Llm-LS +Lmd-LS )(id-LS +i′d-HS )
dt
3
+ ωr ΨF -LS (4.10)
2
′ d
vd-HS =Rs-LS i′d-HS + Lll-LS i′d-HS −ωr Lll-LS i′q-HS
dt
d
+(Llm-LS +Lmd-LS ) (id-LS +i′d-HS )−ωr (Llm-LS +Lmq-LS )(iq-LS +i′q-HS ) (4.11)
dt
(4.12)

136
 ′ d ′
vq-HS =Rs-LS i′q-HS + Lll-LS iq-HS +ωr Lll-LS i′d-HS
dt
d
+(Llm-LS +Lmq-LS ) (iq-LS +i′q-HS )+ωr (Llm-LS +Lmd-LS )(id-LS +i′d-HS )
dt
3
+ ωr ΨF -LS (4.13)
2
where i′d-HS = nnHS
LS
id-HS , i′q-HS = nnHS
LS
′
iq-HS , vd-HS = nnHS
LS ′
vd-HS , and vq-HS = nnHS
LS
vq-HS .
The developed electromagnetic torque of the uneq0-IPMSM can be derived from the
instantaneous power balance equation as,

2 ′
Pe = (vd-LS id-LS + vq-LS iq-LS + vd-HS i′d-HS + vq-HS
′
i′q-HS ) (4.14)
3

By substituting (4.9) - (4.13) in (4.14) and neglecting the resistive drop and transient
terms, the active power output of the machine is obtained as

 
2 3
Pe = ωr (Lmd-LS −Lmq-LS )(iq-LS +i′q-HS )(id-LS +i′d-HS )+ ωr ΨF -LS (iq-LS +i′q-HS )
3 2
(4.15)
Thus, the developed electromagnetic torque of the uneq0-IPMSM in the rotor reference
Pe
frame (d-q) variables is derived by substituting Te = ωr
in (4.15) as,

2P
Te = ((Lmd-LS −Lmq-LS )(id-LS +i′d-HS )+Ψm-LS ) (iq-LS +i′q-HS ) (4.16)
32

where Ψm-LS = 23 ΨF -LS .

The developed torque expressed in terms of parameters of the original three-phase
IPMSM from which uneq0-IPMSM is derived as,

2P
Te = ((Lmd−3φ −Lmq−3φ )(d1 id-LS +d2 id-HS )+Ψm−3φ ) (d1 iq-LS +d2 iq-HS ) (4.17)
32

where d1 = nLSn+n
LS
HS
and d2 = nLSn+n
HS
HS
.
The dynamic equations for an unequal split dual three-phase SMPMSM are obtained by
substituting Lm-LS =Lmd-LS =Lmq-LS and Lm-HS =Lmd-HS =Lmq-HS in the uneq0-IPMSM

137
voltage and torque equations.

4.4.1 Equivalent circuit of the uneq0-IPMSM in the rotor reference frame

The equivalent circuit of the uneq0-IPMSM in the rotor reference frame is derived
from the dynamic model equations (4.9)-(4.13), as shown in Fig. 4.6. It is to be noted
that except first 3 terms, all other terms are the same in the d-axis equations of LS
winding (4.9) and HS winding (4.11), resulting in a model similar to the DTP0-IPMSM
discussed in Section 3.3.1. The q-axis equivalent circuit also has a section common to
both LS and HS winding. Further, a mutual coupling voltage exists between the d-axis
and q-axis equivalent circuits similar to the three-phase IPMSM and the DTP0-IPMSM
discussed earlier.

4.4.2 Phasor diagram of the uneq0-IPMSM

The phasor diagram of the uneq0-IPMSM in the rotor reference frame is derived
considering the steady-state model. The steady-state voltage equations are obtained

id-LS Rs-LS Lll-LS ωr Lll-LSiq-LS

ωr (Llm-LS+Lmq-LS)(iq-LS+i′q-HS )

vd-LS i′d-HS Rs-LS Lll-LS Llm-LS+Lmd-LS

′ (id-LS+i′d-HS )
vd-HS ωr Lll-LSi′q-HS
(a) d-axis equivalent circuit

iq-LS Rs-LS Lll-LS ωr Lll-LSid-LS

ωr (Llm-LS+Lmd-LS )(id-LS+i′d-HS )

vq-LS i′q-HS Rs-LS Lll-LS Llm-LS+Lmq-LS (iq-LS+i′q-HS )

ωr Ψm-LS
′
vq-HS ωr Lll-LSi′d-HS
(b) q -axis equivalent circuit

Fig. 4.6: Equivalent circuit of uneq0-IPMSM in rotor reference frame

138
by neglecting the transient terms in (4.1)-(4.4) as

vd-LS =Rs-LS id-LS −ωr (Lls-LS +Lmq-LS )iq-LS −ωr (Llm +Lmq-LSHS )iq-HS (4.18)

vq-LS =Rs-LS iq-LS +ωr (Lls-LS +Lmd-LS )id-LS +ωr (Llm +Lmd-LSHS )id-HS
3
+ ωr ΨF -LS (4.19)
2
vd-HS =Rs-HS id-HS −ωr (Lls-HS +Lmq-HS )iq-HS −ωr (Llm +Lmq-LSHS )iq-LS (4.20)

vq-HS =Rs-HS iq-HS +ωr (Lls-HS +Lmd-HS )id-HS +ωr (Llm +Lmd-LSHS )id-LS
3
+ ωr ΨF -HS (4.21)
2
The phasor diagram for the LS winding and HS winding of the uneq0-IPMSM in the d-q
frame is constructed using (4.18)- (4.21) as shown in Fig. 4.7 and Fig. 4.8, respectively.

4.4.3 Vector control of the proposed uneq0-IPMSM

The control system for the proposed uneq0-IPMSM with sensored vector control is
shown in Fig. 4.9. The proposed drive is operated in torque control mode since the

ωr (Llm+Lmq-LSHS)iq-HS q-axis
Rs-LSid-LS
jRs-LSiq-LS
ωr (Lls-LS+Lmq-LS)iq-LS
jωr (Lls-LS+Lmd-LS)id-LS j 32 ωr ΨF -LS

jωr (Llm+Lmd-LSHS)id-HS ~r
is-HS jiq-LS
~ir jiq-HS
s-LS

r
~
vs-LS

δ
~r
Ψ d-axis
id-LS θr
id-HS
Stator reference axis
Fig. 4.7: Phasor diagram for LS-winding of uneq0-IPMSM in rotor reference frame

139
 ωr (Llm+Lmq-LSHS)iq-LS q-axis
Rs-HSid-HS
jRs-HSiq-HS
ωr (Lls-HS+Lmq-HS)iq-HS
jωr (Lls-HS+Lmd-HS)id-HS j 32 ωr ΨF -HS

jωr (Llm+Lmd-LSHS)id-LS
~ir
s-HS jiq-LS
~ir jiq-HS
s-LS
r
~
vs-HS

δ
~r
Ψ d-axis
id-LS θr
id-HS
Stator reference axis
Fig. 4.8: Phasor diagram for HS winding of uneq0-IPMSM in rotor reference frame

targeted application is for electric vehicles. The closed loop current control is achieved
using PI controllers. An LUT is used to generate the optimal reference currents for both
windings to achieve MTPA operation. Since the reference currents for FW are also
programmed in the LUT, no additional controller or algorithm is used to implement
FW. During two-winding operation (combined LS-HS), the same reference currents
from ‘id −iq LSHS-LUT’ are given to d-q current controllers of LS and HS winding
by selecting the switch position ‘0’ in Fig. 4.9. During one-winding operation (HS
winding alone), the switch position is changed to ‘1’ to choose the reference currents
from ‘id −iq HS-LUT’.
‘id −iq LSHS-LUT’ is implemented to obtain maximum output torque with minimal
stator current and satisfy the voltage and current limit constraints during the two-
winding operation. Since the LS winding has a higher number of turns than the HS
winding, the terminal voltage of HS winding is inherently controlled within limits
when the terminal voltage of LS winding is limited. Hence, only the terminal voltage
of LS winding needs to be restricted within the maximum inverter capability by field

140
 PI efq-LS
iq-LS-ref vq-LS vα-LS
Te-ref id−iq dq SVPWM
iq-LS Vdc
LSHS- efd-LS αβ Module
PI
id-LS-ref vβ-LS ib-LS
ωr LUT
vd-LS θr ia-LS
id-LS efq-HS
0 PI vq-HS
iq-HS-ref ia-HS
1 vα-HS
id−iq iq-HS efd-HS dq SVPWM ib-HS
HS- 0 id-HS-ref PI vd-HS
αβ Module
LUT 1
vβ-HS
changeover id-HS i iα-LS
pulse a-LS abc αβ id-LS θr
ωr iβ-LS
ib-LS αβ dq iq-LS
id-LS efd-LS speed
iq-LS Feed θr calc
efq-LS iα-HS
forward ia-HS abc αβ i
id-HS efd-HS iβ-HS
d-HS P
ωr
ωm 2
iq-HS calculation efq-HS ib-HS αβ dq i q-HS

Fig. 4.9: Control system of the proposed uneq0-IPMSM with sensored vector control

weakening as motor speed increases. The constraints used to derive ‘id −iq LSHS-LUT’
are as follows:

• Va-LS ≤ Vlim ; Ia ≤ Ilim

• Maximise developed torque (Te )


2P 3
Te = (d1 iq-LS + d2 iq-HS ) ΨF −3φ
32 2

+(Lmd−3φ − Lmq−3φ )(d1 id-LS + d2 id-HS ) (4.22)
 
2P 3
= ΨF −3φ iqs + (Lmd−3φ − Lmq−3φ )ids iqs . (4.23)
32 2

Here, Va-LS is the voltage space vector magnitude of LS winding, Ia is the magnitude
of current space vector, Ilim is the maximum current rating of the machine, iqs is the
q-axis current, and ids is the d-axis current. d1 and d2 are the winding split ratios with
respect to total number of turns, where d1 = nLSn+n
LS
HS
; d2 = nLSn+n
HS
HS
. To achieve maximum
torque per ampere, Ia =Ia-LS =Ia-HS , iqs =iq-LS =iq-HS , ids =id-LS =id-HS . ΨF −3φ , Lmd−3φ
and Lmq−3φ are the PM flux, d-axis magnetising inductance, and q-axis magnetising
inductance, respectively, of the original three-phase IPMSM. Since the rated current

141
flows through the entire winding during two-winding operation, the uneq0-IPMSM has
the same torque-speed characteristics shape and maximum torque as the original three-
phase IPMSM.
‘id −iq HS-LUT’ is derived considering the HS winding operation alone. Since only
HS winding is operated, the machine equations are the same as that of a three-phase
PMSM with nHS turns per phase. Since the rated current flows only through nHS turns,
the torque-speed characteristics is different from the original three-phase PMSM from
which it is derived, with Te-HS as the maximum torque rating. The constraints for ‘id −iq
HS-LUT’ implementation are,
• Va-HS ≤ Vlim ; Ia-HS ≤ Ilim
• Maximise developed torque (Te-HS )
 
2P 3
Te-HS = ΨF -HS iq-HS + (Lmd-HS − Lmq-HS )id-HS iq-HS (4.24)
32 2

where Va-HS is the voltage space vector magnitude of HS winding, Ia-HS is the
current space vector magnitude of HS winding, ΨF -HS =d2 ΨF −3φ ; Lmd-HS =d22 Lmd−3φ ;
Lmq-HS =d22 Lmq−3φ . Here, ΨF -HS , Lmd-HS and Lmq-HS are PM flux, d-axis magnetising
inductance, and q-axis magnetising inductance, respectively, of the uneq0-IPMSM with
HS winding operated alone.
The steady-state voltage equations of LS and HS windings of the proposed uneq0-
IPMSM are given in Section 4.4.2. The coupling voltages of the LS and HS windings
in the synchronous d-q frame are given by

efd-LS = −ωr (Lls-LS +Lmq-LS )iq-LS −ωr (Llm +Lmq-LSHS )iq-HS (4.25)
3
efq-LS = ωr (Lls-LS +Lmd-LS )id-LS +ωr (Llm +Lmd-LSHS )id-HS + ωr ΨF -LS (4.26)
2
efd-HS = −ωr (Lls-HS +Lmq-HS )iq-HS −ωr (Llm +Lmq-LSHS )iq-LS (4.27)
3
efq-HS = ωr (Lls-HS +Lmd-HS )id-HS +ωr (Llm +Lmd-LSHS )id-LS + ωr ΨF -HS (4.28)
2

The coupling voltages are fed forward to the current controller output as shown in
Fig. 4.9 to achieve decoupled current control. SVPWM is used to generate gate pulses

142
from the reference voltages in the α−β stationary reference frame to control the uneq0-
IPMSM drive. A transition algorithm is used to obtain a smooth changeover from two-
winding to one-winding operation when the speed increases and vice versa when the
speed decreases.

4.4.4 Transition from two-winding to one-winding operation and vice versa

A transition from the two-winding to one-winding operation is performed at a

predefined speed (ω12-ch ) when the developed torque of the machine with HS winding
operated alone exceeds the developed torque with combined LS-HS operation. During
one-winding operation, the LS winding should be isolated from inverter-1 using
thyristor switches. However, a direct switch off of the LS winding results in an
overcurrent since both LS and HS winding operate in the field weakening region at
ω12-ch . This condition is similar to a control interruption during the field weakening
operation of a three-phase PMSM, resulting in uncontrolled regeneration. Therefore,
conventionally, the winding transition is performed in the below base speed region of
operation itself in the traditional winding changeover methods (Zhitkova and Hameyer,
2016; Chen and Cheng, 2006; Kume et al., 1991; Swamy et al., 2006), therefore not
able to fully utilise the machine torque capability available with field weakening as
discussed in Section 1.2.13.
The proposed uneq0-IPMSM consists of two windings fed from two independent
inverters. Hence, in the proposed transition algorithm, an additional d-axis current is
injected from the HS winding to compensate for the decrease in current contribution
from the LS winding during the transition. The additional d-axis current that needs to
be injected from the HS winding is − nnHS
LS
id(0) , where id(0) denotes the d-axis current
at the changeover speed producing zero torque output during two-winding operation.
Thus, the uneq0-IPMSM maintains field weakening even during the transition interval
to achieve a smooth LS winding turn off.
The transition process is explained using various reference currents, inverter-1 pulse

143
enable (INV-1 pulse EN), and thyristor enable (Thy EN), as shown in Fig. 4.10. The
transition algorithm is initiated whenever the operating speed crosses the predefined
two to one-winding changeover speed (ω21-ch ). Initially, the d-q reference currents of
the LS windings and the q-axis current reference of the HS winding are changed to
zero at the transition instant, as seen in Fig. 4.10. To maintain the FW operation, the
d-axis current reference of HS winding is changed to − nLSn+n
HS
HS
id(0) . Here, − nnHS
LS
id(0)
component represents the additional current flowing through HS winding to maintain
field weakening and to compensate for the loss of LS winding. Once the actual machine
currents attain near steady-state, the thyristor switches and inverter-1 pulses are turned
off to isolate the LS winding from the inverter-1. The changeover process is completed
by selecting the current reference for the HS winding from ‘id −iq HS-LUT’ to continue
the machine operation in the below base speed with the HS winding operated alone.
It is to be noted that the developed torque becomes zero for a short duration during
the winding transition. However, the performance of the considered application of

Two winding One winding operation Two winding

operation operation
(LS-HS in FW) (HS below base speed) (LS-HS in FW)

speed ω21-ch
ω12-ch
0 time
idq-LS-ref
id-LS-ref iq-LS-ref
0 time

idq-HS-ref id-HS-ref iq-HS-ref

0
time

High ‘−id-HS’
 injection 
nLS+nHS
Thy EN nHS id(0)
&
INV-1
pulse EN 0
time

Fig. 4.10: Transition from two-winding operation to one-winding and vice versa

144
medium/heavy-duty trucks, military vehicles and even passenger cars are not affected
by the short time torque dip due to their large inertia. The duration of the control
transition depends on the current controller bandwidth.
A flow diagram of the proposed transition process from two winding to one winding
and vice versa is provided in Fig. 4.11. It should be noted that during two to one
winding changeover, a delay is provided between disabling the inverter pulses as well
as thyristor pulses and the change of HS winding current references. Similarly, during
one to two winding changeover, a delay is introduced between enabling the inverter
pulses as well as thyristor pulses and the change of HS winding current references.

Start

Measure ωr

Yes If
accelerating && ωr crosses
ω21-chg

Initiate changeover No
2 to 1 winding If No
deccelerating && ωr crosses
ω12-chg
iq-LS-ref =id-LS-ref =0 Yes
iq-HS-ref =0 Initiate changeover 1 to 2 winding
id-HS-ref = nLSn+n
HS
HS
id (0)
Assign
Wait for currents iq-HS-ref =0, id-HS-ref = nLSn+n HS
id (0)
HS
to settle

Disable inv-1 and Wait for currents to settle

thy gate pulses
Enable inv-1 and thy gate pulses
Delay
Select ‘id − iq LS-HS LUT’
Select ‘id − iq HS LUT’ to obtain id-LS-ref , iq-LS-ref ,iq-HS-ref
to obtain Delay
id-HS-ref & iq-HS-ref
Select ‘id − iq LS-HS LUT’
changeover 2 to 1 to obtain id-HS-ref
winding complete
changeover 1 to 2
winding complete

Fig. 4.11: Flow diagram of winding transition from two-winding operation to one-
winding and vice versa for the proposed uneq0-IPMSM

145
This ensures that the high negative current reference of the HS winding is not removed
before the transition process is completed.
The uneq0-IPMSM with parameters given in Table. 4.4 is simulated in
MATLAB/Simulink to evaluate the proposed control, as shown in Fig. 4.12. The
motor speed is ramped up from 100 rpm to 2570 rpm and back to 100 rpm by operating
the load IM in speed control and the uneq0-IPMSM in torque control (which is set to
achieve maximum developed torque at a given speed). The field weakening of the LS
winding is maintained during the winding transition from two to one winding and vice
versa by injecting a high negative d-axis current in the HS winding, as shown in the
inset of Fig. 4.12c. A smooth turn-on and turn-off with minimum current overshoot are
achieved in the LS winding, as observed from the LS winding phase current waveform
given in the inset of Fig. 4.12e.
A basic understanding of IPMSM design for obtaining a specified torque-speed
characteristics is required to appreciate the proposed solution. The variation of the
shape of torque-speed and power-speed characteristics of a conventional three-phase
IPMSM with motor parameters is better understood by representing the location of the
design parameters on a parametric plane. This would give a good insight for machine
designers to understand how the required characteristics can be achieved by varying
the machine parameters to satisfy the load specifications. A detailed discussion of the
parametric plane used for PMSM design is provided in the following section.

4.5 PARAMETRIC PLANE FOR PMSM DESIGN

The torque-speed characteristics of the motor shown in Fig. 4.1 are ideal curves with
a constant torque region followed by a constant power region (hyperbolic torque
curve). However, the shape of the torque-speed characteristics of an IPMSM is
dependent on motor parameters and hence will deviate from the ideal curve. The shape
of the torque-speed and power-speed characteristics of a Surface Mount Permanent
Magnet Synchronous Motors (SMPMSMs), Interior Permanent Magnet Synchronous

146
 3000
2500 Two winding Two winding

speed (rpm)
operation operation
2000
(LS-HS) (LS-HS)
1500
One winding
1000
operation
500 (HS alone)
0
0 5 10 15 20 25 30 35
time (s)
(a) Speed waveform
60 8
6
50 4
Torque (Nm)

2 8
40 0 6
-2 4
30 7.28 7.3 7.32
2
20 0
10 27.7 27.71 27.72
0
0 5 10 15 20 25 30 35
time (s)
(b) Torque waveforms
20
10
Current (A)

0
-10
0 0
-20
-30 -20 -20
-40 -40 -40
7.28 7.3 7.32 27.7 27.71 27.72
-50
0 5 10 15 20 25 30 35
time (s)
(c) d-axis current waveforms
15
12.5
Current (A)

10
15 15
7.5 10 10
5 5 5
0 0
2.5
7.28 7.3 7.32 27.7 27.71 27.72
0
0 5 10 15 20 25 30 35
time (s)
(d) q -axis current waveforms
30 10 10
20 0 0
Current (A)

10 -10 -10
7.28 7.3 7.32 27.7 27.71 27.72
0
-10
-20
0 5 10 15 20 25 30 35
time (s)
(e) LS winding phase current waveform

Fig. 4.12: Simulation results of uneq0-IPMSM with proposed changeover during speed
ramp-up and ramp-down
147
 30 20 20
0 0
20 -20 -20

Current (A)
10 7.28 7.3 7.32 27.7 27.71 27.72

0
-10
-20
0 5 10 15 20 25 30 35
time (s)
(f) HS winding phase current waveform

Fig. 4.12: Simulation results of uneq0-IPMSM with proposed changeover during speed
ramp-up and ramp-down

Motors (IPMSMs) and Synchronous Reluctance Motors (SynRMs) can be completely

characterised using two independent variables; normalised flux linkage (ΨF _n ) and
saliency ratio (ξ) (Soong and Miller, 1993b; Soong and Ertugrul, 2002). The
normalisation is performed by choosing the inverter ratings and motor rated speed as
the base values. The normalised maximum inverter space vector voltage magnitude
(Vlim_n ), normalised rated current vector magnitude (Ilim_n ) and normalised motor base
speed (ωb_n ) are given by

Vlim_n = 1 pu (4.29)

Ilim_n = 1 pu (4.30)

ωb_n = 1 pu (4.31)

It should be noted that even for the same inverter ratings, the rated speed of different
motors would be different. Hence, the base values used for normalisation should also
be changed for different motors to achieve (4.31), in order to represent them in the
common parameter plane.
The operating characteristics like maximum operating speed, constant power speed
range, minimum magnet operating point in B-H curve, and power factor at base speed
are decided by the location of the PMSM design parameters in the parameter plane
(ΨF _n −ξ plane). The parameter plane with locations of various IPMSM designs is

148
 1.2
Design-A
Design-B
1 Design-C
Design-D
0.8

Torque (pu) 0.6

0.4

0.2

0
0 1 2 3 4 5 6
speed (pu)
(a) Torque-speed characteristics
1.2

1
Design-A
0.8 Design-B
Power (pu)

Design-C
0.6 Design-D

0.4

0.2

0
0 1 2 3 4 5 6
speed (pu)
(b) Power-speed characteristics

Fig. 4.14: Variation of power-speed and torque-speed characteristics for various design
on IPM parameter plane

rating, as shown in Fig. 4.14a. This is due to the lower power factor of such machines
during the below base speed operation. Hence, low CPSR IPMSMs (eg: Design-D in
Fig. 4.13) are better suited to extract high output torque with the same inverter due to
their higher power factor in the low-speed region. However, for applications requiring
a wide operating speed range, the location of the machine on the parameter plane has
to be chosen near the optimal FW IPM design line, compromising the low speed power
factor. Else, an overrated low CPSR IPMSM (say Design-D) with either a high inverter
DC bus voltage or an increased inverter switch current rating needs to be used to achieve

150
operation. However, as the operating speed increases, the improvement in output power
and developed torque decreases. Finally, the improvement becomes insignificant at
high-speed operation. The IPM designs located on the ‘optimal FW IPM design line’
with a higher saliency ratio (say ξ = 10) have a better overload performance, as
observed from Fig. 4.15. Hence, a high saliency ratio is necessary for the IPM designs
closer to the ‘optimal FW IPM design line’ to achieve satisfactory overload capability in
the entire speed range. It should be noted that the saliency ratio of practical IPM designs
is limited due to manufacturing complexity and mechanical stability. Saliency ratio in
the range of 7-8 is obtained with axially laminated IPMSMs, which have a complex
manufacturing process compared to radially laminated IPMSMs.
The overload capability of the IPM Design-C, which is located away from the ‘optimal
FW IPM design line’ is shown in Fig. 4.16a and Fig. 4.16b for comparison. It is
observed that the power capability as well as torque capability of Design-C is improved
throughout the operating speed range at twice the rated current. Further, the operating
speed range is also enhanced during overload compared to the rated current operation.
Thus, even though designs that are away from the ‘optimal FW IPM design line’
have lower CPSR and lower maximum speed of operation, they have a good overload
performance throughout the operating speed range. Hence, the machine designer should
decide on the location of the design on the parameter plane to achieve the required
performance at the rated current as well as overload condition, depending on the
application requirement.

4.6 COMPARISON OF THE PROPOSED UNEQ0-PMSM WITH THREE-

PHASE PMSM AND DTP0-PMSM FOR MEDIUM/HEAVY-DUTY
TRUCKS AND MILITARY VEHICLES

Initially, three 3-phase IPMSM design specifications (spec-1, spec-2 and spec-3) with
different CPSR and the same maximum torque (for satisfying high gradability) are
considered, which can be used for medium/heavy-duty trucks and military vehicles.

152
 0.8 Irated Continuous duty 6 Continuous duty
0.7 20 Irated Intermittent duty Intermittent duty
0.6 5
Torque (pu)

Power (pu)
0.5 4
0.4 3
0.3
2 20 Irated
0.2
0.1 1 Irated
0 0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
speed (pu) speed (pu)
(a) Torque-speed characteristics (b) Power-speed characteristics

Fig. 4.18: Characteristics of three-phase IPMSM spec-1 with CPSR 24.7, Vlim = 1 pu

0.8
1.3 Irated Continuous duty 6 Continuous duty
0.7 Intermittent duty Intermittent duty
I
0.6 rated 5
Torque (pu)

Power (pu)

0.5 4
1.3 Irated
0.4 3 I
rated
0.3
2
0.2
0.1 1
0 0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
speed (pu) speed (pu)
(a) Torque-speed characteristics (b) Power-speed characteristics

Fig. 4.19: Characteristics of three-phase IPMSM spec-2 with CPSR 5.87, Vlim = 3.4 pu

0.8 Continuous duty

1.08 I
rated
Continuous duty 6
0.7 Intermittent duty Intermittent duty
Irated
0.6 5
Torque (pu)

Power (pu)

0.5 4
0.4 1.08 Irated
3
0.3 Irated
2
0.2
0.1 1
0 0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
speed (pu) speed (pu)
(a) Torque-speed characteristics (b) Power-speed characteristics

Fig. 4.20: Characteristics of three-phase IPMSM spec-3 with CPSR 4.8, Vlim = 4.47 pu

155
Increasing the power rating of the three-phase IPMSM spec-1 (by increasing the DC
bus voltage or by motor re-rating (Chapman and Krein, 2003)) to satisfy the intermittent
duty requirement will not be a beneficial option compared to other designs, hence not
considered in this thesis. Both the three-phase IPMSM spec-2 and spec-3 can satisfy
all the load requirements for the heavy-duty vehicle, as seen in Fig. 4.19 and Fig. 4.20,
respectively. Here, an overcurrent of 1.3×Irated for the three-phase IPMSM spec-2 and
1.08×Irated for the three-phase IPMSM spec-3 are sufficient to satisfy the intermittent
duty requirements. However, the inverter DC bus voltage needs to be increased to 3.4
times and 4.47 times of spec-1 while using spec-2 and spec-3, respectively. The DC
bus voltage requirement can be reduced to half by using a dual 3-phase PMSM (DTP-
PMSM) configuration with two three-phase inverters. Hence, the DTP-PMSM with
spec-2 and spec-3 has 1.7 times and 2.23 times the DC bus voltage of the three-phase
IPMSM spec-1.
The proposed uneq0-IPMSM can satisfy all the requirements of the military truck
with the same DC bus voltage as three-phase IPMSM spec-1 without any overrating,
as shown in Fig. 4.21 and Fig. 4.22. Fig. 4.21 shows the torque-speed and power-
speed characteristics of the uneq0-IPMSM obtained by reconfiguring the three-phase
IPMSM spec-2 in a 1:12 winding split ratio. Fig. 4.22 shows the characteristics of
the uneq0-IPMSM obtained from the three-phase IPMSM spec-3 reconfiguration in a
1:13 winding split ratio. Here, the winding split ratios are decided to exactly match
the load requirement with minimal overdesign. An overcurrent of 6×Irated is required
for both uneq0-IPMSM spec-2 and spec-3 to satisfy the intermittent operating points.
However, since the overcurrent occurs intermittently, only the IGBTs of the inverter
need to be rated to withstand this short-time high current and not the motor windings.
The cooling provided for continuous operation of the uneq0-IPMSM would be sufficient
for satisfying this short-time overcurrent of the motor windings in most cases. A
comparison of the performance of various IPMSM designs in three-phase, dual three-
phase and uneq0-IPMSM configurations for satisfying the considered military truck

156
 0.8
Continuous duty 6
0.7 Intermittent duty
0.6 5
Torque (pu)

Power (pu)
0.5 4 Continuous duty
Intermittent duty
0.4 6 Irated 3
0.3 6 I
2 rated
0.2 Irated
I
rated
0.1 1
0 0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
speed (pu) speed (pu)
(a) Torque-speed characteristics (b) Power-speed characteristics

Fig. 4.21: Characteristics of uneq0-IPMSM using reconfigured spec-2 with CPSR 5.87,
Vlim = 1 pu

0.8
Continuous duty 6
0.7 Intermittent duty
0.6 5
Torque (pu)

Power (pu)

0.5 4 Continuous duty

Intermittent duty
0.4 3
6 I
rated
0.3 6 Irated
2
0.2 I
Irated
0.1
rated 1
0 0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
speed (pu) speed (pu)
(a) Torque-speed characteristics (b) Power-speed characteristics

Fig. 4.22: Characteristics of uneq0-IPMSM using reconfigured spec-3 with CPSR 4.8,
Vlim = 1 pu

requirement is comprehended in Table. 4.2. The advantages of the proposed uneq0-

IPMSM compared to other topologies are highlighted in grey.

