Florida International University

FIU Digital Commons

FIU Electronic Theses and Dissertations University Graduate School

11-9-2017

Bidirectional Electric Vehicles Service Integration

in Smart Power Grid with Renewable Energy
Resources
Ahmed A S Mohamed Mr
Florida International University, [email protected]

DOI: 10.25148/etd.FIDC004037
Follow this and additional works at: https://digitalcommons.fiu.edu/etd
Part of the Electrical and Electronics Commons, and the Power and Energy Commons

Recommended Citation
Mohamed, Ahmed A S Mr, "Bidirectional Electric Vehicles Service Integration in Smart Power Grid with Renewable Energy
Resources" (2017). FIU Electronic Theses and Dissertations. 3529.
https://digitalcommons.fiu.edu/etd/3529

This work is brought to you for free and open access by the University Graduate School at FIU Digital Commons. It has been accepted for inclusion in
FIU Electronic Theses and Dissertations by an authorized administrator of FIU Digital Commons. For more information, please contact [email protected].
 FLORIDA INTERNATIONAL UNIVERSITY

Miami,

BIDIRECTIONAL ELECTRIC VEHICLES SERVICE INTEGRATION IN SMART

POWER GRID WITH RENEWABLE ENERGY RESOURCES

A dissertation submitted in partial fulfillment of

the requirements for the degree of

DOCTOR OF PHILOSOPHY

in

ELECTRICAL ENGINEERING

by

Ahmed Abdelfatah Abdelaal Said Mohamed

2017
To: Dean John Volakis
College of Engineering and Computing

This dissertation, written by Ahmed Abdelfatah Abdelaal Said Mohamed, and entitled
Bidirectional Electric Vehicles Service Integration in Smart Power Grid with Renewable
Energy Resources, having been approved in respect to style and intellectual contents, is
referred to you for judgment.

We have read this dissertation and recommend its approval.

Sakhrat Khizroev.

Armando Barreto.

Bilal El-Zahab.

Kemal Akkaya.

Osama A. Mohammed, Major Professor.

Date of Defense: November 9, 2017

The dissertation of Ahmed Abdelfatah Abdelaal Said Mohamed is approved.

Dean John Volakis

College of Engineering and Computing

Andrés G. Gil
Vice President for Research and Economic Development
and Dean of the University Graduate School

Florida International University, 2017

ii
Copyright © 2017 by Ahmed Abdelfatah Abdelaal Said Mohamed

All rights reserved.

iii
 DEDICATION

I dedicate this dissertation work to my wife, Asmaa, who has been a constant source of

support and encouragement during the challenges of graduate school and life. I am truly

thankful for having you, our sons, Osama and Mohamed, and our daughter, Sarah, in my

life.

I also dedicated this work to my parents, brother and sisters, who I did not see them

much during my Ph.D. A special feeling of gratitude to my loving parents, Abdelfatah and

Fathya who have always loved me unconditionally and whose good examples have taught

me to work hard for the things that I aspire to achieve. My brother, Yasser, whose words

of encouragement and push for tenacity ring in my ears. My sisters Hanan and Mervat have

never left my side and are very special.

iv
 ACKNOWLEDGMENTS

I owe my deepest gratitude to my advisor, Professor Osama A. Mohammed, for his

invaluable guidance and support throughout my Ph.D. studies. He has by far been the most

influential and inspiring teacher and advisor I have ever had the privilege of working with.

Indeed he has been the predominant source of the academic and professional expertise I

have acquired. Throughout my Ph.D., Professor Mohammed provided me with endless

support, guidance, expert technical assistance and significant contributions through the

period of the research and dissertation development. I will always remember his belief in

my abilities to work hard and to produce at a high level. I do appreciate his many

constructive suggestions while the material was being developed and tested at the energy

systems research laboratory. I am indebted for his willingness to accept me in his laboratory

and for providing significant research facilities to develop this work in addition to his

expert guidance and continued assistance, without which, I wouldn’t have been able to

successfully accomplish my experimental work. I am also grateful to Dr. Sakhrat Khizroev,

Dr. Armando Barreto, Dr. Kemal Akkaya and Dr. Bilal El-Zahab for serving on my

dissertation advisory committee and for the many constructive discussions.

I am also thankful to the Cultural Affairs and Missions Sector, Ministry of Higher

Education, Government of Egypt, which provided me with the financial support as a

doctoral student at Florida International University. I also thank the graduate school at

Florida International University for providing me with the Doctoral Evidence Acquisition

(DEA) Fellowship and the Dissertation Year Fellowship (DYF) during my last year to

complete my doctoral work.

v
 I would also such as to acknowledge all the members and students of the Energy

Systems Research Laboratory, Florida International University for their assistance during

the experimental testing of this research.

vi
 ABSTRACT OF THE DISSERTATION

BIDIRECTIONAL ELECTRIC VEHICLES SERVICE INTEGRATION IN SMART

POWER GRID WITH RENEWABLE ENERGY RESOURCES

by

Ahmed Abdelfatah Abdelaal Said Mohamed

Florida International University, 2017

Miami, Florida

Professor Osama A. Mohammed, Major Professor

As electric vehicles (EVs) become more popular, the utility companies are forced to

increase power generations in the grid. However, these EVs are capable of providing power

to the grid to deliver different grid ancillary services in a concept known as vehicle-to-grid

(V2G) and grid-to-vehicle (G2V), in which the EV can serve as a load or source at the

same time. These services can provide more benefits when they are integrated with

Photovoltaic (PV) generation. The proper modeling, design and control for the power

conversion systems that provide the optimum integration among the EVs, PV generations

and grid are investigated in this thesis.

The coupling between the PV generation and integration bus is accomplished through

a unidirectional converter. Precise dynamic and small-signal models for the grid-connected

PV power system are developed and utilized to predict the system’s performance during

the different operating conditions. An advanced intelligent maximum power point tracker

based on fuzzy logic control is developed and designed using a mix between the analytical

model and genetic algorithm optimization.

vii
 The EV is connected to the integration bus through a bidirectional inductive wireless

power transfer system (BIWPTS), which allows the EV to be charged and discharged

wirelessly during the long-term parking, transient stops and movement. Accurate analytical

and physics-based models for the BIWPTS are developed and utilized to forecast its

performance, and novel practical limitations for the active and reactive power-flow during

G2V and V2G operations are stated. A comparative and assessment analysis for the

different compensation topologies in the symmetrical BIWPTS was performed based on

analytical, simulation and experimental data. Also, a magnetic design optimization for the

double-D power pad based on finite-element analysis is achieved. The nonlinearities in the

BIWPTS due to the magnetic material and the high-frequency components are investigated

rely on a physics-based co-simulation platform. Also, a novel two-layer predictive power-

flow controller that manages the bidirectional power-flow between the EV and grid is

developed, implemented and tested. In addition, the feasibility of deploying the quasi-

dynamic wireless power transfer technology on the road to charge the EV during the

transient stops at the traffic signals is proven.

viii
 TABLE OF CONTENTS

CHAPTER PAGE

Chapter 1 Introduction ..................................................................................................... 1

Problem Statement .................................................................................................. 1
Research Objective ................................................................................................. 6
Original Contributions of the Dissertation .............................................................. 9
Dissertation Organization ..................................................................................... 10

Chapter 2 Dynamic Modeling of Photovoltaic Power System ...................................... 15

Introduction ........................................................................................................... 15
Nonlinear Dynamic Model ................................................................................... 16
2.2.1 PV Panel Dynamic Model ............................................................................ 17
2.2.2 DC-DC Converter with DC Bus Model ........................................................ 20
Small Signal Analysis of PV Power System......................................................... 21
2.3.1 PV Panel Small Signal Model....................................................................... 21
2.3.2 DC-DC Converter and DC Bus Small Signal Model.................................... 25
Small Signal Perturbation Analysis ...................................................................... 26
PV Panel Parameters’ Estimation ......................................................................... 28
2.5.1 Analytical Parameter Estimation .................................................................. 29
2.5.2 GA-Based Parameters Estimation................................................................. 30
2.5.3 Parameters Estimation Results ...................................................................... 31
Conclusion ............................................................................................................ 33

Chapter 3 Design and Implementation of Fuzzy MPPT for PV Power System ............ 34
Introduction ........................................................................................................... 34
MPPT Control Algorithm ..................................................................................... 35
3.2.1 Classical Direct P&O MPPT Algorithm ....................................................... 36
3.2.2 Proposed Fuzzy Direct MPPT Algorithm ..................................................... 38
Design Methodologies of Fuzzy-MPPT Controller .............................................. 40
3.3.1 Analytical Fuzzy Logic Design Strategy ...................................................... 41
3.3.2 Genetic Algorithm-Based Design Optimization ........................................... 42
Hardware Implementation of Fuzzy-MPPT Controller ........................................ 46
Experimental and Simulation Results and Discussion.......................................... 47
3.5.1 Assessment of the Proposed FL Design Strategies ....................................... 47
3.5.2 Experimental and Simulation Results of Fuzzy MPPT ................................ 49
3.5.3 Comparison between the Proposed Fuzzy-MPPT, P&O and INC ............... 51
Conclusion ............................................................................................................ 54

Chapter 4 Analytical Modeling of Different Topologies of BIWPTS in EVs

Applications ...................................................................................................................... 56
Introduction ........................................................................................................... 56
V2G Scheme Description ..................................................................................... 58

ix
 4.2.1 Dual H-Bridge BIWPT System .................................................................... 60
4.2.2 Bidirectional Power-Flow Control ................................................................ 61
4.2.1 Compensation Network ................................................................................ 63
Steady-State Circuit representation of BIWPTS’s Components .......................... 65
4.3.1 HF H-bridge Inverter Model ......................................................................... 66
4.3.2 Magnetic Coupler Model .............................................................................. 66
Steady-State Mathematical Model of Different Topologies of BIWPTS ............. 69
4.4.1 LC-Series Compensation Network ............................................................... 71
4.4.2 LC-Parallel Compensation Network ............................................................. 72
4.4.3 LCL Compensation Network ........................................................................ 73
Fundamental Power-Flow Modeling .................................................................... 78
Conclusion ............................................................................................................ 82

Chapter 5 Hardware Implementation of a Dual H-Bridge BIWPTS Prototype ............ 83

Introduction ........................................................................................................... 83
HF H-Bridge Converter and Driver ...................................................................... 84
5.2.1 Component Selection .................................................................................... 84
5.2.2 Driver’s Circuits Design ............................................................................... 86
5.2.2.1 Sizing the Bootstrap Capacitor (CBOOT) ................................................ 86
5.2.2.2 Sizing the Turn-on Gate Resistance (Rg(on)).......................................... 87
5.2.2.3 Sizing the Turn-off Gate Resistance (Rg(off)) ........................................ 88
5.2.3 Snubber Circuits Design ............................................................................... 88
5.2.4 VSHBC PCB Layout Design ........................................................................ 89
Phase-shift Control and Modulation ..................................................................... 91
Electromagnetic Power Pads................................................................................. 94
Compensation Capacitors ..................................................................................... 96
Design Verification and Results ........................................................................... 97
5.6.1 Evaluating the Phase-shift Circuit Performance ........................................... 99
5.6.2 Evaluating the Driving Circuits Performance ............................................. 101
5.6.3 The Converter’s PCB Layout Performance Assessment ............................ 103
5.6.4 The VSHBC’s Circuit Performance Assessment ........................................ 105
5.6.5 The Entire BIWPTS’s Performance............................................................ 107
5.6.5.1 Full Supply Capacity Test ................................................................... 109
5.6.5.2 Reduced Supply Capacity Test ........................................................... 112
5.6.5.3 Practical Considerations during Testing the BIWPTS........................ 113
Conclusion .......................................................................................................... 114

Chapter 6 Assessment Analysis of Different Compensation Topologies of BIWPTS 116

Introduction ......................................................................................................... 116
Assessment Criteria of Compensation Topologies in BIWPTS. ........................ 117
Performance Evaluation for Different Topologies. ............................................ 119
6.3.1 Assessment of LC-Series Topology............................................................ 120
6.3.2 Assessment of LC-Parallel Topology ......................................................... 124
6.3.3 Assessment of LCL Topology. ................................................................... 125
6.3.4 Power-Flow Assessment for Different Configuration. ............................... 127

x
 Experimental Evaluation for Different Topologies. ........................................... 129
6.4.1 Experimental Validation of LCL Topology Model. ................................... 130
6.4.2 Evaluating the Sensitivity to Misalignment for all Topologies. ................. 133
6.4.3 Summary for the Assessment Analysis....................................................... 138
Conclusion .......................................................................................................... 140

Chapter 7 Dynamic and Physics-based Models of LCL BIWPTS .............................. 142

Introduction ......................................................................................................... 142
Analytical Dynamic Model of BIWPTS ............................................................. 144
7.2.1 HF VSHBC Dynamic Model ...................................................................... 145
7.2.2 Magnetic Coupler and Compensation Network Dynamic Model............... 148
Physics-Based Co-simulation Platform .............................................................. 150
7.3.1 Finite-Element Model in Magnet ................................................................ 151
7.3.2 FE-Simulink Co-simulation ........................................................................ 157
Model Verification and Discussion .................................................................... 158
Error Analysis ..................................................................................................... 159
Harmonics Analysis ............................................................................................ 164
Conclusion .......................................................................................................... 167

Chapter 8 Magnetic Design Strategy of High-Power Symmetrical LCL BIWPTS .... 168
Introduction ......................................................................................................... 168
The Proposed Design Strategy ............................................................................ 169
Magnetic Design of the Power Pads ................................................................... 173
8.3.1 3D Finite Element Model of DD Power Pad .............................................. 174
8.3.2 3D Finite Element Analysis and Optimization of DD Power Pad .............. 176
8.3.3 Design of LCL Compensation Configuration ............................................. 181
Design Verification and Results ......................................................................... 181
Conclusion .......................................................................................................... 188

Chapter 9 Coil Design Optimization in the DD Power Pad Structure......................... 189

Introduction ......................................................................................................... 189
A Detailed 2D FEM of the DD Power Pad ......................................................... 190
The Multi-objective Optimization Problem ........................................................ 192
9.3.1 Optimization Objectives ............................................................................. 193
9.3.1.1 Coupling Performance with Misalignment ......................................... 194
9.3.1.2 Material Cost of the DD Power Pad ................................................... 195
9.3.2 Multi-objective Cost Function and Constraints .......................................... 197
Solving of the Optimization Problem ................................................................. 197
9.4.1 The Proposed Improved Tabu Search (ITS) Algorithm ............................. 198
9.4.2 The Link between the ITS algorithm and FEA ........................................... 200
Results and Discussion ....................................................................................... 201
9.5.1 Optimization Results................................................................................... 202
9.5.2 Assessment of the IPT System’s Performance ........................................... 205
Conclusion .......................................................................................................... 211

Chapter 10 Autonomous Two-Layer Predictive Power-Flow Control for BIWPTS .... 212

xi
 Introduction ......................................................................................................... 212
Detailed Power-Flow Modeling .......................................................................... 214
10.2.1 Analytical Models of Active and Reactive Power-Flow ............................ 215
10.2.1.1 Total Power-Flow Model (PAPM and QRPM) ........................................ 215
10.2.1.2 Fundamental Power-Flow Model (PFAPM and QFRPM) ........................ 216
10.2.1.3 Approximate Fundamental Power-Flow Model (PAFAPM and QAFRPM)216
10.2.1.4 Approximate Power-Flow Model (QARPM) ......................................... 217
10.2.2 Models Verification and Comparison ......................................................... 217
Power-Flow Criteria............................................................................................ 220
10.3.1 Active Power-Flow Criteria ........................................................................ 222
10.3.2 Reactive Power-Flow Criteria..................................................................... 222
10.3.3 Verification for the Power-Flow Criteria .................................................... 223
Proposed Two-Layer Predictive Active Power-Flow Controller........................ 226
10.4.1 Control Parameters’ Design ........................................................................ 227
10.4.2 Online Mutual Inductance Estimation ........................................................ 228
10.4.3 First Control Layer ...................................................................................... 229
10.4.4 Second Control Layer ................................................................................. 234
Control Implementation and Results .................................................................. 235
10.5.1 Testing the Proposed Mutual Inductance Estimation ................................. 237
10.5.2 Verifying the Proposed Controller Performance ........................................ 238
Conclusion .......................................................................................................... 241

Chapter 11 Feasibility Analysis of Quasi-dynamic BIWPTS for EV Applications ...... 243

Introduction ......................................................................................................... 243
Quasi-Dynamic BIWPTS Description ................................................................ 245
Modeling of Wirelessly Connected Light-Duty EV ........................................... 248
11.3.1 Lithium-Ion EV’s Battery Model................................................................ 248
11.3.2 EV’s Drive System Model .......................................................................... 255
11.3.3 Model Verification ...................................................................................... 258
Feasibility Analysis of Implementing QDWPT Systems at Traffic Signals....... 261
11.4.1 Fixed Power Charging Scenario ................................................................. 261
11.4.2 Variable Power Charging Scenario ............................................................. 264
11.4.3 Fixed Power Charging and Discharging Scenario ...................................... 266
11.4.4 Driving Cycle Performance Evaluation ...................................................... 268
Impact of Misalignment on the Feasibility Analysis .......................................... 271
Conclusion .......................................................................................................... 275

Chapter 12 Conclusions ................................................................................................. 277

Chapter 13 Recommendations and Future Works ......................................................... 281

LIST OF REFERENCES ................................................................................................ 283

VITA……… ................................................................................................................... 302

xii
 LIST OF TABLES

TABLE PAGE

Table 2.1. BP 4175T PV Module Specifications at STC. ..................................................31

Table 2.2. Extracted Parameters of BP 4175T PV Module. ..............................................32

Table 3.1. Results of Design Parameters from Analytical Design Method. ......................41

Table 3.2. GA Setup for the Design Optimization Algorithm. ..........................................44

Table 3.3. Results of Design Parameters from GA Design Method. .................................45

Table 3.4. Specifications of Hardware Components. ........................................................47

Table 3.5. MAPE for the Proposed FL-MPPT, P&O and INC Algorithms. .....................54

Table 4.1. γ-Parameter for Different Topologies. ..............................................................80

Table 5.1. Final Converter Design Parameters. .................................................................99

Table 5.2. Performance Parameters during Step Response. ............................................104

Table 5.3. Design Parameters of LCL BIWPTS. .............................................................109

Table 5.4. NRMSE between Simulation and Experimental Results. ...............................113

Table 6.1. Power-Flow Performance for all Topologies under Full-Loading

Conditions. ...............................................................................................................127

Table 6.2. Power-Flow Performance for all Topologies under Light-Loading

Conditions. ...............................................................................................................128

Table 6.3. Large and Small-Scale Power-Flow Analysis. ...............................................133

Table 6.4. Summary of the Assessment Analysis. ...........................................................139

Table 7.1. Switching States of the Primary HF VSHVC. ................................................147

Table 7.2. Dimensions of the Power Pad in 2D-FEM .....................................................153

Table 7.3. Design Parameters of the BIWPTS Model .....................................................159

Table 7.4. %NRMSE at Full and Light Loading Conditions. ..........................................164

Table 7.5. %THD during G2V Operation for Full and Light Loading ............................166

xiii
Table 7.6. %THD during V2G Operation for Full and Light Loading ............................167

Table 8.1. Final Design Parameters of the DD Power Pad. .............................................180

Table 8.2. Design Parameters of LCL BIWPTS with DD pads structure. .......................184

Table 8.3. The Power Flow of the Large and Small Scale Models .................................187

Table 9.1. Dimensions of the 2D FEM for the DD pad structure. ...................................192

Table 9.2. The Misalignment Parameters and the Cost Coefficients during the
Optimization. ...........................................................................................................202

Table 9.3. Boundaries of the Optimization Variables......................................................203

Table 9.4. Settings of the ITS Optimization Algorithm’s Parameters. ............................204

Table 9.5. The Optimum Design Parameters. ..................................................................205

Table 9.6. Cost of the Wireless Coupler (Two Pads) in all Scenarios. ............................209

Table 10.1. Numerical Power-Flow Evaluation using the Different Models. .................220

Table 10.2. Worst Power-Flow Criteria (α=𝛽=180o) for LCL BIWPTS. .......................224

Table 10.3. Different Modes of Power-Flow (α=𝛽=180o) in BIWPTS. .........................225

Table 11.1. Charging and Discharging Coefficients of the Battery Module’s

Analytical Model. ....................................................................................................253

Table 11.2. Design Parameters of an 11.1 kW LCL BIWPTS at 40 and 85 kHz. ...........259

Table 11.3. Benefit Factor Evaluation for all Scenarios (miles/kWh). ............................270

Table 11.4. Driving Performance Considering Misalignment. ........................................275

xiv
 LIST OF FIGURES

FIGURE PAGE

Figure 1.1. DC-bus configuration for G2V and V2G services with wirelessly connected
EVs and PV generation. ...............................................................................................7

Figure 2.1. Block diagram of grid-connected PV power system. ......................................16

Figure 2.2. Equivalent circuit of a grid-connected PV power system through a DC-DC

Cuk converter. ............................................................................................................17

Figure 2.3. Effect of Rp on I-V characteristics of PV module at different climatic

conditions. ..................................................................................................................24

Figure 2.4. Comparison between I-V characteristic for BP 4175T PV module at

STC from measurement and simulated using GA and analytical based parameters. 32

Figure 3.1. Block diagram of grid-connected PV power system including MPPT

control. .......................................................................................................................36

Figure 3.2. Flow-chart of the classical direct P&O MPPT algorithm. .............................37

Figure 3.3. Block diagram of the proposed fuzzy logic system tracking algorithm. ........39

Figure 3.4. Flowchart of GA optimization procedure........................................................43

Figure 3.5. Progression of the GA Fitness function...........................................................45

Figure 3.6. Experimental setup of PV power system.........................................................46

Figure 3.7. Simulation results for the proposed FL system designs ..................................48

Figure 3.8. Experimental and simulation results for FL-MPPT control ............................50

Figure 3.9. Experimental and simulation results for FL-MPPT controller under fast
temperature variation .................................................................................................51

Figure 3.10. Comparison between the proposed fuzzy-MPPT and INC algorithm ...........52

Figure 3.11. Comparison between the proposed fuzzy-MPPT and P&O algorithm .........53

Figure 3.12. Experimental and simulation results for P&O algorithm ..............................54

Figure 4.1. Wired and wireless connected EV through a DC-bus. ....................................58

xv
Figure 4.2. Wired and wireless connected EV though an AC-bus. ...................................59

Figure 4.3. Schematic circuit diagram of BIWPTS. ..........................................................61

Figure 4.4. Phase-shift control of the HF inverters ............................................................62

Figure 4.5. Legs and total inverter output voltages in terms of the phase-shift control
parameters. .................................................................................................................63

Figure 4.6. Different compensation topologies in BIWPTS.. ............................................64

Figure 4.7. T-model representation of wireless coupler ....................................................68

Figure 4.8. π-model of BIWPTS for inverter’s currents in terms of inverter’s voltages. ..69

Figure 4.9. π-model of BIWPTS for coil’s currents in terms of inverter’s voltages. ........70

Figure 4.10. BIWPTS equivalent circuits using T-model mutual coupling for LC-series
topology. ....................................................................................................................71

Figure 4.11. BIWPTS equivalent circuits using T-model mutual coupling for
LC-parallel topology. .................................................................................................72

Figure 4.12. BIWPTS equivalent circuits using T-model mutual coupling for LCL
topology. ....................................................................................................................74