4.7 COMPARISON OF PROPOSED UNEQ0-PMSM WITH EXISTING

WINDING CHANGEOVER TECHNIQUES

Compared to existing winding changeover methods, the proposed uneq0-PMSM

can utilise the field weakening region during the operation of high equivalent
turns configuration. This reduces the large power dip present in the power-speed
characteristics of conventional winding changeover methods. The cause of power

157
Table 4.2: Comparison of various IPMSM designs with three-phase, dual three-phase
and uneq0-IPMSM configurations for military truck

Parameters Spec-1 Spec-2 Spec-3

3-phase 3-phase DTP uneq0- 3-phase DTP uneq0-
PMSM PMSM PMSM IPMSM PMSM PMSM IPMSM
Split ratio 1:1 1:12 1:1 1:13
Requirement satisfied No Yes Yes Yes Yes Yes Yes
DC bus voltage (pu) 1 3.4 1.7 1 4.47 2.23 1
No. 3-phase inverters 1 1 2 2 1 2 2
Over current req. (×Irated ) - 1.3 1.3 6 1.08 1.08 6
Peak power with Irated (pu) 1 3.4 3.4 1.08 4.47 4.47 1.08
Base speed with Irated (pu) 1 3.6 3.6 1.15 4.93 4.93 1.19
Max. operating speed (pu) 35.7 30.49 30.49 26.7 26.52 26.52 27.05

dip in the power-speed characteristics of conventional winding changeover methods

is explained in Section 1.2.13.
In most conventional winding changeover methods, the high equivalent turns winding
is operated in the below base speed region, and the field weakening operation is not
utilised. It is to be noted that some of the methods have demonstrated the field
weakening operation of the high equivalent turns configuration. In such methods, a
transition from the high equivalent turns to low equivalent turns is used to demonstrate
the field weakening operation of the high equivalent turns. However, a reverse transition
from the low equivalent turns to high equivalent turns is not demonstrated in any
literature except by Atiq et al. (2016a). Even though Atiq et al. (2016a) demonstrated
a reverse transition to high equivalent turns, explanation of how the field weakening is
maintained without loss of control during the transition is not provided.
In the proposed uneq0-PMSM, two separate windings (LS and HS winding) are
independently fed from two inverters. Hence, during the reverse transition from ‘low
equivalent turns (HS winding)’ to ‘high equivalent turns (LS winding) operating in
field weakening’, a controlled high negative d-axis current is injected from the HS
winding. This would maintain the field weakening and keep the terminal voltage of
the LS winding within the maximum inverter voltage capability during the transition
process. As a result, the proposed uneq0-PMSM achieves a lower power dip in

158
the combined power-speed characteristics compared to the conventional winding
changeover methods.
A detailed comparison of the proposed uneq0-PMSM with the conventional winding
changeover methods is given in Table. 4.3. The table entries where the proposed uneq0-
PMSM gives an advantage compared to any other topology are highlighted in ‘light
grey’. Also, the major drawbacks of the existing configurations are highlighted in ‘dark
grey’ for easy understanding. The major advantages of the proposed method are

(a) Reduced power dip in the power-speed characteristics due to the operation of the
high equivalent turns winding in the field weakening region.

(b) Fault-tolerant operation during an open-circuit fault in any winding set.

(c) Absence of circulating current compared to delta connected topologies.

(d) Improved degree of design freedom for the machine designer to choose the
required torque-speed characteristics by changing the winding split ratio. Hence,
the proposed method is highly suitable to achieve the high initial starting
torque and wide operating speed range requirement of military vehicles and
medium/heavy-duty trucks.

(e) Smooth reverse transition from the ‘low equivalent turns’ to the ‘high
equivalent turns operating in field weakening’, without resulting in uncontrolled
regeneration.

In winding changeover methods, each bidirectional switch can be realised using two
controlled switches (‘two IGBTs and two diodes’ or two thyristors). To have a one-
to-one comparison of different winding topologies, each IGBT is considered as one
equivalent switch and each bidirectional switch is considered as two equivalent switches
in all the configurations. The total number of equivalent switches in a configuration is
indicative of the system cost and hence can be used for a cost comparison of various
winding changeover configurations.
From the Table. 4.3, it is observed that a lower number of equivalent switches is
required by Kume and Sawa (1995) [68] (14 equivalent switches), Im and Gu (2020)
[56] (12 equivalent switches), and Swamy et al. (2006) [121] (8 IGBTs, 2 diode
bridges and 4 diodes) compared to 16 equivalent switches in the proposed configuration.

159
Table 4.3: Comparison of proposed uneq0-PMSM and existing winding changeover
methods
Winding Ref No. Inv. type No. Total Tmax ratio Pcu-2 FW1 Winding Circ. Var. rev.
changeover IGBT (no.) bi-dir eq. sw (ωb ratio) present OC fault current T −ω trans. to
topology switches tolerance chara high eq.
[69][128] 6 3-leg (1) 6 18 1:0.58
Star-delta 0.58 No Yes Yes No -
[68] 14 3-leg (2) 0 14 (1:1.73)
Series 1:0.5
[146][53] 6 3-leg (1) 9 24 0.25 No No No No -
-parallel (1:2)
[67] 6 6 18
1:x
No
Tapped [121] 8 3-leg (1) Nil - x No No Yes -
1: x1
[56] 12 Nil 12 ND
Combined
1:0.58:x:0.58x

tapped [22][51] 6 3-leg (1) 12 30 0.58 No Yes Yes Yes -
1:1.73: x1 : 1.73
x
and star-delta
1:0.851:0.525
Five-phase [105] 10 5-leg (1) 15 40 0.851 Yes Yes Yes No No
(1:1.17:1.9)
Series-end 1:0.58
[76] 8 4-leg (1) 4 16 0.58 ND Yes No No -
winding (1:2)
[5] 1:0.58 ND
OEW PMSM 12 3-leg (2) 2 16 1 Yes Yes No -
[4] (1:1.73) Yes
1:0.268
[111] 6 18 No - -
Cumulative- (1:3.73)
6 3-leg (1) 1 No No
diff mode [110] 12 30 1:0.58:0.27:0.16
Yes
(1:1.73:3.73:6.46)
uneq0 1:x

12 3-leg (2) 2 16 x Yes Yes No Yes Yes
-PMSM 1 : x1
ND : Not demonstrated
Pcu2 : Copper loss during low equivalent turns operation in PU
Var. T −ω chara : Ability to change the ratio of maximum torque for each winding configuration and ratio of base speeds by
varying the winding turns ratio during design stage
FW1 present : Ability to operate the high equivalent turns in field weakening region before changeover to low equivalent turns
rev. trans. to high eq. : Ability for reverse transition from ‘low eq. turns winding’ to ‘high eq. turns winding in FW region’ without
uncontrolled regeneration
Circ. current : Circulating current during delta connected operation due to triplen harmonic in back-emf

However, none of these methods can achieve all the benefits of the proposed uneq0-
PMSM, including winding open-circuit fault tolerance, reduced power dip in the
combined power-speed characteristics and variable torque-speed characteristics during
design. Even though the proposed method requires 16 equivalent switches which would
increase the inverter cost and require additional gate drivers, it will not be significant
considering the improved reliability, additional power capability and increased design
freedom for critical applications like military vehicles and medium/heavy-duty trucks.
Each of the existing winding changeover techniques given in Table. 4.3 is considered,
and a detailed comparison with the proposed uneq0-PMSM is provided as follows.

(a) Star-delta winding changeover

In the star-delta winding changeover, the low-speed operation is performed using
a star connection, and the high-speed operation is performed using a delta
connection. The delta connection can operate even during a phase winding open-
circuit fault. However, circulating current may flow during delta connection
due to the triplen harmonic components in the back-emf. Here, the maximum
developed torque ratio for each winding configuration (Tmax1 : Tmax2 = 1 : 0.577)
and the ratio of base speeds of operation (ωb1 : ωb2 = 1 : 1.732) are fixed
and cannot be varied during the design to suit the application requiring a very

160
 wide operating speed range. Also, a large power dip is present in the combined
power-speed characteristics, as explained in the previous section. The star-
delta configuration by Kume et al. (1990) [69] and Wang et al. (2010) [128] is
realised using 6 IGBTs and 6 bidirectional switches (18 eq switches), whereas,
Kume and Sawa (1995) [68] use only 14 equivalent switches. Even though
the proposed uneq0-PMSM requires 16 eq switches, it is devoid of circulating
currents, achieves variable torque-speed characteristics during design and has a
reduced power dip in the combined power-speed characteristics in contrast to the
star-delta winding changeover techniques.
(b) Series-parallel winding changeover
In series-parallel winding changeover (Zhitkova and Hameyer, 2016 [146] Huang
and Chang, 1999 [53]), the low-speed operation is performed with a series
connection and the high-speed operation is performed with a parallel connection.
The series-parallel configuration has the lowest copper loss during high-speed
operation since the number of turns is reduced, and the entire winding is
utilised for sharing the current. However, features like winding open-circuit
fault ride through and field weakening during high equivalent turns operation
are not present. Here, the maximum developed torque ratio for each winding
configuration (Tmax1 : Tmax2 = 1 : 0.5) and the ratio of base speeds of operation
(ωb1 : ωb2 = 1 : 2) are fixed and cannot be varied during the design to suit
the application. This configuration is realised using 6 IGBTs and 9 bidirectional
switches, constituting a total of 24 eq. switches. The proposed method uses only
16 eq. switches and has additional features like winding open-circuit fault ride
through, field weakening of high equivalent turns and can have variable torque-
speed characteristics during design.
(c) Tapped winding changeover
Both the tapped winding changeover and the proposed method have similar
resultant torque-speed characteristics and can have variable characteristics during
design. In both methods, a part of winding with high equivalent turns is cut
off during high-speed operation. The tapped winding changeover by Kume
et al. (1991) [67] requires 6 IGBTs and 6 bidirectional switches, constituting
a total of 18 equivalent switches compared to 16 equivalent switches in the
proposed uneq0-PMSM. A tapped winding configuration with a reduced number
of controllable switches (8 IGBTs, two three-phase diode bridges, 4 diodes) is
proposed by Swamy et al. (2006) [121]. Here, the rise in terminal voltage of
unused winding during high-speed operation is limited by shifting the neutral
point from one end to the middle part of the winding. Im and Gu (2020) [56] have
proposed a method to eliminate the snubber used by Swamy et al. (2006) [121],
and uses only 12 equivalent switches. In all these methods, the field weakening
operation of the high equivalent turns winding is not demonstrated, resulting in a
huge power dip in the combined power-speed characteristics. Compared to tapped
winding changeover methods, the proposed uneq0-PMSM can achieve open-
circuit winding fault ride though and have reduced power dip in the combined
power-speed characteristics.
(d) Combined tapped and star-delta winding changeover

161
 Combined tapped and star-delta winding changeover (Chen and Cheng, 2006
[22];Hsieh et al., 2012 [51]) is used to achieve four different winding
configurations with a maximum developed torque ratio of 1 : 0.577 : x : 0.577x
and base speed ratio of 1 : 1.732 : x1 : 1.732 x
. Here, the value of ‘x’ ranges
from ‘0’ to ‘1’ depending on the point of winding tapping. In this method, the
winding is connected in long-star, long-delta, short-star and short-delta to achieve
a wide operating speed range. However, the method requires 6 IGBTs and 12
bidirectional switches, constituting a total of 30 equivalent switches compared to
16 equivalent switches in the proposed uneq0-PMSM. Also, circulating current
would be present during short-delta and long-delta connections, which is absent
in the proposed method. Further, the field weakening of high equivalent turns is
not possible in the ‘combined tapped and star-delta configuration’, resulting in
a large power dip in the combined power-speed characteristics in contrast to the
proposed uneq0-PMSM.
(e) Five-phase winding changeover
Sadeghi et al. (2012) [105] has proposed a winding reconfiguration of a five-
phase machine that can be connected in either star, pentagon or pentacle
configurations. Here, the low-speed operation is performed in star connection
followed by pentagon followed by pentacle in the high-speed operation to achieve
the maximum developed torque ratio of 1:0.851:0.525 and base speed ratio of
1:1.175:1.902. However, 10 IGBTs and 15 bidirectional switches constituting a
total of 40 equivalent switches are required in this configuration compared to 16
equivalent switches in the proposed uneq0-PMSM. Sadeghi et al. (2012) [105]
has also shown the field weakening operation of high equivalent turns winding.
However, a reverse transition from low equivalent turns to high equivalent turns
is not demonstrated, which would have resulted in uncontrolled regeneration.
Compared to the five-phase machine with winding changeover, the proposed
uneq0-PMSM can achieve variable torque-speed characteristics during design
as well as a smooth reverse transition from the ‘low equivalent turns’ operation
to the ‘high equivalent turns operating in field weakening’ without uncontrolled
regeneration.
(f) Series-end winding configuration
The series-end winding PMSM (Li et al., 2022 [76]) uses a 4-leg inverter where
each phase winding is connected between the inverter poles. This configuration
has the same total equivalent switches (16 eq. switches) as the proposed method
since there are 8 IGBTs and 4 bidirectional switches. The resulting characteristics
of series-end winding have a maximum torque ratio of 1:0.577 and a base speed
ratio of 1:2. It is to be noted that variable characteristics during design cannot be
achieved in this method in contrast to the proposed uneq0-PMSM.
(g) OEW-PMSM configuration
An open-end winding PMSM with four different models is proposed by Atiq
et al. (2016b) [5] to achieve a wide operating speed range. In this configuration,
the motor can be connected in a cumulative mode for low-speed operation and
a differential mode for high-speed operation. Similar to the proposed uneq0-
PMSM, the OEW-PMSM reconfiguration also requires two inverters (12 IGBTs)

162
 and two bidirectional switches, constituting a total of 16 equivalent switches.
The same concept is also explored by Atiq et al. (2016a) [4], which includes an
experimental demonstration of the changeover from the low equivalent turns to
the high equivalent turns in the field weakening region. However, an isolated
DC bus is required for the OEW-PMSM compared to a common DC bus in the
proposed uneq0-PMSM. A major drawback of this method compared to other
winding changeover methods is the requirement of rated current injection through
the whole winding, even with the configuration for high-speed operation. This
would result in higher copper loss during high-speed operation compared to all
other winding changeover methods. Consequently, the proposed uneq0-PMSM
has superior performance compared to the OEW-PMSM winding changeover
configuration in all respects.

(h) Cumulative-differential mode based winding changeover

Sin et al. (2020b) [111] has demonstrated a transition between the cumulative
and differential mode using a single inverter (6 IGBTs) and 6 bidirectional
switches constituting a total of 18 equivalent switches. A maximum developed
torque ratio of 1:0.268 and a base speed ratio of 1:3.73 is achieved in this
method. Sin et al. (2020a) [110] combines the cumulative-differential mode
with star-delta winding switch over, by using 30 equivalent switches. Here, a
maximum developed torque ratio of 1:0.577:0.268:0.155 and a base speed ratio
of 1:1.732:3.73:6.46 is achieved. However, the drawbacks of the OEW-PMSM
configuration are also present in these methods also, including the increased
copper loss, presence of a large power dip in the power-speed characteristics,
and lack of winding open-circuit fault ride-through. Hence, the proposed
uneq0-PMSM has superior performance compared to the cumulative-differential
winding changeover configuration in all aspects.

4.8 EXPERIMENTAL RESULTS

The proposed concept is validated on a 1.5 kW uneq0-IPMSM with a 1:3 split ratio
with parameters given in Table. 4.4. The machine is also reconfigured as a 2 kW
DTP0-IPMSM by splitting the windings equally to compare the performance. The

Table 4.4: Parameters of experimental uneq0-IPMSM prototype

Parameter Value Parameter Value Parameter Value

Prated 1.5 kW Rs-LS 4.874 Ω Rs-HS 1.625 Ω
Vdc 200 V ΨF -LS 0.8684 Wb ΨF -HS 0.2894 Wb
Poles 4 Lds-LS-unsat 54 mH Lds-HS-unsat 6 mH
Irated 6A Lqs-LS-unsat 101.54 mH Lqs-HS-unsat 11.284 mH
ωHS-max 2570 rpm Lds-LS-sat 26.1 mH Lds-HS-sat 2.9 mH
Tmax 32 Nm Lqs-LS-sat 94.5 mH Lqs-HS-sat 10.496 mH

163
experimental setup consists of two IGBT based 2-level VSIs, back-to-back thyristor
switches and an IM coupled to the shaft of the uneq0-IPMSM for loading, as shown in
Fig. 4.23. The control algorithm is implemented on Texas Instruments’ TMS320F28335
DSP with 50 µs sampling time, and the switching frequency is selected as 5 kHz. The
uneq0-IPMSM is operated in torque control, whereas the IM is operated with speed
control.
In this thesis, an equivalent current (Ieq ) is defined to represent the contribution of
current flowing through active conductors (conductors with current) out of the total
conductors in the machine. Ieq is given in terms of stator phase current RMS (Iph-RMS )
as
Conducting turns per phase
Ieq = Iph-RMS × . (4.32)
Total turns per phase

Hence, for the DTP0-IPMSM and the uneq0-IPMSM during two-winding operation,
Ieq is equal to Iph-RMS . However, when the HS winding of the uneq0-IPMSM with 1:3

Fig. 4.23: Experimental setup of uneq0-IPMSM fed from two inverters with back-to-
back thyristor switches

164
split ratio is operated alone, Ieq = 14 Iph-RMS .
Fig. 4.24a and Fig. 4.24b show the equivalent current contours plotted in the torque-
speed and power-speed characteristics of the DTP0-IPMSM. The machine has a rated
torque of 32 Nm, 2 kW at 800 rpm and a maximum operating speed of 1500 rpm. It
is observed that a higher current is necessary to operate during high-speed operation
(beyond 800 rpm) compared to low-speed operation to achieve the same developed
torque. This higher current demand is due to the additional d-axis current demanded
to maintain the terminal voltage within inverter limits by field weakening. The
torque-speed and power-speed characteristics when the same three-phase IPMSM is
reconfigured as an uneq0-IPMSM with a 1:3 split ratio are shown in Fig. 4.25a

35 Two winding 2.25 Two winding

operation 5 operation 5
30 2
5.5
5. 5

Equivalent current (A)

1.75 5

Equivalent current (A)

5

25 4.5 4 4
1.5 . 5
Torque (Nm)

4
Power (kW)

4
3 3. 5.5
4

20
5.5 5
3 1.25 3
4.545
5
5.4.54

3.3
High I at
3.

15 3 1 eq
55

High I at 2.5
53

eq 2
.5 4 4.5

2.5 2 high-speed 2
high-speed 0.75
10 2
2 25 5

1.5 0.5
5..
3. 55.

1 5
4 54

1
5..5

1.
1.51
3..53

2.52
3.35
2 5

5 1
4

5 5
1. 1

0.25
55
5

0.5 1 0.5
2
4

0 0
0 500 1000 1500 2000 2500 0 500 1000 1500 2000 2500
speed (rpm) speed (rpm)
(a) Torque-speed characteristics (b) Power-speed characteristics

Fig. 4.24: Experimental results of DTP0-IPMSM with equivalent current contours

35 Two winding One winding (HS) 2.25
Two winding One winding (HS)
5.5 (LS-HS) operation 5 (LS-HS) 5
30 operation 2 operation
5 operation
Equivalent current (A)

1.75
Equivalent current (A)

25 4 Low I at 4
4.5 eq
1.5
Torque (Nm)

Power (kW)

high-speed
5.5 4

4
20 3 1.25 5.5
3
5 4.

Low I at
3. 3

1
5 .5

eq 1
4

15 3.5
5

high-speed 2 3
2 2. .5

5.5 4 3..5523

2.5 2
45

0.75 1
5

10 2
4
5.5 3..55

334.5 5.

1.5
2.5
3.5 24
53

1 0.5
0.5
1 4.

1
2 .51

0.
1
1. 1

4.55
2 52

5
1

1
2
143

0.25
5

0.5 0.5 1 .5
3

0.5 0.5
0 0 11
0 500 1000 1500 2000 2500 0 500 1000 1500 2000 2500
speed (rpm) speed (rpm)
(a) Torque-speed characteristics (b) Power-speed characteristics

Fig. 4.25: Experimental results of proposed uneq0-IPMSM with equivalent current

contours

165
and Fig. 4.25b, respectively. Here, the torque-speed and power-speed characteristics
have different shapes than conventional three-phase or DTP0-IPMSM, as explained
in Section 4.3.1. Even though the initial torque-speed region appears similar to that
of DTP0-IPMSM, the operating region is further extended by changing the winding
configuration by isolating the LS winding and operating the HS winding alone. The
one-winding operation is characterised by a wide below base speed operation with
constant torque, followed by a short FW region, as observed in Fig. 4.25a. Thus,
the resulting power-speed characteristic has two peak power regions during low-speed
and high-speed compared to single peak power in conventional three-phase or DTP0-
IPMSM. Also, the peak power rating of the uneq0-IPMSM has reduced to 1.5 kW
compared to 2 kW in DTP0-IPMSM. However, the uneq0-IPMSM achieves an extended
operating range up to 2570 rpm compared to 1500 rpm for DTP0-IPMSM. Further, in
the uneq0-IPMSM, since the operation with the HS winding alone does not necessitate
field weakening for a wide speed range, the equivalent current in the uneq0-IPMSM
is highly reduced compared to DTP0-IPMSM in high-speed operation, as seen from
Fig. 4.25a and Fig. 4.25b.
Fig. 4.26 and Fig. 4.27 show variation of estimated developed torque (Tb e ), motor speed
(ωm ) and phase currents of the DTP0-IPMSM and the uneq0-IPMSM with 1:3 split,
respectively. In both machines, equal currents are injected into the two windings to
achieve maximum developed torque. For the DTP0-IPMSM, the developed torque
decreases as the motor speed increases beyond the base speed due to field weakening, as
seen in Fig. 4.26. Even with rated current, the developed torque falls to nearly zero when
the motor speed reaches 1500 rpm, limiting the DTP0-IPMSM operation. However, in
the proposed uneq0-IPMSM, the operating speed range is extended to 2570 rpm with the
winding changeover. The developed torque decreases with speed after the base speed of
two-winding operation in the uneq0-IPMSM, similar to the DTP0-IPMSM. However,
after 990 rpm, the LS winding is disconnected from the inverter-1. Further, the HS
winding is operated alone, with the developed torque maintained constant till the base

166
Fig. 4.26: Performance of DTP0-IPMSM during speed increase and decrease at near
full load:- Ch1: ωm ; Ch2: ia1 ; Ch3: Tb e ; Ch4: ia2 ; (X-axis:- 3.8 s/div; Y-
axis:- Ch1: 960 rpm/div; Ch2,Ch4: 10 A/div; Ch3: 7.49 Nm/div)

Fig. 4.27: Performance of uneq0-IPMSM during speed increase at near full load with
winding changeover:- Ch1: ωm ; Ch2: ia-LS ; Ch3: Tb e ; Ch4: ia-HS ; (X-axis:-
1.96 s/div; Zoom X-axis:-20 ms/div; Y-axis:- Ch1: 690 rpm/div; Ch2,Ch4:
12 A/div; Ch3: 6.81 Nm/div)

speed of the HS winding. Thus, the extension of torque-speed characteristics and wide
below base speed operating region of uneq0-IPMSM is demonstrated experimentally.
A smooth transition from two-winding operation to one-winding in Fig. 4.27 is
achieved by injecting a high negative d-axis current (30.8 A) in the HS winding for

167
a short duration (5 ms), as explained in Section 4.4.4. The d-axis current in the HS
winding for a smooth transition is calculated as 4×7.7 A, where 7.7 A corresponds to the
minimum value required to maintain field weakening with zero developed torque while
both windings were operating together. The additional d-axis current is reflected in
the phase current (14.5 A-RMS) as observed from the zoomed waveforms in Fig. 4.27.
The additional d-axis current increases with a higher difference in the winding turns
between the LS and HS windings. Hence, the current and thermal rating of the inverter
switches should be designed to deliver this additional current in the HS winding for a
short duration during the winding transition. In heavy-duty transport trucks and military
vehicles, the inverter would necessarily have a high short-time rating for satisfying the
intermittent load requirements. Hence, the high short-time current rating of inverter
switches for implementing the winding transition with the proposed method will not be
of much concern for such applications.
Variation of the d-axis current of the LS and HS windings during speed increase and
decrease at a constant load of 2.27 Nm are shown in Fig. 4.28 and Fig. 4.29. From
Fig. 4.28, it is observed that the d-axis current increases in both LS and HS winding
as the machine speed increases beyond the base speed of the combined LS-HS winding
operation due to field weakening. However, once the operation changes to one-winding
at 990 rpm, the d-axis current demand decreases drastically since the machine operates
in the below base speed region with only the HS winding activated. The winding
currents of the LS winding become zero, as seen in Fig. 4.28, after it is disconnected
from the inverter-1 using the thyristor switches. As the machine speed increases further,
FW is initiated in the HS winding by injecting a higher d-axis current to reach the final
speed of 2570 rpm. Injection of additional negative d-axis current in the HS winding
during the winding changeover is observed in the zoomed waveforms of Fig. 4.28 and
Fig. 4.29. A seamless transition from the two-winding to one-winding operation and
vice versa with controlled currents is achieved in the proposed uneq0-IPMSM, as seen
from the zoomed waveforms in Fig. 4.28 and Fig. 4.29. The steady-state operation of

168
Fig. 4.28: Performance of uneq0-IPMSM during speed increase at 2.27 Nm load with
winding changeover:- Ch1: id-LS ; Ch2: ia-LS ; Ch3: id-HS ; Ch4: ia-HS ; (X-
axis:- 1.84 s/div; Zoom X-axis:-20 ms/div; Y-axis:- Ch1,Ch3: 10.18 A/div;
Ch2,Ch4: 10 A/div)

Fig. 4.29: Performance of uneq0-IPMSM during speed decrease at 2.27 Nm load with
winding changeover:- Ch1: id-LS ; Ch2: ia-LS ; Ch3: id-HS ; Ch4: ia-HS ; (X-
axis:- 1.88 s/div; Zoom X-axis:-20 ms/div; Y-axis:- Ch1,Ch3: 10.18 A/div;
Ch2,Ch4: 10 A/div)

the uneq0-IPMSM at 990 rpm at full load is shown in Fig. 4.30. At a given speed of
operation, the modulation index of the inverter supplying the LS winding is nearly three
times compared to the inverter supplying the HS winding since the winding split ratio

169
Fig. 4.30: Steady-state operation of uneq0-IPMSM at 990 rpm under full load:- Ch1:
ma-LS ; Ch2: ia-LS ; Ch3: ma-HS ; Ch4: ia-HS ; (X-axis:- 20 ms/div; Y-axis:-
Ch1,Ch3: 1 pu/div; Ch2,Ch4: 5 A/div)

of the uneq0-IPMSM is 1:3. Also, the phase currents of the LS and HS windings are
equal in magnitude and phase and are nice sinusoids without any significant lower-order
harmonics.

4.9 CONCLUSION

In this chapter, an unequal split 0◦ winding displacement dual 3-phase PMSM (uneq0-
PMSM) operated with two inverters having a common DC bus and two back-to-
back thyristor switches was proposed for improved performance of medium/heavy-
duty trucks and military vehicles. The uneq0-PMSM has a better overload capability
than the high CPSR three-phase PMSM and a reduced power rating than the low
CPSR three-phase PMSM for satisfying the same load requirement. Further, the
torque-speed characteristics shape of the uneq0-PMSM can be modified by varying
the winding split ratio to suit the application. The wide speed range of the uneq0-
PMSM with below base speed operation in most operating points also reduces the
additional d-axis current requirement compared to field weakening in conventional
three-phase PMSM. A transition algorithm for smooth winding changeover between

170
the two-winding (combined LS-HS winding) and one-winding operation (HS winding
alone) was also proposed. The proposed concepts were experimentally validated on a
1.5 kW uneq0-IPMSM with a 1:3 winding split ratio.
In the next chapter, the design of a dual three-phase PMSM for a commercial EV is
discussed. An iterative design process is used to achieve the desired motor performance
and hence satisfy all the specifications of the given EV.

171
 CHAPTER 5

MACHINE DESIGN CONSIDERATIONS OF DUAL 3-PHASE

PMSM FOR ELECTRIC AUTOMOTIVE

5.1 INTRODUCTION

The motor design for electric vehicles is different in some aspects compared to the
design for typical applications like fan and pump. They need to be designed to satisfy
various performance requirements like gradability, acceleration time, top speed and
driving range. In this chapter, the guidelines for designing a DTP0-IPMSM (discussed
in Chapter 3) to suit the requirements of an electric vehicle are explained. Further,
most of the design steps explained here are also applicable to three-phase IPMSM
design. It is to be noted that the mechanical structure of the uneq0-IPMSM (discussed
in Chapter 4) is exactly same as the DTP0-IPMSM, except for the difference in winding
split ratio. Even though the design of three-phase IPMSM (Momen et al., 2016)
and split-phase IPMSM (Yang et al., 2019) for electric vehicles are discussed in the
literature, the design of DTP0-IPMSM is not well addressed. In this chapter, the design
of a DTP0-IPMSM suitable for a commercial BEV: Mahindra e-Verito (e-Verito) is
explained. The parameters of the motor under design are derived considering various
vehicle requirements. The next section discusses the procedure of formulating a torque-
speed characteristics to satisfy all the performance requirements of an electric vehicle.

5.2 FORMATION OF TORQUE-SPEED CHARACTERISTICS

CONSIDERING AN ELECTRIC VEHICLE SPECIFICATION

The performance specifications of an EV consist of

• Gradability (Degrees) - Ability of the vehicle to climb a slope road

• Top speed (km/hr) - Maximum operating speed of the vehicle

• Acceleration time requirement (s) - Minimum time required to change the speed
from 0 km/hr to ‘x’ km/hr
 • Driving range (km) - Driving distance of the vehicle using a fully charged battery

The electric motor used in EVs should have torque-speed characteristics with an
efficiency map that satisfies these performance specifications. Here, a dual three-
phase IPMSM is designed considering the specifications of a commercial battery
electric vehicle: Mahindra e-Verito. The performance specifications and parameters
of Mahindra e-Verito are given in Table. 5.1. Torque-speed characteristics of motors
with different ratings which are suitable to meet the specifications of Mahindra e-Verito
are given in Fig. 5.1. The power rating of the EV motor is majorly decided by the
acceleration torque requirement (Tacc in Fig. 5.1). To achieve a given acceleration time,
the motor is either designed as a low power, high torque motor or a high power, low
torque motor. The curves Tacc1 , Tacc2 , .. Tacc5 represent the torque-speed characteristics
that can satisfy the requirement of 0-60 km/hr in 10.7s, considering the base speed of the
vehicle at 10 km/hr, 20 km/hr, . . . 50 km/hr, respectively. The characteristics with a low
base speed have a higher torque rating and low power rating. One of the characteristics
among Tacc1 , Tacc2 , .. Tacc5 or any other characteristics that satisfy the acceleration
requirement can be chosen for the motor design. However, they should also meet the
minimum torque requirement for the gradability specification in the low-speed region
(Tgrad ) and traction requirement for a flat road (Tflat ) at the maximum operating speed.
Here, Tacc1 , Tacc2 , and Tacc3 satisfy all the specifications of the EV given in Table. 5.1.
A higher torque rating design (say Tacc1 ) increases the motor size, whereas a high power
rating design (say Tacc5 ) results in higher converter ratings and costly switches. The

Table 5.1: Specifications and parameters of Mahindra e-Verito (e-Verito; Ehsani et al.,
2018)

Specification Value Parameter Symbol Value

Gradability 9.7◦ Vehicle mass Mveh 1704 kg
Top speed 86 km/hr Wheel radius rd 0.3073 m
Driving range 110 km Gear ratio mG 10.83
Time from Rolling resistance coefficient fr 0.02
10.7 s
0-60 km/hr Aerodynamic drag coefficient CD 0.3
Vehicle front area Af 2.38 m2
Efficiency of drive line ηt 0.9

173
 T flat T grad T acc1 T acc2 T acc3 T acc4 T acc5
300
27.24kW
250

Torque (Nm)
200 29.08kW

150 32.16kW
36.46kW
100
42kW
50

0
0 1000 2000 3000 4000 5000 6000 7000 8000
speed (rpm)
Fig. 5.1: Torque-speed characteristics representing specifications of Mahindra e-Verito
BEV

high torque rating also has the advantages of higher gradability and higher starting
torque. However, the high torque machine has a low base speed, which demands a large
CPSR to satisfy the maximum operating speed requirement of the vehicle. The CPSR
of commercial electric cars lies in the range of 3 to 7.
A wide CPSR means that the machine is operated in the field weakening mode for most
operating regions. For a given drive cycle, frequent field weakening operations would
deteriorate the operating efficiency due to the additional negative d-axis current for the
same torque requirement. Hence, to improve the efficiency of the vehicle, the base
speed is selected to be a higher value, which results in a larger power rating for the
motor and the converter. A high starting torque is necessary for sports cars, off-road
vehicles and vehicles used in hilly areas. Hence, the selection of torque-speed curve for
a given acceleration requirement is decided considering the road conditions, gradability
requirement, vehicle category, CPSR limitation, and overall efficiency during a driving
cycle. In this work, Tacc3 is selected for the DTP0-IPMSM design to match the
specifications of Mahindra e-Verito. The torque-speed and power-speed requirement
given by Tacc3 was achieved by designing a 37 kW, 111 Nm, 3140 rpm DTP0-IPMSM.
The detailed design process of the DTP0-IPMSM is given in the next section.

174
5.3 DESIGN PROCESS OF A DUAL THREE-PHASE IPMSM FOR EV
APPLICATION

A brief flow diagram for the design of a dual three-phase IPMSM for EV application is
given in Fig. 5.2. Most of the design choices are common to all three-phase machines
(IMs (Lipo, 2017), PMSM (Hendershot and Miller, 2010) and synchronous reluctance
motors), whereas some design parameters are specific to high-speed machine design.
Compared to three-phase IPMSM, only the stator winding structure is different for the
symmetric dual three-phase IPMSM.
An iterative design procedure is used to design the dual three-phase IPMSM to satisfy
the given EV requirement, as shown in Fig. 5.2. Various EV motor requirements
like maximum terminal voltage at base speed, rated torque, torque ripple, CPSR, flux
density at various regions, efficiency, and power factor are evaluated at various stages
of the design. The iterative process is continued until all these specifications are met.
In this process, a proper initial value for various parameters is critical to start the design
process. Mostly, only a fine tuning of these parameters is performed during the iterative
process. Thus, a basic understanding of the effect of various motor dimensions and
other design inputs on motor performance is required to start the design process. In the
next section, a brief overview of the design process and some correlation between the
design inputs and the machine performance is provided. Better designs for the same
requirement are obtained by choosing good initial values for the design and by making
proper design choices, which mostly depend on the experience of the machine designer.
Table. 5.2 shows the parameters of the 37 kW DTP0-IPMSM obtained after the iterative
design process.

5.3.1 Machine ratings and dimensions

The rated torque, rated power (Prated ) and base speed of operation for the IPMSM under
design are obtained from the EV requirement as discussed in the previous section. In
this design, the space constraint in the vehicle is used to obtain the outer dimensions of

175
 Start

Set the machine outer dimensions,

Vdc , Torque, Power, base speed,
and no. of poles

Fix maximum flux density,

linear current density,
conductor current density

Compute stator inner diameter

air-gap, rotor outer diameter

Select magnet type, dimension to

obtain desired air gap flux
(with assumed rib & bridge thickness)
Adjust flux density,
electrical loading Choose winding configuation
and cooling
Fix no. of turns to obtain
desired Vlim at ωb
Adjust flux density,
Compute conductor area, slot, electrical loading
teeth & yoke dimensions and cooling
and SWG of wire

Optimize V angle, bridge thickness

for Trated with min current & Tripple
Adjust PM thickness
or length
If N
N Va (FEM) ≈ Vlim
N If
If
Tooth & yoke Torque, CPSR not satisfied
Y in any iteration Y
Y flux density not satisfied
in any iteration If
Torque, CPSR, N
mech stability satisfied
(FEM)
Adjust slot width/slot pitch
Increase current density Y
If
N
Tooth & yoke flux density
within limits
Redeisn to reduce
the major loss components Y
If
efficiency, power factor
N from FEM satisfactory?

Y
end

Fig. 5.2: Design flow diagram of Dual three-phase IPMSM for EV

176
 Table 5.2: Design parameters of forced-air cooled 37 kW DTP0-IPMSM

Parameter Value Parameter Value

Machine ratings Lamination dimensions
Rated power 37 kW Stator ID 170 mm
Rated torque 111 Nm Stator OD 281 mm
DC bus voltage 200 V Stack length 150 mm
Rated current 80 A Rotor OD 168 mm
Rated speed 3163 rpm Rotor ID 60 mm
Maximum speed 11,134 rpm Air gap 1 mm
Poles 8 Lamination thickness 0.35 mm
Phase resistance (Rs ) 7 mΩ Stator wiring
Magnet flux linkage (ΨF ) 0.0704 Wb Number of slots 48
d-axis inductance (Lds ) 0.2282 mH Turns per coil 2
q-axis inductance (Lqs ) 0.4464 mH Wires in parallel 33
Rotor dimensions Slot depth 35.42 mm
Magnet type NdFeB-N35H Slot opening 2 mm
Single magnet size (mm) 10.5×50×4 Stator tooth width 8.02 mm
Magnet overall size (mm) 21×150×4 Stator tooth-tang height 0.5 mm
Total magnets 96 slot width/slot pitch 0.3
Slit gap (gslit ) 1.65 mm Wire size 21 SWG
Rib thickness (trib ) 2 mm
V angle (αv ) 154◦
Design choices
Peak fund. air-gap flux density 0.67 T
Average air-gap flux density 0.57 T
Average teeth flux density 0.92 T
Average yoke flux density 0.68 T
Linear current density RMS 28.8 kA/m
Slot fill factor 0.58
Current density RMS 4.7 A/mm2

the motor. Thus, the stator outer diameter and the total length of the motor are fixed.
The active length is computed considering the space requirement for the cooling fan,
bearing, end cover and end winding. Forced-air cooling is used in this work.

5.3.2 Number of poles

Commercial electric vehicles with distributed stator winding commonly employ 8 to

12 poles (Sarlioglu et al., 2017; Burress, 2016), whereas the ones with concentrated
winding adopt more number of poles (Staunton et al., 2006). A higher number of
poles results in thinner stator back iron due to more division of magnetic flux paths.

177
This would considerably reduce the total weight of the motor improving its power and
torque density. However, it increases the proximity effect, skin effect and magnet eddy
current loss due to the increase in operating frequency. Further, the inverter switching
frequency limit also needs to be considered for a high-speed design with more poles. In
this design, a V-shaped PM configuration with 8 poles is selected.

5.3.3 DC bus voltage

The DC bus is formed by stacking the batteries in series and parallel combinations.
Even though a high DC bus reduces the motor current for a given rated power, complex
cell balancing circuits become necessary for the series connection of batteries (Einhorn
et al., 2011). Hence, the DC bus voltage across the battery terminals is commonly fixed
in the range of 72 V to 403.2 V, as given in Table. 1.1. A 200 V DC bus is used in this
work.

5.3.4 Stator inner diameter, rotor outer diameter, air-gap flux density and linear
current density

The developed torque of the motor is related to its rotor volume, air-gap flux density
and linear current density. Hence, the stator inner diameter (Ds-in ) is obtained as (5.1)
if all other parameters are fixed (Pyrhönen et al., 2008).

s
Si π2 bm
Ds-in = ; where C 0 = √ kw1 AB (5.1)
C0 ls′ ns 2

where Si is the apparent power, C0 is the machine constant, kw1 is the fundamental
b m is the peak air gap flux density,
winding factor, A is the RMS linear current density, B
ns is the mechanical synchronous speed (in rps), and ls′ is the equivalent active length.
An initial value of peak air-gap flux density and linear current density is chosen based
on experience while also considering the cooling method used. Peak air-gap flux density
is selected based on the material used and fundamental operating frequency. A higher
air-gap flux density reduces the amount of core material but results in higher core loss

178
due to saturation.
The rotor outer diameter (Dr ) is given by (5.2) when the stator inner diameter and air
gap (lg ) are fixed.
Dr = Ds-in − 2lg (5.2)

5.3.5 Air gap

The air gap (lg ) of the machine is calculated using empirical formulas. In this work, the
following relation is used to obtain an initial value of the air gap length by considering
high-speed operation (Pyrhönen et al., 2008).

0.4
0.18 + 0.006Prated
lg = 1.6 m (5.3)
1000

where Prated is the rated power of the machine in Watts.

A small air gap is used to achieve higher magnetising inductance to improve the
reluctance torque. However, this also results in poor power factor and necessitates a
higher terminal voltage. A large air gap reduces the harmonics in air gap flux, resulting
in lower harmonic core losses as well as lower torque ripple. The air gap length is
increased from the initial value during the iterative design process if the harmonic core
losses during high-speed need to be reduced.

5.3.6 Slot design and winding

The stator design includes parameters such as the number of slots, slot geometry, and
winding configuration. A lower number of slots reduces the manufacturing cost, enables
easy winding, and reduces space wastage due to the slot insulation and slot opening.
However, more number of slots reduces the mmf harmonics, improves winding cooling
and minimises the leakage inductance. The winding configuration of DTP0-IPMSM is
explained in detail in Section 5.4.

179
5.3.7 Magnet dimensions

After designing the main dimensions, number of stator slots, and winding configuration,
the magnet dimensions are chosen to achieve the desired air-gap flux density.
Neodymium Iron Boron (NdFeB) magnets are most commonly used in commercial
EV and HEV applications (Burress et al., 2011). The cost, availability of specific grade
and feasible dimensions should also be considered during magnet selection. Single
V, double V and delta type configurations are commonly used for IPMSM designs in
commercial EVs (Sarlioglu et al., 2017; Burress et al., 2011). For the same magnet
volume, the delta configuration achieves the highest torque output in the below base
speed operation. Both double V and single V configurations have a wide operating
speed range and low radial forces. However, the single V configuration also offers the
best demagnetisation performance and mechanical stability (Yang et al., 2017). The
single V configuration is selected in this work, considering its simplicity in design
compared to other topologies. The thickness of ribs above the corners of the V-shaped
magnet configuration (trib ) and the slit gap between the two magnets in V-shape (gslit )
are fixed to a minimum value based on a mechanical stress analysis simulation of
the rotor. A higher value of trib and gslit than the minimum required for mechanical
stability results in increased flux leakage, leading to underutilisation of the magnets and
reduction in machine saliency. The rotor structure of the DTP0-IPMSM with V-shaped
magnet configuration and ribs for structural stability is shown in Fig. 5.3.

5.3.8 No-load air-gap flux density

The fundamental no-load air-gap flux density (Bδ1 ) is calculated by performing a Fast
Fourier Transform (FFT) of the normal air gap flux obtained from FEMM (Meeker
et al.). Bδ1 is used to calculate the exact induced emf at base speed for a given number
of turns per phase. The variation of the normal component of flux density in the air gap
for the designed 37 kW DTP0-IPMSM with parameters in Table. 5.2 is given in Fig. 5.4.

180
winding difficulty as well as the skin effect. However, for the same grade, thin wires
have a thinner coating, and lower voltage withstand capability (RR Shramik) according
to the IEC 60317 specifications. Further, due to a large number of parallel conductors
while using thinner wire, more slot space is used by the conductor insulation. Hence,
an appropriate number of parallel paths should be chosen to achieve the required
winding thickness. 15 AWG to 20 AWG enamelled copper wires are commonly used
in commercial EV motors (Burress et al., 2011; Staunton et al., 2006).

5.3.10 Slot area and slot width to slot pitch ratio

The total area of the slot is computed after obtaining the number of conductors in a slot
and fixing the winding fill factor. The winding fill factor is defined as the ratio of the
total area of insulated wire to the area of a non-insulated slot. The ratio lies in the range
of 0.5 to 0.6 for conventional low power machines. However, in the motors for EVs, the
winding fill factor is always kept near the maximum possible limit (near 0.6) to improve
the efficiency by reducing the copper losses.
A parallel tooth with tapered slot is used when round wires of smaller gauge are used
for winding, whereas a tapered tooth with parallel slot is suitable for rectangular/strip
conductors, which improves the slot fill factor. For a given stator volume, smaller slots
give better overall efficiency with most driving cycles than larger slots (Grunditz et al.,
2021). The efficiency improvement occurs since harmonic core loss reduction is more
significant than copper loss increase during slot size reduction. The increased current
density results in higher winding loss, causing reduced overload durations. However,
smaller slots improve the overload capacity by reducing the teeth saturation level and
also reduces the overall material cost due to lower copper volume. Hence, the typical
slot width to slot pitch ratio is selected in the range of 0.3 to 0.4 (Burress et al., 2008)
in electric vehicle motors compared to 0.4 to 0.6 used in conventional machine design.

182
5.3.11 Design optimisation

Once all the motor dimensions are fixed to an initial value, various parameters are
optimised to satisfy all specifications for the given requirement. The effect of magnet
thickness, magnet orientation, and solid bridge between the permanent magnets are
studied in various literature (Chakraborty et al., 2021; Liu et al., 2016; Hetemi et al.,
2010; Hahlbeck and Gerling, 2008). Each of these parameters results in different
torque-speed characteristics of IPMSM. In this thesis, the PM thickness, V angle and
bridge thickness are optimised in an iterative loop to satisfy the terminal voltage at base
speed, developed torque requirement, CPSR, torque ripple and mechanical stability.
The slot width to slot pitch ratio and current density are also varied in an iterative loop
to satisfy the tooth and yoke flux density limits.
The rated torque of the designed motor is the developed torque at MTPA. The MTPA
condition is obtained by computing the developed torque for various current space
vector angles (δ) at the rated current. Here, the current space vector angle is measured
with respect to the rotor d-axis. For SMPMSM, MTPA occurs at δ = 90◦ , whereas
for IPMSM MTPA occurs at δ > 90◦ due to the presence of the reluctance torque
component in addition to the Lorentz torque. The MTPA operation for the designed
37 kW DTP0-IPMSM occurs at 117◦ , and the maximum torque is 111.7 Nm, as shown
in Fig. 5.5.
The flux density distribution of the designed 37 kW DTP0-IPMSM at MTPA with 80 A
RMS current is shown in Fig. 5.6. It is observed that the maximum teeth flux density is
limited to nearly 1.5 T, and the maximum yoke flux density is nearly 1 T.
Finally, the efficiency and power factor of the machine for the entire torque-speed plane
has to be analysed. To obtain the maximum vehicle efficiency for a given driving cycle,
the frequently operated regions of the torque-speed characteristics should coincide with
the high-efficiency region of the designed motor. If the desired efficiency and power
factor are not achieved, further iterations of the complete design flow is performed by
varying the maximum air gap flux density, linear current density and machine cooling

183
flexible than the Nomex paper and hence does not consume more end winding space.
The reduced end winding space reduces the obstruction for rotor insertion after winding.
The operation of the motor when used in EV can be emulated by testing the machine
through a driving cycle. A driving cycle is the vehicle speed profile with time consisting
of starting, stopping, idling, accelerating, decelerating and constant speed cruising for
a limited time span, which represents a model of real-world driving (Agarwal and
Mustafi, 2021). The driving cycle profile varies for different vehicle types (passenger
cars, heavy-duty trucks etc.) and different regions depending on the geography, road
conditions and traffic. The popular standard driving cycles include EPA Federal Test
Procedure (FTP-75), New European Driving Cycle (NEDC), Supplemental Federal
Test Procedure (SFTP-US06), Modified Indian Driving Cycle (MIDC) and Worldwide
Harmonised Light Vehicles Test Cycles (WLTCs).
The performance of the designed 37 kW DTP0-IPMSM with six-step operation is
evaluated by simulating it with the vehicle model of Mahindra e-Verito (e-Verito)
through MIDC (MIDC, 1991) in Matlab/Simulink. However, the model of the electric
vehicle load characteristics is required for the drive cycle testing of the motor. The
modelling of a electric vehicle motor is discussed in Section 5.5, and the drive cycle
simulation is explained in Section 5.6.

5.5 MODELLING OF AN ELECTRIC VEHICLE LOAD

CHARACTERISTICS FOR DRIVE CYCLE TESTING OF THE DTP0-
PMSM

The linear motion equations of the vehicle has to be converted into rotational motion
of the motor in order to test the motor operation while used in an electric vehicle.
Hence, the vehicle speed, vehicle inertia and the tractive effort at the wheels should
be appropriately referred to the motor.

186
The torque equation of an electric motor is expressed as,

dωm
Te = Jeq + Tl (5.4)
dt

where Te is the developed electromagnetic torque of the motor, ωm is the speed of the
motor in rad/s, and Tl is the motor load. The equivalent moment of inertia of the motor
(Jeq ) is given by
 2
Mveh rd
Jeq = (5.5)
ηt mG

where Mveh is the vehicle mass, rd is the effective radius of the wheel, mG is the gear
ratio of the drive, and ηt is the efficiency of the driveline. The tractive effort at the
wheels and the motor load are related by,

F t rd
Tl = (5.6)
m G ηt

where Ft is the tractive effort at the tires.

Since the accelerating force is already included as equivalent inertia in (5.4), only the
remaining load components are used to compute the tractive effort at the wheels. The
tractive effort at the driving wheels (Ft ) is expressed as,

Ft = Fr + Fw + Fg (5.7)
2
= Mveh gfr + 0.5ρAf CD Vveh + Mveh g sin αR (5.8)

where Fr is the rolling resistance, Fw is the aerodynamic drag, Fg is grading resistance,

g is the acceleration due to gravity, fr is the rolling resistance coefficient, Af is the
vehicle front area, CD is the aerodynamic drag coefficient, rho is the air density, αR is
the road angle and Vveh is the vehicle speed.
Similarly, the relationship between Vveh and the motor speed (ωm ) is given by

ωm rd
Vveh = (5.9)
mG

187
Finally, combining (5.4)-(5.9), the dynamic equation of the EV motor relating the
developed electromagnetic torque (Te ) and motor speed (ωm ) in terms of the electric
vehicle parameters is obtained as,

 2  2 !
Mveh rd dωm rd rd 2
Te = + Mveh gfr + 0.5ρAf CD ωm + Mveh g sin αR
ηt mG dt mG ηt mG
(5.10)
The 37 kW DTP0-IPMSM is simulated with equivalent inertia Jeq and load torque Tl ,
calculated using the parameters given in Table. 5.1. The simulation results are discussed
in the next section.

5.6 OPERATION OF THE 37 KW DTP0-IPMSM WITH SIX-STEP

OPERATION THROUGH MODIFIED INDIAN DRIVING CYCLE

The profile of vehicle speed during MIDC is shown in Fig. 5.9. The MIDC consist of an
elementary urban cycle repeated four times and an extra-urban cycle. In this simulation,
one elementary urban cycle combined with the extra-urban cycle from MIDC is used to
evaluate the performance of the DTP0-IPMSM with six-step control. The parameters
of the vehicle used for the simulation are given in Table. 5.1. The performance of
the proposed DTP0-IPMSM with six-step operation is analysed using the motor speed,
developed torque, output power and d-q currents during the MIDC drive cycle. The
developed torque of the proposed DTP0-IPMSM in Fig. 5.10b is able to track the sudden
load changes caused by acceleration and deceleration during the MIDC. As a result,

100 Elementary Extra-urban

Speed (km/hr)

80 Urban Cycle Cycle

60
40
20
0
0 100 200 300 400 500 600 700 800 900 1000 1100 1200
time (s)
Fig. 5.9: Vehicle speed profile in MIDC

188
the motor speed also tracks the reference speed throughout the drive cycle, as seen
in Fig. 5.10a. A high load demand usually occurs in the low-speed operation during
starting and stopping of the vehicle, as observed in Fig. 5.10b. Even though the load
torque demand is low, high power is necessary to operate in the high-speed region. The
maximum power requirement of 30 kW occurs at 10,876 rpm, as shown in Fig. 5.10c.
Hence, the traction motor should have high torque in low-speed region and a wide
constant power speed range for high-speed operation.
The drive operates in field weakening region above 1890 rpm, drawing more negative
d-axis current. An increase in motor speed by 2418 rpm (from 6042 rpm to 8460 rpm in
Fig. 5.10a) results in an increase of d-axis current magnitude by 40 A (from −128 A to
−168 A), as seen in Fig. 5.10d. As the motor speed further increases, the drive operates
in deep field weakening region, demanding more and more negative d-axis current.
Thus, the system efficiency deteriorates at a higher operating speed due to the increased
copper loss. Six-step operation is adapted for motor speed beyond 9668 rpm. Due to the
availability of a higher voltage margin in six-step operation, demand for the negative
d-axis current is reduced. Hence, it is observed that as the motor speed increases by
2416 rpm (from 8460 rpm to 10,876 rpm), the increase in d-axis current magnitude is
only 12 A (−168 A to −180 A). Thus, the six-step operation reduces the d-axis current
demand compared to the linear region as well as achieves reduced switching losses.
The increase in torque ripple during six-step operation is also observed in Fig. 5.10b.
However, the speed oscillation is negligible, as seen from Fig. 5.10a, due to the large
equivalent vehicle inertia.
The operating points while the vehicle executes the MIDC are superimposed with the
torque-speed and power-speed characteristics, as shown in Fig. 5.11 and Fig. 5.12,
respectively. A high load torque requirement (up to 52 Nm) is observed in the low-speed
region, and a high power demand (up to 30 kW) is seen in the high-speed operation. The
designed motor is able to encompass all operating points of the MIDC as well as the
EV requirement characteristics within its torque-speed envelope, validating the motor

189
 dual 3- PMSM with six-step
150 dual 3- PMSM (linear region)
3- PMSM (linear region)
100 MIDC

Torque (Nm)
50

-50

-100

-150

0 2000 4000 6000 8000 10000 12000 14000

speed (rpm)

Fig. 5.11: MIDC operating points of Mahindra e-verito with the proposed DTP0-
IPMSM in torque-speed plane

dual 3- PMSM with six-step

60 dual 3- PMSM (linear region)
3- PMSM (linear region)
40 MIDC
Power (kW)

20

-20

-40

-60
0 2000 4000 6000 8000 10000 12000 14000
speed (rpm)

Fig. 5.12: MIDC operating points of Mahindra e-verito with the proposed DTP0-
IPMSM in power-speed plane

high-speed operation. It is observed from Fig. 5.11 and Fig. 5.12 that the high power
requirement during high-speed operation is satisfied by using the six-step operation.

5.7 CONCLUSION

In this chapter, the design procedure of a DTP0-IPMSM for an EV was detailed. An

iterative process was used to design the DTP0-IPMSM to achieve all the specifications
of a commercial EV: Mahindra e-Verito. Some general correlations between different
design inputs and machine performance were discussed to enable machine designers

191
to make informed decisions during machine design. Finite Element Method (FEM)
simulations were used to analyse the flux density distribution and to calculate the
developed torque of the designed DTP0-IPMSM. Finally, a drive cycle testing of the
machine was performed in Matlab/Simulink to evaluate the performance by emulating
actual road driving conditions. The designed motor was able to satisfy all the
torque-speed characteristic requirements of the target EV including acceleration time,
gradability, maximum operating speed and was able to execute the MIDC driving cycle.
The following two chapters deal with methods to improve the sensorless starting of
three-phase PMSMs. The next chapter presents a simple and reliable starting method
of PMSM with a quick changeover from open-loop control to closed-loop control.
The proposed methods are highly suitable for critical applications like electric ship
propulsion and ‘emergency heat and smoke exhaust’.

192
 CHAPTER 6

OPEN-LOOP STARTING AND POWER FAILURE RIDE

THROUGH IN SENSORLESS VECTOR CONTROLLED
3-PHASE PMSM FOR MARINE PROPULSION

6.1 INTRODUCTION

Various sensorless vector control techniques for PMSMs were discussed in Section 1.3.
It was observed that a quick and reliable starting method together with a simple
sensorless vector control technique is required for certain critical applications like
‘emergency heat and smoke exhaust’ and electric ship propulsion. In this chapter, a
‘pulse-off method’ is proposed to achieve a quick transition from the I-f open-loop
starting to the stator voltage integration based sensorless vector control, which is highly
suitable for the critical applications mentioned above.
In the proposed method, a quick and smooth changeover from I-f to sensorless vector
control is achieved by accurately estimating the rotor angle. Initially, the motor is
started from a stand-still with the I-f open-loop control. The inverter pulses are turned
off for a short duration after the motor reaches enough speed for sensorless vector
control to work. In the pulse-off duration, the rotor position is estimated using the
measured back-emf (using two hall-effect based voltage transducers) after the current
decreases to zero. Hence, the proposed method accurately determines the rotor position
independent of machine parameters like stator resistance and inductance compared to
the conventional direct transition methods. Even though the additional isolated voltage
sensors increase cost and number of hardware components, it will not be significant
compared to the total system cost in high power applications, and also of less concern
for applications where a quick and reliable starting is the priority. Further, the line
voltage sensors can also be used for accurate PM flux estimation during the self
commissioning of PMSM drives. The proposed method is also used for achieving a
smooth on-the-fly start during short time power supply interruptions. Since the motor
terminal voltages are continuously monitored by a sensorless vector control algorithm,
a smooth on-the-fly start is performed without resorting to complex methods based on
signal injection and zero voltage pulse (Wu et al., 2020; Pravica et al., 2018; Jukic et al.,
2020).
An overview of the I-f starting procedure is given in Section 6.2, and the sensorless
vector control using stator voltage integration is explained in Section 6.3 before
detailing the proposed method.