Figure 4.13. Effect of filters inductances on the fundamental impedance frequency

response for LCL-parallel design ..............................................................................76

Figure 4.14. Effect of system inductances on the fundamental impedance frequency

response for LCL-parallel design (40 kHz) ...............................................................77

Figure 4.15. Effect of system resistances on the fundamental impedance frequency

response for LCL-parallel design ..............................................................................78

Figure 5.1. Forward-bias safe operating area of IXFB110N60P3 Si power MOSFET. ....85

Figure 5.2. Bootstrap driving circuit components using FAN7391 IC driver. ..................86

Figure 5.3. Layout of the optimized design converter’s PCB............................................90

Figure 5.4. Layouts of an arbitrary design converter’s PCB. ............................................90

Figure 5.5. Switching signals and output voltage of VSHBC under phase-shift control. .91

Figure 5.6. Hardware implementation of adjustable dead-time circuit. ............................92

Figure 5.7. Logic block diagram of phase-shift control circuit. ........................................93

xvi
Figure 5.8. Phase-shift circuit schematic diagram. ............................................................93

Figure 5.9. The hardware of phase-shift circuit’s PCB. ....................................................94

Figure 5.10. Circular power pad with ferrite core. ............................................................96

Figure 5.11. The hardware implementation of VSHBC’s PCB. ........................................98

Figure 5.12. Measured switching signals (0-5V) of one inverter (4 MOSFETs) during
full inverter voltage ..................................................................................................100

Figure 5.13. Measured switching signals of one leg (2 MOSFETs) during full inverter
voltage showing the dead-time ................................................................................100

Figure 5.14. Measured switching signals (0-5V) of one inverter (4 MOSFETs) during
reduced inverter voltage. ..........................................................................................101

Figure 5.15. Measured gate signals (0-15V) of one inverter (4 MOSFETs) during full
inverter voltage ........................................................................................................102

Figure 5.16. Measured gate signals (0-15V) of one inverter (4 MOSFETs) during
reduced inverter voltage. ..........................................................................................102

Figure 5.17. Measured ringing effect of the optimized and non-optimized PCB
designs. ....................................................................................................................103

Figure 5.18. Measured output inverter voltage with and without snubber circuit. ..........103

Figure 5.19. Measured unit-step response of inverter output voltage..............................104

Figure 5.20. Measured inverter output voltage during full output test ............................105

Figure 5.21. Measured inverter output voltage during reduced output test .....................106

Figure 5.22. Measured inverter output voltage at different frequencies ..........................107

Figure 5.23. Experimental setup of LCL BIWPTS..........................................................108

Figure 5.24. Simulink model of LCL BIWPTS. ..............................................................108

Figure 5.25. Experimental and simulated BIWPTS’s performance under full supply
voltage for G2V operation (α = β =180o, δ = -90o, Vdc=60 V) ................................110

Figure 5.26. Experimental and simulated BIWPTS’s performance under full supply
voltage for V2G operation (α = β =180o, δ = 90o, Vdc=60 V). ................................111

Figure 5.27. Experimental and simulated BIWPTS performance under reduced supply
voltage for G2V operation (α = β =90o, δ = -90o, Vdc=60 V) ..................................112

xvii
Figure 6.1. LC-series topology performance under full supply capacity for G2V
operation (α=β=180o, δ=90o, Vdc=240 V and Vb=200 V)........................................121

Figure 6.2. LC-series topology performance under reduced supply capacity for
charging operation (G2V) (α=β=90o, δ=90o, Vdc=240 V and Vb=200 V) ...............122

Figure 6.3 LC-series topology performance under reduced supply capacity for
discharging operation (V2G) (α=β=90o, δ=-90o, Vdc=240 V and Vb=200 V) .........122

Figure 6.4. LC-parallel topology performance under reduced supply capacity for G2V
operation (α=β=90o, δ=-90o, Vdc=240 V and Vb=200 V) ........................................125

Figure 6.5. LCL-topology performance under reduced supply capacity for charging
operation (G2V) (α=β=180o, δ=-90o, Vdc=240 V and Vb=200 V) ...........................126

Figure 6.6. Experimental and theoretical LCL-topology performance under full supply
capacity for G2V operation (α=β=180o, δ=-90o, Vdc=60 V, Vb=50 V) ...................129

Figure 6.7. Experimental and theoretical LCL-topology performance under full supply
capacity for V2G operation (α=β=180o, δ=90o, Vdc=60 V, Vb=50 V). ....................130

Figure 6.8. Experimental and theoretical LCL-topology performance under reduced

supply capacity for G2V operation (α=β=90o, δ=-90o, Vdc=60 V, Vb=50 V)..........131

Figure 6.9. Experimental and theoretical LCL-topology performance under reduced

supply capacity for V2G operation (α=β=90o, δ=90o, Vdc=60 V, Vb=50 V) ...........132

Figure 6.10. Angle misalignment in IPT system .............................................................134

Figure 6.11. Measured and estimated normalized primary coil current under
misalignment for all topologies (G2V, α=β=180o). .................................................135

Figure 6.12. Measured and estimated normalized power under misalignment for all
topologies (G2V, α=β=180o) ...................................................................................136

Figure 6.13. Theoretical fundamental equivalent impedance under misalignment for

all topologies (G2V, α=β=180o) ..............................................................................137

Figure 7.1. Power electronics circuit diagram of LCL BIWPTS. ....................................146

Figure 7.2. Switching activities and output voltage waveform of primary inverter. .......146

Figure 7.3. Dynamic electric equivalent circuit of symmetrical LCL BIWPTS. ............149

Figure 7.4. Double-D power pad structure using sheet of ferrite core. ...........................151

Figure 7.5. Characteristics of N87 ferrite material ..........................................................152

xviii
Figure 7.6. 2D-FEM of one power pad of the wireless coupler. ......................................153

Figure 7.7. Instantaneous wireless pads currents waveforms. ..........................................155

Figure 7.8. Magnetic flux density distribution of 2D-FEM. ............................................156

Figure 7.9. Co-simulation Simulink-Magnet Platform.....................................................157

Figure 7.10. Coupling between Simulink and Magnet software. .....................................158

Figure 7.11. Introduced dynamics in BIWPTS (Vdc=240V and Vb=200V) ....................160

Figure 7.12. RMS values of BIWPTS’s variables (Vdc=240V and Vb=200V). ...............161

Figure 7.13. Error between analytical and FE co-simulation platform results ................162

Figure 7.14. Active power flow .......................................................................................163

Figure 7.15. FFT analysis during full G2V operation (Vdc=240V, Vb=200V, α=β=180o
and δ=-90o)...............................................................................................................165

Figure 7.16. FFT analysis during full V2G operation (Vdc=240V, Vb=200V, α=β=180o
and δ=90o) ................................................................................................................166

Figure 8.1. Flow-chart for the proposed design methodology. ........................................170

Figure 8.2. Effect of the pad self-inductance on the BIWPTS’s performance (Vdc=
420 V and Vb=350 V) ..............................................................................................172

Figure 8.3. Different structures of power pads in EVs applications ................................173

Figure 8.4. 3D-FEM for DD wireless coupler based on ferrite bars................................175

Figure 8.5. Different views of DD wireless coupler ........................................................175

Figure 8.6. Magnetic field distribution across a vertical section in the middle of the 3D
FEM of the DD power pad.......................................................................................176

Figure 8.7. Magnetic flus density distribution in the ferrite bars of the DD power pad
model. ......................................................................................................................177

Figure 8.8. Effect of coil-width on the evaluation parameters (Vdc=420 V and Vb=350
V) .............................................................................................................................178

Figure 8.9. Effect of Df on the evaluation parameters (Vdc=420 V and Vb=350 V) .......179

Figure 8.10. Effect the pitch on the evaluation parameters (Vdc=420 V and Vb=350
V). ............................................................................................................................180

xix
Figure 8.11. Ferrite bars of N87 material from EPCOS for one DD power pad. ............182

Figure 8.12. Final shape of the developed 8 kW DD power pad. ....................................183

Figure 8.13. Test-bed for a symmetrical LCL BIWPTS based on the DD power pads...184

Figure 8.14. Performance of an 8 kW LCL BIWPTS, during G2V operation (α=β=

180o, δ=-90o, Vdc=420V and Vb=350V), using Simulink and analytical models ....185

Figure 8.15. Performance of the small scale LCL BIWPTS, during G2V operation
(α=β=180o, δ=-90o, Vdc=60V and Vb=50V), using experimental and analytical
models ......................................................................................................................186

Figure 8.16. Performance of the small scale LCL BIWPTS, during V2G operation
(α=β=180o, δ=90o, Vdc=60V and Vb=50V), using experimental and analytical
models ......................................................................................................................187

Figure 9.1. A 2D-FEM of the DD power pad, showing the optimization variables. .......191

Figure 9.2. Different positions of misalignments ............................................................195

Figure 9.3. Top view of DD power pad with components’ dimensions. .........................196

Figure 9.4. Flowchart of the proposed ITS algorithm. ....................................................199

Figure 9.5. Block diagram for the link between ITS algorithm in MATLAB and FEA
in Maxwell. ..............................................................................................................201

Figure 9.6. Progression of ITS optimization algorithm. ..................................................204

Figure 9.7. Four different design scenarios .....................................................................206

Figure 9.8. Magnetic flux distribution for all scenarios (g=130 mm and v=h=θ=0) .......207

Figure 9.9. Coupling factor vs. air gap length with perfect alignment (h=θ=0). .............208

Figure 9.10. Coupling factor vs. horizontal misalignment with g=130 mm and θ=0. .....210

Figure 9.11. Coupling factor vs. rotational misalignment with g=130 mm and h=0. ......210

Figure 10.1. Comparison between the different active power models (α = β = 180o,
and Ps_max is the base)...............................................................................................218

Figure 10.2. Comparison between the different reactive power models (α = β = 180o
and Qs_max is the base). .............................................................................................219

Figure 10.3. BIWPTS’s active and reactive power-flow criteria. ....................................221

xx
Figure 10.4. Power flow criteria with different control parameters (Vdc=60 V). ............224

Figure 10.5. The proposed two-layer predictive active power-flow control. ..................227

Figure 10.6. First layer control principle .........................................................................231

Figure 10.7. The proposed first layer control algorithm. .................................................233

Figure 10.8. The proposed second control layer algorithm. ............................................235

Figure 10.9. Cmod S6 FPGA board. ................................................................................236

Figure 10.10. Implementation of FPGA-based PPM including the dead-time and the
driver. .......................................................................................................................236

Figure 10.11. Mutual inductance estimation performance (Vdc=60V) .............................237

Figure 10.12. PAPF control performance during Abstain and Discharge modes.............238

Figure 10.13. Comparison between the performance of PI and the proposed PAPF
control during Abstain and Discharge modes...........................................................239

Figure 10.14. PAPF control performance during Discharge and Charge modes..............240

Figure 10.15. Comparison between the performance of PI and the proposed PAPF
control during Discharge and Charge modes. ..........................................................241

Figure 11.1. The proposed implementation of Quasi-dynamic WPT system at traffic

signal. .......................................................................................................................247

Figure 11.2. Block diagram of a wirelessly connected EV through the BIWPTS...........248

Figure 11.3. Lithium-ion test battery ...............................................................................249

Figure 11.4. 2nd order dynamic battery equivalent circuit model with EV traction
system. .....................................................................................................................250

Figure 11.5. Curve fitted battery model coefficients ......................................................251

Figure 11.6. Battery model verification tests ...................................................................255

Figure 11.7. FTP-72 Driving profile ...............................................................................257

Figure 11.8. BIWPTS’s performance at 40kHz resonant frequency, during G2V

operation (α=β=180o, δ=-90o). .................................................................................260

Figure 11.9. BIWPTS’s performance at 85kHz resonant frequency, during G2V

operation (α=β=180o, δ=-90o). .................................................................................260

xxi
Figure 11.10. Driving performance under fixed power charging WPT2 (SOCi =80%).
.................................................................................................................................262

Figure 11.11. Driving performance under fixed power charging WPT4 (SOCi =80%).
.................................................................................................................................263

Figure 11.12. Driving performance at different charging levels (SOCi =70%). ..............264

Figure 11.13. Charging rate decision based on SOC. ......................................................265

Figure 11.14. Driving performance under variable power charging level (SOCi =
SOCmax =80%, SOCmin =30%) .................................................................................266

Figure 11.15. Driving performance under charging and discharging operation (WPT4,
SOCi =80%, Ƥ’=5.5 cents/kWh) ..............................................................................268

Figure 11.16. Driving range for all different scenarios ...................................................269

Figure 11.17. Long track primary coupler. ......................................................................272

Figure 11.18. String of pads primary coupler. .................................................................272

Figure 11.19. Driving performance considering a misalignment (WPT4, SOCi =80%).

.................................................................................................................................274

xxii
 LIST OF ABBREVIATIONS AND ACRONYMS

2D FEM Two Dimensional Finite Element Model 13

3D-FEM Three Dimensional Finite Element Model 13

A.M Air Mass 31

AC Alternating Current 6

ADC Analogue to Digital 37

Ah Ampere-Hour 256

b.f Benefit Factor 279

BIWPTS Bidirectional Inductive Wireless Power Transfer System 6

BMS Battery Management System 218

BWPT Bi-directional Wireless Power Transfer 251

DC Direct Current 6

DD Double-D 13

DDQ Double-DQ 99

DWPT Dynamic Wireless Power Transfer 250

EC Equivalent Circuit 68

EMI Electromagnetic Interference 86

EPRI Electric Power Research Institute 1

ESL Equivalent Series Inductance 92

ESS Energy Storage System 2

eV Electron Volts 30

xxiii
EVs Electric Vehicles 1

FCT Full Cycle Time 277

FE Finite Element 168

FEA Finite Element Analysis 10

FEM Finite Element Model 10

FFT Fast-Fourier Transform 140

FIS Fuzzy Inference System 40

FL Fuzzy Logic 36

FLC Fuzzy Logic Control 39

FPGA Field-Programmable Gate Array 10

FTP Federal Test Procedure 264

G2V Grid-to-Vehicle 2

GA Genetic Algorithm 11

HF High Frequency 11

IC Integrated Circuit 104

ICE Internal Combustion Engine 1

ICNIRP International Commission on Non-Ionizing Radiation Protection 98

IEC International Electrotechnical Commission 3

IEEE Institute of Electrical and Electronics Engineers

IGBT Insulated-Gate Bipolar Transistor 20

INC Incremental Conductance 54

IPT Inductive Power Transfer 5

xxiv
ITS Improved Tabu Search 13

ITSs Intelligent Transportation Systems 236

kHz kilo-Hertz 22

kW kilo-Watt 12

kWh kilo-Watt-Hour 252

LEM Liaisons Electroniques et Mecaniques 48

LHS Latin Hypercube Sampling 203

Li Lithium 10

MAP Maximum Active Power 231

MAPE Mean Absolute Power Error 50

MMF Magneto-Motive Force 155

MnZn Manganese and Zinc 70

MOF Multi-Objective Optimization Function 201

MOSFET Metal–Oxide–Semiconductor Field-Effect Transistor 20

MP Maximum Power 45

mpc miles per cycle 277

MPP Maximum Power Point 22

MPPT Maximum Power Point Tracking 8

MRP Maximum Reactive Power 231

NHTSA National Highway Traffic Safety Administration 263

NMSE Normalized Mean Square Error 223

NRMSE Normalized Root Mean Square Error 116

xxv
OCV Open Circuit Voltage 257

ODE Ordinary Differential Equation 21

P&O Perturb and Observe 28

PAPF Predictive Active Power-Flow 217

PCB Printed Circuit Board 11

PF Power Factor 67

PFC Power Factor Correction 13

PI Proportional-Integral 14

PID Proportional-Integral-Derivative 217

PLL Phase-Locked Loop 11

PPM Pulse-Phase Modulation 65

PV Photovoltaic 2

PWM Pulse-Width Modulation 37

QDWPT Quasi-Dynamic Wireless Power Transfer 14

RESs Renewable Energy Sources 2

RK4 Runge-Kutta 154

RMS Root-Mean Square 17

SAE Society of Automotive Engineers 63

SOC State of Charge 218

STC Standard Test Conditions 25

STSL Single-Turn Single-Layer 193

THD Total Harmonic Distortion 169

xxvi
T-S Takagi-Sugeno 41

UPF Unity Power Factor 63

UTS Universal Tabu Search 203

V2G Vehicle-to-Grid 2

VA Volt-Ampere 119

VAR Volt-Ampere-Reactive 226

VSHBC Voltage-Source H-Bridge Converter 87

WPT Wireless Power Transfer 14

WPTAC Wireless Power Transfer Automatic Charging 278

ZCS Zero-Current Switching 121

ZPF Zero Power Factor Operation 231

ZVS Zero-Voltage Switching 121

112362520 20231161317

xxvii
 Problem Statement

Volatile fuel prices, coupled with an increased emphasis on reducing greenhouse and

carbon dioxide emissions, have fostered significant growth in the electric vehicles’ (EVs’)

market over the last ten years. Even though EVs have not been widely adopted, in part

because of technical limitations, social obstacles, and cost compared to conventional

internal combustion engine (ICE) vehicles [1], based on moderate expectations, by 2020

up to 35% of the total vehicles in the U.S. will be EVs, according to the Electric Power

Research Institute (EPRI) [2]. With the large-scale introduction of EVs, the power grid will

face a major challenge to satisfy the load demand. The increasing number of EVs will put

an additional stress on the existing distribution system’s components, such as transformers

and cables, and may perturb their operation, particularly during the peak demand periods.

Several studies have been carried out to investigate the impact of EVs on the power

system in terms of load capacity, power quality, economy and environment [1], [3]–[5].

The impact of EV charging on the grid load capacity was discussed in [6]–[12]. The studies

concluded that the disorder charging will increase grid peak load, so it needed additional

investment on generating electricity and transmission capacity. In terms of power quality,

EVs’ charging behavior will bring the voltage out-of-limits of the distribution network,

reduce the power quality, increase the line loss, reduce the life span of distribution

transformers and increase the harmonics and current faults [3], [13]–[15]. In [16], a review

for the impact of the EVs charging on the residential distribution systems was presented in

1
terms of electricity generation adequacy [11], [17]–[20], transformer aging [14], [21]–[23]

and distribution system power quality [24]–[28]. According to the literature, numerous

techniques are proposed to mitigate the impacts of EVs’ charging on the power grid. These

techniques can be categorized into two groups [16]. In the first approach, the utilities

indirectly manage EVs’ charging using the Time-of-Use pricing [29]–[35]. The low off-

peak electricity rates in a Time-of-Use pricing scenario stimulates EVs’ owners to charge

their vehicles during the off-peak hours. This scheme significantly decreases the peak load

demand and mitigates transformer overloading and heating concerns. In the other

technique, the utilities directly control EVs’ charging rates and charging start time using

smart charging algorithms [24], [7], [36]–[46].

Traditionally, these vehicles have been considered as nonlinear loads for the power

grid, whose impacts on supply’s stability and quality have been well studied. However, as

the penetration rate of EVs into the transportation increases, the idea of utilizing them not

just as loads but also as energy storage systems (ESSs) is explored. This is a very promising

technology since most the light-duty vehicles spend significant time not being operated

and there may be opportunities to utilize their stored energy to support the power grid. This

concept is known as vehicle-to-grid (V2G) and grid-to-vehicle (G2V) service [1], [47]–

[50]. In this concept, an EV can operate in discharge (V2G) mode, as a source, to inject

energy to the grid, or in charge (G2V) mode, as a load, to suck energy from the grid. This

concept has attracted the attention of the grid operators and vehicles’ owners, as EVs are

considered an indispensable component in both “living and mobility” and sustainable

living in the near future [51]. In the V2G concept, EVs as ESSs can provide peak load

shaving and act as a reserve resource against unexpected outages [52]. As the size of the

2
EV fleet increases, the bulk energy storage available can become considerable in size,

which adds to flexibility of actions taken by the power system operator.

The benefits of V2G technology can be maximized if it is being associated with

renewable energy sources (RESs), such as solar and wind generations [53]–[55]. The

intelligent integration between the EVs and RESs can provide an ideal solution for the

issues due to both of them. The RES can be designed to supply the bulk power demand of

EVs charging with minor support from the power grid. On the other hand, EVs (as ESSs)

can be utilized to smoothen the power generation for the intermittent RESs and mitigate

their bad impacts on the power grid, such as voltage and frequency instability issues [56]–

[61]. Unlike conventional loads, the energy in EVs is sizable enough to maintain all solar

and most of the wind generation. Among all RESs, Photovoltaic (PV) power systems are

expected to play a vital role in the future energy efficient and zero emission society. Several

factors have been boosting this: improved generation efficiency of PV modules, flexibility

of implementation and governmental subsidies [62]. However, PV power fluctuates

depending on the environmental conditions, season, and geographic location, and causes

problems, such as voltage fluctuation and large frequency deviation in an electric power

system operation [63]. In order to mitigate these impacts, it is recommended that equipment

with storage capability (such as EVs) should be controlled to compensate the variation in

PV generation [60], [64]–[67]. The integration between EVs and PV generation is

promising due to the massive similarities between them [68]–[71]. The most important

match between them is that EVs will probably be connected to the grid in the same highly

dispersed manner, at the same voltage levels, and in the same range as PV systems. For

example, EVs at residential homes are similar to those typical residential PV systems in

3
parameter values [72]. Also, both PV and EV are integrated to the grid though power

electronic interface with the potential to create intelligent nodes in the system. One vital

difference is that the EV, as an ESS, is bidirectional and can act as either a source or a load,

whereas the PV behaves as a source only. Another dissimilarity is that EVs have both

power and energy limitations, but PV does not have an explicit energy limitation. Due to

these common features between PV and V2G from the grid’s perspective, the grid-support

functions requested for PV will be similar to those required for V2G. In fact, it is possible

that codes and standards presently applied to the PV systems, such as IEEE-1547 and IEC-

62116, may apply to V2G systems with small modification [68]. Therefore, in this study,

the integration between the EVs and PV systems is considered.

As of now, the size, complexity and economics of utility-scale centralized energy

storage control and management might not have been generally acknowledged and

supported. Notwithstanding, with moderate assumptions, the required power and energy

available in the EVs, even with today’s battery technologies, can become available in the

very near future [73]. However, what distinguishes the energy stored in EVs from other

conventional ESSs is their mobile nature. Like other ESSs, accessibility of these resources

relies upon the availability of the primary source of energy (the batteries in this case).

Although unlike other ESSs, accessibility of EVs’ energy depends likewise on the locations

of the vehicles. This complicates matters further by adding another constraint to the

problem. Traditionally, people have been looking at extracting the stored energy, either

from locations where a relatively large number of vehicles exist (e.g., charging stations,

parking lots) [65], [74], or from individual charging devices at residential units [69]. The

common factor in both lies in the fact that the vehicles should be stationary (i.e., parked).

4
While this is a reasonable assumption in many cases, it undermines a great potential source

of energy in the system: vehicles in motion. Furthermore, in order to inject energy back

into the grid, it is necessary for the vehicle to be driven to the charging station/device, and

parked for the duration of the power exchange [75]–[77]. Over those, the driver may need

to change his normal route to get to the charging station. All these could inconvenience the

driver(s) and hence reduce the tendency (or participation level) for participating in V2G

services. Additionally, when the source of energy is needed immediately (e.g., fast

reserves) the distances of the vehicles to the charging stations would serve as a restricting

factor. Finally, for the vehicle to be able to transfer power to the grid, the driver (or an

operator) needs to be physically present to set up an electrical connection between the

vehicle and the station [78], [79]. This obstructs the implementation of supervisory level

control schemes that are managed remotely.

Wireless charging stations (based on inductive power transfer (IPT) technology)

have been proposed as a mean for energy transfer without physical contacts [80]–[91].

Without the need for establishing an actual electrical connection, this technology enables

transfer of energy from the power grid to the vehicle and possibly in the reverse direction

through the usage of magnetic circuits operating at resonance. This power transfer can

occur while the vehicle is in long-term parking (stationary), transient stops (quasi-dynamic)

or moving even close to normal speed (dynamic) [92]–[97]. Utilizing this technology can

turn the potential energy stored in the EVs’ batteries into an accessible source of energy

distributed across the power grid [98]. In addition, a supervisory control scheme can be

designed and implemented to remotely control these individual sources of energy in order

to provide a considerable source of energy storage for the power grid [93]–[95]. Several

5
studies have been conducted, which show comparative analyses for the percentage of EVs

interactivity between the conductive and inductive connections [77], [99], [100]. The

studies emphasized that the conductive connection guarantees about 10% of EVs to interact

successfully with the power grid, while the inductive connection provides about 65% [75],

[100]–[102]. Thus, the bidirectional IPT technology is considered in this study to

accomplish the integration between the EVs, PV generation and power grid, and achieve

G2V and G2V services.

Research Objective

The main objective of this dissertation is to provide a comprehensive study for

wirelessly connected EVs to achieve G2V and V2G services with the large-scale

penetration of PV generation. The system’s configuration under consideration is indicated

in Figure 1.1. In this structure, it is proposed that all the resources should be integrated

through a common direct current (DC)-bus. The alternating current (AC) power grid

connection is established using a bi-directional grid-tied converter, which is responsible

for regulating the DC-link voltage. Each EV is coupled to the DC-bus through its own

bidirectional inductive wireless power transfer system (BIWPTS) to facilitate charging

(G2V) and discharging (V2G) operation. In the G2V mode, control of the BIWPTS can

provide a mechanism to regulate multiple charging scenarios. In the V2G mode, EV’s

energy can be injected back to the AC grid to provide support during peak loading periods

or an outage. As an added benefit, the introduction of a localized DC-bus would help ease

the integration of PV generation, which is accomplished through a unidirectional DC-DC

converter, connected directly to the DC-bus. In addition, the surrounding DC loads can be

6
supplied using another unidirectional DC-DC converter. Through the inclusion of PV

generation, the bulk load demand can reduce grid stress when EVs are in the G2V mode,

while in the V2G mode, the EV’s battery array can aid in smoothening the PV generation.

DC bus

Power grid
BIWPTS
AC

DC
BIWPTS

PV system
DC BIWPTS

DC
DC DC
loads
DC

Figure 1.1. DC-bus configuration for G2V and V2G services with wirelessly connected
EVs and PV generation.

This architecture is flexible and can be physically implemented in several positions,

in which EVs are in long-term parking, such as charging stations, public and private

parking lots and car parks, and short-term parking, such as bus stops and road traffic

signals. Furthermore, the same configuration can be deployed in the powered roads to

provide G2V and V2G services for the in-motion EVs, through the dynamic wireless

charging.

This study is crucial and provides valuable solutions for many issues that impede the

penetration of EVs in transportation and RES in electric energy generation. The study

7
presents modeling, design and control analyses for the power interfaces among the EV, PV

system and utility grid. These interfaces have the capability to provide the necessitate

power-flow between the power grid and the other resources (EVs and PV panel) to achieve

G2V and V2G services. For the PV power system, a unidirectional DC-DC converter is

considered as the power interface to extract the maximum power from the PV generator,

regardless of the environmental conditions, by implementing an intelligent maximum

power point tracking (MPPT) algorithm. For the EV, the duel-side full-bridge BIWPTS is

utilized to manage the two-way power-flow between the EV and DC-bus. Several aspects

have been investigated, analyzing practical issues in the BIWPTS, including developing

accurate analytical and physics-based models, performing effective design optimizations,

developing and implementing new closed-loop controllers, and building prototypes and

testing them.

The outcomes of this study are useful for many organizations, such as utility

companies, automotive companies and consumers. For the power grid, the results help to

minimize the impact of EVs penetration, increase the penetration of PV generation, provide

predictable energy resources, enhance the power grid stability and mitigate its power

quality issues. For EVs manufacturers, it aids to reduce the battery size required by EVs,

improve the operation of EVs through utilizing the energy sources available from other

vehicles and make the vehicle become an energy source to provide energy services to the

grid. Moreover, it is helpful for the consumers by reducing the EV’s price, providing

flexible, reliable and automatic bidirectional interface. Also, it enables the vehicle’s owner

to participate in V2G services, making the vehicle a micro-grid that can be used during

power interruptions and storms.

8
 Original Contributions of the Dissertation

The proper modeling, design and control for the power interfaces among the EVs,

and PV generation, and the power grid through a common DC-bus, is investigated in this

work. The main findings and contributions in this thesis are summarized as follows:

- Developing analytical dynamic modeling and small signal analysis for a grid-connected

PV power system, through a DC-bus.

- Achieving design and optimization analysis for an intelligent maximum power point

tracker in PV systems, based on fuzzy logic, genetic algorithm and analytical models.

The proposed controller is implemented and verified by means of simulation and

experimentally, using the DSPACE microcontroller board.

- Developing accurate steady-state mathematical and computer-based models for the

different topologies of a symmetrical BIWPTS in EV applications. The models were

able to predict the systems’ performance and power-flow; and novel practical

limitations and criteria for the active and reactive power-flow during G2V and V2G

operations are stated.

- Performing a comprehensive performance assessment analysis for the main

configurations of the symmetrical BIWPTS, which aims to define the merits and

demerits of each structure and the most appropriate one for G2V and V2G services.

- Building experimental prototypes for the BIWPTS, including power electronic

converters, compensation networks, power pads and power-flow controller, to verify

the proposed theoretical analyses.

9
- Developing an accurate nonlinear physic-based co-simulation platform that links the

model of the power electronic converters and controllers in Simulink, to the finite-

element model (FEM) of the electro-magnetic coupler.

- Performing magnetic design multi-objective optimization for the power pads in the

wireless interface, based on finite element analysis (FEA) and experimental tests.

- Developing a new, autonomous, two-layer predictive power-flow controller to manage

the bidirectional active power-flow between the EV and DC-bus. The controller is

designed to be placed on the vehicle’s side, and is implemented, using a Field-

Programmable Gate Array (FPGA) board.

- Developing a comprehensive simulation platform for a light-duty wirelessly connected

EV, including the BIWPTS, Lithium (Li)-ion battery and traction system. The platform

was utilized to study the feasibility of implementing the quasi-dynamic wireless

charging systems at the traffic signals on the road, to charge the EVs during the transient

stops at the intersections.

Dissertation Organization

This dissertation is organized as follows:

Chapter 1 introduces the integration between EVs, PV power system and the power

grid, and identifies the G2V and G2V concept based on the wirelessly connected EVs. In

addition, the new findings and contributions in this study are stated.

Chapters 2 and 3 present modeling and control analysis for a grid-connected PV

power system through a DC-bus. In chapter 2, a detailed mathematical nonlinear dynamic

model for the PV power system is developed. Then, this dynamic model is linearized to

10
state the small-signal model for the same system, which is utilized to state the design

considerations for the MPPT algorithm’s parameters. The developed models were verified

by means of simulation and experimental tests. The optimum design for the fuzzy-based

MPPT for the PV system is investigated in chapter 3. Two different design methodologies

are studied in this chapter: the analytical and genetic algorithm (GA)-based method. The

proposed design was implemented and tested by means of simulation and experimentally

in comparison with the existing techniques in the literature.

The following 3 chapters provide assessment analysis for the main topologies of the

symmetrical BIWPTS for EV’s integration. Chapter 4 presents modeling analysis for the

steady-state performance of BIWPTSs based on Fourier series analysis. The analysis is

achieved for the three main compensation configurations of a symmetrical BIWPTS: LC-

series, LC-parallel and LCL-topology. Moreover, the steady-state equivalent circuit-based

mathematical models for all topologies are developed. Finally, general formulas for the

fundamental power-flow (active and reactive) between the EV and DC-bus are developed.

Chapter 5 discusses design and implementation considerations for the entire

BIWPTS’s components. It presents more details about the high frequency (HF) converter

design and optimization, including component selections, driving circuits, snubber circuits,

protection elements and the converter’s printed circuit board (PCB). Additionally, the

hardware implementation of the phase-shift control based on the analog phase-locked loop

(PLL) circuit is investigated and tested. Also, the chapter presents the details of design and

building two symmetrical circular power pads with the compensation network to form the

wireless coupler. The outcomes form the experimental prototype and the simulated model

are presented.

11
 Chapter 6 presents verification analysis for the developed theoretical models in

chapter 4 by means of simulation and using the experimental prototype discussed in chapter

5. Then, these models are utilized to assess the performance of the different topologies

during the bidirectional operation, and present the merits and demerits of each structure.

Also, it presents comparative analysis for the BIWPTS’s performance parameters (power,

power factor and efficiency) among all topologies, under both full and light loading

conditions. Finally, the sensitivity of the different topologies against the misalignment in

the system, based on theoretical models and measurements, is investigated

The next 3 chapters are investigate the nonlinear physics-based model and magnetic

design for the symmetrical LCL BIWPTS. Chapter 7 presents a physics-based co-

simulation platform for LCL BIWPTS in EVs applications. The platform is established

through the coupling between finite element and circuit analysis. In addition, a state-space

dynamic mathematical model for the same system is developed and implemented in the

MATLAB environment. A 1.2 kW LCL BIWPTS is analyzed under different dynamics by

both models, and the results are compared and presented in this chapter. Finally, the effect

of the nonlinearities and the magnetic material characteristics on the system’s performance

is assessed in terms of error and harmonics analysis.

Chapter 8 presents detailed analysis for the choice and design of the passive elements

in a BIWPTS. A clear methodology for achieving the proper design for all the passive

elements in a symmetrical LCL BIWPTS is discussed. A three dimensional finite element

model (3D-FEM) for an 8 kW polarized double-D (DD) power pad is developed and

optimized intuitively. Moreover, the choice of the power factor correction (PFC) capacitors

and the impedance matching coils was investigated. The recommendations from the

12
proposed design methodology are considered to build an entire prototype for a BIWPTS to

meet with WPT2 level and Z2 class in the J2954A standard. The system’s performance is

evaluated based on experimental, simulation and analytical tests.

In chapter 9, an automatic multi-objective optimization for the different design

parameters of the DD power pad structure is performed. A detailed two dimensional finite

element model (2D FEM) for the power pad, considering the separations among the turns,

is developed and linked with an improved Tabu search (ITS) algorithm, for optimization

purposes. The proposed design optimization results are evaluated based on the system’s

coupling performance and cost, in comparison with three other designs that are presented

in the literature.

Chapter 10 presents a new two-layer predictive active power-flow control for the

LCL BIWPTS in EVs applications. The design analysis for the proposed controller is

presented in detail, based on an accurate power-flow analytical model, which is developed

and verified in this chapter. The procedures for executing the two control layers and the

link between them are presented and explained. In addition, a real time mutual inductance

estimation technique that needs to be applied one step before implementing the proposed

controller is discussed. Finally, the proposed controller is implemented and tested using

the FPGA board, and its performance is compared with the classical proportional-integral

(PI) controller.

In chapter 11, the feasibility of implementing the quasi-dynamic wireless power

transfer (QDWPT) system at the traffic signals is explored. A comprehensive modeling

platform for a wirelessly connected EV, including the BIWPTS, EV’s battery, and traction

system, is developed and investigated. The feasibility of implementing a QDWPT system

13
at traffic signals is investigated under three distinct wireless power transfer (WPT)

scenarios: fixed power charging, variable power charging and fixed power charging and

discharging, based on the maximum driving range, duration per cycle, and the additional

distance gained for each consumed kWh. In addition, the effect of coil misalignment in the

WPT system over the driving performance is investigated.

Finally, the conclusions and recommendations for further research are presented in

chapter 12.

14
 Introduction

Considering the DC-bus configuration in Figure 1.1 for integrating the PV

generations with the wirelessly connected EVs, this chapter is investigating the first source

in the structure, which is the grid-connected PV power system. PV utilization is getting

more attention due to the increased progress in power electronics and semiconductor

technologies. A first step to analyze and study a PV power system is to develop an accurate

model that has the ability to predict the system performance. It is also very useful for

control and design purposes. Thus, in this chapter a detailed mathematical nonlinear

dynamic model for grid-connected PV power system is developed. The dynamic model is

used to simulate and predict the entire system performance under different dynamics, such

as variation in environmental and load conditions. It is also helpful to evaluate the proposed

control and maximum power point tracking performance under different irradiance and

temperature conditions. Then, the developed dynamic model is linearized to state the small

signal model for the same system. This small signal model is utilized to state some design

considerations for MPPT algorithm’s parameters. The developed models were verified and

tested by means of simulation and experimental tests.

The PV power system under consideration consists of a PV array connected to a

DC bus through a unidirectional DC-DC converter, as shown in Figure 2.1. In this case,

the DC-DC converter is responsible for boosting the array voltage to the DC bus voltage

level. Also, it is in charge of extracting the maximum power from the PV panel at different

15
environmental and load conditions. The DC bus is coupled to the power grid through a

bidirectional three-phase grid-tie converter. This converter is taking charge of managing

the power flow between the DC bus and the power grid, in addition to regulating the DC

bus voltage to a certain level.

DC
bus

PV array MPPT and Control

Figure 2.1. Block diagram of grid-connected PV power system.

Nonlinear Dynamic Model

In this section, the details of the non-linear dynamic equations for the PV power

system are presented. The formulas are derived based on the electric circuit model of each

component in the system. The individual equivalent circuits are lumped together in one

circuit that represents the entire PV power system, as shown in Figure 2.2. The system

contains a PV array, DC-DC Cuk converter and DC bus. The PV panel is represented by

the single diode (four parameters) model equivalent circuit. The Cuk converter is modeled

by its ideal passive components with neglecting the resistive losses. Since the DC bus

voltage is kept fixed by the grid-tie converter, the DC bus is denoted in the model by a

fixed DC source (Vdc) with a series resistance (Rdc) that emulates its resistive losses. The

details for the dynamic equations of each component are presented in the following

sections.

16
 Rs L1 C1 L2 Rdc
+ + vc1 - +
Isc ipv iL2
Rp vc2 Vdc +
vpv C2 -
Dpv SW
- d
-
PV panel DC-DC Cuk converter DC bus

Figure 2.2. Equivalent circuit of a grid-connected PV power system through a DC-DC

Cuk converter.

2.2.1 PV Panel Dynamic Model

Two advanced equivalent circuit PV models are proposed in the literature: single-

diode and double-diode model. The double diode model features high accuracy since it

considers the carriers’ recombination. The first diode accounts for carriers diffusing across

the P-N junction and recombining in the bulk or at surfaces. The second diode is sometimes

attributed to carrier recombination by traps within the depletion region, or recombination

at an unpassivated cell edge. Even though, this model is rarely used due to its complexity,

high computational cost, and inability to be parameterized based solely on data sheet

information [103]. Single-diode model is the most commonly used in PV systems’ studies,

because it offers reasonable tradeoff between simplicity and accuracy. Also, it has the

possibility to be parameterized based on the provided information on the manufacture

datasheet [104]. A comparative analysis between the performance of the single and double

diode models has been conducted in [105]. The single-diode model shows an average

absolute error = 0.0085 and root-mean square (RMS) error = 1.67%. By using the two-

diode model, it introduced some singular solutions. After solving the issue by adding more

17
complexity to the analysis, it gave an average absolute error = 0.0080 and RMS error =

1.35%. Both the errors are slightly reduced compared with the one-diode model results.

Also, the study showed that, the reverse saturation current of the second diode is indeed

extremely close to zero, whereas other parameters are comparable to their counterparts in

the one-diode model. It was concluded that the one-diode model is good enough to

represent the whole I–V characteristic accurately. Thus, in this work, the PV panel is

modeled based on the single diode equivalent circuit model, as shown in Figure 2.2 and its

I-V characteristic is presented in Equation (2-1).

𝑣𝑝𝑣 +𝑖𝑝𝑣 𝑅𝑠
𝑣𝑝𝑣 +𝑖𝑝𝑣 𝑅𝑠 𝑘𝑇
𝑖𝑝𝑣 = 𝐼𝑠𝑐 − 𝐼𝑟 (𝑒 𝑛𝑉𝑇 − 1) − ( ) ; 𝑉𝑇 = (2-1)
𝑅𝑝 𝑞

where, ipv and vpv are the PV output current and voltage, respectively, Isc is the photon (short

circuit) current, q is the charge of electron (C), k is Boltzmann constant (kg.m2/s2K), T is

cell temperature (K), n is the diode ideality factor, 𝑅𝑠 is the series resistance, Rp is the

parallel resistance, and 𝐼𝑟 is reverse saturation current.

The series resistance (𝑅𝑠 ) stands for the lumped resistive losses in the current path

through the semiconductor material, the metal grid, contacts and current collecting bus.

The parallel resistance (Rp), represents the losses associated with a small leakage of current

through a resistive path in parallel with the intrinsic device. These are due to crystal damage

and impurities in and near the junction. Considering Rp allows the model to provide a logic

behavior under the impact of shading on a string of cells connected in series. The effect of

Rp on the PV array performance is less conspicuous compared to the series resistance, but

it will become noticeable when a number of PV modules are connected in parallel for a

larger system. The recombination in the depletion region of PV cells provides non-ohmic

18
current paths in parallel with the intrinsic PV cell. This is represented by a single diode

with unknown diode ideality factor (n) [106]. This factor (n) ranges from 1 to 2, where 1

ideality factor means ideal diode behavior.

For the purposes of performing small signal analysis, it is better to express the PV

panel output voltage in terms of its current. Thus, by rearranging Equation (2-1), a second

formula for the PV panel I-V characteristic is obtained, as indicated in Equation (2-2).
𝑣𝑝𝑣+𝑖𝑝𝑣 𝑅𝑠
𝐼𝑟 +𝐼𝑠𝑐 −𝑖𝑝𝑣 −( )
𝑅𝑝
𝑣𝑝𝑣 = 𝑛𝑉𝑇 ln [ ] − 𝑖𝑝𝑣 𝑅𝑠 (2-2)
𝐼𝑟

By applying the open circuit conditions on the PV panel (vpv=Voc and ipv=0), an

expression for reverse saturation current in terms of the panel parameters can be obtained,

as in Equation (2-3).

(𝑅𝑝 𝐼𝑠𝑐 +𝑉𝑜𝑐 )

𝐼𝑟 = 𝑉𝑜𝑐 (2-3)
[𝑅𝑝 (𝑒 𝑛𝑉𝑇 −1)]

where, Voc is the open circuit voltage of the PV module.

By inserting Equation (2-3) into Equation (2-2), a third accurate formula for the I-V

characteristic of PV panel is given in Equation (2-4).

𝑉𝑜𝑐
𝑅𝑝 (𝐼𝑠𝑐 −𝑖𝑝𝑣 )+𝑖𝑝𝑣 𝑅𝑠 +𝑣𝑝𝑣
𝑣𝑝𝑣 = 𝑛𝑉𝑇 ln [1 + (𝑒 𝑛𝑉𝑇 − 1) ( )] − 𝑖𝑝𝑣 𝑅𝑠 (2-4)
𝑅𝑝 𝐼𝑠𝑐 −𝑉𝑜𝑐

−𝑉𝑜𝑐
By applying the approximation of (𝑒 𝑛𝑉𝑇 ≪ 1), a simple formula for PV output

voltage is obtained in Equation (2-5).

𝑅𝑝 𝐼𝑠𝑐 −𝑖𝑝𝑣 (𝑅𝑝 −𝑅𝑠 )+𝑣𝑝𝑣

𝑣𝑝𝑣 = 𝑉𝑜𝑐 + 𝑛𝑉𝑇 ln [ ] − 𝑖𝑝𝑣 𝑅𝑠 (2-5)
𝑅𝑝 𝐼𝑠𝑐 −𝑉𝑜𝑐

19
 2.2.2 DC-DC Converter with DC Bus Model

DC-DC Cuk converter is selected in this work to boost up the PV panel voltage to

the DC link level. It is also controlled to extract the maximum power form the PV panel

regardless of the irradiance level and cell temperature. The circuit diagram of Cuk

converter is shown in Figure 2.2. It consists of an input inductor (L1) and capacitor (C1),

power electronic switch (SW) [metal-oxide semiconductor field-effect transistor

(MOSFET) or insulated-gate bipolar transistor (IGBT)], which is controlled by the duty

cycle (d), power semiconductor diode (Dc), and output filter capacitor (C2) and inductor

(L2) to smooth the output voltage and current. Under the assumption that the inductor

current is always positive, when the switch SW is ON, the diode is OFF and the capacitor

C1 is discharged by the inductor L2 current. When the switch is OFF, the diode conducts

the current of the inductors L1 and L2 whereas capacitor C1 is charged by the inductor L1

current. The Cuk converter offers several advantages, such as providing capacitive

isolation, which protects against switch failure, continuous input current, and almost ripple-

free output current, which is important for efficient systems. The passive components (L1,

L2, C1 and C2), are designed based on the switching frequency, and the output voltage and

current ripple.

Following the state-space representation theory, the dynamic performance of the

Cuk converter and the DC bus is represented by four non-linear ordinary differential

equations (ODEs) that describe the inductors’ current (iL1=ipv and iL2) and capacitors’

voltage (vC1 and vC2). The final set of nonlinear ODEs are stated in Equation (2-6).

20
 1
𝑝𝑖𝑝𝑣 = 𝐿 [𝑣𝑝𝑣 − (1 − 𝑑)𝑣𝑐1 ]
1
1
𝑝𝑣𝐶1 = 𝐶 [𝑑 𝑖𝐿2 + (1 − 𝑑)𝑖𝑝𝑣 ]
1
1 (2-6)
𝑝𝑖𝐿2 = 𝐿 (−𝑑 𝑣𝐶1 − 𝑣𝐶2 )
2
1 𝑣 𝑉
𝑝𝑣𝐶2 = 𝐶 (𝑖𝐿2 − 𝑅𝐶2 + 𝑅𝑑𝑐 )
2 𝑑𝑐 𝑑𝑐

𝑑
where, the symbol 𝑝, represents the differential operator (𝑑𝑡).

Equations (2-2) and (2-6) represent the non-linear state-space dynamic model for the

entire PV power system. These equations are implemented and analyzed in MATLAB

environment to simulate the PV system performance. Equation (2-6) is analyzed by means

of the numerical integration technique (Runge-Kutta) and Equation (2-2) is solved by the

iterative technique (Newton-Raphson).

Small Signal Analysis of PV Power System

For proper converter control system design, it is necessary to model the system

dynamic behavior. In particular, it is required to determine how the variations in the

converter duty cycle affect the PV power and voltage. Unfortunately, understanding

converter dynamic behavior is hampered by the nonlinear time-varying nature of the

switching and pulse-width modulation process. These difficulties can be overcome through

the use of waveform averaging and small-signal modeling techniques [107]. The dynamic

model that is presented in section 2.2 is linearized to state the small signal model for the

PV power system.