6.2 OPEN-LOOP I-F STARTING

In the open-loop I-f method, the motor is accelerated with controlled stator currents.
After the motor speed crosses a minimum speed for the sensorless algorithm to operate
satisfactorily, a transition to the sensorless vector control is performed. The I-f control
is a speed open-loop and current closed-loop method. In this method, a preset reference
speed profile is used to increase the motor speed. The expression for the I-f ramp
reference speed (ωi ) is given as
ωi = Kw t (6.1)

where Kw is the slope of the speed ramp.

The reference frame transformation angle during I-f control (θi ) is obtained as

Z
θi = ωi dt (6.2)

The stator currents are referred to a synchronously rotating d′ −q ′ reference frame with
the transformation angle θi , as shown in Fig. 6.1. The q ′ -axis current (i′q ) is controlled
at the rated value (Ilim ), and the d′ -axis current (i′d ) is maintained at zero to obtain full
torque capability during starting. Thus, the stator current vector (~is ) is oriented along the
q ′ -axis. Since the d′ −q ′ reference frame may be shifted from the rotor reference frame
(d-q), the machine draws both flux producing d-axis current (ids ) and torque producing

194
 q ′-axis
q-axis
~is ωr d-axis
iqs ~r
Ψ
δ d′-axis
N ωi
θi θr

S stator axis

Fig. 6.1: Synchronously rotating reference frame and rotor reference frame during I-f
open-loop control of PMSM

q-axis current (iqs ) depending on the load.

Fig. 6.2 shows a typical frequency ramp-up curve. The conventional I-f control is
divided into the following intervals.

• Kick-off interval

• Ramp-up interval

• Constant speed interval

kick ramp const

off up speed
ω2
ωi
(rad/s)
ω1

0 t1 t2
time (s)
Fig. 6.2: Starting frequency profile with I-f control

In the kick-off interval, the reference frequency is maintained at a low value to overcome
the initial stiction and cogging torque. However, an initial jerk resulting in a reverse
rotation may occur depending on the initial angle between the stator current space vector
and the rotor flux space vector (represented as load angle ‘δ’). Since the reference
frequency is small, the rotor latches on to the stator flux when the load angle (δ) reaches
a sufficient value, in spite of the initial reverse motion. The motor speed would match
the reference speed by the end of the kick-off interval. Further, the reference frequency

195
is increased in the ramp-up interval according to (6.1). The ramp-up interval is followed
by a constant speed interval in the conventional I-f control. However, when the I-f
method is used for starting, the changeover can be performed either in the ramp-up or
in the constant speed interval once the machine picks up sufficient speed for sensorless
vector control to operate.
Loss of synchronism may occur during the ramp-up interval if the total load torque
becomes greater than the developed torque of the motor due to high acceleration. The
developed electromagnetic torque (Te ) of an SMPMSM is given by

Te = kt Ilim sin δ (6.3)

where kt is the torque constant, Ilim is the magnitude of rated current vector and δ
is the load angle (angle between the current vector and the rotor flux vector). Under
stable operation, the developed torque is balanced by the total load demand. Also,
the reference speed (ωi ) is assumed to be same as the rotor speed (ωr ) for large-signal
variations. Thus, the developed torque is expressed as

2 2 d
kt Ilim sin δ = Tl + Tf + B ωi + J ωi (6.4)
P P dt

where RHS of (6.4) represents the total load demand, Tl is the load torque, Tf is
the frictional torque, J is the moment of inertia, ωi is the electrical reference speed
(elec rad/s), P is the number of poles and B is the viscous friction coefficient.
Here, loss of synchronism is prevented even under the worst-case loading condition
by fixing the reference frequency slope to a proper value. Let the slope of reference
frequency be denoted by a constant ‘Kw ’ as given in (6.1). The load angle is 90◦ under
worst-case loading. Hence, the reference frequency slope during the ramp-up interval
to maintain stability even under the worst-case total load demand is derived from (6.4)
as
 
P kt Ilim − Tl(max) − Tf − B P2 ωchg
Kw = Ks (6.5)
2 J

196
where Tl(max) is the maximum expected load torque during the ramp-up, and ωchg is the
changeover speed (elec rad/s). A safety factor Ks is included to incorporate the effect
of error in the estimated parameters (Ks < 1).
Speed and current oscillations occur during the frequency ramp-up interval of I-f
controlled PMSM motors if the damping in the system is insufficient. These oscillations
can even lead to instability.

6.2.1 Damping of speed oscillations during I-f starting

Mid-frequency instability is an inherent instability of open-loop controlled PMSMs

without damper windings (Kan and Tzou, 2012). In mid-frequency instability, the open-
loop control of PMSM becomes unstable beyond a certain operating frequency. The
instability may or may not occur depending on the machine parameters and the type
of load connected. For machines with large viscous friction or if the connected load
is a function of rotor speed, the chance for mid-frequency instability is less due to the
high damping in the system. The mid-frequency instability in open-loop I-f controlled
PMSM drive is stabilised by modulating the reference speed with perturbations in input
power (Boldea et al., 2015; Perera et al., 2003; Borisavljevic et al., 2010; Haichao et al.,
2017), thus adding damping to the overall system. The power perturbation algorithm is
included in this work to improve the damping of speed and current oscillations during
open-loop I-f control. A detailed analysis of mid-frequency instability in PMSMs and
its stabilisation technique is provided in Chapter 7.
When the machine reaches sufficient speed in the ramp-up interval, a changeover to
the sensorless vector control is performed. The sensorless vector control operation
of PMSM for the medium and high-speed range of operation is explained in the next
section. A stator voltage integration technique is used to implement the sensorless
vector control.

197
6.3 STATOR VOLTAGE INTEGRATION BASED SENSORLESS VECTOR
CONTROL OF 3-PHASE PMSM

In this work, an inset-PMSM (parameters given in Table. 6.1) is used to implement the
direct transition method and the proposed changeover method. Since Lds ≈ Lqs , the
machine can be considered similar to an SMPMSM, and the control of the SMPMSM
~ s ) is obtained
can be applied here. In stator voltage integration, the stator flux vector (Ψ
by integrating the stator terminal voltage with the resistance drop reduced.

Z
~s =
Ψ (~vs − ~is Rs )dt (6.6)

where ~vs = vαs + jvβs is the stator voltage vector, Rs is the stator resistance, and
~is = iαs + jiβs is the stator current vector.
~ r ) is obtained as
The rotor flux vector (Ψ

~r = Ψ
Ψ ~ s − Ls~is (6.7)

where Ls is the stator inductance. Here, Ls is chosen as the average of Lds and Lqs .
Sine and cosine of the estimated rotor position (b
θr ) are given by

Ψb αr
cos b
θr = (6.8)
~ r|
|Ψ
Ψb βr
sin b
θr = (6.9)
~ r|
|Ψ

where Ψb αr and Ψ
b βr are the α and β components of the estimated rotor flux vector (Ψ
~ r)
q
~ r| = Ψ
and |Ψ b2 + Ψ b2 .
αr βr

The rotor speed is estimated from the estimated rotor position as

d d
b r = cos b
ω θr (sin b
θr ) − sin b
θr (cos b
θr ) (6.10)
dt dt

However, the offset in the measured current results in a drift in the output when a

198
pure integrator is used. To overcome the drift and stabilise the integration action, a
small negative feedback (Hurst et al., 1998) is provided, as shown in Fig. 6.3. Thus,
the resulting system acts similar to an LPF. By choosing the corner frequency of LPF
(ωLPF ) to be a small value, the transfer function approximates a pure integrator during
high-speed operation. In the low-speed range, the operating frequency nears the corner
frequency of the LPF, causing phase and magnitude errors in the estimated rotor flux
linkage. Further, inaccuracy in the estimated stator resistance and stator inductance
also adds to this error. Hence, to overcome the low-speed issues of the stator voltage
integration technique, an open-loop I-f method is used for starting and low-speed
operation (0-8 Hz range). The back-emf based sensorless vector control is used only
in the medium and high-speed range (above 8 Hz).

b αs
Ψ
vαs b αr
Ψ

Rs iαs ωLPF Lsiαs

Fig. 6.3: Sensorless position estimation using stator voltage integration

6.4 QUICK AND RELIABLE TRANSITION FROM OPEN-LOOP I-F TO

SENSORLESS VECTOR CONTROL FOR CRITICAL APPLICATIONS

Increased acceleration rate and current profiling based methods discussed in Section 1.5
are used to achieve a smooth changeover from I-f control to sensorless vector control.
However, for critical applications requiring a quick start-up, the direct transition method
discussed in Section 1.4.1, which is devoid of a transition interval, is more suited. The
changeover to sensorless vector control in such applications is usually performed at
about 5−10% of the rated speed based on the assumption that the estimated position
has already converged. To achieve a smooth transition, the transformation angle and the
q-axis current reference should be properly initialised. However, the convergence of the
estimated position highly depends on the system parameters. Further, speed oscillations

199
during I-f ramp-up due to mid-frequency instability also add to estimation error in the
direct transition method. Hence, the direct transition is performed at a constant speed,
where the speed oscillations have already settled (Baratieri and Pinheiro, 2013). A
detailed simulation study of the effect of position estimation error at the changeover
instant in the direct transition method is provided in Section 6.5.

6.5 ANALYSIS OF DIRECT TRANSITION METHOD

The direct transition method is implemented in the system shown in Fig. 6.4 (assume
that it is devoid of the proposed method block) and simulated in MATLAB/Simulink.
A 25 kW inset-PMSM with parameters given in Table. 6.1 is used for the study. The
simulation results while using the direct transition method are shown in Fig. 6.5. A
constant load torque of 25 Nm is considered for all the simulations. The machine
is started with a kick-off interval of 2 s, where the machine speed is maintained at
1 Hz. After the kick-off interval, the machine speed is ramped up at 5 Hz/s to reach
a steady-state speed of 10 Hz. In spite of the power perturbation algorithm, damped
oscillations are observed in the speed and current waveforms during both kick-off and
ramp-up intervals. The magnitude of oscillations would have been more prominent with
a longer settling time if the I-f control was performed without the power perturbation

0
PMSM
Ilim efd 2 1 2
1
efq
Inverter
V
Vdc
Speed PI Current PI
1 iq−ref vqs V
ωref vαs 2 φ vas
2 dq
vbs PWM vab
ω̂r iqs Current PI vds αβ vβs 3 φ vcs Module vbc
id−ref=0
ids θc
iβs 2 φ ias
dq
ωi iαs i
θi αβ 3 φ bs
t 1 2
θi θbr
Reference speed
ramp-up vab Estimation θr∗
sin θbr ids
from ωr∗
vαs,vβs efd vbc back-emf iq load
Sensorless cos θbr iqs Feed forward
iαs,iβs estimation ω̂r calculation efq Proposed pulse-off method
br
ω

Fig. 6.4: Control system for sensorless vector control of PMSM with I-f starting

200
 Table 6.1: Inset-PMSM parameters

Rated power 25 kW Permanent-magnet flux 0.185 Wb

Rated speed 3000 rpm d-axis inductance 0.168 mH
Rated frequency 400 Hz q-axis inductance 0.178 mH
Rated torque 80 Nm Stator resistance 29 mΩ
RMS line voltage 400 V Inertia 2 kg m2
Rated current 35 A Pole pairs 8

stabilisation. Once the oscillations are settled, a transition to the sensorless vector
control is performed at 5 s. The transition is performed by changing all the switch
positions in Fig. 6.4 from 1 to 2. The reference frame transformation angle (θc ), which
was θi during the I-f control is changed to θ̂r during the sensorless vector control, as
shown in Fig. 6.5d. ‘iq_load ’ in Fig. 6.5b represents the q-axis current demanded by the
load. iq_load is 25.3 A for the considered machine with 25 Nm load. q ′ -axis current is
maintained at a constant reference during the I-f control, thereby maintaining the phase
currents at the rated value, as observed from Fig. 6.5b and Fig. 6.5c. The actual q-axis
current (iqs ) oscillates about iq_load in the kick-off interval and has a mean value above
iq_load for acceleration in the ramp-up interval. Since the actual q-axis current (iqs ) is
proportional to the developed torque for SMPMSM, iqs represents the developed torque
itself.
Even though the sensorless algorithm uses accurate machine parameters, oscillations
in speed and current are observed after the transition to sensorless vector control. The
oscillations arise due to the error in the estimated rotor position at the transition instant,
as observed in Fig. 6.5d. The control system transformation angle (θc ) is initialised to
the sensorless estimated rotor position (θ̂r ) instead of the actual rotor angle (θr ). The
small error in the estimated position is due to the use of LPF instead of a pure integrator
for stator voltage integration. The magnitude of error is proportional to the proximity
of the operating frequency to the corner frequency of the LPF (1 Hz). The error in
the estimated rotor position causes an error in the estimated speed and the initial value
of the q-axis current reference (iq−ref ) used for sensorless vector control initialisation

201
of Fig. 6.5b. Further, error in estimated speed also adds to additional oscillations in
iq−ref due to the proportional action of the speed controller after the changeover.
Variation in motor parameters such as stator resistance variation due to temperature and
saturation of d-axis inductance with loading, causes additional magnitude and phase
error in the estimated rotor flux in the sensorless vector control. Further, the effect of
dead-band, cogging torque at low-speed and inaccurate inverter modelling cause low-
frequency oscillations (5th and 7th harmonic) in the estimated rotor flux. Compensation
techniques are normally used to compensate for stator resistance variation, effect of
dead-time and modelling errors of the inverter. However, the resulting sensorless
algorithm becomes highly complex while using such techniques. Also, the effects of
d-axis inductance saturation and cross-coupling magnetisation are difficult to predict,
as explained before. Low-frequency oscillations during I-f starting also result in
inaccuracies in the estimated rotor position. In this work, the stator voltage integration
based sensorless algorithm is used without additional compensation techniques for
implementing both the direct transition method and the proposed pulse-off method.
To incorporate the effect of parameter variation, inverter dead-time, cogging torque,
inaccurate inverter modelling, and error due to speed oscillations, an error of 15% is
added to the sine of the sensorless estimated position at the transition instant. Let
sin θerr be the sine of estimated sensorless position with 15% error. The resulting
√
erroneous cosine of the estimated angle is computed as 1 − sin θerr . Fig. 6.6 shows
the simulation result of the direct transition method with an additional error of 15%
added to the estimated position at the transition instant. A larger magnitude of speed and
current oscillations is observed after transition compared to the direct transition method
without additional error. iq−ref is initialised to 71.7 A instead of 25.3 A demanded by
the load torque, as seen from the inset of Fig. 6.6b. The higher value of initial iq−ref
increases the developed torque momentarily, accelerating the machine to a higher speed
even though the machine is already at the reference speed. Consequently, the machine
takes a longer time to settle to the reference speed due to the slow integral control action

203
of the speed PI controller, as observed in Fig. 6.6a.
The proposed pulse-off method estimates the rotor position from the measured back-
emf and does not depend on the sensorless estimated position at the transition. Hence, a
constant speed interval for speed oscillations to settle is not mandatory for the proposed
method. The proposed method is discussed in detail in the next section.

6.6 PROPOSED PULSE-OFF METHOD

In the proposed pulse-off method, the rotor position at changeover is obtained from
the line voltage measurement instead of relying on the sensorless vector control. Thus,
an accurate rotor position is obtained independent of machine parameter variation and
speed oscillations at low-speed to achieve a quick and smooth transition. In this method,
the inverter pulses are disabled for a short duration when the motor speed reaches
the changeover speed. Consequently, the PMSM regenerates through the freewheeling
diodes for a short duration, forcing the stator currents to zero. The internal back-emf of
the machine is reflected in the measured line voltage when the current reduces to zero.
Assuming a sinusoidal back-emf, the back-emf vector (~eb ) and rotor flux vector are
related by
~r
dΨ
~eb = (6.11)
dt

The rotor position of PMSM is the same as the rotor flux position. Thus, the rotor
position is estimated in the pulse-off interval from the measured line voltages.
The assumption of sinusoidal back-emf is applicable only if the harmonics in back-
emf are not significant. Hence, the proposed method is only applicable to sinusoidal
back-emf machines with less amount of other harmonics. Methods like skewing and
fractional slot windings (Jahns and Soong, 1996) are conventionally used to reduce the
higher-order harmonics (due to slot harmonics) in the back-emf. The back-emf of the
25 kW inset-PMSM with fractional slot winding used in this work is shown in Fig. 6.7.

eb = eα + jeβ ) is computed from the two measured line voltages

The back-emf vector (~

205
Fig. 6.7: Measured line-line back-emf voltage of the 25 kW inset-PMSM: Ch4 - vab
(X-axis:- 50 ms/div; Y-axis:- 5 V/div)

as

vbc
eα = vab + (6.12)
√ 2
3
eβ = vbc (6.13)
2

Sine and cosine of back-emf vector position are given by

eα
cos(θemf ) = (6.14)
|~e|
eβ
sin(θemf ) = (6.15)
|~e|

where θemf is the position of the back-emf vector with respect to the stationary a-phase
q
axis, and |~e| = e2α + e2β is the magnitude of the back-emf vector. Let θr∗ denote the
estimated rotor position using the pulse-off method. The rotor flux vector lags back-emf
by 90◦ . Hence, during anticlockwise rotation θr∗ = θemf − 90◦ , as shown in Fig. 6.8a.
However, during clockwise rotation, θr∗ = θemf + 90◦ , as shown in Fig. 6.8b.
Sine and cosine of the rotor flux position during anticlockwise rotation are expressed as

sin(θr∗ ) = sin(θemf − 90◦ ) = − cos θemf (6.16)

206
 Anticlockwise
rotation Clockwise
rotation
~eb
~eb

θemf θr∗
~r
Ψ θemf
θr∗
~r
Ψ
Stator axis Stator axis
(a) Case-1 (b) Case-2

Fig. 6.8: Relative position of back-emf and rotor flux vectors

cos(θr∗ ) = cos(θemf − 90◦ ) = sin θemf (6.17)

Sine and cosine of the rotor flux position during clockwise rotation are obtained as

sin(θr∗ ) = sin(θemf + 90◦ ) = cos θemf (6.18)

cos(θr∗ ) = cos(θemf + 90◦ ) = − sin θemf (6.19)

The estimated rotor speed using pulse-off method (ωr∗ ) is given by

d d
ωr∗ = cos θr∗ (sin θr∗ ) − sin θr∗ (cos θr∗ ) (6.20)
dt dt

An accurate estimate of the PM flux linkage (ΨF ) is necessary to obtain the

machine back-emf for the control system feedforward calculations. However, due
to demagnetisation and ageing, ΨF reduces from its original value. In the proposed
method, ΨF is estimated in the pulse-off duration using the measured back-emf and
estimated speed. The estimated PM flux linkage using the pulse-off method (Ψ∗F ) is
obtained as (6.21).
|~eb |
Ψ∗F = (6.21)
ωr∗

Thus, ΨF used in the feedforward calculation of the control system is updated in every

207
drive start-up during the pulse-off interval.
The control is transferred from the I-f method to the sensorless vector control by
enabling the inverter pulses once the rotor position, rotor speed and q-axis current
demand of load are computed. The estimated rotor position is used to initialise the
integrator in the sensorless module. Further, the estimated speed during the pulse-off
interval is used to initialise the speed filter. However, the speed controller also should
be initialised appropriately with ‘the q-axis current demanded by the load torque at the
transition’ to achieve a smooth changeover. A wrong initialisation of the integral control
produces a higher or lower developed torque resulting in unwanted oscillations in speed
and current, as discussed in Section 6.5. The computation of the q-axis current demand
by the load for speed PI controller initialisation is discussed in the next section.

6.6.1 Computation of the q-axis current demanded by the load

The developed torque of PMSM supplies the total load torque (Tl + Tf + B P2 ωr ) and is
also used for acceleration, as observed from (6.4). Hence, the q-axis current demanded
by the load (iq_load ) before the pulse-off interval (which is devoid of the acceleration
component) is obtained as

Te − J P2 dtd ωr
iq_load = (6.22)
kt
J 2 d
= Ilim sin δ − ωr (6.23)
kt P dt

From Fig. 6.1, the load angle is given by

δ = 90◦ + θi − θr∗ (6.24)

where θi is the synchronously rotating reference frame angle in the I-f control and θr∗ is
the rotor angle estimated from the back-emf.
However, θi and θr∗ cannot be computed at the same time instant. θi is used for
controlling the current vector before the start of the pulse-off interval, whereas θr∗ is

208
computed only by the end of the pulse-off interval. Hence, θr∗ at the end of the pulse-off
∗ ∗
interval (θr1 ) is interpolated to the start of the pulse-off interval (θr0 ) to estimate the
q-axis current demanded by the load. The rotor speed is assumed as constant during the
pulse-off duration due to the large inertia of the considered machine. θr∗ at the beginning
∗
of pulse-off interval (θr0 ) is given by

∗ ∗
θr0 = θr1 − ωr∗ toff (6.25)

where toff is the pulse-off duration.

The q-axis current demanded by the load torque for speed controller initialisation when
the changeover is performed in the ramp-up interval is obtained as,

∗ J 2
iq_init = Ilim cos(θi0 − θr0 )− Kw (6.26)
kt P

where θi0 is the synchronous reference frame transformation angle at the instant before
the pulse-off interval.
It should be noted that the acceleration torque is not present when the changeover is
performed in the constant speed interval after the speed ramp-up. Hence, the q-axis
current demanded by the load torque in the constant speed interval is given by

∗
iq_init = Ilim cos(θi0 − θr0 ) (6.27)

6.6.2 Simulation of the proposed method

The control block diagram of the proposed method is shown in Fig. 6.4. The proposed
method is implemented on a 25 kW inset-PMSM drive with the parameters in Table. 6.1
and simulated in MATLAB/Simulink to evaluate its performance. Similar to the direct
transition method, the I-f control is used to ramp up the speed initially. However, instead
of directly changing all the switch position from 1 to 2 in Fig. 6.4, a pulse-off interval
of 1 ms is provided at the transition instant. The rotor position, rotor speed and q-

209
axis current demanded by the load are estimated, and the corresponding integrators are
initialised with the proposed pulse-off method. A smooth transition is observed in the
speed waveform in Fig. 6.9a and the current waveforms in Fig. 6.9b and Fig. 6.9c. The
rotor position is estimated accurately from the sensed line voltages, and the sensorless
position is corrected at the end of the pulse-off interval, as seen in Fig. 6.9d. From
the inset of Fig. 6.9b, it is observed that the reference q-axis current is also initialised
to iq_load accurately in the proposed method. As a result, the oscillations in speed and
phase currents are minimised. The small oscillations are due to the inaccuracies in
sensorless vector control after the changeover. However, the oscillations are reduced
significantly compared to both direct transition method and the direct transition method
with an additional error. A comparison of the performance of the proposed pulse-off
method with the direct transition method and the direct transition method with 15%
additional error is shown in Table. 6.2 for better clarity.
Table 6.2: Performance comparison of various transition methods

Magnitude of maximum overshoot after transition

in motor speed (ωr ) in q-axis current (iqs ) in a-phase current (ias )
Direct transition
method 9.5 rpm 21.7 A 14.5 A
Direct transition
method with 15%
17 rpm 34.7 A 23.1 A
additional error
Proposed pulse-off
method 3.2 rpm 8.3 A 5.5 A

6.6.3 On-the-fly starting for power failure ride through

The proposed method is also extended to achieve on-the-fly start of PMSM.

Conventionally, during a short-time power failure, the inverter pulses are disabled to
prevent the DC bus discharge and consequent under voltage trip. When the supply
is restored after the short time power failure, the motor will be in motion due to the
stored kinetic energy. In industries, the motor is started on-the-fly with sensorless
vector control to prevent monetary loss due to a complete system restart after waiting

210
Similar to the pulse-off method, the motor is started directly in sensorless vector control
when the supply is restored by properly initialising the speed controller, speed filter and
the integrator in the sensorless module. However, for directly starting with sensorless
vector control, the motor speed should not fall below the changeover speed of the pulse-
off method during the short-time supply failure. The motor deceleration during supply
failure depends on the motor inertia and the load torque.

6.7 DETERMINATION OF PULSE-OFF TIME

A finite time is required for the stator currents to decay to zero depending on the DC
bus voltage, instantaneous motor currents and stator inductance during pulse-off. Also,
accurate position estimation is obtained with the proposed method only when the back-
emf is estimated at zero current. Hence, the pulse-off duration should be selected such
that all currents becomes zero even in the worst case before re-enabling the pulses. In
this section, the worst-case time required for current decay is analytically derived. The
pulse-off time is fixed by providing an extra margin to the worst-case current decay
time.
The stator currents of the PMSM are marked in Fig. 6.10. It is assumed that ias is
positive, and ibs and ics are negative at the pulse-off instant. When the pulses are
disabled, the highlighted diodes in Fig. 6.10 conduct, and the stator currents decay
to zero, as shown in Fig. 6.11. The resulting equivalent circuit is shown in Fig. 6.12a.

ias
ibs N
Vdc S
ics
PMSM

Fig. 6.10: PMSM supplied from inverter

212
 Pulse-off
ias interval (toff)

tdecay
Stator
currents

0 t1 t2
ibs time

ics Interval−1 Interval−2

(tdecay 1) (tdecay 2)

Fig. 6.11: Stator current decay during pulse-off interval

ibs ics ics

vbs vcs vcs

Vdc Vdc

vas vas
ias ias
(a) Interval-1 (b) Interval-2

Fig. 6.12: Equivalent circuit of PMSM under pulse-off interval

This initial time interval with three non zero currents is termed as interval-1. However,
the smallest current first decays to zero resulting in an equivalent circuit shown in
Fig. 6.12b. Interval-2 denotes the time interval with only two non-zero currents.
Depending on the instantaneous currents at the pulse-off instant, either interval-1 alone
or interval-2 alone or both interval-1 and interval-2 may be present. The time duration
for interval-1 and interval-2 are analytically derived in the following section.

6.7.1 Calculation of interval-1 time duration

During stator current decay in the pulse-off duration, full DC bus voltage gets applied
across the stator terminals. Since the motor speed is very low at the transition interval,

213
the back-emf voltage of PMSM is small compared to the DC bus voltage (Vdc ). Hence,
to simplify the calculation of stator current decay time, the stator resistance and back-
emf are neglected in the stator voltage equations of PMSM. The resulting equation is
given by

dias dibs dics

vas = (Lls + Lms ) − 0.5Lms − 0.5Lms (6.28)
dt dt dt
dias dibs dics
vbs = −0.5Lms + (Lls + Lms ) − 0.5Lms (6.29)
dt dt dt
dias dibs dics
vcs = −0.5Lms − 0.5Lms + (Lls + Lms ) (6.30)
dt dt dt

where Lls is the leakage inductance and Lms is the magnetising inductance.
From the equivalent circuit in Fig. 6.12a, it is observed that vb = vc and ias +ibs +ics =0.
Hence, the relation between the derivatives of stator currents is obtained by equating


(6.29) and (6.30) and using didtas = − didtbs + didtcs as

dias dibs dics

= −2 = −2 (6.31)
dt dt dt

The DC bus voltage equation is derived by using (6.31), (6.28) and (6.29) as

3 dias
Vdc = vbs − vas = −1.5(Lls + Lms ) (6.32)
2 dt
dias
= −1.5 L0 (6.33)
dt

where L0 = Lls + 32 Lms .

Let ‘ibs ’ be the current with smallest magnitude at the pulse-off instant. The slope of
the smallest current at the pulse-off instant in interval-1 (m1_small ) is obtained by using
(6.33) and (6.31) as
dibs Vdc
m1_small = = (6.34)
dt 3L0

214
Hence, the time duration of interval-1 (tdecay_1 ) is given by

|ismall (0)| |ibs (0)|

tdecay_1 = = (6.35)
|m1_small | |m1_small |
3L0 |ibs (0)|
= (6.36)
Vdc

where ismall (0) is the value of ismall (the smallest current at the instant of pulse-off) at the
instant ‘0’ and ibs (0) is the b-phase current at the instant of pulse-off.

6.7.2 Calculation of interval-2 time duration

When the smallest current at the pulse-off instant decays to zero, the equivalent circuit
changes to Fig. 6.12b, initiating interval-2 operation. From Fig. 6.12b, it is observed
dias
that in interval-2, ias = −ics and hence dt
= − didtcs .
The DC bus voltage expression is obtained using (6.28) and (6.30) as

Vdc = vcs − vas (6.37)

dias
= −2L0 (6.38)
dt

The slope of the largest current at the pulse-off instant (ias in this case) during interval-2
(m2_large ) is obtained from (6.38) as

dias Vdc
m2_large = =− (6.39)
dt 2L0

Thus, the time duration of interval-2 (tdecay_2 ) is given by

|ilarge (t1 )| |ias (t1 )|

tdecay_2 = = (6.40)
|m2_large | |m2_large |
2L0 |ias (t1 )|
= (6.41)
Vdc

where |ias (t1 )| = |ias (0)| − 2|m1_small | tdecay_1 . Here ‘ias (t1 )’ denotes the magnitude of
a-phase current at the instant t1 , and ‘ilarge (t1 )’ is the magnitude of ilarge (largest current

215
at the pulse-off instant) at the instant t1 .

6.7.3 Calculation of total stator current decay time

Finally, the total time duration of stator current decay is obtained using (6.36) and (6.41)
as

tdecay = tdecay_1 + tdecay_2 (6.42)

L0
= (2 |ilarge (0)| − |ismall (0)|) (6.43)
Vdc
L0
= (2 |ias (0)| − |ibs (0)|) (6.44)
Vdc

For IPMSM, L0 varies between d-axis self inductance (Lds ) and q-axis self inductance
(Lqs ) depending on the rotor position. Further, ilarge (0) and ismall (0) are the magnitude
of instantaneous currents at the pulse-off instant.
Variation of instantaneous stator currents with the current vector angle (ǫ) is shown in
Fig. 6.13. The a-phase, b-phase or c-phase cyclically becomes the largest or the smallest
phase current in every 30◦ interval. Hence, the limiting case of tdecay is captured when
ǫ is varied from 0◦ to 30◦ . When ǫ = 30◦ , the smallest current at the pulse-off instant
ismall (0) = 0. This corresponds to the condition where only interval-2 is present (t1 =
0). When ǫ = 0◦ , the largest current becomes zero by the end of interval-1 (ilarge (t1 ) =
0). This represents the condition where only interval-1 is present (t2 = 0). The plot
of current decay time obtained using (6.44) for ǫ variation in the range 0◦ − 30◦ with

ias ibs ics

ilarge
Stator
currents
0
30◦ 60◦ 90◦ 120◦ 150◦ 180◦ 210◦ 240◦ 270◦ 300◦
ismall ǫ

Fig. 6.13: Variation of stator currents with current vector angle

216
both L0 = Lds and L0 = Lqs is shown in Fig. 6.14. The stator current decay time for
a particular operation could be anywhere between the two limits. However, the pulse-
off duration should be selected considering the worst-case decay time. The maximum
current decay time (38.57 µs) occurs at ǫ = 30◦ with the current vector aligned along the
rotor q-axis (L0 = Lqs ). Finally, the pulse-off time is fixed at 500 µs for the experiments
considering a safety margin and the additional time for the computations. Since the
moment of inertia is large enough, the pulse-off time will not cause a significant speed
reduction. The speed reduction during the pulse-off interval is calculated in Section 6.8.

40
L ds

38 L qs

36

34

32

30
0 5 10 15 20 25 30

Fig. 6.14: Stator current decay time variation during pulse-off

6.8 EXPERIMENTAL VALIDATION

The proposed pulse-off method is validated on a 25 kW inset-PMSM drive whose

parameters are given in Table. 6.1. The PMSM is coupled to a DC generator for loading,
as shown in Fig. 6.15. TMS320F28335 DSP is used as the control platform, and the
inverter switching frequency is selected as 5 kHz. A kick-off interval of 0.5 s with
a constant frequency of 2 Hz is provided. After the kick-off interval, the machine is
ramped up at 20 Hz/s, and a changeover is performed at 8 Hz. A pulse-off duration

217
 Fig. 6.15: Experimental setup of 25 kW inset-PMSM drive

of 500 µs is provided when the reference speed reaches 8 Hz. The a-phase current
and speed waveform under loaded and no-load conditions are shown in Fig. 6.16 and
Fig. 6.17, respectively. In both loaded and unloaded cases, the rated current is drawn
by the machine during I-f startup since the current vector magnitude is held constant by
the control. Even after the transition, the machine draws full rated current to quickly
ramp up to the reference speed with full rated torque. During steady-state operation
with sensorless vector control, the machine draws only the q-axis current proportional
to load torque, and the d-axis current is controlled at zero. Hence, at the steady-state
operation of sensorless vector control, a higher current is drawn by the machine in the
loaded case compared to the unloaded case as observed in Fig. 6.16 and Fig. 6.17. A
smooth changeover is observed in the stator current with the proposed method. Even
though the developed torque falls to zero for a short duration in the pulse-off interval,
a dip in the motor speed is not observed in both loaded and no-load conditions. The
moderately high inertia of the machine (2 kg m2 ) is sufficient to maintain the machine
speed in the pulse-off duration.
Since the stator currents decay to zero in the pulse-off interval, the developed torque
also becomes zero. Hence, a reduction in speed occurs depending on the load present
and the machine inertia. Since the speed reduction is reflected in the back-emf vector

218
Fig. 6.16: Speed and stator current waveforms with pulse-off method
under loaded condition. Ch3: ω b r , Ch4: ias (X-axis:- 0.7 s/div;
Y-axis:- Ch3: 240 mech rpm/div, Ch4: 50 A/div)

Fig. 6.17: Speed and stator current waveforms with pulse-off method
under no-load. Ch3: b r , Ch4:
ω ias (X-axis:- 0.36 s/div;
Y-axis:- Ch3: 240 mech rpm/div, Ch4: 50 A/div)

position, the proposed method remains unaffected by the speed dip. However, in a given
system, during the pulse-off interval of the proposed method, the motor speed should
not fall below the minimum speed required by the sensorless algorithm. Hence, the
worst-case dip in speed during pulse-off interval under full load should be assessed for

219
every system, and the changeover speed should be selected accordingly. For the 25 kW
inset-PMSM, the worst-case speed dip is obtained as

Tl(max) toff 80 × (500 × 10−6 )

ωdip = = (6.45)
J 2
= 0.02 rad/s (6.46)

Thus, a negligible speed dip occurs, as verified by the experimental results.

The d and q-axis current waveforms under loaded condition are shown in Fig. 6.18.
After the pulse-off duration, the integrator in the speed controller is initialised with an
accurate value of q-axis current demanded by the load. Thus, the transients in the q-axis
current reference quickly settle after the transition. The control system transformation
angle (θc ) is equal to the synchronously rotating reference frame angle (θi ) till the
transition instant, as observed in Fig. 6.19. During the pulse-off interval, the rotor
position is estimated from the back-emf, and the integrator in the sensorless algorithm
is initialised appropriately. After the transition to the sensorless vector control, the
control system transformation angle (θc ) is the same as the sensorless vector control

Fig. 6.18: Variation of d-axis and q-axis currents with pulse-off method under loaded
condition. Ch1: iq−ref , Ch2: i′q , Ch3: i′d , Ch4: ias (X-axis:- 0.74 s/div; Y-
axis:- Ch1,Ch2,Ch3: 40 A/div, Ch4: 50 A/div)

220
Fig. 6.19: Variation of control system transformation angle with the pulse-off method
at the transition instant. Ch1: θi , Ch3: θc , Ch4: pulse-off signal (X-axis:-
50 ms/div; Y-axis:- Ch1,Ch3,Ch4: 0.4 unit/div

angle (b
θr ).
The 3-phase current waveforms in the pulse-off interval are shown in Fig. 6.20. Nearly
33 µs is required for the stator currents to decay, which is within the computed range
given in Fig. 6.14. Also, the two intervals of the pulse-off process mentioned in
Section 6.7 are clearly visible in Fig. 6.20. The a-phase current and speed waveforms

Fig. 6.20: Stator current waveforms in the pulse-off duration. Ch2: ias , Ch3: ibs , Ch4:
ics (X-axis:- 20 µs/div; Y-axis:- Ch2,Ch3,Ch4: 20 A/div

221
during on-the-fly starting of PMSM under loaded and no-load conditions are shown
in Fig. 6.21 and Fig. 6.22. As soon as the supply is restored, the machine is started
directly with sensorless vector control. The currents are quickly increased to the rated
value without any oscillations. The quick current control action is achieved by accurate

Fig. 6.21: Speed and stator current waveforms during on-the-fly start with pulse-off
method under loaded condition. Ch3: ωb r , Ch4: ias (X-axis:- 1.04 s/div; Y-
axis:- Ch3: 240 mech rpm/div, Ch4: 50 A/div)

Fig. 6.22: Speed and stator current waveforms during on-the-fly start with pulse-off
b r , Ch4: ias (X-axis:- 0.76 s/div; Y-axis:- Ch3:
method under no-load. Ch3: ω
240 mech rpm/div, Ch4: 20 A/div)

222
position estimation using the proposed method.
Fig. 6.23 and Fig. 6.24 show d-axis and q-axis current waveforms during on-the-fly
starting under loaded and no-load conditions, respectively. The q-axis current is quickly
regulated at the rated value, and the d-axis current is maintained at zero after the

Fig. 6.23: Variation of d-axis and q-axis currents during on-the-fly start with pulse-off
method under loaded condition. Ch1: iq−ref , Ch2: i′q , Ch3: i′d , Ch4: ias
(X-axis:- 0.92 s/div; Y-axis:- Ch1,Ch2,Ch3: 40 A/div, Ch4: 50 A/div)

Fig. 6.24: Variation of d-axis and q-axis currents during on-the-fly start with pulse-off
method under no-load. Ch1: iq−ref , Ch2: i′q , Ch3: i′d , Ch4: ias (X-axis:-
0.68 s/div; Y-axis:- Ch1,Ch2,Ch3: 40 A/div, Ch4: 50 A/div)

223
transition.

6.9 CONCLUSION

This chapter proposed a quick and reliable starting method, the ‘pulse-off method’,
for PMSMs used for critical applications like ship propulsion and emergency heat
and smoke exhaust. The method is highly suitable for systems with huge inertia and
using I-f open-loop control to achieve full starting torque. Since the rotor position
is estimated using line voltage sensors at zero current, the method is independent of
machine parameters. Even though two additional line voltage sensors used in the
proposed method increase the system cost, it will not be a significant concern for
critical applications where reliability and quick starting are of utmost importance. The
proposed method enables a recalibration of the rotor flux linkage during every starting
in order to accommodate the parameter variation due to ageing. The proposed method
is extended to perform a reliable and seamless on-the-fly start during short time power
supply interruptions, without resorting to any complex algorithms. The effectiveness of
the proposed method is validated on a 25 kW inset-PMSM drive.
In this chapter, the frequency ramp slope was selected to be a minimum value to
maintain stability. However, this would delay the control changeover to the sensorless
vector control, hence will not achieve the best starting performance. However,
increasing the slope of the frequency ramp to a higher value may lead to instability
during loading. The next chapter deals with the stability of PMSM while using
the I-f starting. Here, the starting performance is improved by varying the slope of
the frequency ramp dynamically depending on the load torque. Thus, the starting
performance is improved while maintaining stability. Another inherent instability of
PMSM during open-loop control, the ‘mid-frequency instability’, is also discussed in
the following chapter.

224
 CHAPTER 7

STABILITY OF PMSM IN OPEN-LOOP CONTROL

7.1 INTRODUCTION

The shape and slope of the frequency ramp in the ramp-up interval are critical in
deciding the stability during starting for open-loop V/f or I-f controlled PMSM drives,
as discussed in Section 1.5. The conventional methods use a trial and error approach
to fix the frequency ramp slope. A high value of frequency slope may result in pole
slipping when the system has high inertia and when the load demand is high. Hence,
most of the conventional methods use a very low frequency slope considering the
worst-case loading, which deteriorates the starting performance of the drive. Here,
the available developed torque is not utilised for quick acceleration even during light
load conditions.
In this work, an approach to vary the frequency ramp slope dynamically depending
on the moment of inertia and load torque present is proposed, to achieve a quick and
stable ramp-up during I-f open-loop control. The causes of instability during frequency
ramp-up are analysed in detail. The total torque demand is maintained well within the
machine capability by dynamically controlling the reference frequency slope. A torque
controller is proposed to modulate the reference frequency slope depending on the load
torque and machine inertia, thus preventing pole slipping. Since the small-signal model
of PMSM is insufficient to capture the dynamics corresponding to pole slipping during
frequency ramp-up, a nonlinear large-signal dynamic model of PMSM with I-f control
is adopted to analyse the proposed method. An additional slope attenuator is also
proposed to reduce the speed and current oscillations at the transition instant from the
ramp-up to the constant speed interval when the drive needs to be operated continuously
in I-f control. The performance of the proposed methods are validated in simulation as
well as experimentally on a 25 kW SMPMSM drive. The proposed method is highly
suitable for critical applications like marine propulsion and ‘emergency heat and smoke
exhaust’, where a quick and reliable starting is necessary.

7.2 INSTABILITY DURING OPEN-LOOP CONTROL OF PMSM

7.2.1 Pole slipping during frequency ramp-up

Pole slipping occurs when the developed torque is insufficient to maintain the rotor in
synchronism with the rotating stator magnetic field. It causes pulsating torque on the
rotor, resulting in severe acceleration and deceleration. Finally, the motor comes to a
halt.
The maximum torque capability (Tmax ) of an SMPMSM is represented as (7.1),

2P
Tmax = ΨF Ilim (7.1)
32

where P is the number of poles, Ilim is the magnitude of the rated current vector, and
ΨF is the PM flux linkage. The developed electromagnetic torque (Te ) of an SMPMSM
is derived in terms of load angle (δ) as

Te = Tmax sin δ (7.2)

The total torque demand (Td ) is expressed as

2 2 d
Td = Tl + B ωr + Tf + J ωr (7.3)
P P dt

where J is the moment of inertia (kg m2 ), B is the viscous friction coefficient (Nms),
Tf is the frictional torque (Nm), ωr is the electrical rotor speed, and Tl is the load torque
(Nm). Te is equal to Td during stable operation. Under the assumption that the motor
speed (ωr ) is equal to the reference speed (ωi ) for large-signal variation, the equation

226
governing the rotor dynamics of the SMPMSM is expressed as (7.4).

2 2
Tmax sin δ = Tl + B ωi + Tf + J αi (7.4)
P P

dωi
where αi is the slope of reference speed (αi = dt
).
The variation of developed torque with load angle is plotted as shown in Fig. 7.1. During
motoring operation, δ varies from 0◦ −180◦ . When a sudden load is applied, rotor speed
(ωr ) decreases instantaneously, leading to an increase in δ. In region-1 of Fig. 7.1, Te
increases when δ increases. Hence, the motor reaches a stable equilibrium following a
load torque disturbance. However in region-2, Te decreases when δ increases, leading
to instability. Consequently, stable operation during load transients is achieved by
restricting the operation to region-1 (below 90◦ ) of torque characteristics. In other
words, the total torque demand (Td ) should not be more than the maximum torque
capability (Tmax ) during the frequency ramp-up for ensuring stable operation.

zone of zone of
operation non-operation
Tmax
ks Tmax
Te

Region-1 Region-2
0 30 60 90 120 150 180
δ(◦ )
Fig. 7.1: Variation of the developed torque with the load angle

A large increase in load torque during the ramp-up interval may result in a condition
where Td exceeds Tmax . Here, the operating point shifts to region-2, causing pole
slipping. Analysing (7.4), it is observed that the only variables that can be reduced
to restrict the operation to region-1 during an increase in load are the reference speed
(ωi ) and frequency ramp slope (αi ). However, since a ramp-up operation is intended, ωi
should not be decreased. Therefore, stability of I-f control is conventionally achieved
by controlling the slope of the frequency ramp (αi ) to a minimal value (Yang et al.,

227
2015), limiting Td below Tmax . Nevertheless, reliable starting is achieved at the cost
of poor dynamic performance. The proposed solution for pole slipping is discussed in
Section 7.4. Here, the frequency ramp slope is dynamically adjusted to ensure stable
frequency ramp-up under all load conditions within the motor capability.

7.2.2 Mid-frequency instability and its small-signal stabilisation using reference

frequency modulation

In addition to pole slipping during frequency ramp-up, an inherent instability called

“mid-frequency instability” exists during the I-f and V/f control of PMSM in the
absence of damper windings (Kan and Tzou, 2012). In mid-frequency instability, the
open-loop I-f control of PMSM becomes unstable beyond a certain operating frequency.
This is due to the low damping present in the system. Even though the total system is
nonlinear, the mid-frequency instability is commonly analysed using a linearized small-
signal model of the open-loop I-f controlled PMSM (Kan and Tzou, 2012; Borisavljevic
et al., 2010; Haichao et al., 2017; Boldea et al., 2015; Perera et al., 2003), which is
discussed as follows.
The SMPMSM rotor flux oriented model in state variable form is given by

vs cos(φ + δ) ids
pids = − + ωr iqs (7.5)
Ls τs
vs sin(φ + δ) iqs ΨF
piqs = − + ωr (ids + ) (7.6)
Ls τs Ls
 2
21 P Bωr P
pωr = ΨF iqs − − Tl (7.7)
3J 2 J 2J
pδ = ωi − ωr (7.8)

where ids is the d-axis current, iqs is the q-axis current, vs is the magnitude of the stator
Ls
voltage vector, τs is the stator time constant ( R s
), Ls is the stator inductance, Rs is the
stator resistance, and φ is the power factor angle. Since the machine model is nonlinear,
the model is linearized about a steady-state operating point. The linearized machine

228
model is derived as
    
−1 − Vs sin(φ+δ 0)
∆ids   ∆ids 
ω0 Iqs
 τs  
Ls
    
∆iqs  −ω0 − τ1s − Ids + ΨF Vs cos(φ+δ0 )  ∆iqs 
   Ls Ls  
p  = 
 
 ∆ω   21 P 2   ∆ω 
 r  0 3J 2
ΨF − BJ 0   r
    
∆δ 0 0 −1 0 ∆δ
 
cos(φ+δ0 )
 Ls 0 
 
 sin(φ+δ ) 
 0 
0  ∆vs 
 Ls
+   (7.9)
 0 P 
− 2J
  ∆Tl
 
0 0

Here, Ids , Iqs , ω0 , and δ0 denote the steady-state values of d-axis current, q-axis current,
motor speed and load angle, respectively. Vs represents the steady-state voltage vector
magnitude. Equation (7.9) is of the form

∆ẋ = A∆x + B∆u (7.10)

where A is the state transition matrix, B is the input matrix, ∆x is the state variable
perturbation vector, and ∆u is the input perturbation vector. The eigenvalues of A(X)
are categorised into stator and rotor poles, representing the electrical and mechanical
system dynamics, respectively. Fig. 7.2 represents the eigenvalue plot with a variation
of rotor frequency up to 100 Hz for a PMSM with parameters in Table. 7.1. It is
observed that the rotor poles move to the right side of the s-plane (at 17.2 Hz) and return
to the left side for a range of rotor frequencies. The movement of poles to the unstable
region is due to weak coupling between the electrical stator poles and the mechanical
rotor poles (Perera et al., 2003). This inherent instability of PMSM without damper
windings during open-loop control is termed as mid-frequency instability.
Colby and Novotny (1988) achieved stable operation by modulating the reference
frequency with variations in the dc-link current. However, the dynamics during low-

229
 Table 7.1: Motor and controller parameters

Motor Value Controller Parameters Value

Parameters
Prated 25 kW Switching frequency 3 kHz
Vll 400 V Sampling time 50 µs
ωrated 3000 rpm Current controller BW 106 Hz
P 16 speed controller BW 3 Hz
Rs 29 mΩ Torque controller BW 0.16 Hz
ΨF 0.185 Wb Feedback gain constant (CG ) 1.05
Lds 0.27 mH HPF time constant (τh ) 0.159 s
Lqs 0.285 mH Torque controller 0.8
J 2 kg m2 stability factor (ks )

speed operation with dc-link current modulation were unsatisfactory. Hence, in

most of the recent literature (Borisavljevic et al., 2010; Haichao et al., 2017; Perera
et al., 2003; Boldea et al., 2015), the reference frequency modulated with a component
corresponding to input power perturbations is used to add damping to the system, thus
achieving stable operation. The frequency component added to the reference frequency
is given in (7.11).
CG
∆ωi = −ke ∆Pe ; ke = (7.11)
ωi

50
Stator Poles 100 Hz
Imaginary axis (s )
-1

17.2 Hz

Rotor Poles

-50
-100 -50 0
-1
Real axis (s )
Fig. 7.2: Eigenvalue loci of state transition matrix as a function of ωi with conventional
I-f control

230
where ∆Pe is the perturbation component of the input power obtained by high-pass
filtering of the input power (Pe ). ‘ke ’ is the feedback gain, which is varied in inverse
proportion to the reference speed (Perera et al., 2003) to achieve a nearly constant
damping factor (constant real part for rotor poles). CG is the feedback gain constant,
which is designed to achieve stable operation.
Input power is computed using voltage and currents in the α−β frame as

2
Pe = (vαs iαs + vβs iβs ) (7.12)
3

The modified linearized model after adding reference frequency modulation is given by

∆ẋ′ = A′ ∆x′ + B′ ∆u (7.13)

where      
 ∆ẋ   ∆x   B 
∆ẋ′ =  ′
 ; ∆x = 
′
; B =   (7.14)
∆ω̇i ∆ωi [0]1×2
 
T
 A [0 0 0 1] 
A′ =   (7.15)
[A′51 A′52 A′53 A′54 ] A′55

The elements of A′ matrix are

 
cos(φ + δ0 )
A′51 = C ω0 sin(φ + δ0 ) + (7.16)
τs
 
sin(φ + δ0 )
A′52 = C − ω0 cos(φ + δ0 ) (7.17)
τs
ΨF sin(φ + δ0 )
A′53 = C (7.18)
Ls
  
ω0 ΨF Iqs
A′54 = C cos(φ + δ0 ) ω0 Ids + + (7.19)
Ls τs
 
−Ids
+ sin(φ + δ0 ) + Iqs ω0 (7.20)
τs
1
A′55 = C [Ids sin(φ + δ0 ) − Iqs cos(φ + δ0 )] − (7.21)
τh

231
where C = 32 ke Vs , τh is the High-Pass Filter (HPF) time constant and Vs is the steady-
state voltage vector magnitude. Here Ids , Iqs , ω0 , and δ0 denote the steady-state values
of corresponding variables. The eigenvalue plot of the linearized PMSM model with
reference frequency modulation is shown in Fig. 7.3a. The eigenvalues are restricted
to the left side of the s-plane, thus achieving stable operation for the entire speed range
of operation. Both the HPF time constant (τh ) and the feedback gain constant (CG )
are control system variables, which are adjusted to obtain different characteristics of
eigenvalue plot with operating frequency variation. The values are designed so as to
restrict the eigenvalue variation with operating frequency to the left side of the s-plane,
and hence achieve stable operation. The HPF is used to extract the oscillations in power
and remove the low-frequency components.
Initially, the HPF time constant (τh ) is fixed at 0.159 s (1 Hz), and the feedback gain

50 50
100 Hz
Imaginary axis (s )

Imaginary axis (s )

Stator Poles
-1

-1

Stator Poles
100 Hz

0 0

Rotor Poles Rotor Poles

-50 -50
-100 -50 0 -100 -50 0
-1 -1
Real axis (s ) Real axis (s )
(a) CG = 0.1 (b) CG = 0.5
600
Imaginary axis (s )

400
-1

26 Hz
200
Stator Poles
0
-200
-400
Rotor Poles
-600
-150 -100 -50 0
-1
Real axis (s )
(c) CG = 3

Fig. 7.3: Eigenvalue loci of state transition matrix as a function of ωi with reference
frequency modulation for various CG , where τh is fixed at 0.159 s

232
(CG ) is varied to obtain stable operation. Another approach for design is by fixing CG
and varying τh to achieve stable operation. The plot of eigenvalue loci of A′ for various
CG is shown in Fig. 7.3. Finally, CG is chosen as 0.5 to obtain a nearly constant real
part for rotor poles, as observed in Fig. 7.3b. Hence, the small-signal stabilisation of
PMSM using I-f control is achieved by restricting the rotor pole movement to the left
side of the s-plane by improving system damping.
In the second approach, the LPF time constant is varied for a fixed CG . Fig. 7.4 shows
the eigenvalue loci of the state transition matrix with variation in reference frequency
for various values of τh (CG is fixed at 0.5). It is observed that any value of τh above
0.0051 s (31 Hz) gives a stable operation in the entire speed range. For τh above 0.159 s,
the damping factor remains almost constant. Further, it should be noted that for lower
values of τh , higher will be the attenuation to low-frequency variations. Finally, τh is
chosen as 0.159 s to obtain a nearly constant damping factor (constant real part for rotor
poles) and sufficient low-frequency rejection.
To analyse the pole slipping instability, the total system model which includes the Te −δ
curve is required. Hence, the small-signal model used for analysing the mid-frequency
instability, where the machine model is linearized about a particular operating point,
becomes insufficient. As a result, the nonlinear large-signal model of PMSM with
open-loop control is needed to analyse the pole slipping during frequency ramp-up.
To simplify the analysis, a large-signal model with only the rotor poles is considered.
The large-signal model considered in the study includes the Te −δ curve capturing the
pole slipping phenomenon. The effect of large-signal dynamics is well captured by
considering the rotor poles alone.

7.3 LARGE-SIGNAL DYNAMIC MODEL OF PMSM WITH I-F CONTROL

As the system is nonlinear, a linearized model of PMSM with I-f control will not
suffice to perform a large-signal stability analysis. Hence, a nonlinear large-signal
dynamic model is simulated in MATLAB/Simulink to observe the pole slipping during

233
 50 100 Hz 50 100 Hz

Imaginary axis (s )

Imaginary axis (s )
-1

-1
Stator Poles Stator Poles

0 0
Rotor Poles Rotor Poles

-50 -50
-100 -50 0 -100 -50 0
-1 -1
Real axis (s ) Real axis (s )
(a) τh = 15.92 s (b) τh = 1.592 s

50 50
100 Hz
Imaginary axis (s )

Imaginary axis (s )
Stator Poles
-1

-1
100 Hz
Stator Poles
0 0
Rotor Poles
Rotor Poles
-50 -50
-100 -50 0 -100 -50 0
-1 -1
Real axis (s ) Real axis (s )
(c) τh = 0.159 s (d) τh = 0.0159 s

50 100 Hz 50 100 Hz
Imaginary axis (s )

Imaginary axis (s )
-1

-1

18 Hz
Stator Poles Stator Poles
0 0
Rotor Poles Rotor Poles

-50 -50
-100 -50 0 -100 -50 0
-1 -1
Real axis (s ) Real axis (s )
(e) τh = 0.0051 s (f) τh = 0.0016 s

Fig. 7.4: Eigenvalue loci of state transition matrix as a function of ωi with reference
frequency modulation for various τh , where CG is fixed at 0.5

frequency ramp-up. The block diagram of the large-signal dynamic model of PMSM
with conventional I-f control is shown in Fig. 7.5a. The load torque is ramped up with
time to a near rated value of the motor during the speed ramp-up interval as shown
in Fig. 7.6a. As long as Td <Tmax , the system remains stable. When the load torque
increases, the electromagnetic torque also increases to maintain a stable equilibrium.
Since there is no feedback to control the frequency ramp slope, the motor must provide

234
 d
d ωr ωr
ω
P 1 dt r 1 ωr Tl
P 1 dt 1
Tl 2J s Te
2J s
Te
Tmax
Tmax δ θr 1
δ θr sin s
1
sin s ωref 1 π
1 π s 2
s 2
θi
θi ωi
ωi ∆ωi ke HPF
(a) Conventional I-f control (b) With frequency perturbation

Fig. 7.5: Large-signal dynamic model of PMSM with open-loop control

200

ω (elec rad/s)
Torque (Nm)

Te Tl Tmax 200 ωr
100
100 ωref
0
0
-100
0 1 2 3 4 0 1 2 3 4
time (s) time (s)
(a) Torque variation (b) Speed variation

Fig. 7.6: Pole slipping during frequency ramp-up with conventional I-f control

a constant acceleration torque independent of the load torque. Whenever the total
torque demand exceeds the maximum motor torque capability (Td >Tmax ), pole slipping
occurs, and the speed drops to zero, as shown in Fig. 7.6b.
Speed and torque oscillations with the open-loop I-f control are stabilized by adding
damping to the system using a reference frequency modulation module, as shown in
Fig. 7.5b. Smooth starting of the motor without speed and torque oscillations in the
initial light load region is observed in Fig. 7.7a and Fig. 7.7b. The reference frequency

200
Torque (Nm)

ω (elec rad/s)

Te Tl Tmax 400 ωr
100 ωref
200
0
-100 0
0 2 4 0 2 4
time (s) time (s)
(a) Torque variation (b) Speed variation

Fig. 7.7: Pole slipping with frequency modulated I-f control

235
modulation module reduces the reference speed during a sudden increase in power
demand. However, if a continuous high torque demand occurs during the ramp-up
interval, the stabilization loop becomes passive due to the action of the HPF. Finally,
pole slipping occurs, as shown in Fig. 7.7a and Fig. 7.7b, once Td exceeds Tmax . Thus,
the small-signal stabilization of the I-f control with reference frequency modulation is
insufficient to avoid pole slipping during ramp-up. Therefore, an additional control
together with reference frequency modulation is required to guarantee a reliable start of
PMSM using the I-f control.

7.4 PROPOSED DYNAMIC FREQUENCY SLOPE CONTROL USING

TORQUE CONTROLLER

In this work, a torque controller is proposed to dynamically vary the frequency slope
depending on the load torque. As analysed previously, pole slipping occurs whenever Td
exceeds Tmax during the ramp-up interval. Using the proposed torque controller, during
an increased load torque demand, the J P2 αi component of (7.4) is reduced by reducing
the frequency slope. The slope of the frequency ramp (αi ) is dynamically varied to
limit Td within a safe level (below ks Tmax shown in Fig. 7.1) to ensure stable operation.
Here, ‘ks ’ is a constant, whose value lies in the range 0<ks <1. The reference frequency
is obtained by integrating the frequency slope output from the torque controller. Small-
signal stabilization using reference frequency modulation should also be included in
the control for overcoming the mid-frequency instability. The total control system for a
quick and stable I-f control with the proposed dynamic slope variation together with the
reference frequency modulator and slope attenuator is shown in Fig. 7.8. The function
of the slope attenuator will be discussed in Section 7.5.