2.3.1 PV Panel Small Signal Model

The main function of MPPT in PV system is to keep the PV panel working at the

21
maximum power point (MPP) regardless of the irradiance level and temperature. Thus, in

the PV systems that contain MPPT, it can be considered that the operating point is the

MPP. When the PV panel is operating at MPP, the variation of the operating point around

the MPP due to the fast control action (10-20 kHz) can be linearized using the small-signal

analysis. This analysis can be performed by assuming small variations in the different

dynamics that the PV panel is usually subjected to. The disturbance in PV panel output is

mainly due to the variation of temperature, radiation and load conditions. Thus, the small

variation in the PV output voltage (𝑣̂𝑝𝑣 ) will be affected by the small variation in PV current

(𝑖̂𝑝𝑣 ), radiation level (𝑔̂), and cell temperature (𝑡̂) [108]. If these conditions slightly change

around the operating point, it can be written as: 𝐺 = 𝐺𝑄 + 𝑔̂, 𝑇 = 𝑇𝑄 + 𝑡̂ and 𝑖𝑝𝑣 = 𝐼𝑝𝑣 𝑄 +

𝑖̂𝑝𝑣 , where, G , 𝑇 and 𝑖𝑝𝑣 are the dynamic irradiance level, temperature and PV current,

respectively, GQ, 𝑇𝑄 and 𝐼𝑝𝑣 𝑄 are the average of these variables at the quiescent point (Q),

and 𝑔̂, 𝑡̂ and 𝑖̂𝑝𝑣 are their small variations. Thus, the small signal representation of the PV

voltage at any operating point (e.g. MPP) is described in Equation (2-7).

𝑣̂𝑝𝑣 = 𝐾𝑖𝑣 𝑖̂𝑝𝑣 + 𝐾𝑔𝑣 𝑔̂ + 𝐾𝑡𝑣 𝑡̂ (2-7)

where, 𝐾𝑖𝑣 is the variation of the PV voltage with respect to the variation of PV current,

which represents the slope of I-V characteristic at the operating point; 𝐾𝑔𝑣 is the variation

of the PV voltage with respect to the variation of the irradiance level; and 𝐾𝑡𝑣 is the

variation of the PV voltage with respect to the variation of the cell temperature.

These small signal coefficients (Kiv, Kgv and Ktv) can be evaluated using an explicit

PV voltage equation as function of current, radiation and temperature. Small approximation

needs to be applied to Equation (2-5) to put it in an explicit form. This form can be obtained

22
by neglecting the shunt resistance effect in Equation (2-5) by assuming very large

resistance (Rp ≈ ∞). Typically, the PV power systems are operating at MPP due to the

MPPT in the system. Thus, the small signal analysis of PV panel is typically applied at the

MPP. In [109], the effect of Rp on the MPPT control performance is studied for a certain

PV module. It shows that at high values of Rp (≥ 601.34Ω), its effect is negligible. Based

on this study, we conduct a study for the effect of Rp on the I-V characteristics of the used

PV module (BP 4175T PV array) in this work at different climatic conditions, as shown in

Figure 2.3. It can be noticed that for high values of Rp (≥ 100Ω) its impact on the slope at

MPP is negligible, even at different climatic conditions. Thus, its effect on the small signal

analysis will be unnoticeable. The estimated Rp for the used PV module in the work is 2.8

kΩ, as will be demonstrated in section 2-5, which is much larger than 100 Ω [109].

By applying this approximation on Equation (2-5), an explicit PV model that is

appropriate for the small signal analysis is obtained in Equation (2-8).

𝑖𝑝𝑣
𝑣𝑝𝑣 = 𝑉𝑜𝑐 + 𝑛𝑉𝑇 ln [1 − 𝐼 ] − 𝑖𝑝𝑣 𝑅𝑠 (2-8)
𝑠𝑐

where, Voc, Isc and VT are function of the cell temperature and irradiance level as described

in Equations (2-9), (2-10) and (2-11), respectively.

𝐺
𝑉𝑜𝑐 (𝑇, 𝐺) = 𝑉𝑜𝑐 𝑜 [1 + 𝛽(𝑇 − 𝑇𝑜 )] + 𝑉𝑇 𝑜 ln (𝐺 ) (2-9)
𝑜

𝐺
𝐼𝑠𝑐 (𝑇, 𝐺) = 𝐼𝑠𝑐 𝑜 (𝐺 ) [1 + 𝛼(𝑇 − 𝑇𝑜 )] (2-10)
𝑜

𝑇
𝑉𝑇 (𝑇) = 𝑉𝑇 𝑜 (𝑇 ) (2-11)
𝑜

23
 Rp=  Rp= 100  Rp= 5  Rp= 1 
6

5
PV Current (A)
4
G=1 kW/m 2 & T=25 Co

1 G=0.5 kW/m 2 & T=50 Co

0
0 10 20 30 40
PV Voltage (V)

Figure 2.3. Effect of Rp on I-V characteristics of PV module at different climatic

conditions.

By differentiating Equation (2-8) with respect to the PV current, an expression for

the coefficient 𝐾𝑖𝑣 at Q-point, is obtained in Equation (2-12).

𝑉𝑜𝑐 𝑄
−
𝑛𝑉𝑇 𝑄
𝑛𝑉𝑇 𝑄 (1−𝑒 )
𝑑𝑣𝑝𝑣
𝐾𝑖𝑣𝑄 = | =− 𝑉𝑜𝑐 𝑄 (2-12)
𝑑𝑖𝑝𝑣
𝑄 −
𝑛𝑉𝑇 𝑄
[𝐼𝑠𝑐 𝑄 −𝐼𝑝𝑣 (1−𝑒 )]
𝑄

where, the subscript (Q) denotes the variables at the operating point (Q-point).

For the radiation coefficient, 𝐾𝑔𝑣 , Equation (2-8) is differentiated with respect to the

radiation level (G) as given in Equation (2-13).

24
 𝑑𝑣𝑝𝑣 𝑑𝑣𝑂𝐶 𝑑 𝑖𝑝𝑣 𝑑𝑖𝑝𝑣
𝐾𝑔𝑣𝑄 = | = | + 𝑛𝑉𝑇 𝑄 𝑑𝐺 [ln (1 − 𝐼 )]| − 𝑅𝑠 . | (2-13)
𝑑𝐺 𝑄 𝑑𝐺 𝑄 𝑠𝑐 𝑄 𝑑𝐺 𝑄

𝑑𝑖𝑝𝑣 𝑑𝑖𝑠𝑐
Using Equations (2-9), (2-10) and (2-11), with the assumption of, | ≈ | at
𝑑𝐺 𝑄 𝑑𝐺 𝑄

MPP, the final formula for the radiation coefficient at Q-point is given in Equation (2-14).

Similarly, the temperature coefficient at Q-point, 𝐾𝑡𝑣 , is presented in Equation (2-15).

𝑑𝑣𝑝𝑣 𝑉𝑇 𝑜 𝐼𝑠𝑐 𝑛𝑉𝑇 𝑄

𝐾𝑔𝑣𝑄 = | = − ( 𝐺 𝑜 ) [1 + 𝛼(𝑇𝑄 − 𝑇𝑜 )] ( 𝐼 −𝑅𝑠 ) (2-14)
𝑑𝐺 𝑄 𝐺𝑄 𝑜 𝑠𝑐 𝑄

𝑑𝑣𝑝𝑣 𝛼𝐺𝑄 𝐼𝑠𝑐 𝑜 𝑛𝑉𝑇 𝑄 𝑛𝑘 𝐼𝑝𝑣

𝑄
𝐾𝑡𝑣𝑄 = | = 𝛽𝑉𝑜𝑐 𝑜 − ( ) (𝑅𝑠 + )+ ln (1 − 𝐼 ) (2-15)
𝑑𝑇 𝑄 𝐺𝑜 𝐼𝑠𝑐 𝑄 𝑞 𝑠𝑐 𝑄

where, the subscript (o) distinguishes the variables at standard test conditions (STC), which

are available in the PV module manufacture datasheet.

2.3.2 DC-DC Converter and DC Bus Small Signal Model

The DC-DC converter and DC bus are modeled based on the state-space averaging

technique in [107]. Each variable in Equation (2-6) is replaced by the superposition of the

average value and the variations; 𝑓 = 𝐹𝑄 + 𝑓̂, where 𝑓 = {𝑖𝑝𝑣 , 𝑣𝑝𝑣 , 𝑣𝑐1 , 𝑖𝐿2 , 𝑣𝑐2 , 𝐷}, FQ is

the average value of the variable f at the Q-point and 𝑓̂ is the variation of f around the Q-

point. For example, 𝑖𝑝𝑣 = 𝐼𝑝𝑣 𝑄 + 𝑖̂𝑝𝑣 . In this analysis, the DC components and the second-

order small variation terms are neglected. Thus, final small signal representation of the DC-

DC Cuk converter with the DC bus is given in Equation (2-16).

25
 1
𝑝𝑖̂𝑝𝑣 = 𝐿 [𝑣̂𝑝𝑣 − (𝐷𝑄 − 1)𝑣̂𝑐1 + 𝑉𝑐1 𝑄 𝑑̂ ]
1
1
𝑝𝑣̂𝐶1 = 𝐶 [𝐷𝑄 𝑖̂𝐿2 + (1 − 𝐷𝑄 )𝑖̂𝑝𝑣 + (𝐼𝐿2 𝑄 − 𝐼𝑝𝑣 𝑄 ) 𝑑̂ ]
1
1
(2-16)
𝑝𝑖̂𝐿2 = 𝐿 (−𝐷𝑄 𝑣̂𝐶1 − 𝑣̂𝐶2 −𝑉𝑐1 𝑄 𝑑̂)
2
1 𝑣̂
𝑝𝑣̂𝐶2 = 𝐶 (𝑖̂𝐿2 − 𝑅𝐶2 )
2 𝑑𝑐

By mixing the PV array in Equations (2-7), (2-12), (2-14), and (2-15), with the Cuk

converter and the DC bus models in Equation (2-16), the final small signal model for grid-

grid connected PV system is obtained. Based on these equations, the state-space

representation for the system is given Equation (2-17).

ẋ = A x + B u; y=Cx+Du (2-17)

where, x = [𝑖̂𝑝𝑣 𝑣̂𝐶1 𝑖̂𝐿2 𝑣̂𝐶2 ]𝑇 ; u = [𝑑̂ 𝑔̂ 𝑡̂ ]𝑇 ; and y = [𝑖̂𝑝𝑣 𝑣̂𝑝𝑣 ]𝑇 .

𝐾𝑖𝑣 𝐷𝑄 −1 𝑉𝑐1 𝑄 𝐾𝑔𝑣

0 0 𝐾𝑡𝑣
𝐿1 𝐿1 𝐿1 𝐿1 𝐿1
1−𝐷𝑄 𝐷𝑄 𝐼𝐿2 𝑄 −𝐼𝑝𝑣
0 0 𝑄
A=
𝐶1 𝐶1
; B= 0 0
−𝐷𝑄 −1 𝐶1
0 0 −𝑉𝑐1 𝑄
𝐿2 𝐿2 0 0
1 −1 𝐿2
[ 0 0 𝐶2 𝑅𝑑𝑐 𝐶2 ] [ 0 0 0]

1 0 0 0 0 0 0 0 0
C=[ ]; D = [0 𝐾 𝐾𝑡𝑣 ]
𝐾𝑖𝑣 0 0 0 0 0 𝑔𝑣

Small Signal Perturbation Analysis

The maximum power point tracker in PV system is responsible for adjusting the duty

cycle of the DC-DC converter to force the operating point to match with the MPP. Thus,

the main effect of the controller in MPPT is applying a small variation in duty cycle. Thus,

by considering fixed temperature and irradiance level, the effect of the small variation in

26
duty cycle on the PV output (voltage, current and power) at steady-state can be investigated

by following the state-space averaging technique in [107]. By considering the variation in

duty cycle only as an input in Equation (2-17), the state-space model can be rewritten as in

Equation (2-18).

ẋ = 𝑨̀ x + 𝑩̀ 𝑑̂ (2-18)

𝐾𝑖𝑣 𝐷𝑄 −1 𝑉𝑐1 𝑄
0 0
𝐿1 𝐿1 𝐿1
1−𝐷𝑄 𝐷𝑄 𝐼𝐿2 𝑄 −𝐼𝑝𝑣
0 0 𝑄
𝐶1 𝐶1
where, 𝑨̀ = −𝐷𝑄 −1
; 𝑩̀ = 𝐶1
0 0 −𝑉𝑐1 𝑄
𝐿2 𝐿2
1 −1 𝐿2
[ 0 0 𝐶2 𝑅𝑑𝑐 𝐶2 ] [ 0 ]

By applying the averaging technique in the model in Equation (2-18), the average

small variation in state-space variables can be evaluated as in Equation (2-19).

x = -𝑨−𝟏 B 𝑑̂ (2-19)

By solving Equation (2-19), the average small variation of the PV current 〈𝑖̂𝑝𝑣 〉 with

respect to the perturbation in the duty cycle is stated in Equation (2-20).

〈𝑖̂𝑝𝑣 〉 = 𝐼𝑓 〈𝑑̂ 〉 (2-20)

𝐷𝑄 𝑉𝑐1 𝑄 +𝑅𝑑𝑐 (1−𝐷𝑄 )(𝐼𝑝𝑣 −𝐼𝐿2 𝑄 )

𝑄
where, 𝐼𝑓 = − , 𝐷𝑄 , 𝑉𝑐1 𝑄 , 𝐼𝐿2 𝑄 and 𝐼𝑝𝑣 𝑄 are the steady-state
𝐾𝑖𝑣 𝐷𝑄 2 −𝑅𝑑𝑐 (1−𝐷𝑄 )2

values of the duty-cycle, capacitor (𝐶1 ) voltage, inductor (𝐿2 ) current and PV current at the

MPP, respectively.

The average PV voltage perturbation is obtained by combining Equations (2-7) and

(2-20), as in Equation (2-21).

27
 〈𝑣̂𝑝𝑣 〉 = 𝐾𝑖𝑣 𝐼𝑓 〈𝑑̂ 〉 (2-21)

By applying the small-signal concept and neglecting the second order variation

terms, the PV power small variation 〈𝑝̂𝑝𝑣 〉 is deduced in Equation (2-22) [110].

〈𝑝̂𝑝𝑣 〉 = 〈𝑣̂𝑝𝑣 〉𝐼𝑝𝑣 + 𝑉𝑝𝑣 〈𝑖̂𝑝𝑣 〉 (2-22)

𝑄 𝑄

Thus, the perturbation in the PV power with respect to the perturbation in the duty

cycle is obtained by inserting Equations (2-20) and (2-21) into Equation (2-22), as given

in Equation (2-23).

〈𝑝̂𝑝𝑣 〉 = (𝐾𝑖𝑣 𝐼𝑝𝑣 + 𝑉𝑝𝑣 ) 𝐼𝑓 〈𝑑̂ 〉 (2-23)

𝑄 𝑄

Equations (2-20), (2-21) and (2-23) represent the key design formulas that can be

used to choose the proper perturbation size in the classical perturb and observe (P&O)

algorithm. Equation (2-21) is useful when the control is based on the voltage observation,

while Equation (2-20) is used for the current observation. These formulas are utilized in

this work to achieve the proper design for MPPT algorithm based on fuzzy logic control,

as will be discussed in detail in the next chapter.

PV Panel Parameters’ Estimation

In order to solve the equivalent circuit model of the PV panel, the model parameters

need to be known. The single diode model has four parameters that need to be estimated to

match with the practical I-V characteristics [111]. These parameters are the series resistance

(Rs), parallel resistance (Rp), diode ideality factor (n), and reverse saturation current (Ir). In

this work, two different parameter estimation techniques are used and compared to extract

28
the PV module parameters. The first one is based on mathematical equations and the

measurements of the I-V characteristic’s slope at open-circuit voltage and short-circuit

current [112]. The second technique is performed by using GA optimization procedure.

The details of these methodologies and results are discussed in this section.

2.5.1 Analytical Parameter Estimation

This methodology is mainly based on measuring the slope of the I-V characteristic

of the PV module (given in the datasheet) and some analytical formulas. The first parameter

(Ir) is estimated at STC by neglecting the parallel resistance effect in Equation (2-3), as

given in Equation (2-24). Since, Ir, is a temperature-dependent parameter, its value at a

given temperature (T) is updated using Equation (2-25) [113].

𝐼𝑠𝑐 𝑜
𝐼𝑟 𝑜 = (2-24)
(𝑒𝑉𝑜𝑐 𝑜 /𝑛𝑉𝑇 −1)

3
−𝐸𝑔
𝑇 𝑛 (𝑉𝑇 − 𝑉𝑇𝑜 )
𝐼𝑟 | 𝑇 = 𝐼𝑟 | 𝑇𝑜 . (𝑇 ) . 𝑒 𝑛 (2-25)
𝑜

where, Eg is the band gap energy; it is the energy that an electron must acquire to jump

across the forbidden band to the conduction band, and it is equal 1.12 eV for Silicon.

Until this point, the diode ideality factor is unknown and must be estimated. Ideal

diode (n =1) is assumed in this stage until a more accurate value is estimated later by trial

and error. The series resistance of the PV module has a large impact on the slope of the I-

V curves near the open-circuit voltage. Hence, the value of Rs is calculated by evaluating

𝑑𝑉
the slope ( 𝑑𝐼 ) of the I-V curve at Voc, as indicated in Equation (2-26) [114].

29
 𝑑𝑉 𝑛𝑉𝑇
𝑅𝑠 = − 𝑑𝐼 | −( 𝑉 ) (2-26)
𝑉𝑜𝑐 ( 𝑜𝑐 )
𝐼𝑟 𝑒 𝑛𝑉𝑇

The parallel resistance of the PV module has a large impact on the slope of the I-V

curves near the short circuit current. Thus, the value of Rp is calculated by evaluating the

𝑑𝑉
slope ( 𝑑𝐼 ) of the I-V curve published on the datasheet at Isc and substituting in Equation

(2-27) [114]. The last parameter is the diode ideality factor, which is estimated by trail and

error such that its value attains the best match with the I-V curves on the datasheet.

𝑑𝑉
−( | +𝑅𝑠 )
𝑑𝐼 𝐼𝑠𝑐
𝑅𝑝 = 𝑅 .𝐼 (2-27)
𝐼 𝑑𝑉 ( 𝑠 𝑠𝑐 )
1+ 𝑟 .( | +𝑅𝑠 ).𝑒 𝑛𝑉𝑇
𝑛𝑉𝑇 𝑑𝐼 𝐼𝑠𝑐

2.5.2 GA-Based Parameters Estimation

GA is a stochastic search algorithm that emulates biological evolutionary theories to

solve optimization problems. It enables parallel search from a population of points. Based

on the literature, GA shows massive success in estimating the system parameters in many

different applications due to the following reasons [115], [116]:

- It enables parallel search from a population of points. Therefore, it has the ability to

avoid being trapped in a local optimal solution, unlike traditional methods, which

search from a single point.

- It supports probabilistic selection rules, not deterministic ones.

- It behaves well in the case of noisy or stochastic objective function.

- It does not require explicit definitions.

- It is very efficient in case of cheap cost function (which is our case).

30
 Therefore, GA has been used in this study to extract the best combination of the

parameters such that the error between the measured (from the datasheet) and the simulated

I-V characteristics is minimized.

2.5.3 Parameters Estimation Results

The two techniques of parameters estimation are utilized to extract the parameters of

the commercial BP 4175T PV module [117]. The electrical characteristics and

specifications of this module at STC [Go=1000 W/m2 at the air mass (A.M) = 1.5 and To =

25 oC] are given in Table 2.1.

Table 2.1. BP 4175T PV Module Specifications at STC.

Electrical characteristic Value

Open circuit voltage (Voc) 43.6V

Short circuit current (Isc) 5.45A

Voltage at maximum power (Vmp@STC) 35.4V

Current at maximum power (Imp@STC) 4.94A

Maximum power (Pmax@STC) 175W

Temperature coefficient of Isc (0.065 ± 0.015)%/ oC

Temperature coefficient of Voc -(0.5 ± 0.05)%/ oC

Module efficiency 14%

The two techniques are implemented in MATLAB environment and applied for BP

4175T PV module. The final extracted parameters from each technique are presented and

31
compared in Table 2.2. Moreover, the I-V characteristic of PV module is plotted at STC

using the parameters estimated by each technique and compared with the measured I-V

characteristic in Figure 2.4. It can be observed that GA technique shows more accurate

results than the analytical one. GA-based parameters give 4.77% of current error, while the

analytical technique presents 6.45%. The parameters that provide the least error (GA-

based) are considered in this work to model the performance of the PV module.

Table 2.2. Extracted Parameters of BP 4175T PV Module.

Method Rs (Ω) Rp (Ω) n Ir (A) Current Error (%)

Analytical 0.0068 1000.8 1.6 1.45*10-6 6.45

GA 0.00657 2.8*103 1.51 7.0685*10-7 4.77

5
PV Current (A)

2
Exp.
GA
1
Analytical

0
0 10 20 30 40 50
PV Voltage (V)

Figure 2.4. Comparison between I-V characteristic for BP 4175T PV module at STC from
measurement and simulated using GA and analytical based parameters.

32
 Conclusion

In this chapter detailed dynamic model for grid-connected PV power system is

presented. Four ODEs are developed to represent the entire system’s dynamics including a

PV generator, DC-DC Cuk converter and DC bus. Moreover, the dynamic model is

linearized to develop a small-signal model for the same system. The small-signal model is

utilized to investigate the effect of duty-cycle perturbation on the performance of the PV

generator in terms of power, current and voltage. Final design formula for MPPT algorithm

parameters are derived, which will be utilized to design the proposed MPPT algorithm in

the next chapter.

33
 Introduction

When a PV array is directly connected to a load, the operating point will be at the

intersection of their I-V curves. This point may be anywhere on the curve, and it is seldom

to be at MPP of PV array, thus the PV array will not be able to produce its maximum power.

Even if the system was designed to match the operating point with MPP, this operation is

not guaranteed since the MPP changes as the module temperature and the solar irradiance

change. This mismatch in the maximum power causes further over-sizing of the PV array

and the whole system as a result. To mitigate this problem, a MPPT must be added to the

system to follow up the MPP of PV panel regardless of changing climatic conditions and/or

loads.

MPPT consists of hardware (DC-DC converter) and software (MPPT algorithm)

parts. Several MPPT algorithms were presented, experimentally tested and compared in

the literature [118], [119]. The most widely used in PV power applications is the Perturb

and Observe (P&O) algorithm, due to the simplicity, ease of implementation, and not

requiring previous knowledge for the system characteristics [120]. In this technique, the

perturbation action can be applied directly [120], [121] or indirectly [122]. The direct

perturbation algorithm is preferred, as it does not require PI controller, and provides higher

energy utilization efficiency and less noise and oscillations [123]. The main drawbacks of

P&O algorithm include its failure under fast variation in climatic conditions and the steady-

state swinging around the MPP [124]. Some solutions were presented in literature to deal

with these issues. Variable perturbation size or adaptive techniques are proposed to

34
enhance the tracking response and minimize the steady-state mismatch [125]–[127]. Most

of these techniques are based on analytical models and need deep knowledge about the

system. Other nonlinear tracking techniques based on fuzzy logic (FL) are presented in

[128]–[131]. These techniques suffer from complexity due to the lack of the design

strategy, which is mostly done by experience.