7.4.1 Torque controller

The torque controller is realised using an integral control together with a unidirectional
proportional control. The total torque demand is estimated and fed to the torque
controller to control it within the maximum torque capability. To prevent pole slipping

236
 Vdc

vd′ vαs
i′d ref =0 PI d′ q ′ Inverter
SVPWM
vq′ vβs Module
i′q ref =Ilim PI αβ
i′q iαs ias
d′ q ′ αβ ibs
i′d iβs
αβ abc
iαs
∆ωi Reference iβs
frequency vαs
slope vβs PMSM
attenuator modulator
ωi
θi ωref ωi βA ki Td iαs
Torque iβs
s
αres αi sw Estimator vαs
0 kp
ksTmax vβs
Unidirectional PI
Fig. 7.8: Control system with proposed torque controller and slope attenuator

of PMSM, Td is controlled below a safe torque level ks Tmax (where 0<ks <1) by the
torque controller. The effect of parameter variations and measurement errors on the
estimated torque should be considered while choosing ks . The region 0<Te <ks Tmax
is defined as the zone of operation, and the region ks Tmax <Te <Tmax is the zone of
non-operation, as shown in Fig. 7.1. In the zone of operation, the control action is
to increase the frequency slope such that Te increases to reach ks Tmax . Whereas in
the zone of non-operation, the control action is to decrease the frequency slope and
maintain stability. Hence, the larger the zone of non-operation (lower value of ks ), the
better will be the system stability with poor dynamic performance. Consequently, ks
should be chosen judiciously depending on the error in the estimated torque to prevent
the total torque demand (Td ) from exceeding Tmax . At the steady-state operation of
the torque controller, the machine accelerates with a developed torque of ks Tmax . The
estimated total torque demand at steady-state is given by (7.17).

Tb d = ks Tmax (7.17)

237
The steady-state frequency slope (αi(ss) ) is obtained as (7.18) by substituting (7.17) in
(7.3).
P (ks Tmax − Tl − P2 Bωi − Tf )
αi(ss) = (7.18)
2J

The unidirectional proportional control is used not to achieve a fast response but to
restrict operating region within the zone of operation. The noise in the estimated total
torque demand (Tb d ) feedback can cause overshoot in the frequency slope due to the
proportional control action, leading to pole slipping. Hence, the proportional control
is disabled when the estimated total torque demand (Tb d ) is less than the safe operating
limit ks Tmax . The slow action of the integral control filters out the high-frequency noise
in Tb d feedback from affecting the output frequency slope. Thus, the frequency slope
rises from zero to the steady-state value given by (7.18).
The system model from the controller point of view is derived for designing the
proposed torque controller. For synchronous machines under stable operation and
neglecting the high-frequency variation,

dωr
ωi = ωr ; αi = (7.19)
dt

The motor model required for the torque controller design is obtained as

2 2
Te = Tl + B ωi + Tf + J αi (7.20)
P P

Considering B P2 ωi , Tf and Tl as disturbances to the controller, the system model

reduces to
2
Te = J αi (7.21)
P

Even though the total large-signal model is nonlinear, the system model required for
controller design reduces to a linear system model as given by (7.21). Hence, the control
system is designed using the concepts of linear control theory. The total system model
with integral control and unidirectional proportional control is shown in Fig. 7.9. After

238
 1 2
kp
sw
0 Tl
ksTmax ki αi
J Td
s

Fig. 7.9: Control system with proposed torque controller

designing the control system, the stability of the total nonlinear system is verified by
simulation of the large-signal model with the proposed controller.
The proposed torque controller is designed such that it should not interfere with
the reference frequency perturbation, which is already present to overcome the mid-
frequency instability of the system. Hence, the integral control bandwidth is set to be
lower than the lowest frequency of small-signal stabilization. The simulation of the
large-signal model with the open-loop I-f control shows that the frequency of torque
and speed oscillations is 2.5 Hz, as observed from Fig. 7.6a and Fig. 7.6b, respectively.
The integral control bandwidth is chosen as 0.16 Hz so that it will not respond to
2.5 Hz oscillations from the reference frequency modulator. The transfer function of
the closed-loop system with integral control has a first order response with switch ‘sw’
connected to 2 in the system model shown in Fig. 7.9. The integral gain (ki ) is given by
(7.22).
1
ki = (7.22)
Jτfo

where τfo is the equivalent first-order system time constant. The frequency response plot
of the closed-loop system transfer function with the proposed integral control is shown
in Fig. 7.10. It is observed that the oscillations in reference frequency modulation have
an attenuation of −23 dB. Thus, the proposed integral control is passive to the action of
the reference frequency modulator.
The limits of the torque controller decide the maximum and minimum slope of the
reference frequency curve. Since the torque controller is intended to operate in the
“ramp-up" interval, it should not cause a “ramp down" (negative frequency slope)
during its operation. Hence, the lower limit is kept as zero to prevent speed decrease

239
where kp is the gain of proportional control. The proportional control should be
designed to ensure stability in the ramp-up interval even under worst-case loading.
When Tb d hits the actual stability limit of the drive (which is the torque capability Tmax ),
the output of proportional control should fully cancel the integrator output, forcing the
frequency slope to zero. Hence, the proportional control gain is derived by equating
the maximum integrator output (7.23) to the proportional control output (7.24) under
worst-case loading (substitute Tb d = Tmax ).

P ks Tmax P ks
kp = = (7.25)
2J(Tmax − ks Tmax ) 2J(1 − ks )

Finally, the torque controller output after integration (ωi ) and the output of the reference
frequency perturbation module (∆ωi ) are combined to obtain the reference speed for the
I-f control as (7.26).
ωref = ωi + ∆ωi (7.26)

The large-signal dynamic model of PMSM using I-f control with the torque controller
as shown in Fig. 7.11 is simulated. The value of ks is selected as 0.8 for simulation
and hardware experiments. The machine accelerates with high slope values in the
initial light load region, ensuring good dynamic performance, as shown in Fig. 7.12a
and Fig. 7.12b. Whenever the developed torque exceeds ks Tmax , the torque controller
limits the frequency slope for maintaining developed torque at ks Tmax . Hence, a stable
operation is achieved even during large load torque by accelerating with a reduced
frequency slope.

7.4.2 Total torque demand estimation

An accurate estimation of Td is critical for the torque PI controller. During stable

operation, the developed torque is equal to the total torque demand. Hence, the total

241
 d
ωr ωr
P 1 dt 1
Tl 2J s
Te
Tmax θr
δ 1
sin s
αi 1 ωi ωref 1 π
ksTmax PI s s 2
θi
Proposed controller ∆ωi ke HPF

Fig. 7.11: Large-signal dynamic model of PMSM using I-f control with torque
controller

ω (elec rad/s)
Torque (Nm)

400 40

slope (Hz/s)
50 ωr
Tl Tmax 200 αi 20
ωref
Te ks Tmax
0 0 0
5 10 15 20 5 10 15 20
time (s) time (s)
(a) Torque variation (b) Speed variation

Fig. 7.12: Stable operation during frequency ramp-up using I-f control with torque
controller

torque demand for SMPMSM is estimated as (7.27).

2P b
Tb d = ΨF iq (7.27)
32

where biq is the estimated q-axis current. The estimated q-axis current is obtained
by transforming the stator current into the rotor reference frame using an estimated
rotor position. Here, the rotor position is estimated by the stator voltage integration
technique (Hurst et al., 1998), which is commonly used in sensorless vector control. In
the proposed dynamic frequency slope control, the sensorless algorithm is used only
to estimate the rotor position for calculating the estimated torque demand and not
for sensorless vector control (for obtaining the control system transformation angle).
Hence, the open-loop I-f control of the PMSM drive operation remains unaffected by

242
the low-speed issues of the sensorless vector control while using the proposed method.
The stator flux space vector is computed as

Z
~s =
Ψ (~vs − ~is Rs )dt (7.28)

where ~vs is the stator voltage space vector, and ~is is the stator current space vector.
~ r ) is derived from the stator flux space vector as
The rotor flux space vector (Ψ

~r = Ψ
Ψ ~ s − Ls~is (7.29)

The cosine and sine of the rotor position are estimated from the α-component of the
b αr ) and β-component of the rotor flux space vector (Ψ
rotor flux space vector (Ψ b βr ),

respectively.
Ψb αr Ψb βr
cos θbr = ; sin θbr = (7.30)
~ r|
|Ψ ~ r|
|Ψ
~ r | is the magnitude of rotor flux space vector.
where |Ψ
An integrator with feedback is used by Hurst et al. (1998) to minimizes the output drift
problem and stabilize the integrator action. However, at low speeds of operation, the
Rs~is drop becomes comparable to the applied voltage, which may result in a significant
error in the estimated position. Also, in the low-speed region, there will be an error
in output magnitude and phase due to the proximity of the operating point to the filter
cut-off frequency. In order to overcome the above difficulties, a flux estimation scheme
for IM proposed by Bhattacharya and Umanand (2006) is adapted for operation with
PMSM in the present work. When an error is present in the rotor flux estimation,
the feedback loop acts to minimize the error. The feedback loop is eliminated and
the resulting system acts as a pure integrator whenever the error becomes zero. An
e αr and Ψ
independent method is used to estimate the rotor flux position (Ψ e βr ), which is
b αr and Ψ
compared with the output of stator voltage integration (Ψ b βr ) and then used as

the feed back for the integrator. Fig. 7.13 shows the computation of the α-component

243
 vαs -Rs iαs b αr
Ψ

Lsiαs
kc

e αr
Ψ
Fig. 7.13: Flux estimation scheme

e αr ). Independent calculation of the rotor flux position for

of rotor flux position (Ψ
e αr and Ψ
comparison (Ψ e βr ) is obtained from

e αr = ΨF cos θbr ; Ψ
Ψ e βr = ΨF sin θbr (7.31)

where ΨF is the PM flux linkage which is a constant parameter of the machine and
~ r |). θbr represents the estimated
equal to the magnitude of the rotor flux space vector (|Ψ
rotor position using sensorless algorithm. Hence, the method stabilizes the integrator
action with the correction mechanism.
The estimation error of stator voltage integration during low-speed operation should be
taken care of by choosing a proper value of ks in the torque controller.

7.5 SLOPE ATTENUATOR FOR SMOOTH TRANSITION FROM

ACCELERATION INTERVAL TO CONSTANT SPEED INTERVAL

The back-emf sensorless vector control cannot operate in drives requiring a continuous
low-speed operation. For these systems, I-f control with a constant speed interval can
be used. In such cases, an oscillation in speed is observed during the transition from the
ramp-up interval to constant speed operation (Perera et al., 2003; Haichao et al., 2017).
The oscillation in speed results in higher current and torque oscillations depending on
the magnitude of change in frequency slope during the transition.

244
The speed oscillation is produced as a result of unwanted interference of the reference
frequency modulation control at the transition instant. The output of the reference
frequency modulation module is represented as (7.11).
The input power is expressed as

2 2 2
Pe = (Tl + B ωi + Tf + J αi ) ωi (7.32)
P P P

Let the frequency ramp slope at the end of the ramp-up interval be α1 . During the
transition from the ramp-up interval to the constant speed interval, αi suddenly drops
from α1 to zero, resulting in a sudden decrease of input power as given in (7.32).
Therefore, the reference frequency modulation module reacts with an increase in ∆ωi .
The quick increase in ωref according to (7.26) results in the speed oscillation at the
transition. Even though the reference frequency modulation module imparts small-
signal stability to the system, it also causes speed oscillation during the transition. The
magnitude of oscillations depends on the frequency slope at the end of the ramp-up
interval (α1 ). The higher the value of α1 , the larger is the oscillation in current and
torque. To study this effect, the large-signal dynamic model of PMSM using I-f control
with the torque controller in Fig. 7.11 is simulated with 50 Nm continuous load torque.
Whenever the motor speed reaches the reference speed, the frequency slope suddenly
drops to zero. This results in oscillation of both developed torque and motor speed, as
observed from Fig. 7.14a and Fig. 7.14b.

Tmax ks Tmax
ω (elec rad/s)
Torque (Nm)

100 400 40
slope (Hz/s)

50 200
ωr 20
αi
Tl Te ωref
0 0 0
5 10 15 20 5 10 15 20
time (s) time (s)
(a) Torque variation (b) Speed variation

Fig. 7.14: Speed and torque oscillations at the transition interval using I-f control with
torque controller

245
A slope attenuator is proposed in this work, which ensures a seamless transition from
the ramp-up interval to the constant speed interval. The slope attenuator is activated
when the motor speed reaches 70% of the intended speed of steady-state operation. Let
the intended speed of steady-state operation be denoted as ωfinal . Once the reference
speed (ωref ) crosses 0.7ωfinal , the slope value (αi ) is multiplied by an attenuation
generated by the slope attenuator (βA ), as shown in Fig. 7.8. βA varies from 1 to 0
linearly as speed increases from 0.7ωfinal to ωfinal . Therefore, the resultant slope of the
frequency curve (αres ) decreases smoothly and finally reaches zero at the instant when
speed reaches ωfinal . The trajectory of slope decrease during slope attenuation depends
on the output of the torque controller module. The proposed slope attenuator is included
in the large-signal model of PMSM (Fig. 7.11) and simulated. A smooth variation of
motor speed and developed torque during the transition from the ramp-up interval to the
constant speed interval is achieved by attenuating the slope after reaching 70% of ωfinal ,
as observed from Fig. 7.15a and Fig. 7.15b.
Torque (Nm)

Tmax ks Tmax
ω (elec rad/s)

100

slope (Hz/s)
400 40
ωr
50 200
αi
20
ωref
Tl Te
0 0 0
5 10 15 20 5 10 15 20
time (s) time (s)
(a) Torque variation (b) Speed variation

Fig. 7.15: Smooth transition from ramp-up interval to constant speed interval using
slope attenuator

7.6 EXPERIMENTAL RESULTS

Hardware validation of the proposed dynamic slope control and the slope attenuator
is carried out on a 25 kW, 400 V, 3000 rpm, 16-pole, 3-phase star connected PMSM
drive. A DC generator is coupled to the PMSM for loading. The laboratory setup of
the 25 kW inset-PMSM drive is shown in Fig. 7.16. The parameters of the motor and
the controller are given in Table. 7.1. Texas Instrument’s TMS320F28335 DSP is used

246
Fig. 7.21: Stable operation of I-f control with reference frequency modulation
b r ; Ch4 - ias (X-axis:- 1.44 s/div;
stabilization: Ch1- ωref Ch2 - Pe ; Ch3 - ω
Y-axis:- Ch1,Ch3: 37.7 elec rad/s/div; Ch2: 8 kW/div; Ch4: 40 A/div)

together with the reference frequency perturbation algorithm, as shown in Fig. 7.24.
The torque controller ensures stable operation by dynamically varying the frequency
slope depending on the load torque. Td is maintained at ks Tmax by the torque controller
using the estimated total torque demand (Tb d ), as shown in Fig. 7.25. The proportional
control gets activated whenever Tb d goes beyond ks Tmax during an increase in load
torque. As observed from Fig. 7.25, when a sudden load is applied, the frequency slope
is decreased by the fast action of proportional control, thereby reducing Tb d . Stable
operation of the drive is achieved since Tb d is always controlled below the maximum
torque capability of the PMSM. After the machine reaches sufficient speed, the control
is transferred to sensorless vector control using the pulse-off method proposed in
Chapter 6. After changeover to sensorless vector control, the machine accelerates with
Tb d equal to Tmax . Since the drive is operated in closed-loop sensorless vector control
during the medium and high-speed range, pole slipping is not present in that speed
range.
As discussed before, the machine is operated continuously with the I-f control, without
changeover to sensorless control in some applications. In such cases, speed oscillation
and a consequent current oscillation occur during the transition from the ramp-up to

250
 CHAPTER 8

HARDWARE ORGANISATION

In this thesis, two hardware setups were used to demonstrate the proposed methods.
The first setup consists of a 25 kW inset-PMSM drive which is used to demonstrate
the performance of sensorless starting methods and stabilisation. The other setup
includes a low power custom-designed reconfigurable IPMSM drive (configured as
either a 3 kW DTP0-IPMSM (Vdc =300 V) or a 1.5 kW (Vdc =200 V) uneq0-IPMSM)
which was used to validate the proposed methods pertaining to EVs. The complete
hardware organisation of the reconfigurable IPMSM drive is provided in this chapter.
Out of the various drive configurations possible, the 1.5 kW uneq0-IPMSM setup is used
to explain the hardware organisation in detail since it includes the maximum number
of components. The hardware setup for the DTP0-PMSM forms a subset of the uneq0-
PMSM drive.

8.1 POWER CIRCUIT

The complete power circuit of the 1.5 kW uneq0-IPMSM is given in Fig. 8.1. The
power circuit consists of a custom designed 1.5 kW uneq0-IPMSM, an 11 kW IM, two
numbers of 2-level VSIs, a 7.5 kW loading drive, a 10 kVA autotransformer, a thyristor

Pre-charging Diode 7.5kW

circuit bridge 10kVA 11kW 3-φ
Si IGBT 1.5kW Induction Danfoss
rectifier 2-level VSIs uneq0-PMSM motor Drive
ias1
ibs1
Vdc S N IM
Grid ias2

10kVA ibs2 Brake

Auto DSP chopper
transformer Controller
position Thyristor
switch box
Grid

Fig. 8.1: Complete power circuit of 1.5 kW uneq0-IPMSM drive test bench
switch box and a pre-charging circuit. The photograph of the complete 1.5 kW uneq0-
IPMSM setup is shown in Fig. 4.23 in Chapter 4. Each component in the power circuit
is explained in the following sections.

8.1.1 Custom designed reconfigurable 0 deg displacement 4-split IPMSM

The reconfigurable IPMSM was developed by replacing only the rotor of a 5.5 kW
standard IM with a PM rotor. The stator winding was also rewound to form four sets
of three-phase windings for the reconfigurable IPMSM. The IPMSM rotor was custom
designed considering the inner diameter of the original stator and outer diameter of the
original shaft. The test bench showing reconfigurable 4-split IPMSM coupled to the
11 kW IM is shown in Fig. 8.2.
Various parts of the custom designed reconfigurable 0◦ displacement 4-split IPMSM are
described as follows.

Fig. 8.2: Custom designed reconfigurable 4-split IPMSM coupled to an IM

Stator structure

The original IM was disassembled to obtain the stator core. The original windings were
removed, and the slots were cleaned. The stator core of the original IM is shown in
Fig. 8.3a, which has 48 slots. The stator has parallel teeth with round bottom slots. The
slots were inserted with 5 mil Nomex insulation for slot lining, as shown in Fig. 8.3b.

256
 (a) Cleaned stator core (b) Stator with slot insulation

Fig. 8.3: Stator structure of 4-split IPMSM

Winding diagram

The stator winding of the IPMSM machine is made in a 1:1:1:1 split ratio, so that it
can be configured as either three-phase IPMSM, DTP0-IPMSM, or uneq0-IPMSM by
changing the terminal connection. The three-phase IPMSM is realised by connecting all
the winding sets in series (1+1+1+1) for each phase. The DTP0-IPMSM is obtained by
connecting the windings to obtain a 1:1 split (1+1:1+1), whereas a 3:1 split (1+1+1:1)
is used in the uneq0-IPMSM.
The winding diagram for one set out of the four sets is shown in Fig. 8.4. The other
winding sets are placed one over the other in the same slot with a phase insulation
separation between them. A single layer full pitch winding is used for each winding
set. Each winding set consists of 72 turns per phase with 9 turns per coil. Each coil is
made of two 22 SWG dual coated enamelled copper wires (Grade-2).
Phase separators made of 5 mil Nomex insulation are used for the separation of winding
sets within the slot as shown in Fig. 8.5a. Fibreglass cloth insulation is used as the end
winding separator, as shown in Fig. 8.5b. The terminal box with termination for start
and end windings of four three-phase winding sets, constituting a total of 24 terminals,
is shown in Fig. 8.6.

257
 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48

W1−end U1−start V1−start W1−start U1−end V1−end

W2−end U2−start V2−start W2−start U2−end V2−end
W3−end U3−start V3−start W3−start U3−end V3−end
W4−end U4−start V4−start W4−start U4−end V4−end
U1 = 9 turns/coil (2 terminals) V1 = 9 turns/coil (2 terminals) W1 = 9 turns/coil (2 terminals)
U2 = 9 turns/coil (2 terminals) V2 = 9 turns/coil (2 terminals) W2 = 9 turns/coil (2 terminals)
U3 = 9 turns/coil (2 terminals) V3 = 9 turns/coil (2 terminals) W3 = 9 turns/coil (2 terminals)
U4 = 9 turns/coil (2 terminals) V4 = 9 turns/coil (2 terminals) W4 = 9 turns/coil (2 terminals)

Total = 24 terminals

Fig. 8.4: Winding diagram for the 4-split IPMSM

(a) Phase insulation (b) End winding insulation

Fig. 8.5: Insulation at various parts of 4-split IPMSM

Rotor structure

The designed IPMSM rotor with four poles is shown in Fig. 8.7a. The rotor is made
of M36 grade steel laminations with 0.35 mm thickness. The laminations are stacked
together and riveted to form a solid structure. The complete rotor with N35AH grade
NdFeB magnets inserted is shown in Fig. 8.7b. The individual magnets are shown in
Fig. 8.8. A total of eight magnets are required since each pole is formed by stacking
two magnets along the length of the machine. The shaft of the original IM was removed

258
 Fig. 8.6: Terminal box of 4-split IPMSM (top-view)

(a) Without magnets (b) With magnets inserted

Fig. 8.7: Rotor structure of 4-split IPMSM

and inserted into the PM rotor with a hydraulic press.

Accessories

A 10-bit absolute encoder (Autonics: EPM50S-8-10-13-B-S-24) is coupled to the non-

driving end of the IPMSM for obtaining the rotor position during sensored vector
control. The absolute encoder is interfaced with the DSP using Synchronous Serial
Interface (SSI). The absolute encoder and its mounting arrangement are shown in
Fig. 8.9a. A 220 V, 3-phase, 2900 rpm, 529 CFM axial cooling fan (SEM:8AF3B2)

259
 (a) (b)

Fig. 8.8: N35AH grade NdFeB magnet

(a) Rotary absolute encoder (b) Axial cooling fan

Fig. 8.9: Accessories for IPMSM motor

is also fixed outer to the absolute encoder, as shown in Fig. 8.9b.

8.1.2 Loading setup

The shaft of the uneq0-IPMSM is coupled to an 11 kW, three-phase induction machine,

which is driven by a 7.5 kW AC drive (Danfoss make FC302P7K5) for loading. The
two motors are coupled using a jaw coupling and mounted on a base plate, as shown
in Fig. 8.10a. The induction machine operates as an induction generator to act as a
load for the uneq0-IPMSM. Since the DC bus of the AC drive is derived using a diode
bridge rectifier, the power generated by the induction machine cannot be fed back to
the grid. To restrict the DC bus voltage to a safe operating limit during regeneration, a

260
 (a) 11 kW Induction machine (b) FC302P7K5 Danfoss AC drive

(c) Brake resistor

Fig. 8.10: Loading setup

brake chopper (Resistor with a series IGBT) is connected across the DC bus. The brake
chopper is shown in the power circuit given in Fig. 8.1. The brake resistor bank consists
of five resistor modules connected in parallel to form a 4.4 kW, 63 Ω load. Photographs
of the 11 kW induction machine, Danfoss AC drive and brake resistor are shown in
Fig. 8.10a, Fig. 8.10b and Fig. 8.10c, respectively.

8.1.3 IGBT based 2-level VSI

Two custom made 10 kVA Si IGBT based 2-level VSIs with a common DC link are used
to supply the uneq0-IPMSM, as shown in Fig. 8.1. Semikron make SKM100GB12T4
IGBT (100 A, 1200 V) modules constitute each leg of the 2-level VSI. A three-phase

261
diode bridge rectifier (Semikron:SKD110/12) is used to derive the DC bus from the
three-phase grid. The photograph of the two 2-level VSIs with the diode bridge rectifier
is shown in Fig. 8.11. A sandwich bus bar is used in the inverters to reduce the parasitic
inductances between the DC link capacitors and the IGBTs. Each inverter also includes
hall effect based current and voltage sensors. LEM make LA 100-P is chosen as the
current sensor, and LV-25P is used for sensing the DC bus voltage and the line-line
voltages. Appropriate burden resistances are provided at the output of hall effect current
and voltage sensors to convert the output current signal to a voltage signal. The burden
resistors are chosen such that 100 A current through the current sensor and 1000 V
across the voltage sensor corresponds to 5 V across the corresponding burden resistance.
An encoder interface board is also provided to level shift the 10 V single-ended signal
from the DSP board to 5 V differential signal for the absolute rotary encoder and vice-
versa. A photograph of the current sensors, voltage sensors and encoder interface board
is shown in Fig. 8.12a.
As shown in Fig. 8.12b, a pre-charging circuit is provided at the input of the diode
bridge rectifier to limit the initial charging current of the DC bus capacitor when the

Fig. 8.11: Two 2-level VSI with common DC bus

262
 (a) Current sensors, voltage sensors and encoder interface board

(b) Pre-charging circuit (c) SMPS

Fig. 8.12: Inverter components

system is initially powered up. When the drive is powered up, the switches in the pre-
charging circuit are kept open. Hence, the pre-charging resistors would be connected in
series with the input of the diode bridge rectifier, increasing the charging time constant
of the DC bus capacitor. When the DC bus capacitors are charged to a sufficiently
high voltage (near the steady-state voltage), the switches in the pre-charging circuit are
closed to bypass the resistors. Here, the pre-charging resistors will be present only
during the initial charging of the DC bus capacitors and will not interfere with the drive
operation.
DIN rail mount Switched Mode Power Supplies (SMPS) (+15 V, −15 V, and +5 V) are
used to power the control boards, gate driver boards, current sensors, voltage sensors,
encoder and encoder interface board. MEAN WELL make HDR-60-15, HDR-30-15,

263
and HDR-30-5, as shown in Fig. 8.12c, are used to derive +15 V, −15 V, and +5 V,
respectively.

8.1.4 Thyristor switch box

Back-to-back connected thyristors are used as bidirectional switches to disconnect

the LS winding of the uneq0-IPMSM during high-speed operation. The thyristor
switch box which includes the back-to-back connected thyristors, gate drivers, isolation
transformers and cooling fan is shown in Fig. 8.13. Semikron make SKKT 162/22EH4
thyristor modules were used to realise the bidirectional switches. The thyristors have
an average current rating of 101 A and 2.2 kV blocking voltage. The high voltage rating
of the thyristor switches is essential to block the high induced emf of the LS winding
during high-speed operation. Since each thyristor module consists of two thyristors
connected in series, the anode of the bottom thyristor is shorted to the cathode of the
top thyristor to form a back-to-back configuration. An RC snubber (Rsnb = 100 Ω,
dv
30 W; Csnb = 10 µF) is connected across the thyristors to limit the voltage spike and dt

during reverse recovery.

A custom-made gate driver board is used to drive the thyristor switches. Each gate
driver board is capable of driving two thyristors. A high voltage optoisolator (OPI1268)
is used to provide isolation between the receiver circuit and the driving circuit of the

Fig. 8.13: Back-to-back thyristor switch box

264
gate driver board. The isolated power supply for the driving circuit is derived using a
230/18V isolation transformer (10kV isolation between primary and secondary), whose
output is rectified using a diode bridge rectifier and regulated using a voltage regulator
IC (L7815AB). The gate current for the thyristor is provided using a 31.3 Ω resistor
connected in series with a 15 V supply and a high gain power transistor. The minimum
gate current pulse duration of SKKT 162/22EH4 is 3 µs. Hence, gate pulses of 25 kHz
with 15% duty (6 µs on-time) is applied to the thyristor gate driver during the thyristor
on period. The pulsed waveform reduces the heat generation in the gate resistors
compared to continuous gating and also ensures reliable turn-on of the thyristors.

8.1.5 Autotransformer

A 10 kVA autotransformer as shown in Fig. 8.14 is used to step down the grid voltage
and regulate the DC bus voltage to 200 V. The autotransformer is connected between
the grid and the pre-charging circuit, as shown in Fig. 8.1.

Fig. 8.14: 10 kVA autotransformer

8.2 CONTROL CIRCUIT

All the PMSM drives used in this thesis are controlled using a DSP-based control card
(DSP-FPGA control platform) and a protection board. The details of the DSP-FPGA
control platform and protection board are discussed as follows.