This work presents an optimum design for fuzzy-based MPPT controller to solve the

classical P&O drawbacks. Two different design methodology are considered in this work:

analytical method and GA-based methods. In both techniques, the small signal analysis

that was presented in the previous chapter are utilized during the design. The proposed

design was implemented and tested by means of simulation and experimentally. The

system performance was studied at different environmental conditions and compared with

the existed techniques in the literature.

MPPT Control Algorithm

MPPT algorithm is the software part of the power conditioning component, which is

responsible for calculating the converter duty-cycle to control the PV voltage and current.

The main objective of MPPT algorithm is to extract the maximum power from the PV

panel regardless of the fluctuations of environmental and/or load conditions. The entire

system block diagram showing the MPPT control is depicted in Figure 3.1. The controller

measures the output voltage and current of the PV panel, which are converted to digital

signals using analogue to digital converters (ADCs). The digital measurements are passed

to the MPPT algorithm to estimate the converter duty cycle that achieves the maximum

35
power operation. The duty cycle is translated to switching signal using a pulse-width

modulation (PWM) generator. The output of the PWM generator is applied to the driver

circuit. The driver circuit boosts the switching signal to the voltage and power level to drive

the semiconductor switch of the DC-DC converter.

DC-DC
converter
Ipv Vpv Pulses
Driver DC bus
PV
array ADC

PWM generator
D
Power
D conditioning
MPPT algorithm and duty components
cycle adjustment

Figure 3.1. Block diagram of grid-connected PV power system including MPPT control.

3.2.1 Classical Direct P&O MPPT Algorithm

In the direct P&O method, two measurements (𝑣𝑝𝑣 and 𝑖𝑝𝑣 ) are required to calculate

the PV power (𝑝𝑝𝑣 ). The duty cycle is perturbed with a fixed step size (𝐷𝑠𝑡𝑒𝑝 ) and the

variation of power (Δ𝑝𝑝𝑣 ) and voltage or current (Δ𝑣𝑝𝑣 or Δ𝑖𝑝𝑣 ) are observed. If the power

increases, the algorithm continues to perturb the system in the same direction; otherwise

the system is perturbed in the opposite direction until it reaches the MPP. The voltage or

current observation are utilized to define the direction of the power increase. The classical

36
direct P&O MPPT algorithm is indicated in Figure 3.2. At the MPP, the algorithm cannot

stay at that point but keeps swinging around it, which causes steady-state error. Moreover,

under the fast fluctuation of climatic conditions, the algorithm fails to track the MPP, since

it cannot identify the reason of the power variation, either it is due to the normal control

action or the radiation variation. Thus, wrong actions are taken in this situation and the

tracking will be lost. This represents a serious problem, especially in inclement weather

conditions [120].

Begin P&O
Algorithm

Measure:
Vpv(k), Ipv(k)

Ppv(k) = Vpv(k) x Ipv(k)

ΔPpv= Ppv(k) – Ppv(k-1)
ΔVpv= Vpv(k) – Vpv(k-1) or
ΔIpv= Ipv(k) – Ipv(k-1)

NO YES
ΔPpv > 0

YES ΔVpv > 0 or ΔVpv > 0 or YES

ΔIpv > 0 ΔIpv > 0
NO NO
Decrease Increase Decrease Increase
D D D D

Update History
V(k-1) = V(k)
Ppv(k-1) = Ppv(k)

Figure 3.2. Flow-chart of the classical direct P&O MPPT algorithm.

37
 3.2.2 Proposed Fuzzy Direct MPPT Algorithm

An effective solution for mitigating the problems of the classical P&O algorithm is

to apply the nonlinear fuzzy logic for successful tracking [123]. The FL-based MPPT

provides an adaptive variation of the duty-cycle based on the location of the operating point

with respect to the MPP. This helps to minimize the steady-state error and the malfunction

problem during the fast variations of climatic conditions. The main limitation of the fuzzy

logic control (FLC) is the requirement of high-speed microprocessor ability with large

memory size [123]. In addition, the design of FLC system parameters shows tradeoff

between complexity and accuracy [132]. The contributions in this work appear in the

design strategy of the fuzzy inference system (FIS).

The proposed design aims to deal with the abovementioned drawbacks of the

classical MPPT algorithm without losing the simplicity of the control. The proposed FIS

is indicated in Figure 3.3. It consists of two inputs (Δ𝑝𝑝𝑣 and Δ𝑣𝑝𝑣 ) and one output (𝐷𝑠𝑡𝑒𝑝 ).

The universe of discourse of each input is described by three membership functions

[Negative (N), Zero (Z) and Positive (P)]. The proposed FIS design outlines to achieve

both simplicity and efficiency are summarized as follows:

- A minimum number of membership functions for each input variable was considered

(three functions; N, Z, and P).

- Linear shapes of membership functions were used (Triangle and Trapezoidal).

- The linear Z and S shapes were assigned to N (red) and P (blue) membership functions,

respectively, to define the limits of the universe of discourse.

- A trapezoidal function was allocated to Z-membership function (green), which is

responsible for minimizing the swinging around MPP and improving the steady-state

38
 tracking performance. Using the trapezoidal shape increases the region at which Z-

membership function equals unity around the MPP. Thus, the controller will not take

action in this region, which is desirable since the system is working already at the MPP.

Fuzzification

N Z P

ΔVpv
ΔP
ΔV N Z P Defuzzification
ΔV
-aΔv N Dstep 0 -Dstep
-bΔv 0 bΔv aΔv ∑ zi wi
-Dstep 0 Dstep
Dstep =
Z ∑wi Dstep
N Z P
P -Dstep 0 Dstep
ΔPpv Takagi-Sugeno
Fuzzy Rules

ΔP Fuzzy Inference
-aΔp -bΔp 0 bΔp aΔp Engine

Figure 3.3. Block diagram of the proposed fuzzy logic system tracking algorithm.

- The fuzzy rules were set to achieve the tracking in all possible modes of operation as

follows:

1. The blue rules are set to achieve the tracking through the I-V characteristic of the

PV panel during the starting and variation of the load conditions, while the climatic

conditions are fixed.

2. The red rules are designed for tracking during the variation of climatic conditions

(radiation and temperature).

39
 3. The green rules are for keeping the system working at the MPP when there are no

variations.

- Based on the inputs and the rules table, the output of the fuzzy system was calculated

by the Takagi-Sugeno (T-S) formula, which requires the least computation effort among

all other FIS [133].

- Moreover, in order to keep the control efficient enough, the locations and the boundaries

of the input membership functions and T-S function coefficients are assigned using the

developed small-signal model and GA optimization procedure. The details of the

parameters’ generation procedure are presented in the next section.

Design Methodologies of Fuzzy-MPPT Controller

One of the major drawbacks of FLC is the lack of the design strategies. Most of fuzzy

system parameters are determined intuitively based on the designer and/or the operator

experience [134]. In this section, general design considerations for any MPPT controller

are deduced. In our particular case, the Cuk converter is used, however, these

considerations can be extended to any other converter while the dynamic model is

developed and linearized, as was presented in chapter (2). For the proposed fuzzy-MPPT

controller, the most effective parameters on the tracking performance are the locations and

the boundaries of the input membership functions (𝑎∆𝑣 , 𝑏∆𝑣 , 𝑎∆𝑝 and 𝑏∆𝑝 ), and the

coefficient of T-S function (𝐷𝑠𝑡𝑒𝑝 ). Thus, for accurate design parameters, two different

approaches are proposed to choose these parameters. The first one is pure analytical based

on the small-signal model presented in chapter (2). The other one depends on the same the

small-signal model with a combination of a stochastic searching technique based on GA.

40
 3.3.1 Analytical Fuzzy Logic Design Strategy

This design technique is based on formulas in Equations (2-21) and (2-23), in which

the FL design parameters (𝑎∆𝑣 , 𝑏∆𝑣 , 𝑎∆𝑝 , 𝑏∆𝑝 and 𝐷𝑠𝑡𝑒𝑝 ), are estimated as following:

- Assume maximum (∆𝑃𝑚𝑎𝑥 ) and minimum (∆𝑃𝑚𝑖𝑛 ) PV power perturbation such that,

𝑎∆𝑝 = ∆𝑃𝑚𝑎𝑥 and 𝑏∆𝑝 = ∆𝑃𝑚𝑖𝑛 .

- Evaluate the corresponding variation of duty cycle 𝐷𝑠𝑡𝑒𝑝 𝑚𝑎𝑥 and 𝐷𝑠𝑡𝑒𝑝 𝑚𝑖𝑛 , respectively

based on Equation (2-23).

- Assign, 𝐷𝑠𝑡𝑒𝑝 = 𝐷𝑠𝑡𝑒𝑝 𝑚𝑖𝑛 , since the smaller step size provides more accuracy.

- Estimate the maximum (∆𝑉𝑚𝑎𝑥 ) and the minimum (∆𝑉𝑚𝑖𝑛 ) voltage perturbation using

Equation (2-21).

- Assign, 𝑎∆𝑣 = ∆𝑉𝑚𝑎𝑥 and 𝑏∆𝑣 = ∆𝑉𝑚𝑖𝑛 .

Based on the intuitive assumption that, ∆𝑃𝑚𝑎𝑥 = 0.1 W and ∆𝑃𝑚𝑖𝑛 = 0.02 W, the

design parameters are estimated and presented in Table 3.1.

Table 3.1. Results of Design Parameters from Analytical Design Method.

Parameter Value Parameter Value

𝑫𝒔𝒕𝒆𝒑 0.0015 𝒂∆𝑷 0.1

𝒂∆𝑽 0.01 𝒃∆𝑷 0.02

𝒃∆𝑽 0.002

41
 3.3.2 Genetic Algorithm-Based Design Optimization

The second proposed strategy for generating the FL design parameters is based on

combination between analytical and artificial intelligent techniques. GA is used in this

work to optimize the proposed FL system design parameters and reduce time consumption

comparatively to the trial and error method. GA is a random search methodology based on

population genetics, and is used for optimization purposes. Based on the literature, using

GA for tuning FL system parameters exhibited huge success in many different applications

[115], [116]. Such an optimal FLC could provide ideal control performance and achieve

the desired MPPT performance. GA was used to select the optimum combination of the FL

system design parameters (𝑎∆𝑣 , 𝑏∆𝑣 , 𝑎∆𝑝 , 𝑏∆𝑝 and 𝐷𝑠𝑡𝑒𝑝 ) based on the optimization

procedure described in Figure 3.4.

GA generates an initial random population within the given ranges of variables.

Then, it calls and runs the PV power dynamic simulation to calculate the integral absolute

PV power error, as given in Equation (3-1), which represents the fitness function (F). The

optimization constraints are defined in Equation (3-2).

𝑡
𝐹 = 𝑀𝑖𝑛. {∫𝑡 𝑓 |𝑝𝑝𝑣_𝑡ℎ (𝑡) − 𝑝𝑝𝑣_𝑎𝑐𝑡 (𝑡)| 𝑑𝑡} (3-1)
𝑜

𝑎∆𝑣 𝑚𝑖𝑛 < 𝑎∆𝑣 < 𝑎∆𝑣 𝑚𝑎𝑥

𝑏∆𝑣 𝑚𝑖𝑛 < 𝑏∆𝑣 < 𝑏∆𝑣 𝑚𝑎𝑥
𝐶𝑜𝑛𝑠𝑡𝑟𝑎𝑖𝑛𝑡𝑠 = 𝑎∆𝑝 𝑚𝑖𝑛 < 𝑎∆𝑝 < 𝑎∆𝑝 𝑚𝑎𝑥 (3-2)
𝑏∆𝑝 𝑚𝑖𝑛 < 𝑏∆𝑝 < 𝑏∆𝑝 𝑚𝑎𝑥
{𝐷𝑠𝑡𝑒𝑝 < 𝐷𝑠𝑡𝑒𝑝 < 𝐷𝑠𝑡𝑒𝑝 𝑚𝑎𝑥

where, 𝑝𝑝𝑣_𝑡ℎ is the theoretical maximum power (MP) that can be extracted at certain

irradiance, and 𝑝𝑝𝑣_𝑎𝑐𝑡 is the actual PV power extracted by the proposed MPPT.

42
 Start GA

Generate Initial Population

(Dstep, a∆v, b∆v, a∆p and b∆p)

Evaluate Fitness
YES
Pass variables to Simulink
a∆v≥b∆v and NO
a∆p≥b∆p?

- Run Simulation. Apply a large

- Evaluate ppv_act. penalty factor
- Evaluate ppv_th.
- Evaluate Fitness Function (F)

YES
NO Termination
Selection Best Individual
Generate criteria?
New
Population Crossover Translate

Mutation End

Figure 3.4. Flowchart of GA optimization procedure.

The objective of GA optimization is to minimize the cost function by accurately

choosing the design parameters of the FL system. GA setup for the proposed design

optimization procedure is presented in Table 3.2.

The selection of optimization constrains in Equation (3-2) (lower and upper

boundaries of the optimization variables) are typically assigned by experience. However,

these boundaries can be better configured using the developed formulas (2-21) and (2-23).

The step size (𝐷𝑠𝑡𝑒𝑝 ) when ∆𝑝𝑈𝑝𝑝𝑒𝑟 = 0.3 𝑊 is ∆𝑑@0.3𝑊 = 3.87𝑥10−4, and the change of

43
the voltage is ∆𝑣@0.3𝑊 = 0.0195. Thus, the upper boundaries of 𝐷𝑠𝑡𝑒𝑝 , and ∆𝑣 are

calculated as in Equation (3-3).

∆𝑑𝑈𝑝𝑝𝑒𝑟 = 5. ∆𝑑@0.3𝑊 ≈ 0.002

(3-3)
∆𝑣𝑈𝑝𝑝𝑒𝑟 = 5. ∆𝑣@0.3𝑊 ≈ 0.1

These values are used to set the upper boundaries for the GA optimization variables

𝑎∆𝑣 , 𝑎∆𝑝 and 𝐷𝑠𝑡𝑒𝑝 . The lower boundaries of the variables are set to be very close to zero

for considering the maximum range of the design optimization. The optimization upper

boundaries of 𝑏∆𝑣 and 𝑏∆𝑝 are heuristically set, as indicated in Table 3.2.

Table 3.2. GA Setup for the Design Optimization Algorithm.

Parameter Value

Variable [𝐷𝑠𝑡𝑒𝑝 , 𝑏∆𝑣 , 𝑎∆𝑣 , 𝑏∆𝑝 , 𝑎∆𝑝 ]

Bounds Lower [0.00001, 0, 0, 0, 0]

Upper [0.002, 0.01, 0.1, 0.05, 0.3]

Populations 20

Mutation Uniform 80%

Crossover Single point

Iterations 20

Number of running 5

In order to avoid the infeasible optimization solutions, the parameters a∆v and a∆p

should always be greater than or equal to b∆v and b∆p, respectively. A large penalty factor

44
is appended to the cost function if the solution does not satisfy these criteria, as indicated

in Figure 3.4. This penalty factor helps to guide the next iterations to the feasible region

after evaluating the fitness, as indicated in the results presented in Figure 3.5. The fitness

function shows smooth descending progress without any infeasible solution. The final

optimum FL system design parameters using GA optimization are presented in Table 3.3.

7
x 10
2
Best fitness
Mean fitness
1.5
Fitness value

0.5

0
0 2 4 6 8 10 12 14 16 18 20
Generation

Figure 3.5. Progression of the GA Fitness function.

Table 3.3. Results of Design Parameters from GA Design Method.

Parameter Value Parameter Value

𝑫𝒔𝒕𝒆𝒑 9.7353 x 10-4 𝒂∆𝑷 0.08417

𝒂∆𝑽 0.00779 𝒃∆𝑷 0.01271

𝒃∆𝑽 3.1832 x 10-4

45
 Hardware Implementation of Fuzzy-MPPT Controller

The system described in Figure 3.1 was simulated and experimentally tested under

fast variations of climatic conditions. The simulations were performed in Simulink

MATLAB and compared with the mathematical model presented in chapter (2). The

experimental test setup is shown in Figure 3.6. It consists of a commercial BP 4175T PV

array connected to a DC bus through a DC-DC Cuk converter and its control. The PV array

was emulated by XR375-15.9 Programmable DC MAGNA power supply. The proposed

FL MPPT algorithm was implemented using the commercial microcontroller DSpace

1104. The DC-DC converter is built using HGTG30N60C3D IGBT power switch and

FFPF15S60S power diode. The commercial SKHI 22A H4 R gate driver was considered

to drive the IGBT. Hall effect LEM voltage and current sensors are utilized to measure the

PV voltage and current. The nominal values for the chosen system’s components are given

in Table 3.4. The converter switching frequency is 10 kHz and the perturbation rate is 50

variations/sec.

PV Panel DC Bus

XR375-15.9 Programmable DC supply in parallel

DC MAGNA Power Supply DC-DC Cuk Converter PCB
with resistive load

LEM Sensors Gate

SKHI 22A R Driver
Sensors Driver
Board
DSPACE 1104
MPPT
Figure 3.6. Experimental setup of PV power system.

46
 Table 3.4. Specifications of Hardware Components of the DC-DC Converter.

Component Model/Rating

Power IGBT HGTG30N60C3D, 600 V, 63 A

Power Diode FFPF15S60S, 600 V, 15 A

Gate Driver Hybrid Dual SKHI 22 A / B H4 (R)

Current Sensor LA25-NP 713194

Voltage Sensor LV25-P 712100

Experimental and Simulation Results and Discussion

The PV power system with the FL MPPT is evaluated in this section. Several tests

were applied to the system for verification purposes. The first category of tests are made

to validate the proposed design analysis. The second class of measurements are achieved

to verify the robustness of the proposed FL MPPT under different environmental

conditions. Finally, comparative analysis is realized between the performance of the

proposed MPPT and the most common techniques in the literature.

3.5.1 Assessment of the Proposed FL Design Strategies

The effect of the two design techniques presented in section 3.3 on the MPPT

performance is indicated in this section. The FL system is analyzed and compared using

the parameters resulting from the analytical and GA-based procedure. The comparison

results are described in Figure 3.7. It shows the PV variables (𝑝𝑝𝑣 , 𝑖𝑝𝑣 and 𝑣𝑝𝑣 ) under

47
certain irradiance profile during 4.5 sec in which two step changes were applied: from 1 to

0.6 kW/m2 , and then turns back to 1 kW/m2. Figure 3.7(a) presents the PV power using

the two combinations of design parameters (analytical and GA-based) and compared with

the theoretical MP, which is supposed to be generated from the PV panel. Both design

techniques exhibit successful tracking performance but the analytical one shows slower

transient response, as indicated in Figure 3.7(c). By evaluating the Mean Absolute Power

Error (MAPE) for both combinations, GA-based provides better accuracy with 2.36% than

the analytical one, which shows 3.15%. Thus, the parameters from GA optimization are

considered for the rest of the results.

GA-based Analytical Theoritical

200
Ppv (W)

100 (a)
0
0 1 2 3 4
6
Ipv (A)

4
2 (b)
0
0 1 2 3 4
50
(V)

(c)
pv
V

0
0 1 2 3 4
Time (sec)

Figure 3.7. Simulation results for the proposed FL system designs. (a) PV power. (b) PV
current. (c) PV voltage.

48
 3.5.2 Experimental and Simulation Results of Fuzzy MPPT

In this section, a comparison between the theoretical, experimental and simulation

results are presented to evaluate the dynamic and steady-state performance of the proposed

FL MPPT controller. The entire system performance is analyzed by simulation and

compared with the experimental results. A pulse irradiance profile is applied for 10 sec,

which shows a step change from 1 to 0.6 kW/m2 and turns back to 1 kW/m2.

The PV variables are indicated in Figure 3.8 (𝑝𝑝𝑣 , 𝑣𝑝𝑣 and 𝑖𝑝𝑣 ). The figure shows

how the control successfully tracked the MPP regardless of the irradiance variation.

Generally speaking, the figures show very good correlation between the experimental,

simulation and theoretical results. Larger ripples in the simulation results than the

experimental can be noticed. This is due to using digital low-pass filters after the voltage

and current transducers to minimize the noise and disturbances due to the PV emulator.

These filters help the MPPT control algorithm to work properly, however, they smooth the

normal ripple due to the swinging around the MPP. That is why these oscillations appear

in the experimental results during the whole operating period, as shown in Figure 3.8,

unlike the simulation results, which show oscillations during the transient region only.

Also, another reason for the experimental oscillations is the low acquisition resolution of

the experimental tests (5 kHz) compared with the simulation (0.5 MHz). The slow dynamic

performance of the PV emulator can be observed during the irradiance step down and up

transient response of the experimental results in Figure 3.8.

49
 Simulated Experimental Theoritical

Ppv (W)
200
(a)
100
0 1 2 3 4 5 6 7 8 9 10
Ppv (W)

(b)
100 100
2 3 7 8
6
Ipv (A)

4
(c)
2
0 1 2 3 4 5 6 7 8 9 10
40
(V)

(d)
pv
V

20
0 1 2 3 4 5 6 7 8 9 10
Time (sec)

Figure 3.8. Experimental and simulation results for FL-MPPT control. (a) PV power. (b)
PV power zoomed. (c) PV current. (d) PV voltage.

The effectiveness of the proposed FL controller under the fast temperature variation

is investigated in Figure 3.9. Typically, the variation of the temperature is slow compared

to the radiation variation. However, a dramatic temperature variation is assumed in this test

to confirm the robustness of the proposed controller. The temperature stepped from 25 oC

to 50 oC, then returned back to 25 oC. As it can be noticed, increasing the temperature

slightly increases the PV current and significantly decreases the PV voltage, which results

in a small reduction of the PV power. The effect of the temperature on the PV power is

small compared with the radiation impact. The proposed MPPT shows robust tracking

50
performance under the fast temperature variation with good agreement between

experimental, simulated and theoretical results.

Simulated Experimental Theoritical

200
Ppv (W)

150
(a)
100
2 4 6 8 10
6
Ipv (A)

4 (b)
2
2 4 6 8 10
(V)

40
(c)
pv

20
V

2 4 6 8 10
Time (sec)

Figure 3.9. Experimental and simulation results for FL-MPPT controller under fast
temperature variation. (a) PV power. (b) PV current. (c) PV voltage.

3.5.3 Comparison between the Proposed Fuzzy-MPPT, P&O and INC Algorithm

In order to verify the effectiveness of the proposed FL-MPPT technique, a

comparative study has been conducted between the proposed algorithm and the two most

commonly used hill-climbing MPPT algorithms: P&O and Incremental Conductance

(INC) [135]. Figure 3.10 presents the tracking performance of the PV power system under

51
the proposed FL-MPPT and INC using the same irradiance profile used in Figure 3.8. The

proposed technique shows faster transient response and less steady-state error. Also, loss-

tracking is noticed in INC performance during the low irradiance level, which results in

less tracking efficiency.

Proposed fuzzy INC Theoritical

200
Ppv (W)

100 (a)

0
0 1 2 3 4
5
Ipv (A)

(b)

0
0 1 2 3 4
50
V pv (V)

(c)

0
0 1 2 3 4
Time (sec)

Figure 3.10. Comparison between the proposed fuzzy-MPPT and INC algorithm. (a) PV
power. (b) PV current. (c) PV voltage.

Figure 3.11 describes a comparison between the proposed FL-MPPT and the

conventional P&O algorithm. P&O algorithm exhibits steady-state error due to swinging

52
around the MPP during the whole operating period, unlike the proposed algorithm, in

which the error appears only during the transient. Also, longer transient can be noticed with

P&O during the step variation of irradiance due to the wrong control actions. The MAPE

has been evaluated for the three algorithms, as presented in Table 3.5. The proposed

algorithm shows the least power error among all techniques. The conventioal P&O

algorithm is implemented experimentally and the results are presented in Figure 3.12. Good

agreement is noticed between simulation and experimental results.

Proposed fuzzy P&O Theoritical

200
Ppv (W)

100 (a)

0
0 1 2 3 4
5
Ipv (A)

(b)

0
0 1 2 3 4
50
(V)
pv

(c)
V

0
0 1 2 3 4
Time (sec)

Figure 3.11. Comparison between the proposed fuzzy-MPPT and P&O algorithm. (a) PV
power. (b) PV current. (c) PV voltage.

53
 Table 3.5. MAPE for the Proposed FL-MPPT, P&O and INC Algorithms.

Variable P&O INC Fuzzy-MPPT

MAPE % 7.1951 13.5513 2.3605

Ppv (W)

Simulated Experimental Theoritical

200 (a)
150
100
0 2 4 6 8 10
Ppv (W)

100 (b)

6 4
Ipv (A)

4
(c)
2
0 2 4 6 8 10
Vpv (V)

40
30 (d)

0 2 4 6 8 10
Time (sec)
Figure 3.12. Experimental and simulation results for P&O algorithm. (a) PV power. (b) PV
power zoomed. (c) PV current. (d) PV voltage.

Conclusion

In this chapter, a comprehensive study for grid-connected PV power system is

presented. The system dynamics are accurately modeled by integrating 4th order ODE (for

54
DC-DC Cuk converter and DC bus) with a non-linear equation representing the PV panel.

General MPPT control design considerations were developed based on the small signal

analysis. Then, proper FL-MPPT control design was performed by means of combining

GA and the analytical design formulas. This technique is used for determining the location

and the boundaries of the FL membership functions. The proposed intelligent MPPT

control algorithm was simulated and experimentally tested under fast variation of

irradiance and temperature conditions, and compared with the P&O and INC algorithms.

The proposed control shows faster transient response, less steady state error and robust

tracking performance. In all the tests, a comparative analysis was conducted between the

simulation, experimental and theoretical results. The results show very good agreement

between all of them.

55
 Introduction

The Smart grids are converting the traditional power system into more efficient and

reliable networks. Using EVs as active elements to serve the power grid is a promising

service in smart grid applications, which is knowing as V2G service [136]. This kind of

application requires a flexible, automatic, safe and reliable interface between the EVs and

the power grid. The inductive wireless power technology is an ideal choice for these

services as [92], [137]: 1) the vehicle does not need to be parked in a certain location for a

long time, 2) avoiding the manual plug provides safe operation in the presence of water,

rain or dust, 3) it is reliable during the environmental disaster such as hurricanes, storms

and earthquakes, and 4) it is automatic and does not need the driver intervention. This

wireless interface needs to be bidirectional to allow two-way power flow between the EV

and power grid.

According to the literature, the BIWPTS has been proposed in [51], [138]. These

systems have been produced for aircraft application, in which the leakage inductance of a

transformer forms a resonant circuit with a series capacitor to allow the bidirectional power

transfer while operating as a voltage source. Such a system would not be appropriate for

V2G applications, where a fleet of EVs needs to be supplied at the same time [139]. A

dual-side full-bridge BIWPTS for charging and discharging EVs was proposed in [137],

[140]. Several analytical models and assessment analyses for IPT systems can be found in

the literature. Most of these analyses were developed for unidirectional IPT systems, such

56
as in [82], [86], [89], [141]. These studies cannot be simply applied for bidirectional

operation, since in this case the system is designed such that each side should be able to

work as a source and a load at the same time. Moreover, in most of these models, only the

fundamental frequency component was considered, which leads to inaccurate analysis

especially in the topologies where both the voltage and the current waveforms are non-

sinusoidal. For BIWPTS, steady-state mathematical models were presented in [75], [101],

[142], a dynamic analytical model was developed in [143], and a physics-based model was

proposed in [144], [145]. In these models, the harmonics components were considered,

however, they were developed for LCL topology only and no assessment analysis was

presented. Simple mathematical models for LC-series and LCL topologies were presented

in [146]. In this study, only the fundamental frequency component was considered, and

brief sensitivity analysis was performed. Up to the author knowledge, a comprehensive

modeling and evaluation analysis for the different compensation topologies in BIWPTS

has not been presented yet.

This chapter presents modeling analysis for the steady-state performance of BIWPTS

in EVs implementations based on Fourier series analysis considering the harmonic contents

in the system’s performance. The analysis is presented for the three main compensation

configurations of a symmetrical BIWPTS: LC-series, LC-parallel and LCL-topology.

Moreover, the steady-state equivalent circuit-based mathematical models for all topologies

are developed. These models were used to precisely determine the system’s response

during V2G and G2V operations based on Fourier series. Generalized fundamental power

flow formulas (active and reactive), to evaluate the power flow performance in the different

topologies, are developed.

57
 V2G Scheme Description

The description of G2V and V2G concepts including the wireless power interface are

depicted in Figure 1.1. The figure demonstrates a DC-bus integration arrangement, in

which all the elements are connected together through a common DC-link. Typically, the

integration between EVs and the power grid can be achieved through a DC or an AC-bus.

The DC-bus configuration offers more benefits than the AC-bus structure, whether the EV

was connected by wires or wirelessly, or both of them (which is the typical case in

commercial EVs that support wireless chargers nowadays) [147]. In this sense, two block

diagrams are prepared to indicate the wired and wireless connection of an EV to a DC and

an AC-bus. In Figure 4.1, which shows the DC-bus connection, the EV carries a wireless

power pad, high-frequency inverter (for wireless) and DC-DC converter (for wired). Also,

during the installation of the primary side of the wireless system (which is still the

responsibility of the automotive company), the company does not need to worry about the

grid-tie inverter, since it will work directly with the available DC-bus.

Inside EV
HF HF EV’s
DC Bus inverter inverter Battery
Grid-tie
inverter

Wireless connection Inside EV

Power EV’s Battery

grid

Wired connection

Figure 4.1. Wired and wireless connected EV through a DC-bus.

58
 On the AC-bus connection, which is shown in Figure 4.2, the EV carries the same

devices as those that are in the DC-bus connection, in addition to a grid-tie inverter for the

wired connection, which means more EV’s weight and cost, and less utilization efficiency.

Moreover, during the installation for the primary side of the wireless system, the company

will be responsible for installing a second grid-tie inverter for the wireless option. This

means more cost and effort for the automotive companies, which will be reflected in the

price of EV.

Inside EV
Grid-tie HF HF EV’s
AC Bus inverter inverter inverter Battery

Wireless connection Inside EV

Grid-tie inverter EV’s Battery
Power
grid

Wired connection

Figure 4.2. Wired and wireless connected EV though an AC-bus.

Based on this discussion, we can conclude that the DC-bus architecture provides

more benefits for EVs manufactures and users, in both the wired and wireless connection.

It needs less components inside the EV in the wired connection and on the road for the

wireless integration. This results in less EV weight and cost, more utilization efficiency,

59
and EV manufacturing companies can get rid of installing the grid-tie converter. Also, it is

more convenient for integrating RESs and energy storage systems [148].

4.2.1 Dual H-Bridge BIWPT System

Typically, the dual H-bridge BIWPTS consists of two sides: primary (grid or ground)

and secondary (vehicle). The former is connected to the DC-bus and is implemented on the

road underneath the vehicle. The latter is coupled to the EV battery and is placed inside the

vehicle. Each side consists of a HF H-bridge inverter, controller, compensation circuit and

the wireless pad, as shown in Figure 4.3. The two sides are weakly coupled by magnetic

induction through a large air gap (100-200 mm), according to the Society of Automotive

Engineers (SAE) J2954A standard [149]. During G2V (charging) operation, the power

flows from the DC-bus to charge the EVs. The DC power is converted to HF AC (81.38-

90 kHz) by the primary inverter of the wireless entity to supply the primary circuit. The

primary power is transferred by magnetic induction to the secondary circuit through the

air-gap with the same frequency. Then, the secondary power is converted to DC by the

secondary converter to supply the EV battery. The compensation capacitors are essential

to resonate with the wireless coupler coils and provide reactive power compensation and

unity power factor (UPF) operation in the primary and secondary sides. These capacitors

result in minimizing the required VA from the supply and maximizing the utilization

efficiency [150]. In the discharging mode (V2G), the power is transferred from the EVs

and the DC-bus through the same path.

60
 Primary inverter Secondary inverter

13 21 23
11
Wireless coupler
M
Vpa + + Vsa
Vpi Lpc Lsc Vsi
Vdc Vb
Vpb - - Vsb

Primary Secondary
14 comp. comp. 22 24
12

S11-S14 S21-S24
primary secondary
controller controller

Figure 4.3. Schematic circuit diagram of the dual-side full-bridge BIWPTS.

4.2.2 Bidirectional Power-Flow Control

The bidirectional power-flow operation between the EVs and DC-bus is managed by

the two HF H-bridge inverters in the system. These inverters are driven by two

synchronized controllers, as indicated in Figure 4.3. In this work, both the primary and

secondary controllers are utilizing the phase-shift technique. In this technique, the system’s

currents and power are managed by adjusting both the magnitude and phase of the two

inverters’ voltage. This voltage control is achieved by changing the phase-shift between

legs’ voltage for the same inverter and the two inverters’ voltage. The primary controller

generates the phase-shift between the two primary inverter’s legs (α). This parameter (α)

is used to adjust the magnitude of the primary inverter output voltage based on the reference

signal, which is typically the rated current that the primary circuit can support [see Figure

61
4.4(a)] [137]. The secondary controller produces the phase-shift between the secondary

inverter’s legs (β), and the phase shift between the primary and secondary inverters’ output

voltage (δ). Typically, the secondary control parameters (β and δ) are adjusted to achieve

the required power flow magnitude and direction, respectively [Figure 4.4(b)] [139]. The

phase-shift parameters, from both the controllers (α, β and δ), are then translated to low-

level switching signals based on the pulse-phase modulation (PPM) technique, as indicated

in Figure 4.4. These signals are boosted up using the driver circuits to drive the eight

switches of the two HF inverters. By applying these switching signals, the inverter voltage

magnitude and phase are adjusted to achieve the required power-flow in the system. The

variation of the inverter voltages in terms of the phase-shift control parameters is described

in Figure 4.5.

sin >0 s11

ωt
Ipc_ref NOT s12
α
α
Ipc - Primary α + >0 s13
+ - sin
controller
(a) δ NOT s14
δ

+- sin >0 s21

ωt
β
Pout_ref NOT s22
β
Pout - Secondary δ -+
+ sin >0 s23
controller -
β NOT s24
(b)

Figure 4.4. Phase-shift control of the HF inverters. (a) Primary controller. (b) Secondary
controller.

62
 Ts
Vpa
Vdc Time
α
Vpb Vdc

Vpi Vdc
Vdc
δ
Vsa
Vb
β
Vsb Vb

Vsi Vb
Vb

Figure 4.5. Legs and total inverter output voltages in terms of the phase-shift control
parameters.

4.2.1 Compensation Network

The large air-gap between the two power pads in the IPT system decreases the system

magnetizing reactance and increases the leakage reactance. Therefore, large magnetizing

current will be drawn from the supply in terms of reactive power. Compensation PFC

capacitors are typically used in BIWPTS to provide reactive power restitution and improve

the operating power factor in the primary and secondary circuits. Improving the power

factor (PF) leads to higher utilization efficiency and power transfer capability [151]. These

63
capacitors can be connected either in series or in parallel based on the applications. In

BIWPTS, each side may act as a supply or a load at the same time, thus the two sides

should be treated equally. Therefore, in this study it is assumed that, the BIWPTS under

consideration is symmetrical. Therefore, there is no a specific side that needs to be designed

to work as a source or a load. Both sides should be able to supply and collect the power at

the same time.

Based on this assumption, there are four compensation topologies for symmetrical

BIWPTS in the literature: LC-series, LC-parallel, LCL-topology and CLCL-topology, as

indicated in Figure 4.6. In LC-series structure, the capacitors (Cp and Cs) are connected in

series with the pad coils (Lpc and Lsc) [Figure 4.6(a)], and in parallel for LC-parallel

structure [Figure 4.6(b)]. In LCL-topology, L-filters (Lpi and Lsi) are added between the

inverters and the resonance tank in LC-parallel topology, as shown in Figure 4.6(c).

M Lpi M Lsi

Cp Lpc Cs Cp Lpc
Lsc Lsc Cs

(a) (c)
M Lpi M Lsi
Cpi Csi
Cp Lpc Lsc Cs Cp Lpc Lsc Cs

(b) (d)

Figure 4.6. Different compensation topologies in BIWPTS. (a) LC-series. (b) LC-parallel.
(c) LCL-topology.

64
 In CLCL structure, shown in Figure 4.6(d), DC blocking capacitors (Cpi and Csi) are

inserted in series with the filter inductances, in order to remove the DC current component.

The best design for these capacitors is such that the equivalent reactance of the capacitors

and the filters’ inductance in each side is equal to the filter reactance in LCL-topology. In

this case, the mathematical model and the performance of CLCL-topology is equivalent to

that of LCL architecture, as was discussed in [142]. Hence, in this work, the authors

decided to focus on the three main topologies: LC-series, LC-parallel and LCL-topology.

Steady-State Circuit representation of BIWPTS’s Components

BIWPTS in EV applications consists of three main elements: EV’s battery, HF

inverters and the wireless coupler, which includes the power pads and the compensation

capacitors. As indicated in Figure 1.1, a grid-tie converter is used in the framework to keep

the DC-bus voltage level fixed and to manage the power-flow between the EV and the grid.

Hence, the DC-link is represented mathematically as a fixed DC voltage source (Vdc). Also,

due to the huge difference between the dynamics of the charging circuit and Li-ion battery,

which is commonly used in EVs, the battery side is ideally represented as a second fixed

DC voltage source (Vb). The steady-state mathematical model for each component in the

system based on its equivalent circuit (EC), including HF inverters, power pads and

compensation network, are presented in this section. In addition, the mathematical link

between these components is stated to predict the entire system’s performance.

65
 4.3.1 HF H-bridge Inverter Model

Two HF inverters are used in the system, to supply the wireless coupler with

controlled square waves at the resonant frequency, as depicted in Figure 4.3. Both inverters

generate periodic square or quasi-square wave voltages (Vpi and Vsi). The steady-state

inverter output voltages are represented, in terms of the system design and control

parameters (α, 𝛽, and δ), based on Fourier series analysis, as given in Equation (4-1) [75].

Only odd harmonics appear in the equations due to the symmetry around the origin.

4 1 𝑛𝛼 𝑛𝛼
𝑉𝑝𝑖 = 𝜋 𝑉𝑑𝑐 ∑∞
𝑛=1,3,… 𝑛 cos (𝑛𝜔𝑠 𝑡 − ) sin ( 2 )
2
4 1 𝑛𝛼 𝑛𝛽
(4-1)
𝑉𝑠𝑖 = 𝜋 𝑉𝑏 ∑∞
𝑛=1,3,… 𝑛 cos (𝑛𝜔𝑠 𝑡 − + 𝑛𝛿) sin ( 2 )
2

where, n is the number of harmonics, α is the PPM parameter of the primary inverter, 𝛽

and δ are the PPM parameters of the secondary inverter, and ωs is the switching frequency

for both primary and secondary circuits, which is equal to the resonant frequency ωr.

The resonant frequency is chosen based on the self-inductance of the coupler coils to

be independent on the mutual inductance and the misalignment between the two sides

during the operation, as given in Equation (4-2).

𝜔𝑠 = 𝜔𝑟 = 1⁄√𝐿𝑝𝑐 𝐶𝑝 = 1⁄√𝐿𝑠𝑐 𝐶𝑠 (4-2)

where, the subscript ‘p’ stands for the primary side and ‘s’ for the secondary circuit, Lpc

and Lsc are the pad self-inductances, and Cp and Cs are the compensation capacitors.

4.3.2 Magnetic Coupler Model

Typically, for high power applications (e.g. EVs), the power pads consists of copper

coils and magnetic ferrite martials. The magnetic coupling between the primary and

66
secondary power pads is demonstrated in this work by the T-model representation of the

mutual coupling. In this representation, the effect of the magnetic material losses and

nonlinearities are neglected, and only the resistive losses of the coils are considered. These

are accepted assumptions for inductive power transfer systems, due to the following

reasons [147]:

- Typically, the magnetic material used in the design of the IPT system is a composition

of different soft ferrites. The most popular combinations are manganese and zinc

(MnZn) magnetic material. This compound exhibits good magnetic properties and has

a rather high intrinsic electric resistivity. These materials can be used up to very high

frequencies without laminating (up to 10 MHz) [152]. As an example that is considered

in this work, the industrial type ferrite N87, which has very soft behavior and near-

insulating character, makes it ideal for uses in 10 kHz-10 MHz range of frequencies in

the design of high-power applications. This kind of material shows low hysteresis and

eddy current losses [152].

- The operating frequency of IPT systems for EV applications is less than 100 kHz (81.38-

90 kHz) [149], which is relatively low compared with the maximum operating

frequency that this material can handle (10 MHz). Thus, this relatively low-frequency

range leads to insignificant magnetic losses.

- The magnetic losses mainly depend on the volume of the magnetic core. The core that

is typically used in IPT systems is small compared with the entire volume of the system.

In contrast with the conventional transformer, the core is huge and it introduces large

magnetic losses, which cannot be neglected. This small core results in negligible

magnetic losses compared with the other losses in the system.

67
 Lpc-M Rpc Lsc-M Rsc Cs
+ Ipc Isc +
Vpc M Vsc
- -

Figure 4.7. T-model representation of wireless coupler.

The mutual inductance is described in terms of the circuit parameters and the

coupling factor (k), as given in Equation (4-3).

𝑀 = 𝑘√𝐿𝑝𝑐 𝐿𝑠𝑐 (4-3)

Although the analysis are achieved for symmetrical BIWPTS, in which the two sides

are identical, the two sides are represented in the mathematical model by different

parameters to present the general unsymmetrical case. The T-model equivalent circuit, as

a recognized model, is used to represent the mutual inductance in the system and represent

the system with one equivalent circuit instead of two isolated circuits with mutual

inductance. In the T-model representation, although the direct correlation with turns-ratio

is missing, its effect on the system is included in the model parameters (Lpc, Lpc and M)

[153]. These parameters are considered as constant in the model and were measured using

the RLC meter. Thus, general turns ratio (other than one) was internally considered in the

model by using different symbols for each side. A different number of turns will be

reflected on the values of self and mutual inductances of the wireless coupler. The T-model

68
representation is a general structure that can be used to represent unsymmetrical coils with

mutual coupling. It has been used in the literature for unsymmetrical IPT system, such as

in [154].

Steady-State Mathematical Model of Different Topologies of BIWPTS

In the steady-state equivalent circuit, the two HF inverters are represented as two

controlled AC voltage sources with output voltages given in Equation (4-1). The DC-bus

and the EV’s battery are represented by two fixed DC sources (Vdc and Vb, respectively).

These sources are linked with the coupler model equivalent circuit and the compensation

elements to develop the whole steady-state equivalent circuit model. This circuit is deduced

for the three main compensation topologies: LC-series, LC-parallel and LCL topology.

Using electric circuits’ analysis, the equivalent circuit of each topology is analyzed and

two π-models are developed to investigate the system’s currents in terms of the inverter

voltages. The first model is shown in Figure 4.7, which represents inverters’ currents (Ipi

and Isi) in terms of their voltages (Vpi and Vsi). The admittance matrix of this model is stated

mathematically in Equation (4-4).

Ipi Zb Isi

Vsi
Vpi
Zp Zs

Zp_eq Zs_eq

Figure 4.8. π-model of BIWPTS for inverter’s currents in terms of inverter’s voltages.

69
 (𝑍𝑝 +𝑍𝑏 )
𝐼𝑝𝑖 −1⁄𝑍𝑏 𝑉𝑝𝑖
𝑍𝑝 𝑍𝑏
[ ]=[ ][ ] (4-4)
𝐼𝑠𝑖 (𝑍𝑠 +𝑍𝑏 ) 𝑉𝑠𝑖
−1⁄𝑍𝑏 𝑍𝑠 𝑍𝑏

where, Zp, Zb and Zs are the model parameters which are different for each configuration,

as is explained in detail in the next sections.

The second π-model demonstrates the pad currents (Ipc and Isc) as functions of Vpi

and Vsi, as indicated in Figure 4.8 and represented mathematically in Equation (4-5).

Ipc λYps Isc

Vpi Vsi

Figure 4.9. π-model of BIWPTS for coil’s currents in terms of inverter’s voltages.

𝐼𝑝𝑐 𝑌𝑝𝑝 𝑌𝑝𝑠 𝑉𝑝𝑖

[ ] = 𝜆[ ][ ] (4-5)
𝐼𝑠𝑐 𝑌𝑠𝑝 𝑌𝑠𝑠 𝑉𝑠𝑖

where, λ, Ypp, Yss, Yps and Ysp are the model parameters which are different in each structure.

This model is based on the harmonic contents, and all the presented parameters and

variables need to be solved at each harmonic component and added together. By knowing

the admittance matrix models, the system steady-state response can be evaluated and many

70
variables can be calculated, such as the active and reactive power, the charging and

discharging efficiency, and primary and secondary PF. Also, the equivalent impedance can

be easily obtained from the π-model in Figure 4.7, by applying Thevenin’s theory, as given

in Equation (4-6). This impedance model is very useful for tuning and evaluating the

compensation capacitors [102].

𝑍𝑝_𝑒𝑞 = 𝑍𝑝 𝑍𝑏 ⁄(𝑍𝑝 + 𝑍𝑏 )
(4-6)
𝑍𝑠_𝑒𝑞 = 𝑍𝑠 𝑍𝑏 ⁄(𝑍𝑠 + 𝑍𝑏 )

4.4.1 LC-Series Compensation Network

In this arrangement, the capacitor is connected in series with the pad on both the

primary and secondary sides. The steady-state equivalent circuit is shown in Figure 4.9.

The inverter and pad currents are the same and are represented by the first model given in

Equation (4-4). The π-model parameters for this configuration are described in Equation

(4-7).

Cp Lpc-M Rpc Lsc-M Rsc Cs

+ Ipi = Ipc Isc = Isi +

Vpi Vpc M Vsc Vsi
- -

Figure 4.10. BIWPTS equivalent circuits using T-model mutual coupling for LC-series
topology.

71
 𝑍𝑏 = (𝜔𝑠 4 𝑀2 𝐶𝑝 𝐶𝑠 − 𝑍𝑝1 𝑍𝑠1 )⁄(𝑗𝜔𝑠 3 𝑀𝐶𝑝 𝐶𝑠 )
𝑍𝑝 = (𝑍𝑝1 𝑍𝑠1 − 𝜔𝑠 4 𝑀2 𝐶𝑝 𝐶𝑠 )⁄[𝑗𝜔𝑟 𝐶𝑝 (𝜔𝑠 2 𝑀𝐶𝑠 + 𝑍𝑠1 )] (4-7)
𝑍𝑠 = (𝑍𝑝1 𝑍𝑠1 − 𝜔𝑠 4 𝑀2 𝐶𝑝 𝐶𝑠 )⁄[𝑗𝜔𝑟 𝐶𝑠 (𝜔𝑠 2 𝑀𝐶𝑝 + 𝑍𝑝1 )]

where, 𝑍𝑝1 = 1 − 𝜔𝑠2 𝐿𝑝𝑐 𝐶𝑝 + 𝑗𝜔𝑠 𝐶𝑝 𝑅𝑝𝑖 ,

𝑍𝑠1 = 1 − 𝜔𝑠2 𝐿𝑠𝑐 𝐶𝑠 + 𝑗𝜔𝑠 𝐶𝑠 𝑅𝑠𝑖 .