265
8.2.1 DSP-FPGA control platform

The DSP-FPGA control platform consists of Texas instruments’ TMS320F28335 DSP,

and Altera make Cyclone IV Field Programmable Gate Array (FPGA), as shown in
Fig. 8.15. Since the DSP in the DSP-FPGA control platform was sufficient to satisfy all
the control requirements proposed in this thesis, the FPGA section was not activated.
All the control algorithms for the three-phase PMSM, DTP0-PMSM, and uneq0-PMSM
drive were executed within an interrupt service routine with 50 µs execution time in the
DSP. The main features of the DSP section in the control platform are given below.
• GPIO pins of the DSP are configured to form 13 digital inputs and 13 digital
outputs. The digital outputs are level shifted from 3.3 V at the DSP side to 15 V
at the connectors. Similarly, the digital inputs are level shifted from 15 V at
the connectors to 3.3 V at the DSP. The digital input signals include controller
start/stop and fault input from the protection board. The thyristor enable/disable
signal is generated using a digital output.
• Both uneq0-PMSM and DTP0-PMSM require four current sensors and a DC bus

Fig. 8.15: DSP-FPGA board for drive control

266
 voltage sensor, whereas the ‘pulse-off’ method for sensorless starting of three-
phase PMSM requires two current sensors, two line voltage sensors and a DC bus
voltage sensor. 16 analog inputs available at the DSP side of the control platform
are sufficient for realising all configurations proposed in this thesis. A signal
conditioning circuit is used to convert ±10 V at the control platform connectors
to 0-3 V at the DSP.

• 12 PWM signals can be generated from the DSP side of the control platform
with provision for both single-ended output and differential output, both at 15 V
level. The dual inverter configuration used for both DTP0-PMSM and uneq0-
PMSM drive requires 12 PWM signals. The PWM signals are passed through the
protection card before applying to the IGBT gate drivers.

• Four Digital to Analog Converters (DACs) are provided in the control platform
to monitor the variables involved in the control. Serial communication using
the I2C protocol is established with the DSP using a serial interfacing DAC IC
(AD5625R). The output signals of the DAC IC are level shifted to ±10 V and can
be probed using a Digital Storage Oscilloscope (DSO).

• The absolute rotary encoder is interfaced with the DSP through SSI
communication using the SPI pins of the DSP. The DSP acts as master and the
encoder acts as slave during the communication.

• A JTAG communication is implemented in the control platform for programming

and monitoring the control variables from a computer using Texas Instruments’
Code Composer Studio.

8.2.2 Protection board

The protection board consists of an Altera make MAX II Complex Programmable

Logic Device (CPLD) (EPM1270T144C5N), burden resistors, slide and push-button
switches, potentiometers and signal conditioning circuits, as shown in Fig. 8.16. The
main purpose of this board is to provide protection against overvoltage, undervoltage
and over current faults. The board can sense up to eight analog inputs. In the uneq0-
IPMSM drive, the analog inputs are configured to measure 4 line currents and the DC
bus voltage. The sensed signals are compared with a reference signal (corresponding
to fault limit) using an op-amp comparator. Whenever the sensed signal crosses the
fault limit, the output fault signal is latched using a logic circuit in the CPLD. LEDs
are provided for fault annunciation. The references for setting fault limits are generated
using potentiometers.

267
 Fig. 8.16: Protection board

The PWM signals generated from the control platform at 15 V are level shifted to 3.3 V
and passed through the CPLD for protection. Further, it is again level shifted back to
15 V before passing to the IGBT gate drivers. Whenever a fault occurs, PWM signals
to all the inverters are disabled. The protection card also controls the relay in the pre-
charging circuit depending on the sensed DC bus voltage. The pre-charging resistors are
active when the DC bus voltage is less than the set ‘Under Voltage trip limit (UV limit)’.
Once the DC bus capacitors charge above the UV limit, the pre-charging resistors are

268
bypassed by closing the contactors across them. The 25 kHz PWM with 15% duty
for controlling the back-to-back thyristor switches is also generated by the CPLD. The
PWM is applied to the thyristor gate driver only if the enable signal from the DSP is
present; else, the thyristor pulse is disabled.

8.3 CONCLUSION

A detailed description of the hardware setup used for validating the proposed methods
was provided in this chapter. A custom designed and manufactured 4-split winding
IPMSM, which can be reconfigured as either a 3 kW DTP0-IPMSM (Vdc =300 V) or a
1.5 kW uneq0-IPMSM (Vdc =200 V) was used for testing. The test rig consisted of the
reconfigurable IPMSM coupled to a standard 11 kW IM, which was fed from a 7.5 kW
AC drive. Two custom made 2-level IGBT based VSIs and back-to-back thyristor
switches were used to drive the uneq0-IPMSM. Various stages during the manufacturing
of the reconfigurable IPMSM were also explained with photographs.

269
 CHAPTER 9

CONCLUSION

9.1 BACKGROUND

Permanent magnet synchronous motors (PMSMs) are widely used in EVs, HEVs,
marine propulsion, railway traction, and aerospace applications due to their high torque
density, high power density and high efficiency compared to other motors. Most of the
present day EVs use a DC-DC converter to boost the low battery voltage to achieve
a wide operating speed range for the PMSM. The DC-DC converters increase the
converter size and overall weight due to the associated bulky inductors. Also, certain
applications like medium/heavy-duty trucks require a wide constant power range of
nearly 25 times their base speed. The present work proposes solutions based on dual
three-phase topologies to enhance the power rating as well as achieve wide operating
speed range for the PMSM used in EVs.
The thesis also deals with sensorless starting issues of 3-phase PMSM drives. Certain
critical applications like marine propulsion and ‘emergency heat and smoke exhaust’
require only medium and high-speed operation and do not necessitate high dynamic
performance in the low-speed region of operation. I-f open-loop starting together with
closed-loop sensorless vector control for medium and high-speed operation provides
an elegant solution to satisfy such requirements. A direct changeover from the open-
loop I-f control to the sensorless algorithm can be performed to achieve quick starting,
assuming that the sensorless algorithm has already converged. However, this may
result in speed and current oscillations due to the error in position estimation caused by
parameter variations. A pulse-off starting method is proposed in this thesis to achieve
a quick and seamless transition from I-f to sensorless control, which is independent of
machine parameters. Further, a method to improve the stability of I-f starting during
the ramp-up interval is also proposed.
9.2 SUMMARY OF PRESENT WORK

The major contribution of the thesis can be summarised as follows.

• 0◦ displacement dual 3-phase PMSM (DTP0-PMSM) with six-step operation

was proposed to achieve double the power rating and operating speed extension
compared to conventional three-phase PMSM. The DTP0-PMSM can be obtained
by winding reconfiguration of a three-phase PMSM and feeding it with two
similar rated VSIs. The proposed configuration can eliminate the requirement for
DC-DC converters used in conventional commercial EVs, thus achieving reduced
converter size and reduction in overall vehicle weight. Further, the circulating
current associated with the split-phase PMSM (used for achieving double the
power rating compared to three-phase PMSM) while operated in six-step mode is
eliminated in the DTP0-PMSM.

• Unequal split 0◦ winding displacement dual 3-phase PMSM (uneq0-PMSM)

was proposed to achieve the wide operating speed requirements of medium and
heavy-duty trucks. The proposed configuration achieves all the performance
requirements of the vehicle, including high starting torque, high overload
capability and wide operating speed range. The DC bus voltage requirement
is reduced using the proposed method compared to the three-phase PMSM
and DTP0-PMSM configurations when designed for wide CPSR requirements.
Further, the designer has an additional degree of freedom to choose a suitable
torque-speed characteristic from different options available by varying the split
ratio of the uneq0-PMSM.

• The thesis also discusses the design of a DTP0-PMSM considering an EV

application. Initially, a target torque-speed characteristics was chosen to
satisfy all the specifications of EV, including acceleration time, gradability and
maximum speed of operation. An iterative design procedure in conjunction with
FEM simulation was followed to satisfy various design requirements. Various
considerations while deciding on different design choices during the design
procedure was explained in detail. Finally, the designed motor was simulated with
the target vehicle executing the modified Indian driving cycle to check whether
all the operating points are met by the designed motor.

• A ‘pulse-off’ method for a quick and seamless transition from I-f starting to stator
voltage integration based sensorless control is also proposed. In the proposed
method, all the inverter pulses are disabled for a short duration. Two line
voltage sensors are used to measure the back-emf after the stator currents fall
to zero. The rotor position and motor speed are directly estimated from the
measured back-emf. Thus, a smooth changeover is performed to sensorless
algorithm by estimating the rotor position independent of machine parameter
variations. Further, the same method can also be used to perform on-the-fly
starting of PMSM for implementing a power failure ride through control. The
measured back-emf also enables accurate PM flux linkage estimation during self
commissioning of PMSM drives.

271
 • A dynamic frequency slope control is also proposed to improve the stability
during the ramp-up interval of open-loop I-f controlled PMSM drives. The
proposed method uses a torque controller to maintain the load torque within the
maximum torque capability of PMSM by controlling the slope of the frequency
ramp. The frequency ramp slope is increased during light loads to achieve quick
starting, and the slope is reduced during heavy loads to prevent pole slipping.
The proposed method is used in conjunction with the pulse-off method to satisfy
the quick and reliable starting requirements for critical applications like marine
propulsion and ‘emergency heat and smoke exhaust’.

9.3 FUTURE SCOPE

The possible improvements and future scope of the presented work are as follows:-

• The proposed speed range extension with dual 3-phase PMSM was performed
with sensored vector control using an absolute position encoder. The sensorless
vector control options for the dual three-phase PMSM would improve the
reliability of the drive and hence can be considered as a possible extension of
the presented work.

• All the proposed methods related to dual 3-phase PMSMs were implemented on
a low power 3kW IPMSM prototype. The designed 37 kW air-cooled DTP0-
IPMSM for EV application is currently under manufacturing. The proposed
concepts will be tested on the 37 kW DTP0-IPMSM, and performance will be
evaluated to improve the design further in future iterations. Also, the design of
a liquid-cooled DTP0-IPMSM and liquid-cooled uneq0-PMSM for EV can be
considered for future work.

• All the presented work considered the motoring operation of PMSM with the
maximum possible load as the rated load. Operation under overloads for a short
duration and regeneration operation can also be thought of.

• Even though the effect of saturation was included in the LUTs used for optimal
control of DTP0-IPMSM and uneq0-IPMSM, the effect of cross-saturation and
temperature were not considered. These effects can also be included while
formulating the LUT to improve the drive performance and achieve better
efficiency.

9.4 CONCLUSION

This thesis has proposed equal and unequal split configurations with dual three-phase
PMSM to eliminate the conventional DC-DC converters from the electric vehicle
power train and to achieve a wide operating speed range for electric automotive and
medium/heavy-duty trucks. The thesis also proposed methods for improving the

272
starting performance and stability of three-phase PMSMs for marine propulsion systems
and other critical applications. The proposed methods for equal and unequal split dual
three-phase PMSMs were validated experimentally on a custom made 3 kW IPMSM
prototype, whereas the starting techniques of 3-phase PMSM were validated on a 25 kW
PMSM drive.

273
 APPENDIX A

Derivation of maximum generated torque per ampere in the

uneq0-IPMSM

Consider a three-phase IPMSM with ‘nLS + nHS ’ number of turns per phase. Assume
that the Maximum Torque Per Ampere (MTPA) is achieved at the rated current, when
the current space vector is at an angle ‘β1 ’ with respect to the q-axis. The generated
torque of a three-phase IPMSM can be expressed as

 
2P 3
Te−3φ = iqs_MTPA ΨF + (Lmd − Lmq )ids_MTPA (A.1)
32 2

where iqs_MTPA = Ilim cos β1 , ids_MTPA = Ilim sin β1 . Here, Ilim is the magnitude of
the rated current space vector. Also, ΨF , Lmd and Lmq are permanent magnet flux,
d-axis magnetising inductance, and q-axis magnetising inductance, respectively, of the
original three-phase IPMSM. The uneq0-IPMSM is derived by splitting the turns of the
original three-phase IPMSM in the ratio nLS : nHS . Here, ‘nLS ’ is the number of turns
of low-speed (LS) winding and ‘nHS ’ is the number of turns of the high-speed (HS)
winding. The torque equation of the uneq0-IPMSM when both the LS and HS winding
are operating is given by

  
2P nLS nHS 3
Te = iq-LS + iq-HS × ΨF + (Lmd − Lmq )
32 nLS + nHS nLS + nHS 2
 
nLS nHS
id-LS + id-HS (A.2)
nLS + nHS nLS + nHS

Both the three-phase IPMSM and the uneq0-IPMSM can produce the same output
torque since they have the same rated current, winding distribution and rotor structure.
Assuming that both three-phase IPMSM and uneq0-IPMSM are operated to produce
the same output torque, (A.1) and (A.2) can be equated as

nLS nHS
ids_MTPA = id-LS + id-HS (A.3)
nLS + nHS nLS + nHS
nLS nHS
iqs_MTPA = iq-LS + iq-HS (A.4)
nLS + nHS nLS + nHS

Rewriting (A.3) gives

(nLS + nHS )ids_MTPA = nLS id-LS + nHS id-HS (A.5)

Similarly,
(nLS + nHS )iqs_MTPA = nLS iq-LS + nHS iq-HS (A.6)

The resultant mmf vector of the original three-phase IPMSM is given by

# »
mmf3φ = (nLS + nHS )(ids_MTPA + jiqs_MTPA ) (A.7)

= (nLS + nHS )ids_MTPA + j(nLS + nHS )iqs_MTPA (A.8)

The location of the mmf space vector of the original three-phase IPMSM in ids − iqs
plane is shown in Fig. A.1.

# »
mmf3φ

β1

Fig. A.1: Location of the mmf space vector during MTPA operation of the three-phase
IPMSM

Substituting (A.6) and (A.5) in (A.8) gives

# »
mmf3φ = nLS (id-LS + jiq-LS ) + nHS (id-HS + jiq-HS ) (A.9)

275
 # » # »
= mmfLS + mmfHS , if Te−3φ = Te (A.10)

Conversely, both the three-phase IPMSM and the uneq0-IPMSM produces the same
output torque when they have the same resultant mmf vector. Thus, to achieve MTPA
in the uneq0-IPMSM, it is sufficient to maximise the resultant mmf per total stator
copper loss.
The objective function that needs to be maximised is given by

# » # »
mmfLS + mmfHS
F = (A.11)
|~is-LS |2 nLS Rs + |~is-HS |2 nHS Rs

where, ~is-LS and ~is-HS are the current space-vectors of the LS and HS windings,
respectively.
# » # »
Let βLS and βHS represent the angle of mmfLS and mmfHS , respectively, from the q-axis.
The resultant mmf of the uneq0-IPMSM can be generated by various combinations of
# » # »
mmfLS and mmfHS as shown in Fig. A.2. Case:1, Case:2, and Case:3 represent three
ways to realise the resultant mmf of the uneq0-IPMSM to match the mmf of the 3-
phase IPMSM out of the infinite combinations possible. However, the maximum value
of resultant mmf with the minimum magnitude of individual mmf is obtained when
both the individual components are aligned (algebraic addition of vector magnitudes),
as shown in Fig. A.2a. Hence, (A.12) should be satisfied in the uneq0-IPMSM to
achieve the maximum resultant mmf (equivalent to maximum generated torque) with
# » # »
the minimum magnitude of mmfLS and mmfHS .

βLS = βHS (A.12)

Since βLS = βHS , the vector equation in the objective function given in (A.11) reduces

276
 # » # » # »
mmf3φ mmf3φ mmfHS
# »
mmfHS # » βHS # » βHS
mmfHS mmf3φ
# »
β1 = βLS = βHS β1 β1 mmfLS
# » # »
mmfLS mmfLS
βLS βLS

(a) Case:1 (b) Case:2 (c) Case:3

Fig. A.2: Location of mmf space vector during operation of uneq0-IPMSM to produce
equivalent mmf as three-phase IPMSM

to an algebraic equation as

nLS Ia-LS + nHS Ia-HS

Fa = 2 2
(A.13)
Ia-LS nLS Rs + Ia-HS nHS Rs

where Ia-LS and Ia-HS represent the magnitude of the space vector current in the LS and
HS winding, respectively. Even with βLS = βHS , many combinations of Ia-LS and Ia-HS
are possible to generate the required resultant mmf vector. Let the numerator of the
objective function Fa be defined as

nLS Ia-LS + nHS Ia-HS = C (A.14)

where, C is a constant.
Now the objective function can be redefined as ‘Minimise Fb ’ instead of ‘Maximise
Fa ’, where ‘Fb ’ is the denominator of Fa , and is given by

2 2
Fb = Ia-LS nLS Rs + Ia-HS nHS Rs (A.15)

Minimising ‘Fb ’ is equivalent to achieving ‘Minimum copper loss for a given generated
torque’.
From (A.14),
C − nLS Ia-LS
Ia-HS = (A.16)
nHS

277
Substituting (A.16) in (A.15) gives

 2
2 C − nLS Ia-LS
Fb = Ia-LS nLS Rs+ nHS Rs (A.17)
nHS
2 (C − nLS Ia-LS )2
= Ia-LS nLS Rs + Rs (A.18)
nHS

Differentiating (A.18) with respect to Ia-LS and equating to zero gives

dFb (C − nLS Ia-LS )

= 2Ia-LS nLS − 2nLS =0 (A.19)
dIa-LS nHS

Taking the second derivative of Fb with respect to Ia-LS gives

d2 Fb (0 − nLS )
2
= 2nLS − 2nLS (A.20)
dIa-LS n
 HS
nLS
= 2nLS 1 + (A.21)
nHS
= +ve

Since the second derivative is positive, (A.19) represents a minimum solution of the
optimisation function. Rewriting (A.19),

C = nLS Ia-LS + nHS Ia-LS (A.22)

Comparing (A.14) and (A.22) gives (A.23) as the condition for achieving minimum Fb
for a given Ia-LS .
Ia-LS = Ia-HS (A.23)

From (A.12) and (A.23), it is observed that both LS and HS windings should have
equal current space vector magnitude and angle to achieve MTPA condition during two
winding operation of the uneq0-IPMSM. Hence, the condition for MTPA in uneq0-

278
IPMSM can be simplified as

id-LS = id-HS = ids_MTPA (A.24)

iq-LS = iq-HS = iqs_MTPA (A.25)

where, id-LS and id-HS are the d-axis currents of the LS and HS winding, respectively.
iq-LS and iq-HS are the q-axis currents of the LS and HS winding, respectively. ids_MTPA
and iqs_MTPA correspond to the d-axis and q-axis current of the original three-phase
PMSM to achieve MTPA operation (current space vector at β1 from q-axis).
It can be concluded that MTPA operation of the uneq0-IPMSM occurs when currents
in both the LS and HS windings are controlled to be equal and in phase. Hence, the
same q-axis current reference obtained from ‘id − iq LS-HS LUT’ in the control system
shown in Fig. 4.9 (from Chapter 4) is given to the current controllers of both windings
(LS and HS winding). Similarly, the d-axis current reference is also the same for both
the LS and HS winding to ensure MTPA operation.

279
 REFERENCES

1. Agarwal, A. K. and N. N. Mustafi (2021). Real-world automotive

emissions: Monitoring methodologies, and control measures. Renewable
and Sustainable Energy Reviews, 137, 110624. ISSN 1364-0321,
doi:https://doi.org/10.1016/j.rser.2020.110624.

2. Allen, M., O. Dube, W. Solecki, F. Aragón-Durand, W. Cramer, S. Humphreys,

M. Kainuma, J. Kala, N. Mahowald, Y. Mulugetta, R. Perez, M. Wairiu, and
K. Zickfeld (2018). Framing and context. in: Global warming of 1.5°c. an IPCC
special report on the impacts of global warming of 1.5°c above pre-industrial levels
and related global greenhouse gas emission pathways, in the context of strengthening
the global response to the threat of climate change, sustainable development, and efforts
to eradicate poverty, [Masson-Delmotte, V., P. Zhai, H.-O. Pörtner, D. Roberts, J. Skea,
P.R. Shukla, A. Pirani, W. Moufouma-Okia, C. Péan, R. Pidcock, S. Connors, J.B.R.
Matthews, Y. Chen, X. Zhou, M.I. Gomis, E. Lonnoy, T. Maycock, M. Tignor, and
T. Waterfield (eds.)]. in press. URL https://apps.who.int/iris/handle/
10665/107332.

3. Amerise, A., L. Rovere, A. Formentini, M. Mengoni, L. Zarri, and P. Zanchetta

(2020). Electric drive based on an open-end winding surface pm synchronous machine
with a floating capacitor bridge. IEEE Transactions on Industry Applications, 56(3),
2709–2718, doi:10.1109/TIA.2020.2977596.

4. Atiq, S., T. A. Lipo, and B.-i. Kwon (2016a). Experimental verification of

winding switching technique to enhance maximum speed operation of surface mounted
permanent magnet machines. IET Electric Power Applications, 10(4), 294–303.

5. Atiq, S., T. A. Lipo, and B.-I. Kwon (2016b). Wide speed range operation of non-
salient pm machines. IEEE Transactions on Energy Conversion, 31(3), 1179–1191,
doi:10.1109/TEC.2016.2547421.

6. Baratieri, C. L. and H. Pinheiro (2013). An I-F starting method for smooth and fast
transition to sensorless control of BLDC motors. In 2013 Brazilian Power Electronics
Conference. doi:10.1109/COBEP.2013.6785212.

7. Baratieri, C. L. and H. Pinheiro (2014). Sensorless vector control for PM brushless

motors with nonsinusoidal back-EMF. In 2014 International Conference on Electrical
Machines (ICEM). doi:10.1109/ICELMACH.2014.6960290.

8. Barcaro, M., N. Bianchi, and F. Magnussen (2010). Analysis and tests of a dual
three-phase 12-slot 10-pole permanent-magnet motor. IEEE Transactions on Industry
Applications, 46(6), 2355–2362.

9. Bellur, D. M. and M. K. Kazimierczuk (2007). Dc-dc converters for electric vehicle

applications. In 2007 Electrical Insulation Conference and Electrical Manufacturing
Expo. doi:10.1109/EEIC.2007.4562633.

280
10. Bhattacharya, T. and L. Umanand (2006). Improved flux estimation and stator-
resistance adaptation scheme for sensorless control of induction motor. IEE
Proceedings - Electric Power Applications, 153(6), 911–920. ISSN 1350-2352,
doi:10.1049/ip-epa:20050452.

11. Boldea, I., A. Moldovan, and L. Tutelea (2015). Scalar V/f and I-f control of AC motor
drives: An overview. In 2015 Intl Aegean Conference on Electrical Machines Power
Electronics (ACEMP), 2015 Intl Conference on Optimization of Electrical Electronic
Equipment (OPTIM) 2015 Intl Symposium on Advanced Electromechanical Motion
Systems (ELECTROMOTION). doi:10.1109/OPTIM.2015.7426739.

12. Bolognani, S., S. Calligaro, R. Petrella, and F. Pogni (2011). Flux-weakening in ipm
motor drives: Comparison of state-of-art algorithms and a novel proposal for controller
design. In Proceedings of the 2011 14th European Conference on Power Electronics
and Applications.

13. Bolognani, S., R. Oboe, and M. Zigliotto (1999). Sensorless full-digital pmsm drive
with ekf estimation of speed and rotor position. IEEE Transactions on Industrial
Electronics, 46(1), 184–191, doi:10.1109/41.744410.

14. Borisavljevic, A., H. Polinder, and J. A. Ferreira (2010). Realization

of the I/f control method for a high-speed permanent magnet motor. In
The XIX International Conference on Electrical Machines - ICEM 2010.
doi:10.1109/ICELMACH.2010.5607892.

15. Burress, T. (2012). Benchmarking of competitive technologies. URL

http://energy.gov/sites/prod/files/2014/03/f10/ape006_
burress_2012_p.pdf.

16. Burress, T. (2016). Benchmarking ev and hev technologies. URL https:

//www.energy.gov/sites/prod/files/2016/06/f32/edt006_
burress_2016_o_web.pdf.

17. Burress, T. A., S. L. Campbell, C. L. Coomer, C. W. Ayers, A. A. Wereszczak,

J. P. Cunningham, L. D. Marlino, L. E. Seiber, and H. T. Lin (2011). Evaluation of
the 2010 toyota prius hybrid synergy drive system. URL https://doi.org/10.
2172/1007833.

18. Burress, T. A., C. L. Coomer, S. L. Campbell, L. E. Seiber, L. D. Marlino, R. H.

Staunton, and J. P. Cunningham (2008). Evaluation of the 2007 toyota camry hybrid
synergy drive system. doi:10.2172/928684.

19. Chakraborty, S., S. V. Nair, and K. Hatua (2021). Effect of v angle variation and
strengthening rib on performance characteristics of ipmsm motor for electric vehicles.
In 2021 1st International Conference on Power Electronics and Energy (ICPEE).
doi:10.1109/ICPEE50452.2021.9358664.

20. Chapman, P. and P. Krein (2003). Motor re-rating for traction applications - field
weakening revisited. In IEEE International Electric Machines and Drives Conference,
2003. IEMDC’03., volume 3. doi:10.1109/IEMDC.2003.1210633.

281
21. Che, H. S. and W. P. Hew (2015). Dual three-phase operation of single neutral
symmetrical six-phase machine for improved performance. In IECON 2015 - 41st
Annual Conference of the IEEE Industrial Electronics Society.

22. Chen, C.-H. and M.-Y. Cheng (2006). Design of a multispeed winding for a brushless
dc motor and its sensorless control. IEE Proceedings - Electric Power Applications,
153, 834–841(7). ISSN 1350-2352. URL https://digital-library.
theiet.org/content/journals/10.1049/ip-epa_20060073.

23. Chen, C.-H., M.-Y. Cheng, and M.-S. Tsai (2005). Study on a wide speed range
integrated electrical transmission system. In 2005 International Conference on Power
Electronics and Drives Systems, volume 1. doi:10.1109/PEDS.2005.1619791.

24. Cheng, B. and T. R. Tesch (2010). Torque feedforward control technique for
permanent-magnet synchronous motors. IEEE Transactions on Industrial Electronics,
57(3), 969–974, doi:10.1109/TIE.2009.2038951.

25. Colby, R. S. and D. W. Novotny (1988). An efficiency-optimizing permanent-magnet

synchronous motor drive. IEEE Transactions on Industry Applications, 24(3), 462–469.
ISSN 0093-9994, doi:10.1109/28.2897.

26. Consoli, A., G. Scarcella, and A. Testa (2001). Industry application of

zero-speed sensorless control techniques for PM synchronous motors. IEEE
Transactions on Industry Applications, 37(2), 513–521. ISSN 0093-9994,
doi:10.1109/28.913716.

27. Corley, M. J. and R. D. Lorenz (1998). Rotor position and velocity estimation for
a salient-pole permanent magnet synchronous machine at standstill and high speeds.
IEEE Transactions on Industry Applications, 34(4), 784–789. ISSN 0093-9994,
doi:10.1109/28.703973.

28. e-Verito (). Mahindra e-verito techincal specification. URL https://www.

mahindraverito.com/everito/pdf/everito_brochure.pdf.

29. Ehsani, M., Y. Gao, S. Longo, and K. M. Ebrahimi (2018). Modern Electric, Hybrid
Electric, and Fuel Cell Vehicles. CRC Press, Taylor & Francis Group, Florida.

30. Einhorn, M., W. Roessler, and J. Fleig (2011). Improved performance

of serially connected li-ion batteries with active cell balancing in electric
vehicles. IEEE Transactions on Vehicular Technology, 60(6), 2448–2457,
doi:10.1109/TVT.2011.2153886.

31. Emadi, A., S. Williamson, and A. Khaligh (2006). Power electronics

intensive solutions for advanced electric, hybrid electric, and fuel cell vehicular
power systems. IEEE Transactions on Power Electronics, 21(3), 567–577,
doi:10.1109/TPEL.2006.872378.

32. Estima, J. O. and A. J. Marques Cardoso (2012). Efficiency analysis of drive

train topologies applied to electric/hybrid vehicles. IEEE Transactions on Vehicular
Technology, 61(3), 1021–1031, doi:10.1109/TVT.2012.2186993.

282
33. Fajri, P., R. Ahmadi, and M. Ferdowsi (2012). Equivalent vehicle
rotational inertia used for electric vehicle test bench dynamic studies. In
IECON 2012 - 38th Annual Conference on IEEE Industrial Electronics Society.
doi:10.1109/IECON.2012.6389231.

34. Fatu, M., R. Teodorescu, I. Boldea, G. D. Andreescu, and F. Blaabjerg (2008). I-F
starting method with smooth transition to EMF based motion-sensorless vector control
of PM synchronous motor/generator. In 2008 IEEE Power Electronics Specialists
Conference. ISSN 0275-9306, doi:10.1109/PESC.2008.4592146.

35. Gao, Y., H. Maghbelli, M. Ehsani, G. Frazier, J. Kajs, and S. Bayne (2003).
Investigation of proper motor drive characteristics for military vehicle propulsion. In
Future Transportation Technology Conference & Exposition. SAE International. ISSN
0148-7191, doi:https://doi.org/10.4271/2003-01-2296.

36. Ghaderi, A. and T. Hanamoto (2011). Wide-speed-range sensorless vector control

of synchronous reluctance motors based on extended programmable cascaded low-pass
filters. IEEE Transactions on Industrial Electronics, 58(6), 2322–2333. ISSN 0278-
0046, doi:10.1109/TIE.2010.2066537.

37. Giulii Capponi, F., G. Borocci, G. De Donato, and F. Caricchi (2015). Flux regulation
strategies for hybrid excitation synchronous machines. IEEE Transactions on Industry
Applications, 51(5), 3838–3847, doi:10.1109/TIA.2015.2417120.

38. Gopakumar, K., V. T. Ranganthan, and S. R. Bhat (1993). Split-phase induction

motor operation from pwm voltage source inverter. IEEE Transactions on Industry
Applications, 29(5), 927–932.

39. Grambsch, A. (2001). Climate change and air quality, the potential impacts of climate
change on transportation. URL https://www.transportation.gov/sites/
dot.gov/files/docs/grambsch_CC_Air_Quality.pdf.

40. Grunditz, E. A., T. Thiringer, J. Lindström, S. Tidblad Lundmark, and M. Alatalo

(2021). Thermal capability of electric vehicle pmsm with different slot areas
via thermal network analysis. eTransportation, 8, 100107. ISSN 2590-1168,
doi:https://doi.org/10.1016/j.etran.2021.100107.

41. Guglielmi, P., M. Pastorelli, and A. Vagati (2006). Cross-saturation effects in ipm
motors and related impact on sensorless control. IEEE Transactions on Industry
Applications, 42(6), 1516–1522, doi:10.1109/TIA.2006.882646.

42. Hadiouche, D., L. Baghli, and A. Rezzoug (2006). Space-vector pwm techniques
for dual three-phase ac machine: analysis, performance evaluation, and dsp
implementation. IEEE Transactions on Industry Applications, 42(4), 1112–1122.

43. Hadiouche, D., H. Razik, and A. Rezzoug (2000). Study and simulation of space
vector pwm control of double-star induction motors. In 7th IEEE International
Power Electronics Congress. Technical Proceedings. CIEP 2000 (Cat. No.00TH8529).
doi:10.1109/CIEP.2000.891389.

283
44. Hadiouche, D., H. Razik, and A. Rezzoug (2004). On the modeling and design of
dual-stator windings to minimize circulating harmonic currents for vsi fed ac machines.
IEEE Transactions on Industry Applications, 40(2), 506–515.
45. Hahlbeck, S. and D. Gerling (2008). Design considerations for rotors with embedded
v-shape permanent magnets. In 2008 18th International Conference on Electrical
Machines. doi:10.1109/ICELMACH.2008.4800192.
46. Haichao, F., S. Boyang, and G. Lizhen (2017). A closed-loop I/f
vector control for permanent magnet synchronous motor. In 2017 9th
International Conference on Modelling, Identification and Control (ICMIC).
doi:10.1109/ICMIC.2017.8321595.
47. Hatua, K. (2003). Direct Torque Control Schemes for Split Phase Induction Machine.
Master’s thesis, Department of Electrical Engineering, IISC Banglaore, Bangalore –
560012.
48. He, Y., Y. Wang, J. Wu, Y. Feng, and J. Liu (2010). A simple current sharing scheme
for dual three-phase permanent-magnet synchronous motor drives. In 2010 Twenty-Fifth
Annual IEEE Applied Power Electronics Conference and Exposition (APEC).
49. Hendershot, J. R. and T. J. E. Miller (2010). Design of Brushless Permanent-Magnet
Machines. Motor Design Books LLC, Venice, Florida. ISBN 978-0-9840687-0-8.
50. Hetemi, F., G. Dajaku, and D. Gerling (2010). Influence of magnet
thickness and magnet orientation on the performance of ipmsm. In
The XIX International Conference on Electrical Machines - ICEM 2010.
doi:10.1109/ICELMACH.2010.5607796.
51. Hsieh, M.-F., F.-S. Hsu, and D. G. Dorrell (2012). Winding changeover permanent-
magnet generators for renewable energy applications. IEEE Transactions on Magnetics,
48(11), 4168–4171, doi:10.1109/TMAG.2012.2196266.
52. Hu, Y., Z. Q. Zhu, and M. Odavic (2017). Comparison of two-individual
current control and vector space decomposition control for dual three-phase
pmsm. IEEE Transactions on Industry Applications, 53(5), 4483–4492,
doi:10.1109/TIA.2017.2703682.
53. Huang, H. and L. Chang (1999). Electrical two-speed propulsion by motor winding
switching and its control strategies for electric vehicles. IEEE Transactions on
Vehicular Technology, 48(2), 607–618, doi:10.1109/25.752586.
54. Hurst, K. D., T. G. Habetler, G. Griva, and F. Profumo (1998). Zero-
speed tacholess IM torque control: simply a matter of stator voltage integration.
IEEE Transactions on Industry Applications, 34(4), 790–795. ISSN 0093-9994,
doi:10.1109/28.703975.
55. Iepure, L. I., I. Boldea, and F. Blaabjerg (2012). Hybrid I-f starting and
observer-based sensorless control of single-phase BLDC-PM motor drives. IEEE
Transactions on Industrial Electronics, 59(9), 3436–3444. ISSN 0278-0046,
doi:10.1109/TIE.2011.2172176.

284
56. Im, S.-H. and B.-G. Gu (2020). A snubberless solid-state tap changer for permanent
magnet synchronous motors. IEEE Transactions on Power Electronics, 35(11), 12143–
12152, doi:10.1109/TPEL.2020.2988933.

57. Jahns, T. M., G. B. Kliman, and T. W. Neumann (1986). Interior permanent-

magnet synchronous motors for adjustable-speed drives. IEEE Transactions on Industry
Applications, IA-22(4), 738–747, doi:10.1109/TIA.1986.4504786.

58. Jahns, T. M. and W. L. Soong (1996). Pulsating torque minimization techniques

for permanent magnet ac motor drives-a review. IEEE Transactions on Industrial
Electronics, 43(2), 321–330.

59. Jukic, F., L. Pravica, T. Barisa, and D. Sumina (2020). Flying-start and continuous
operation of a permanent-magnet wind generator based on discontinuous currents,
discrete second-order sliding-mode observer and phase-locked loop. IET Renewable
Power Generation, 14(1), 90–99.

60. Kan, K.-S. and Y.-Y. Tzou (2012). A sensorless I/f control method for single-phase
BLDC fan motors with efficiency optimization by power factor control. In Proceedings
of The 7th International Power Electronics and Motion Control Conference, volume 4.
doi:10.1109/IPEMC.2012.6259257.

61. Karttunen, J., S. Kallio, P. Peltoniemi, P. Silventoinen, and O. Pyrhönen

(2012). Dual three-phase permanent magnet synchronous machine supplied
by two independent voltage source inverters. In International Symposium on
Power Electronics Power Electronics, Electrical Drives, Automation and Motion.
doi:10.1109/SPEEDAM.2012.6264448.

62. Kim, H., K.-K. Huh, R. D. Lorenz, and T. M. Jahns (2004). A novel
method for initial rotor position estimation for IPM synchronous machine drives.
IEEE Transactions on Industry Applications, 40(5), 1369–1378. ISSN 0093-9994,
doi:10.1109/TIA.2004.834091.

63. Kim, H., J. Son, and J. Lee (2011). A high-speed sliding-mode observer for the
sensorless speed control of a pmsm. IEEE Transactions on Industrial Electronics, 58(9),
4069–4077, doi:10.1109/TIE.2010.2098357.

64. Kim, S. and J. Seok (2013). Maximum voltage utilization of IPMSMs using
modulating voltage scalability for automotive applications. IEEE Transactions on
Power Electronics, 28(12), 5639–5646, doi:10.1109/TPEL.2013.2253802.

65. Krause, P., O. Wasynczuk, S. Sudhoff, and S. Pekarek (2013). Analysis of Electric
Machinery. John Wiley & Sons, Inc.

66. Krishnan, R. and R. Ghosh (1989). Starting algorithm and performance of a PM DC

brushless motor drive system with no position sensor. In 20th Annual IEEE Power
Electronics Specialists Conference. doi:10.1109/PESC.1989.48564.

67. Kume, T., T. Iwakane, T. Sawa, T. Yoshida, and I. Nagai (1991). A

wide constant power range vector-controlled ac motor drive using winding

285
 changeover technique. IEEE Transactions on Industry Applications, 27(5), 934–939,
doi:10.1109/28.90350.

68. Kume, T. and T. Sawa (1995). A static winding changeover technique. Japanese Patent
Hei 7-99959.

69. Kume, T., T. Sawa, M. Miyazato, M. Sawamura, and M. Zenke (1990). Inverter
driving method for induction motors. U.S. Patent 4916376.

70. Kung, Y. and Risfendra (2016). ModelSim/Simulink co-simulation of a

sensorless control for PMSM drives based on I-F startup and EKF. In
2016 International Conference on Applied System Innovation (ICASI).
doi:10.1109/ICASI.2016.7539749.

71. Kung, Y., Risfendra, Y. Lin, and L. Huang (2016). FPGA-based sensorless
controller for PMSM drives using sliding mode observer and phase locked
loop. In 2016 International Conference on Applied System Innovation (ICASI).
doi:10.1109/ICASI.2016.7539798.

72. Kung, Y.-S. and M.-H. Tsai (2007). Fpga-based speed control ic for pmsm drive with
adaptive fuzzy control. IEEE Transactions on Power Electronics, 22(6), 2476–2486,
doi:10.1109/TPEL.2007.909185.

73. Lee, H., J. Kim, J. Hong, and K. Nam (2014). Torque control for ipmsm in the
high speed range based on voltage angle. In 2014 IEEE Applied Power Electronics
Conference and Exposition - APEC 2014. doi:10.1109/APEC.2014.6803655.

74. Lee, Y. and J.-I. Ha (2015). Hybrid modulation of dual inverter for open-end permanent
magnet synchronous motor. IEEE Transactions on Power Electronics, 30(6), 3286–
3299, doi:10.1109/TPEL.2014.2325738.

75. Leonhard, W. (2001). Symmetrical Three—Phase AC Machines, 163–214. Springer

Berlin Heidelberg, Berlin, Heidelberg. ISBN 978-3-642-56649-3. URL https://
doi.org/10.1007/978-3-642-56649-3_11.

76. Li, A., D. Jiang, X. Sun, and Z. Liu (2022). Online drive topology conversion
technology for pmsm speed range extension. IEEE Transactions on Power Electronics,
1–1, doi:10.1109/TPEL.2022.3140184.

77. Li, Y., H. Guo, Q. Xie, and P. Yuan (2010). Research on the control method for the
start of microturbine generation system. In The 2010 IEEE International Conference
on Information and Automation. doi:10.1109/ICINFA.2010.5512061.

78. Liang, Z., D. Liang, P. Kou, and Q. Ze (2020). Modeling and bumpless
switching control of symmetrical dual three-phase pmsm in two operating
modes. Electric Power Components and Systems, 48(3), 304–319,
doi:10.1080/15325008.2020.1758844.

79. Lipo, T. A. (2017). Introduction to AC machine design. Wiley, Hoboken, New Jersey.
ISBN 9781119352167.

286
80. Liu, X., H. Chen, J. Zhao, and A. Belahcen (2016). Research on the performances and
parameters of interior pmsm used for electric vehicles. IEEE Transactions on Industrial
Electronics, 63(6), 3533–3545, doi:10.1109/TIE.2016.2524415.

81. Lu, K., M. Vetuschi, P. O. Rasmussen, and A. E. Ritchie (2010). Determination

of high-frequency d- and q-axis inductances for surface-mounted permanent-magnet
synchronous machines. IEEE Transactions on Instrumentation and Measurement,
59(9), 2376–2382.

82. Luo, Z., D. Liang, and W. Ding (2015). Modelling of dual three-phase permanent
magnet brushless machine and drive using matrix and tensor approach. IET Electric
Power Applications, 9(1), 30–43.

83. Marouani, K., L. Baghli, D. Hadiouche, A. Kheloui, and A. Rezzoug (2008). A new
PWM strategy based on a 24-sector vector space decomposition for a six-phase vsi-
fed dual stator induction motor. IEEE Transactions on Industrial Electronics, 55(5),
1910–1920.

84. Meeker, D. et al. (). Finite element method magnetics. URL https://www.femm.
info/wiki/HomePage.

85. MIDC (1991). Modified indian driving cycle. URL https://www.araiindia.

com/CMVR_TAP_Documents/Part-14/Part-14_Chapter03.pdf.

86. Mohammadi, A. S. (2018). Hybridisation ratio for hybrid excitation

synchronous motors in electric vehicles with enhanced performance. IET
Electrical Systems in Transportation, 8, 12–19(7). ISSN 2042-9738. URL
https://digital-library.theiet.org/content/journals/10.
1049/iet-est.2017.0029.

87. Momen, F., K. Rahman, Y. Son, and P. Savagian (2016). Electrical propulsion system
design of chevrolet bolt battery electric vehicle. In 2016 IEEE Energy Conversion
Congress and Exposition (ECCE). doi:10.1109/ECCE.2016.7855076.

88. Monajemy, R. and R. Krishnan (1999). Performance comparison for six-step voltage
and constant back emf control strategies for pmsm. In Conference Record of the
1999 IEEE Industry Applications Conference. Thirty-Forth IAS Annual Meeting (Cat.
No.99CH36370), volume 1. doi:10.1109/IAS.1999.799949.

89. Morimoto, S., Y. Inoue, T.-F. Weng, and M. Sanada (2007). Position sensorless pmsm
drive system including square-wave operation at high-speed. In 2007 IEEE Industry
Applications Annual Meeting. doi:10.1109/07IAS.2007.107.

90. Morimoto, S., Y. Takeda, T. Hirasa, and K. Taniguchi (1990). Expansion of

operating limits for permanent magnet motor by current vector control considering
inverter capacity. IEEE Transactions on Industry Applications, 26(5), 866–871,
doi:10.1109/28.60058.

91. Niu, L., J. Zhai, X. Liu, Y. Wang, and A. K. Gupta (2016). A

smooth and fast transition method for PMSM SMO based sensorless control.

287
 In 2016 IEEE 2nd Annual Southern Power Electronics Conference (SPEC).
doi:10.1109/SPEC.2016.7845995.

92. Oleschuk, V., G. Griva, F. Profumo, and A. Tenconi (2007). Synchronized pwm
control of symmetrical six-phase drives. In 2007 7th Internatonal Conference on Power
Electronics.

93. Ortombina, L., F. Tinazzi, and M. Zigliotto (2017). An effective start-up algorithm
for sensorless synchronous reluctance and IPM motor drives. In 2017 IEEE 12th
International Conference on Power Electronics and Drive Systems (PEDS). ISSN 2164-
5264, doi:10.1109/PEDS.2017.8289167.

94. Pan, D., F. Liang, Y. Wang, and T. A. Lipo (2014). Extension of the operating
region of an IPM motor utilizing series compensation. IEEE Transactions on Industry
Applications, 50(1), 539–548, doi:10.1109/TIA.2013.2270223.

95. Parsa, L. (2005). On advantages of multi-phase machines. In 31st

Annual Conference of IEEE Industrial Electronics Society, 2005. IECON 2005..
doi:10.1109/IECON.2005.1569139.

96. Perera, P. D. C., F. Blaabjerg, J. K. Pedersen, and P. Thogersen (2003). A

sensorless, stable V/f control method for permanent-magnet synchronous motor drives.
IEEE Transactions on Industry Applications, 39(3), 783–791. ISSN 0093-9994,
doi:10.1109/TIA.2003.810624.

97. Poulain, F., L. Praly, and R. Ortega (2008). An observer for permanent magnet
synchronous motors with currents and voltages as only measurements. In 2008 47th
IEEE Conference on Decision and Control. doi:10.1109/CDC.2008.4738868.

98. Pravica, L., D. Sumina, T. Barisa, M. Kovacic, and I. Colovic (2018). Flying start of a
permanent magnet wind power generator based on a discontinuous converter operation
mode and a phase-locked loop. IEEE Transactions on Industrial Electronics, 65(2),
1097–1106.

99. Pyrhönen, J., T. Jokinen, and Hrabovcová (2008). Design of Rotating Electrical
Machines. John Wiley & Sons, Ltd, Witshire. ISBN 978-0-470-69516-6.

100. Qiao, Z., T. Shi, Y. Wang, Y. Yan, C. Xia, and X. He (2013). New
sliding-mode observer for position sensorless control of permanent-magnet
synchronous motor. IEEE Transactions on Industrial Electronics, 60(2), 710–719,
doi:10.1109/TIE.2012.2206359.

101. Quang, N. K., N. T. Hieu, and Q. P. Ha (2014). Fpga-based sensorless pmsm speed
control using reduced-order extended kalman filters. IEEE Transactions on Industrial
Electronics, 61(12), 6574–6582, doi:10.1109/TIE.2014.2320215.

102. Rahman, Z., M. Ehsani, and K. L. Butler (2000). An investigation of

electric motor drive characteristics for ev and hev propulsion systems. In Future
Transportation Technology Conference & Exposition. SAE International. ISSN 0148-
7191, doi:https://doi.org/10.4271/2000-01-3062.

288
103. Ranganathan, V. T. (2004). Course Notes on Electric Drives. Indian Institute of
Science.

104. RR Shramik (). RR Shramik product brochure: Fine and ultra fine enamelled copper
wire. URL http://www.rrshramik.com/wp-content/uploads/sites/
2/2018/06/Fine-and-Ultra-Fine-Enamelled-Copper-Wire.pdf.

105. Sadeghi, S., L. Guo, H. A. Toliyat, and L. Parsa (2012). Wide operational speed
range of five-phase permanent magnet machines by using different stator winding
configurations. IEEE Transactions on Industrial Electronics, 59(6), 2621–2631,
doi:10.1109/TIE.2011.2164771.

106. Salem, A. and M. Narimani (2019). A review on multiphase drives for automotive
traction applications. IEEE Transactions on Transportation Electrification, 5(4), 1329–
1348, doi:10.1109/TTE.2019.2956355.

107. Sarlioglu, B., C. T. Morris, D. Han, and S. Li (2017). Driving toward accessibility:
A review of technological improvements for electric machines, power electronics, and
batteries for electric and hybrid vehicles. IEEE Industry Applications Magazine, 23(1),
14–25, doi:10.1109/MIAS.2016.2600739.

108. Schiferl, R. and T. Lipo (1990). Power capability of salient pole permanent magnet
synchronous motors in variable speed drive applications. IEEE Transactions on
Industry Applications, 26(1), 115–123, doi:10.1109/28.52682.

109. Sebastian, T. and G. R. Slemon (1987). Operating limits of inverter-driven permanent

magnet motor drives. IEEE Transactions on Industry Applications, IA-23(2), 327–333,
doi:10.1109/TIA.1987.4504909.

110. Sin, S., M. Ayub, and B.-I. Kwon (2020a). Investigation study of multi-mode multi-
speed operation method for surface-mounted permanent magnet synchronous machines.
IEEE Access, 8, 169470–169485, doi:10.1109/ACCESS.2020.3024183.

111. Sin, S., M. Ayub, and B.-I. Kwon (2020b). Operation method of non-salient permanent
magnet synchronous machine for extended speed range. IEEE Access, 8, 105922–
105935, doi:10.1109/ACCESS.2020.3000256.

112. Singh, G., V. Pant, and Y. Singh (2003). Voltage source inverter driven multi-phase
induction machine. Computers & Electrical Engineering, 29(8), 813 – 834. ISSN
0045-7906, doi:https://doi.org/10.1016/S0045-7906(03)00036-3.

113. Slemon, G. (1995). Achieving a constant power speed range for pm

drives. IEEE Transactions on Industry Applications, 31(2), 368–372,
doi:10.1109/28.370286.

114. Soong, W. and N. Ertugrul (2002). Field-weakening performance of interior

permanent-magnet motors. IEEE Transactions on Industry Applications, 38(5), 1251–
1258, doi:10.1109/TIA.2002.803013.

289
115. Soong, W. and T. Miller (1993a). Practical field-weakening performance of the five
classes of brushless synchronous ac motor drive. In 1993 Fifth European Conference
on Power Electronics and Applications.

116. Soong, W. and T. Miller (1993b). Theoretical limitations to the field-weakening

performance of the five classes of brushless synchronous ac motor drive. In 1993 Sixth
International Conference on Electrical Machines and Drives (Conf. Publ. No. 376).

117. Staunton, R. H., T. A. Burress, and L. D. Marlino (2006). Evaluation of 2005

Honda Accord hybrid electric drive system. URL https://doi.org/10.2172/
891260.

118. Stirban, A., I. Boldea, G. Andreescu, D. Iles, and F. Blaabjerg (2010). Motion
sensorless control of BLDC PM motor with offline FEM info assisted state observer.
In 2010 12th International Conference on Optimization of Electrical and Electronic
Equipment. ISSN 1842-0133, doi:10.1109/OPTIM.2010.5510439.

119. Stojan, D., D. Drevensek, Z. Plantic, B. Grcar, and G. Stumberger (2012). Novel
field-weakening control scheme for permanent-magnet synchronous machines based on
voltage angle control. IEEE Transactions on Industry Applications, 48(6), 2390–2401,
doi:10.1109/TIA.2012.2227133.

120. SubbaRao, M. (2022). GISTEMP Climate Spiral. URL https://svs.gsfc.

nasa.gov/4975.

121. Swamy, M., T. Kume, A. Maemura, and S. Morimoto (2006). Extended high-speed
operation via electronic winding-change method for ac motors. IEEE Transactions on
Industry Applications, 42(3), 742–752, doi:10.1109/TIA.2006.873657.

122. Tsai, M., M. Chou, and C. Chu (2001). Control of a variable-winding

brushless motor with the application in electric scooters. In IEMDC 2001.
IEEE International Electric Machines and Drives Conference (Cat. No.01EX485).
doi:10.1109/IEMDC.2001.939430.

123. US DRIVE (2017). Electrical and electronics technical team roadmap. URL
https://www.energy.gov/eere/vehicles/downloads/us-drive-
electrical-and-electronics-technical-team-roadmap.

124. Vas, P. (1998). Sensorless vector and direct torque control. Oxford University Press.

125. Vaseghi, B., N. Takorabet, J. P. Caron, B. Nahid-Mobarakeh, F. Meibody-Tabar,

and G. Humbert (2011). Study of different architectures of fault-tolerant actuator using
a two-channel pm motor. IEEE Transactions on Industry Applications, 47(1), 47–54,
doi:10.1109/TIA.2010.2090930.

126. Venugopal, S. and G. Narayanan (2006). An overmodulation scheme for vector

controlled induction motor drives. In 2006 International Conference on Power
Electronic, Drives and Energy Systems. doi:10.1109/PEDES.2006.344382.

290
127. Wang, M., Y. Xu, J. Zou, and H. Lan (2017). An optimized I-F
startup method for BEMF-based sensorless control of SPMSM. In 2017 IEEE
Transportation Electrification Conference and Expo, Asia-Pacific (ITEC Asia-Pacific).
doi:10.1109/ITEC-AP.2017.8080874.

128. Wang, M.-S., N.-C. Hsu, C.-Y. Chiang, S.-H. Wang, and T.-C. Shau (2010). A novel
changeover technique for variable-winding brushless dc motor drives. In Proceedings
of SICE Annual Conference 2010.

129. Wang, T., F. Fang, X. Wu, and X. Jiang (2013). Novel filter for stator harmonic
currents reduction in six-step converter fed multiphase induction motor drives. IEEE
Transactions on Power Electronics, 28(1), 498–506.

130. Wang, W., Z. Li, and X. Xu (2014). A novel smooth transition strategy for BEMF-
based sensorless drive startup of PMSM. In Proceeding of the 11th World Congress on
Intelligent Control and Automation. doi:10.1109/WCICA.2014.7053435.

131. Wang, Z., K. Lu, and F. Blaabjerg (2012). A simple startup strategy
based on current regulation for back-EMF-based sensorless control of PMSM.
IEEE Transactions on Power Electronics, 27(8), 3817–3825. ISSN 0885-8993,
doi:10.1109/TPEL.2012.2186464.

132. Welchko, B. (2005). A double-ended inverter system for the combined propulsion
and energy management functions in hybrid vehicles with energy storage. In 31st
Annual Conference of IEEE Industrial Electronics Society, 2005. IECON 2005..
doi:10.1109/IECON.2005.1569110.

133. Welchko, B. and J. Nagashima (2003). The influence of topology selection on the
design of ev/hev propulsion systems. IEEE Power Electronics Letters, 1(2), 36–40,
doi:10.1109/LPEL.2003.821033.

134. WHO Regional Publications (1999). Monitoring ambient air quality for health impact
assessment. URL https://apps.who.int/iris/handle/10665/107332.

135. WHO Regional Publications (2000). Air quality guidelines for europe. URL https:
//apps.who.int/iris/handle/10665/107335.

136. Wu, T., D. Luo, S. Huang, X. Wu, K. Liu, K. Lu, and X. Peng (2020). A
fast estimation of initial rotor position for low-speed free-running ipmsm. IEEE
Transactions on Power Electronics, 35(7), 7664–7673.

137. Xu, L. and C. Wang (1998). Implementation and experimental investigation

of sensorless control schemes for PMSM in super-high variable speed operation.
In Conference Record of 1998 IEEE Industry Applications Conference. Thirty-
Third IAS Annual Meeting (Cat. No.98CH36242), volume 1. ISSN 0197-2618,
doi:10.1109/IAS.1998.732349.

138. Xu, L. and L. Ye (1995). Analysis of a novel stator winding structure minimizing
harmonic current and torque ripple for dual six-step converter-fed high power ac
machines. IEEE Transactions on Industry Applications, 31(1), 84–90.

291
139. Yang, J., W. Huang, R. Cao, and X. Jiang (2015). A closed-loop I/f sensorless
control based on current vector orientation for permanent magnet synchronous motors.
In 2015 18th International Conference on Electrical Machines and Systems (ICEMS).
doi:10.1109/ICEMS.2015.7385298.

140. Yang, R., N. Schofield, N. Zhao, and A. Emadi (2019). Dual three-
phase permanent magnet synchronous machine investigation for battery electric
vehicle power-trains. The Journal of Engineering, 2019(17), 3981–3985,
doi:https://doi.org/10.1049/joe.2018.8178.

141. Yang, Y., S. M. Castano, R. Yang, M. Kasprzak, B. Bilgin, A. Sathyan,

H. Dadkhah, and A. Emadi (2017). Design and comparison of interior
permanent magnet motor topologies for traction applications. IEEE Transactions on
Transportation Electrification, 3(1), 86–97, doi:10.1109/TTE.2016.2614972.

142. Yazdani, D., S. Ali Khajehoddin, A. Bakhshai, and G. Joos (2009). Full utilization
of the inverter in split-phase drives by means of a dual three-phase space vector
classification algorithm. IEEE Transactions on Industrial Electronics, 56(1), 120–129.

143. Yu, Y., D. Chang, X. Zheng, Z. Mi, X. Li, and C. Sun (2018). A stator current
oriented closed-loop I–f control of sensorless SPMSM with fully unknown parameters
for reverse rotation prevention. IEEE Transactions on Power Electronics, 33(10), 8607–
8622. ISSN 0885-8993, doi:10.1109/TPEL.2017.2780191.

144. Zhan, H., Z. Q. Zhu, M. Odavic, Z. Wu, and A. S. Thomas (2017).

Robust initial phase correction strategy for zero sequence back EMF based
sensorless drive under transition from current regulation control. In 2017
IEEE International Electric Machines and Drives Conference (IEMDC).
doi:10.1109/IEMDC.2017.8002035.

145. Zhang, Z., H. Guo, Y. Liu, Q. Zhang, P. Zhu, and R. Iqbal (2019). An improved
sensorless control strategy of ship ipmsm at full speed range. IEEE Access, 7, 178652–
178661.

146. Zhitkova, S. and K. Hameyer (2016). Realization of a wide speed range for an
agricultural tractor. In 2016 XXII International Conference on Electrical Machines
(ICEM). doi:10.1109/ICELMACH.2016.7732802.

147. Zhou, C., G. Yang, and J. Su (2016). PWM strategy with minimum harmonic
distortion for dual three-phase permanent-magnet synchronous motor drives operating
in the overmodulation region. IEEE Transactions on Power Electronics, 31(2), 1367–
1380.

148. Zhu, Y., W. Gu, K. Lu, and Z. Wu (2020). Vector control of asymmetric dual three-
phase pmsm in full modulation range. IEEE Access, 8, 104479–104493.

149. Zhu, Z. Q., Y. Li, D. Howe, C. M. Bingham, and D. Stone (2007). Influence of
machine topology and cross-coupling magnetic saturation on rotor position estimation
accuracy in extended back-emf based sensorless pm brushless ac drives. In 2007 IEEE
Industry Applications Annual Meeting.

292
 CURRICULUM VITAE

1. NAME : Sandeep V Nair

2. DATE OF BIRTH : 18 Nov 1989

3. EDUCATIONAL QUALIFICATIONS

2011 Bachelor of Technology (B. Tech.)

Institution : NSS College of Engineering, Palakkad
Specialization : Electrical and Electronics Engineering

2015 Master of Technology (M. Tech.)

Institution : Indian Institute of Technology Bombay
Specialization : Electrical Engineering

Doctor of Philosphy
Institution : Indian Institute of Technology Madras
Specialization : Electrical Engineering
Registration Date : 14 July 2015

293
 DOCTORAL COMMITTEE

CHAIRPERSON : Dr. Mahesh Kumar

Professor
Department of Electrical Engineering

GUIDE : Dr. Kamalesh Hatua

Associate Professor
Department of Electrical Engineering
MEMBERS : Dr. N. Lakshminarasamma
Professor
Department of Electrical Engineering

Dr. K Shanti Swarup

Professor
Department of Electrical Engineering

Dr. T. Sundararajan
Institute Chair Professor
Department of Mechanical Engineering

294