By knowing the currents, the voltage across the coils can be estimated using Equation

(4-8).

𝑗
𝑉𝑝𝑐 = 𝑉𝑝𝑖 + 𝜔 𝐼𝑝𝑖
𝑠 𝐶𝑝
𝑗
(4-8)
𝑉𝑠𝑐 = 𝑉𝑠𝑖 + 𝜔 𝐼𝑠𝑖
𝑠 𝐶𝑠

4.4.2 LC-Parallel Compensation Network

In this structure, the PFC capacitors are attached in parallel with the power pads. The

steady-state equivalent circuit is given in Figure 4.10. In this case, the inverter and pad

currents are different. They are described by the two admittance-matrix model parameters

given in Equations (4-9) and (4-10).

Ipi Lpc-M Rpc Lsc-M Rsc Isi

+ Ipc Isc +
Vpi Vpc Vsc Vsi
Cp M
Cs
- -

Figure 4.11. BIWPTS equivalent circuits using T-model mutual coupling for LC-parallel
topology.

72
 𝑍𝑏 = [𝜔𝑠 2 𝑀2 + (𝑗𝜔𝑠 𝐿𝑠𝑐 + 𝑅𝑠𝑐 )(𝑗𝜔𝑠 𝐿𝑝𝑐 + 𝑅𝑝𝑐 )]⁄(𝑗𝜔𝑠 𝑀)
𝑍𝑝 = (𝑗𝜔𝑠 𝑀𝑍𝑏 )⁄[𝑗𝜔𝑠 (𝐿𝑠𝑐 − 𝑀) + 𝑅𝑠𝑐 − 𝜔𝑠 2 𝑀𝐶𝑝 𝑍𝑏 ] (4-9)
𝑍𝑠 = ( 𝑗𝜔𝑠 𝑀𝑍𝑏 )⁄[𝑗𝜔𝑠 (𝐿𝑝𝑐 − 𝑀) + 𝑅𝑝𝑐 − 𝜔𝑠 2 𝑀𝐶𝑠 𝑍𝑏 ]

−𝑗𝜔𝑠 𝑀𝑉𝑠𝑖 + (𝑗𝜔𝑠 𝐿𝑠𝑐 + 𝑅𝑠𝑐 )

𝑌𝑝𝑝 = 2 𝑀2
𝜔𝑠 + (𝑗𝜔𝑠 𝐿𝑠𝑐 + 𝑅𝑠𝑐 )(𝑗𝜔𝑠 𝐿𝑝𝑐 + 𝑅𝑝𝑐 )
−𝑗𝜔𝑠 𝑀𝑉𝑝𝑖 + (𝑗𝜔𝑠 𝐿𝑝𝑐 + 𝑅𝑝𝑐 )𝑉𝑠𝑖 (4-10)
𝑌𝑠𝑠 = 𝜔𝑠 2 𝑀2 + (𝑗𝜔𝑠 𝐿𝑠𝑐 + 𝑅𝑠𝑐 )(𝑗𝜔𝑠 𝐿𝑝𝑐 + 𝑅𝑝𝑐 )

where, 𝜆 = 1, 𝑌𝑠𝑝 = 𝑌𝑝𝑠 = 0.

In this case, the coil voltages are the same as the inverter voltages. The direct

𝑑𝑣
connection of the voltage source inverters leads to very large ( 𝑑𝑡 ) across the capacitors,

which results in large current spikes to be drawn from the inverter. These spikes are very

damaging for the inverter and lead to poor system’s performance. Thus, L-filters need to

be added to each inverter output to block the HF harmonics in the currents and remove the

current spikes. This modification leads to the LCL-compensation network, which is

analyzed in the next section.

4.4.3 LCL Compensation Network

In this configuration, a second coil is inserted in series to each inverter output in the

LC-parallel topology (Figure 4.11). This filter converts the voltage-source inverter into a

current source supply, which is desirable for EV’s battery operation. The filter inductance

is chosen to reduce the harmonics in the inverter current and allow a certain amount of

active and reactive power transfer. The admittance matrix model parameters for LCL-entity

are given in Equations (4-11) and (4-12).

73
 Lpi Rpi Lpc-M Rpc Lsc-M Rsc Lsi Rsi

Ipi + Ipc Isc + Isi

Vpi Vpc M Vsc Vsi
Cp Cs
- -

Figure 4.12. BIWPTS equivalent circuits using T-model mutual coupling for LCL
topology.

𝑍𝑏 = (𝑍𝑝𝑚 𝑍𝑠𝑚 )⁄(𝑗𝑀𝜔𝑠 ) + 𝑍𝑝𝑚 𝑍𝑠2 + 𝑍𝑠𝑚 𝑍𝑝2

𝑍𝑝 = (𝑍𝑝𝑚 𝑍𝑏 )⁄(𝑍𝑏 (𝜔𝑠2 𝑀𝐶𝑝 + 𝑍𝑝2 )) − 𝑍𝑠𝑚 − 𝑍𝑝𝑚 (4-11)
𝑍𝑠 = (𝑍𝑠𝑚 𝑍𝑏 )⁄(𝑍𝑏 (𝜔𝑠2 𝑀𝐶𝑠 + 𝑍𝑠2 )) − 𝑍𝑝𝑚 − 𝑍𝑠𝑚

where, 𝑍𝑝2 = 1 − 𝜔𝑠2 𝐿𝑝𝑖 𝐶𝑝 + 𝑗𝜔𝑠 𝐶𝑝 𝑅𝑝𝑖 ,

𝑍𝑠2 = 1 − 𝜔𝑠2 𝐿𝑠𝑖 𝐶𝑠 + 𝑗𝜔𝑠 𝐶𝑠 𝑅𝑠𝑖 ,

𝑍𝑝𝑚 = 𝑍𝑝1 [𝑗𝜔𝑠 (𝐿𝑝𝑐 − 𝑀) + 𝑅𝑝𝑐 ] + 𝑗𝜔𝑠 𝐿𝑝𝑖 + 𝑅𝑝𝑖 ,

𝑍𝑠𝑚 = 𝑍𝑠1 [𝑗𝜔𝑠 (𝐿𝑠𝑐 − 𝑀) + 𝑅𝑠𝑐 ] + 𝑗𝜔𝑠 𝐿𝑠𝑖 + 𝑅𝑠𝑖 .

𝜆 = 1⁄[𝑍𝑏 (𝜔𝑠2 𝑀2 + 𝑍𝑠𝑐 + 𝑍𝑝𝑐 )]

𝑌𝑝𝑝 = 𝑍𝑠𝑐 [𝑍𝑏 − (1 − 𝑍𝑏 ⁄𝑍𝑝 )𝑍𝑝𝑖 ] − 𝑗𝜔𝑠 𝑀𝑍𝑠𝑖
𝑌𝑝𝑠 = 𝑍𝑝𝑖 𝑍𝑠𝑐 + 𝑗𝜔𝑠 𝑀[𝑍𝑠𝑖 (1 + 𝑍𝑏 ⁄𝑍𝑝 ) − 𝑍𝑏 ] (4-12)
𝑌𝑠𝑝 = 𝑍𝑠𝑖 𝑍𝑝𝑐 + 𝑗𝜔𝑠 𝑀[𝑍𝑝𝑖 (1 + 𝑍𝑏 ⁄𝑍𝑝 ) − 𝑍𝑏 ]
𝑌𝑠𝑠 = 𝑍𝑝𝑐 [𝑍𝑏 − (1 − 𝑍𝑏 ⁄𝑍𝑝 )𝑍𝑠𝑖 ] − 𝑗𝜔𝑠 𝑀𝑍𝑝𝑖

where, 𝑍𝑝𝑐 = 𝑅𝑝𝑐 + 𝑗𝜔𝑠 𝐿𝑝𝑐 ,

74
 𝑍𝑠𝑐 = 𝑅𝑠𝑐 + 𝑗𝜔𝑠 𝐿𝑠𝑐 ,

𝑍𝑝𝑖 = 𝑅𝑝𝑖 + 𝑗𝜔𝑠 𝐿𝑝𝑖 ,

𝑍𝑠𝑖 = 𝑅𝑠𝑖 + 𝑗𝜔𝑠 𝐿𝑠𝑖 .

In this case, the coil’s voltage is different from the inverter’s voltage, due to the

voltage drop on the filter. By knowing the inverter current, the voltage across the coils can

be estimated using Equation (4-13).

𝑉𝑝𝑐 = 𝑉𝑝𝑖 − 𝑗𝜔𝑠 𝐿𝑝𝑖 𝐼𝑝𝑖

(4-13)
𝑉𝑠𝑐 = 𝑉𝑠𝑖 − 𝑗𝜔𝑠 𝐿𝑠𝑖 𝐼𝑠𝑖

In this topology, the compensation capacitors are designed based only on the

resonance frequency and the coupler self-inductances, as given in Equation (4-2). The filter

parameters (Lpi, Lsi, Rpi and Rsi) are assigned based on the power transfer capabilities. Even

though, the BIWPTS performance is very sensitive to the variation of the filter inductances

with respect to the coupler inductances, as indicated in Figure 4.12. The figure shows the

fundamental equivalent impedance seen from the primary side (Zp-eq) with respect to the

operating frequency at different filter designs. In this study, the system is designed to

resonate at 40 kHz. As it can be noticed, the resonance frequency matches the design

frequency only when the filter and the coil inductances are equal (Lpi = Lpc) and by

assuming a symmetrical system (Lsi = Lsc). In this case, the system exhibits zero impedance

angle which means UPF operation. Any deviation of Lpi from Lpc results in resonance

frequency deviation. In this sense, the operating frequency needs to be adjusted to make

the system work at resonance. For more clear analysis, the effect of the variation of Lpi

75
and Lpc on the system impedance is investigated in Figure 4.13. As it can be observed, the

resonance always occurs at the operating frequency (40 kHz) when Lpi = Lpc. Thus, in order

to keep the operating frequency of BIWPTS the same as the resonant design frequency, the

filter inductances should match with the pad inductances (i.e 𝐿𝑝𝑖 = 𝐿𝑝𝑐 and 𝐿𝑠𝑖 = 𝐿𝑠𝑐 )

[76], [102]. In addition, this synthesis allows the system to behave as a current source

topology, supplying a fixed current regardless of the loading conditions [137]. This

characteristic is very desirable for EV charging operation, in which the battery is charged

at the constant current mode.

|Zp-eq | ( )

1000

500 (a)
Lpi = 0.5 Lpc

0 Lpi = Lpc
36 37 38 39 40 41 42 43L 44L
= 1.5
100
Angle (Degree)

pi pc
Lpi = 2 Lpc

0 (b)

-100
36 38 39 40 41 42 43 44
Frequency (kHz)

Figure 4.13. Effect of filters inductances on the fundamental impedance frequency

response for LCL-parallel design. (a) Impedance magnitude. (b) Impedance phase angle.

76
 4000
| ( )
2000 (a)
p-eq
|Z

0
4 4
3 3
-5
2 2
x 10 1 1 -5
Lpc (mH) x 10
Lpi (mH)
Angle (Degree)

100

0 (b)

-100
4 4
3 3
2 2
-5 1 1 -5
x 10 x 10
Lpc (mH) Lpi (mH)

Figure 4.14. Effect of system inductances on the fundamental impedance frequency

response for LCL-parallel design (40 kHz). (a) Impedance magnitude. (b) Impedance phase
angle.

The effect of the filters’ resistances (Rpi and Rpc) on the system performance is

indicated in Figure 4.14. Generally speaking, the filters’ resistances impact on the

resonance performance is negligible compared with the pads’ resistances. Varying the

system resistances affects the impedance magnitude but does not affect the location of the

resonance (resonant frequency). It also has a slight effect on the impedance phase angle

[Figure 14(b)].

77
 |Zp-eq | ( ) 1000

(a)
500

0
39 39.5 40 40.5 41

100 Rpi = Rpc = 20 m

Rpi = 30 m & Rpc = 20 m
Angle (Degree)

50
R = 20 m & R = 30 m
pi pc

0
(b)
-50

-100
39 39.5 40 40.5 41
Frequency (kHz)

Figure 4.15. Effect of system resistances on the fundamental impedance frequency

response for LCL-parallel design. (a) Impedance magnitude. (b) Impedance phase angle.

Fundamental Power-Flow Modeling

The main feature of the BIWPTS is the ability to transfer power from EV to the grid

and vice versa, such that EVs can be used to inject or absorb active and reactive power to

and from the grid, respectively. Generalized simple formulas for active and reactive power-

flow in all structures of BIWPTS are presented in this section. Moreover, the effect of the

control parameters (α, 𝛽, and δ) on the energy-flow between EV and the grid, is described.

The power-flow formulas were derived for the fundamental primary active (Pp_1) and

78
reactive (Qp_1) power. Harmonic components in Equation (4-1) are neglected and formulas

for the fundamental component of Vpi (Vpi-1) and Vsi (Vsi-1) are obtained in Equation (4-14).

The π-model in Equation (4-4) is solved for Ipi in terms of Vpi and Vsi for each system’s

topology. Then, a simple formula for the fundamental component of Ipi (Ipi-1) is obtained,

by neglecting the higher order harmonics’ components and the resistive losses, and applying

the resonance condition given in Equation (4-2). This formula is the same for all topologies,

as given in Equation (4-15). By substituting Equations (4-14) and (4-15) in the basic power-

flow formulas in Equation (4-16), the final mathematical model for the fundamental

primary active and reactive power is described in Equation (4-17).

4 𝛼 𝛼
𝑉𝑝𝑖_1 = (𝜋) 𝑉𝑑𝑐 cos (𝜔𝑠 𝑡 − 2 ) sin ( 2 )
4 𝛼 𝛽
(4-14)
𝑉𝑠𝑖_1 = (𝜋) 𝑉𝑏 cos (𝜔𝑠 𝑡 − 2 + 𝛿) sin ( 2 )

𝑗𝑀 𝛾
𝐼𝑝𝑖_1 = 𝜔 𝑉𝑠𝑖_1 (4-15)
𝑠 𝐿𝑝 𝐿𝑠

𝑃𝑝_1 = 𝑅𝑒𝑎𝑙(𝑉𝑝𝑖_1 𝐼𝑝𝑖_1 ∗ )

(4-16)
𝑄𝑝_1 = 𝐼𝑚𝑎𝑔(𝑉𝑝𝑖_1 𝐼𝑝𝑖_1 ∗ )

8 𝑀𝛾 𝛼 β
𝑃𝑝_1 = − (𝜋2 ) 𝜔 𝑉𝑑𝑐 𝑉𝑏 sin ( 2 ) sin (2 ) sin(δ)
𝑠 𝐿𝑝 𝐿𝑠
8 𝑀𝛾 𝛼 β
(4-17)
𝑄𝑝_1 = − (𝜋2 ) 𝜔 𝑉𝑑𝑐 𝑉𝑏 sin ( 2 ) sin (2 ) cos(𝛿)
𝑠 𝐿𝑝 𝐿𝑠

where, γ is a real value which is a function of the system parameters and is defined for each

composition in Table 4.1, Lp and Ls are the primary and secondary inductances, which are

accurately defined in the same table for each structure.

79
 Table 4.1. γ-Parameter for Different Topologies.

Configuration Inductances γ- equation

LC-series Lp=Lpc; Ls=Lsc − 𝐿𝑝𝑐 𝐿𝑠𝑐 ⁄𝑀2 = − 1⁄𝑘 2

LC-parallel Lp=Lpc; Ls=Lsc 𝐿𝑝𝑐 𝐿𝑠𝑐 ⁄(𝐿𝑝𝑐 𝐿𝑠𝑐 − 𝑀2 )=1⁄(1 − 𝑘 2 ) ≈ 1

LCL-topology Lp=Lpi; Ls=Lsi 1

As it can be noticed in Equation (4-17), the power-flow in BIWPTS depends on the

DC voltage level on both sides and the system design parameters (resonant frequency, coil

inductances and mutual coupling between coils). After the design stage, these parameters

become fixed except the mutual inductance, which varies based on the vehicle position.

Also, the power-flow is a function of the control parameters of the primary and secondary

inverters (α, β and δ). Typically, the angle δ is kept fixed at ±90o to minimize the reactive

power-flow and achieve UPF operation. The sign of δ is decided to control the power-flow

direction to charge or discharge the EV and switch between the various modes of operation

(G2V and V2G). When δ is negative, the power flows from the primary to the secondary

for charging the EV (G2V). However, when δ is positive, the power transfer occurs from

the secondary to the primary for supplying the grid (V2G). This δ sign notation is

applicable for LC-parallel and LCL structures, but for the LC-series, the opposite notation

is the correct, due to the negative sign in γ-equation (see Table 4.1).

The power-flow magnitude can be controlled by varying either the primary side

voltage magnitude Vpi (using α) or the secondary side voltage magnitude Vsi (using 𝛽) while

80
keeping the relative phase angle (δ) at ±90o. However, controlling Vpi to regulate the

amount of power delivery to a particular secondary is not possible in multiple secondary

systems as it affects the power delivered to other EVs. In single secondary IPT systems,

the power-flow can be controlled through the modulation of either Vpi or Vsi but the former

is preferable, as IPT systems are usually designed for a given primary current. Moreover,

the modulation of Vpi changes the primary current and, hence, the induced voltage in the

secondary. Consequently, low primary currents at light loads will result in a very high

quality factor, Q, (the ratio between the effective AC output voltage and induced voltage)

for a given secondary output voltage. Such an operation with high and variable Q is

undesirable as the system becomes too sensitive. Therefore Vsi of the secondary side

converter is modulated to control both forward and reverse power flow while using Vpi to

maintain a constant primary coil current [137]. Thus, the parameter β is used to control the

value of the power-flow, which ranges from 0 to maximum when β changes from 0 to 180o,

respectively. The relative power-flow value between different configurations can be

observed from the γ parameter equations in Table 4.1. For the same design parameters, in

LC-parallel and LCL assembly, the power are almost equal, since γ is almost the same and

very close to unity. However, for LC-series arrangement, γ is inversely proportional to the

square of the coupling factor. When the coupling factor changes from 0.3 to 0.1 (which is

the typical range in the IPT system), 𝛾 varies from 11.11 (≈ 10) to 100, respectively.

Therefore, the power of the LC series network ranges from 10 to 100 times the power of

the other two structures as well. This indicates that LC-series structure is very sensitive to

the coupling factor and the misalignment, which is one of the main limitations of this

topology.

81
 Conclusion

In this chapter, a comprehensive steady-state performance analysis for various

configurations of the bidirectional inductive power transfer system in EVs applications is

presented. The analysis was based on theoretical harmonics representation for each

topology, using Fourier series analysis. For each topology, two different π-models are

developed, which represent the system’s currents in terms of the inverter’s voltages. The

developed models provide information for inverter currents, inverter voltages, coil currents

and coil voltages on both system’s sides. It also provides a simple derivation for the

system’s equivalent impedance, which is very useful to evaluate the design parameters. In

addition, the analytical models are utilized to derive simple generic formulas for

fundamental active and reactive power-flow in terms of the design and control parameters.

The utilization of these models to evaluate the different compensation configurations and

compare them is demonstrated in chapter (6).

82
 Introduction

The dual H-bridge BIWPTS under consideration consists of four main components;

HF H-bridge converters, power pads, compensation network and controllers, as depicted

in Figure 4.3. Building such a kind of system for EVs’ applications presents huge

challenges. All the system’s components require robust and reliable design to be able to

handle the high-power and high-frequency operation. The semiconductor switches need to

be carefully chosen to support the high-voltage and current stresses during the resonance

operation. The power pads must be designed to provide maximum coupling performance

and less sensitivity for the misalignment in the system. This chapter discusses design

considerations for the entire BIWPTS’s component. It presents more details about the HF

converter design and optimization. An effective design strategy for the HF H-bridge

resonant converter is investigated. The driving and power component selections and design

issues are discussed in detail. The driver and the inverter circuits were implemented

together in one PCB. The PCB layout was optimized for reducing the electromagnetic

interference (EMI). Moreover, two snubber circuits are developed and added to the inverter

PCB to minimize the output voltage ringing phenomenon. Also, the proposed structure

includes voltage and current protection for both the DC and AC sides. Additionally, the

hardware implementation of phase-shift control based on the analog PLL circuit is

investigated and tested. In addition, two symmetrical circular power pads are built to form

the wireless coupler. The practical design considerations for the power pad and the

83
compensation network are discussed as well. All the components are linked together and

tested during both the charging and discharging operation.

HF H-Bridge Converter and Driver

One of the most critical components in IPT system is the HF inverter. Many inverter

topologies were proposed in the literature for the power supply design [87]. The most

popular one is the voltage-source H-bridge converter (VSHBC) structure. VSHBC is a

bidirectional topology, which has the capability to operate in both inverting and rectifying

modes. It provides flexible management for the power-flow in the system. Design aspects

of the HF inverter for induction heating applications are evaluated in [155], [156]. These

works presented some design notes without going into detail about the inverter circuit

design such as component selection, driver circuit, PCB layout, etc., especially under the

hard switching activities. Typically, VSHBC consists of four semiconductor power

switches. Power MOSFETs are still preferred for automotive manufactures due to the

positive temperature coefficient and their ability to handle high switching frequency, as

well as considerations related to the breakdown voltage, cost and reliability requirements.

The entire inverter design process is investigated in this section.

5.2.1 Component Selection

All the inverter’s components are selected based on the power handling and

switching requirements of the inverter. Also, the cost and the size are considered during

84
the selection process. IXFB110N60P3 Si power MOSFET is chosen for this design. It

supports up to the absolute maximum 600V and 110A with 56 mΩ ON resistance [157]. It

provides very wide safe operating area in terms of voltage and current capability, as shown

in Figure 5.1. To drive the gates of the four switches of VSHBC, two FAN7391 IC drivers

are used. They are monolithic high and low-side gate drive IC, which can drive high-speed

MOSFETs that operate up to 600V. They have high current capabilities up to 4.5 A [158].

Surface mounted ceramic capacitors and the hyper-fast diodes are selected for the driver

circuit design to handle the HF operation and the size limitations.

Figure 5.1. Forward-bias safe operating area of IXFB110N60P3 Si power MOSFET [157].

85
 5.2.2 Driver’s Circuits Design

The FAN7391 IC driver is attached to the Bootstrap circuit to drive the power

MOSFETs. Two driving circuits are built to drive the entire VSHBC. Each circuit is

managing one converter’s leg (two MOSFETs). The components of the driving circuit for

one leg are shown in Figure 5.2, and the design steps are indicated below [159]. The details

of sizing each component in the circuit are discussed in this section.

DC SUB 15 V Driver circuit DC IN

Rboot Dboot Power circuit
Dst Rst
SH
1 14 Power
MOSFET
2 13
Rgon
SL
3 12
FAN7391

R
4 11
Cboot Rgoff TO LOAD
5 10

6 9 Rsnubber
7 8 Csnubber
Rgon
C
R
Rgoff

Figure 5.2. Bootstrap driving circuit components using FAN7391 IC driver.

5.2.2.1 Sizing the Bootstrap Capacitor (CBOOT)

CBOOT is designed based on the maximum allowable voltage drop (∆𝑉𝐵𝑂𝑂𝑇 ), which

depends on the minimum gate-drive voltage for the upper-side switch. Considering the

minimum gate-source voltage (𝑉𝐺𝑆𝑀𝐼𝑁 ), ∆𝑉𝐵𝑂𝑂𝑇 is calculated using Equation (5-1).

86
 ∆𝑉𝐵𝑂𝑂𝑇 = 𝑉𝐷𝐷 − 𝑉𝐹 − 𝑉𝐺𝑆𝑀𝐼𝑁 (5-1)

where, 𝑉𝐷𝐷 is the supply voltage of the gate driver and 𝑉𝐹 is the bootstrap diode forward

voltage-drop.

The bootstrap capacitance is estimated using Equation (5-2).

𝑄𝑇𝑂𝑇𝐴𝐿
𝐶𝐵𝑂𝑂𝑇 = ∆𝑉𝐵𝑂𝑂𝑇 (5-2)
𝑄𝑇𝑂𝑇𝐴𝐿 = 𝑄𝐺𝐴𝑇𝐸 + (𝐼𝐿𝐾𝐶𝐴𝑃 + 𝐼𝐿𝐾𝐺𝑆 + 𝐼𝑄𝐵𝑆 + 𝐼𝐿𝐾 + 𝐼𝐿𝐾𝐷𝐼𝑂𝐷𝐸 )𝑡𝑂𝑁 + 𝑄𝐿𝑆 𝑄𝐺𝐴𝑇𝐸

where, 𝑄𝑇𝑂𝑇𝐴𝐿 is the total charge supplied by the capacitor, 𝑄𝐺𝐴𝑇𝐸 is the total maximum

gate-charge of the MOSFET switch, 𝐼𝐿𝐾𝐺𝑆 is the switch gate-source leakage current,

𝐼𝐿𝐾𝐶𝐴𝑃 is the bootstrap capacitor leakage current, 𝐼𝑄𝐵𝑆 is the bootstrap circuit quiescent

current of the driver, 𝐼𝐿𝐾 is the bootstrap circuit leakage current and 𝑄𝐿𝑆 is the charge

required by the internal level shifter, which is usually 3nC for all high voltage gate drivers.

5.2.2.2 Sizing the Turn-on Gate Resistance (Rg(on))

This parameter is very critical and can be determined based on the required output

voltage rate of variation (𝑑𝑉𝑜𝑢𝑡 /𝑑𝑡), as indicated in Equation (5-3).

𝑅𝑔(𝑜𝑛) = 𝑅𝑇𝑜𝑡𝑎𝑙 − 𝑅𝑅𝐷𝑉(𝑜𝑛)

𝑉𝐷𝐷 (𝑉𝐷𝐷 −𝑉𝐺𝑆(𝑡ℎ) ) (5-3)
𝑅𝑅𝐷𝑉(𝑜𝑛) = , 𝑅𝑇𝑜𝑡𝑎𝑙 = 𝑑𝑉
𝐼𝑆𝑂𝑈𝑅𝐶𝐸 (𝐶𝑔𝑑(𝑜𝑓𝑓) ∗ 𝑜𝑢𝑡 )
𝑑𝑡

where, 𝐼𝑆𝑂𝑈𝑅𝐶𝐸 is the source-driver current, 𝑉𝐺𝑆(𝑡ℎ) is the gate-drive voltage of the switch,

𝑅𝐷𝑅𝑉(𝑜𝑛) is the equivalent ON resistance of the gate-driver, and 𝐶𝑔𝑑(𝑜𝑓𝑓) is the miller effect

capacitor of the switch.

87
 5.2.2.3 Sizing the Turn-off Gate Resistance (Rg(off))

This element is calculated at the worst case scenario, i.e. when the drain of the

MOSFET is in turn-off state, as given in Equation (5-4).

𝑉𝑔𝑠(𝑡ℎ)
𝑅𝑔(𝑜𝑓𝑓) ≤ 𝑑𝑉𝑜𝑢𝑡 − 𝑅𝐷𝑅𝑉(𝑜𝑓𝑓)
𝐶𝑔𝑑 ∗
𝑑𝑡 (5-4)
𝑉𝐷𝐷
𝑅𝐷𝑅𝑉(𝑜𝑓𝑓) = 𝐼𝑆𝐼𝑁𝐾

5.2.3 Snubber Circuits Design

During the HF operation with hard switching activities, a snubber RC circuit is

required to minimize the ringing caused by the recovery-induced oscillations and the

conducted EMI due to the stray inductances and capacitances. Two RC snubber circuits are

designed and added to the inverter circuit, as depicted in Figure 5.2. By knowing the

parasitic capacitance (Cp) from the MOSFET data sheet, the snubber capacitance (Csnubber)

will be within the range of (0.5-2) Cp [160]. The ringing frequency (Fp) is measured and

used in Equation (5-5) to estimate the snubber resistance (Rsnubber). This resistance needs to

be carefully designed to handle the high power flow through it. The wattage of Rsnubber is

calculated based on the switching frequency (Fsw) and the snubber voltage (Vsnubber), as

given in Equation (5-6).

1
𝑅𝑠𝑛𝑢𝑏𝑏𝑒𝑟 = 4𝜋𝐹 (5-5)
𝑝 𝐶𝑝

1
𝑃𝑠𝑛𝑢𝑏𝑏𝑒𝑟 = 2 𝐶𝑠𝑛𝑢𝑏𝑏𝑒𝑟 𝑉𝑠𝑛𝑢𝑏𝑏𝑒𝑟 2 𝐹𝑠𝑤 (5-6)

88
 5.2.4 VSHBC PCB Layout Design

In the proposed design, all the inverter’s components and circuits are implemented

in one PCB, including the power switches, snubber, driver and protection elements. This

design helps to minimize the inverter size and weight, and the EMI effect. The details of a

double-layer converter PCB are presented in this section.

PCB layout design is a critical stage in the inverter circuit planning process. It is

important to maximize the power transfer capability for all the stages and minimize the

switching losses. Moreover, it can significantly reduce the magnitude and the duration of

the ringing on the inverter outputs. In the proposed design, the components are arranged in

a way to minimize the tracks’ length and avoid creating loops and junctions between the

top and bottom layers. The components of each inverter’s leg are equally distributed to

achieve symmetrical square output voltage waveforms. The power tracks are assigned to

be straight, short and continuous to minimize the conducted and radiated EMI in the circuit.

The node length between the negative terminal of the input capacitor and the source pin of

the MOSFET is diminished to reduce the parasitic elements’ effect. Also, ceramic input

capacitors are considered to provide low equivalent series inductance (ESL). A comparison

between the optimized PCB layout and an arbitrary (non-optimized) one is presented in

Figures 5.3 and 5.4, respectively.

Two protection schemes are implemented in this design: over-current and over-voltage

protection. The former is achieved by using current-limiting fuses in the DC power input

ports. The latter is performed by using a varistor in series with a current-limiting fuse across

the AC output terminals.

89
Figure 5.3. Layout of the optimized design converter’s PCB.

Figure 5.4. Layouts of an arbitrary design converter’s PCB.

90
 Phase-shift Control and Modulation

In HF applications, like IPT, it is difficult to use the conventional modulation

techniques to generate the switching signals of the power MOSFETs. Typically, other

modulation techniques, in which the switching frequency is the same as the power

frequency, are considered [161]–[166]. The most commonly used approach in IPT systems

is the pulse-phase modulation (phase-shift) technique [102], [162], [165]. In this procedure,

the switching signals have a fixed 50% duty cycle at the resonant frequency, as indicated in

Figure 5.5. The inverter output voltage is controlled by adjusting the phase-shift between

the switching signals of the two legs (α). The figure also shows the dead-time periods (td)

that need to be introduced between the switching of two MOSFETs in the same leg.

Ts

S11 Time
td
S12
α
S13
td
S14

Vout Vdc
Vdc

Figure 5.5. Switching signals and output voltage of VSHBC under phase-shift control.

91
 An analog circuit is designed and implemented in this work to achieve the phase-shift

control algorithm. The circuit is able to supply the switching signals for driving two

synchronized inverters at the same time. It is able to provide variable modulation parameters

(α and 𝛽) for each inverter. Also, it supplies the phase-shift between the two inverters’

output voltages (δ). By varying these parameters, the power flow magnitude and direction

can be controlled, as was indicated in chapter (4). Moreover, the circuit includes four

adjustable dead-time circuits, to assign the dead-time between the switches in the same leg.

Each dead-time circuit is simply variable RC branch with AND gate, as shown in Figure

5.6.

signal

Figure 5.6. Hardware implementation of adjustable dead-time circuit.

A logic block diagram for the phase-shift analogue circuit is depicted in Figure 5.7. It

consists of four dead-time, two NOT circuits and one PLL circuit. The NOT gates are used

to generate 180o phase shift between two trains of pulses. The PLL circuit is used to provide

the 90o phase-shift (δ-parameter). It is designed to lock with frequency ranges from 0-40

kHz. The cascaded NOT gates, shown in Figure 5.7, are used for isolation purpose. The

92
schematic diagram that shows the physical connections and components of the phase-shift

circuit is described in Figure 5.8; and the implemented PCB of the same circuit is depicted

in Figure 5.9.

dead time
S 11
circuit #1
NOT # 1
Signal S 12
generator dead time
circuit #2
S 13 INV 1
90 Degree
S 14
phase-shift
isolation
dead time S 21
circuit #3 S22
NOT # 2
dead time S 23 INV 2
circuit #4
S 24

α=β=180o α=β=90o
δ=±90o δ=±90o

Figure 5.7. Logic block diagram of phase-shift control circuit.

Signal Generator

5V
S11
R3 S12
1 16 1 14

2 16
1 14
R7 2 13
R9 S13
AND_74LS08

2 13
NOT_SN7404N

R5 3 12
MC14046BCP

3 14
R1 3 12
4 13
R4 4 11
S14
4 11 R6 R8
5 12 5 10

C1 5 10 R10
6 11 6 9
6 9
S21
C2
7 10 R2 7 8
7 8

8 9
S22

S23

90o Phase Shifter 180o Phase Shifter Deadtime S24

Output

Figure 5.8. Phase-shift circuit schematic diagram.

93
 AND

NOT 180o Phase-shift 90o Phase-shift

PLL

Figure 5.9. The hardware of phase-shift circuit’s PCB.

Electromagnetic Power Pads

Magnetic power pad is the most important component in an IPT system, since it

dominates the design of the power electronics, maximum power, efficiency, and power

transfer distance. Each IPT system contains two power pads. The first one in placed in the

ground and is called the primary, ground or transmitter pad. The second pad exists in the

vehicle and is called the secondary, vehicle, pick-up or receiver pad. The two pads can be

symmetrical or unsymmetrical. Although, the symmetrical pads provide better coupling

performance, the international standard SAE J2954A recommends smaller size for the

vehicle pad due to space limitations inside the vehicle [149]. The primary pad generates

the HF AC magnetic flux that is linked to the secondary pad. Secondary coil, capturing the

94
applied field, induces voltage that is then converted to a regulated DC voltage to charge

the EV’s battery. Typically, the design of magnetic pads targets high coupling factors,

smaller size and cost, confining the magnetic field between pads without having

considerable fringe/stray magnetic and electric fields, and achieving inductance values that

requires reasonable DC-bus voltage [167]. The typical design of power pads in EVs’

applications contains copper coils, ferrite magnetic material and aluminum shield. The

copper coils are responsible for transmitting and picking up the magnetic field between the

two system’s sides. Each pad may contain one or more coils that are electrically and

magnetically connected. Litz wire is preferred for the coils in order to minimize the internal

resistance, skin effect losses, and increase the quality factor.

Magnetic materials are typically used in order to shape and direct the magnetic field,

and maximize the coupling performance. They also help to minimize the leakage flux in

the system and decrease the reluctance of the magnetic field path. All these effects lead to

maximizing the magnetic coupling coefficient k, which is the most important factor in an

IPT system. An IPT system with a higher coupling coefficient means a system that can

transfer power more efficiently than others. There may, however, be a cost penalty, as

magnetic materials such as ferrite are expensive and fragile. In circumstances where a large

lateral tolerance to misalignment is needed, the cost may be excessive [84]. Aluminum

shield is used in the system to minimize the stray field around the power pads to comply

with the standard limits defined by the International Commission on Non-Ionizing

Radiation Protection (ICNIRP) guidelines [168].

Several studies for different planar pad configurations for EV charging, such as

circular, DD, bi-polar, double-DQ (DDQ) and tri-polar are presented in the literature [84],

95
[85], [169]. In this prototype, the circular design is considered. Two symmetrical power

pads are built. Each pad consists of one stranded copper coil of radius 200 mm, attached to

seven ferrite rods with a pole shoe. Each rod consists of three standard ferrite cores

(I93x28x16) of N87 magnetic material from EPCOS, as indicated in Figure 5.10. More

details about the design process and how to optimize this design are presented in chapters

(8) and (9).

Figure 5.10. Circular power pad with ferrite core.

Compensation Capacitors

As discussed in chapter (4), the compensation capacitor plays a very important role

in the operation of the IPT system. It is responsible for creating the resonance in the system.

96
The design of these capacitors is challenge, since they must be able to handle the high-

voltage and current stresses. The theoretical value of the capacitance is estimated based on

the self-inductance of the power pad and the resonant frequency using Equation (4-2).

Typically, this calculated value is close to the required one but not exactly the same. This

deviation is due to errors in measurements and parameters’ estimation. Thus, the estimated

capacitance must be corrected to achieve the desired resonance performance. In addition,

the resonance network is affected by the parasitic elements of the wires and HF converter’s

components. Thus, in this work, the value of the compensation capacitor is decided through

two steps. The first step is to calculate the capacitance value using Equation (4-2), and find

a capacitor or a combination of capacitors that match or are near this value. The second

step is achieved by connecting one side of the system (HF inverter, power pad and the

compensation network) and operating it at the resonant frequency. Then, the value of the

capacitor is modified, such that the system hits the resonance by showing pure resistive

performance (the inverter’s current and voltage are in phase). Even in the symmetrical

system, small differences are expected between the two sides. Thus, these steps must be

repeated for the other side as well. Metallized Film Capacitors are typically used in IPT

system applications, since they provide high operating voltage (700 VRMS), current (800

ARMS) and power (400 KVA). They also support HF operation, up to 700 kHz, and show

good conduction cooling [170].

Design Verification and Results

For design verification purposes, two different designs of PCB for VSHBC were

developed, built, tested and compared. A snubber circuit is developed and included in the

97
optimum inverter design and tested. The final shape and dimensions of the proposed

inverter, including drivers, protections, snubber circuits and cooling, are indicated in Figure

5.11 [171]. The details of the used components and the estimated design parameters are

presented in Table 5.1. The proposed inverter performance is studied at different

modulation and operating conditions. The different components are linked together to form

a LCL BIWPTS. The entire system is tested and compared with a Simulink model results.

Ultra-fast driver diode

MOSFETs cooling system

OC protection
6"

OV
Varistor

AC OUT

DC IN

DC
First leg signals SUB OC protection Second leg signals
9"

Figure 5.11. The hardware implementation of VSHBC’s PCB.

98
 Table 5.1. Final Converter Design Parameters.

Component Value Commercial name Type

Power MOSFET 600V, 110A IXFB110N60P3 Si Power MOSFET

IC Driver 4.5 A FAN7391 High & Low side

Diodes (Dboot & Dst) 8 A, 600 V VS-ETX0806-M3 Hyper fast

Rboot 9Ω ------ (5-10 Ω ), 1-2 W

Cboot 470 nF ------ Ceramic

Rg-on 10 Ω ------ 1-2 W

Rg-off 33 Ω ------ 1-2 W

R 2.2 k Ω ------ 1-2 W

C 8 μF ------ Ceramic, > 10 Cboot

Csnubber 100nF, 630V R0614 MMKP

5.6.1 Evaluating the Phase-shift Circuit Performance

The main objective of the phase-shift circuit is to generate the switching signals (0-

5V) of the inverters with the necessary dead-times. Two tests are performed: in the first,

the inverter supplies the full capacity with a phase-shift parameter (α) of 180o, while in the

second test, the phase-shift parameter is adjusted to 90o to reduce the RMS inverter’s output

voltage. The switching signals for the 4 MOSFETs of one inverter during the first test are

depicted in Figure 5.12. As it can be noticed, the phase-shift between S11 and S13 is 180o.

Also, there is dead-time between the switches in the same leg (S11, S12 and S13, S14),

which is clarified in Figure 5.13. The assigned time in this case is 250 μsec at each

99
transition. The switching signals during the reduced inverter’s output voltage are described

in Figure 5.14. As it can be observed, the phase phase-shift between S11 and S13 is 90o.

Volt. (V) Volt. (V) Volt. (V)

4
2 (a)
0
0 20 40 60 80 100
6
4
2 (b)
0
0 20 40 60 80 100
6
4
2 (c)
0
0 20 40 60 80 100
6
Volt. (V)

4
2 (d)
0
0 20 40 60 80 100
Time (usec)

Figure 5.12. Measured switching signals (0-5V) of one inverter (4 MOSFETs) during full
inverter voltage. (a) S11. (b) S12. (c) S13. (d) S14.

S11 S12 (a)

6
Volt. (V)

4
2
0
0 40 60 20
80 100
Time (usec)
6 6
4 4
2 (b) (c)
2
0 0
35.5 36 36.5 3772.5 73 73.5 74
Time (usec) Time (usec)

Figure 5.13. Measured switching signals of one leg (2 MOSFETs) during full inverter
voltage showing the dead-time. (a) S11 and S12. (b) Left dead-time. (c) Right dead-time.

100
 6

Volt. (V) Volt. (V) Volt. (V) Volt. (V)

4
2 (a)
0
0 20 40 60 80 100
6
4
2 (b)
0
0 20 40 60 80 100
6
4
2 (c)
0
0 20 40 60 80 100
6
4
2 (d)
0
0 20 40 60 80 100
Time (usec)

Figure 5.14. Measured switching signals (0-5V) of one inverter (4 MOSFETs) during
reduced inverter voltage: (a) S11. (b) S12. (c) S13. (d) S14.

5.6.2 Evaluating the Driving Circuits Performance

The next step, after generating the switching signals, is to boost them to the voltage

and the power levels of the power MOSFETs. This is achieved using the bootstrap circuit

with FAN7391 integrated circuit (IC) gate drivers. The driving circuits receive the signals

with 0-5V level and generate 0-15V, which represent the gate-voltage level of the

IXFB110N60P3 Si power MOSFET. Also, the driving circuits provide the required

currents to drive the gate of the MOSFETs. The gate signals (output of drivers), during the

full and the reduced output inverter voltage tests, are presented in Figure 5.15 and 5.16,

respectively.

101
 20

Volt. (V) Volt. (V) Volt. (V)

10 (a)
0
0 20 40 60 80 100
20
10 (b)
0
0 20 40 60 80 100
20
10
(c)
0
0 20 40 60 80 100
20
Volt. (V)

10 (d)
0
0 20 40 60 80 100
Time (usec)

Figure 5.15. Measured gate signals (0-15V) of one inverter (4 MOSFETs) during full
inverter voltage. (a) S11. (b) S12. (c) S13. (d) S14.

20
Volt. (V) Volt. (V) Volt. (V) Volt. (V)

10 (a)
0
0 20 40 60 80 100
20
10 (b)
0
0 20 40 60 80 100
20
10 (c)
0
0 20 40 60 80 100
20
10 (d)
0
0 20 40 60 80 100
Time (usec)

Figure 5.16. Measured gate signals (0-15V) of one inverter (4 MOSFETs) during reduced
inverter voltage. (a) S11. (b) S12. (c) S13. (d) S14.

102
 5.6.3 The Converter’s PCB Layout Performance Assessment

The effect of the PCB layout design optimization on the ringing of the inverter output

voltage is shown in Figure 5.17. As it can be noticed, the ringing amplitude is reduced and

also the voltage reaches to the steady-state faster in the optimized design compared with

the non-optimized one. The voltage profile of the optimum design is improved during both

the rising and falling edges. The effect of the snubber circuit design on the inverter

performance is indicated in Figure 5.18. It can be observed that, the snubber circuit cleaned

the ringing in the inverter’s output without affecting the rise or the fall time.

40 42
38
20 34
Voltage (V)

0
-38
-20
-42
Optimized
-40 Non-Optimized -46
10 15 20 25 30 35 40 45
Time (usec)

Figure 5.17. Measured ringing effect of the optimized and non-optimized PCB designs.

40
Voltage (V)

20

-20
No-Snubber
-40 Snubber

10 15 20 25 30 35 40
Time (usec)
Figure 5.18. Measured output inverter voltage with and without snubber circuit.

103
 Transient performance analysis is conducted for a step variation of the output voltage

of each inverter design: the non-optimized, optimized without snubber and the optimized

with snubber. The unit step output voltage response for each case is described in Figure

5.19. The performance evaluation parameters, such as settling time, rise time, overshoot and

undershoot are estimated and compared for all cases in Table 5.2. It can be noticed the

improvement of the inverter performance by following the proposed design, which leads to

less transient oscillations amplitude and time.

4 Opt. without Snubber

Non-Opt. without Snubber
Optimized with Snubber
Voltage (V)

3
2
1
0
0 5 10 15
Time (usec)
Figure 5.19. Measured unit-step response of inverter output voltage.

Table 5.2. Performance Parameters during Step Response.

Parameters Non-opt. Opt. without Snubber Opt. with Snubber

without Snubber
Rise time (μsec) 0.15 0.1336 0.1437

Settling time (μsec) 6.71 6.52 0.8

Overshoot (%) 285 274 55.2

Undershoot (%) 64 64 0

Peak (pu) 3.86 3.736 1.55

Peak time (μsec) 0.3 0.28 0.32

104
 5.6.4 The VSHBC’s Circuit Performance Assessment

The output voltage of each inverter’s leg (Va and Vb) and the final AC output voltage

(Vab) are measured during the full and reduced output tests and investigated in Figure 5.20

and 5.21. As it can be noticed, each leg generates train of pulses with an amplitude equal

to the DC input level and a specific phase-shift based on the PPM parameters. The AC

output voltage represents the subtraction of the two legs’ voltages (Vab= Va - Vb). In Figure

5.20, the phase-shift is 180o, thus the AC output voltage is full-square waveforms, as given

in Figure 5.20(c). In Figure 5.21, the phase-shift is 90o, therefore the inverter generates

quasi-square waveforms [see Figure 5.21(c)]. In this case, a zero voltage level appears in

the voltage profile to reduce the RMS voltage values, which is useful for power-flow

control in IPT systems.

20
V (V)

10 (a)
a

0
0 20 40 60 80 100
20
V (V)

10 (b)
b

0
0 20 40 60 80 100
20
V ab (V)

0 (c)
-20
0 20 40 60 80 100
Time (usec)

Figure 5.20. Measured inverter output voltage during full output test: (a) Voltage of the
first leg. (b) Voltage of the second leg. (c) AC inverter voltage.

105
 20

V (V)
10 (a)

a
0
0 20 40 60 80 100

20
V (V)

10 (b)
b

0
0 20 40 60 80 100
20
V ab (V)

0 (c)
-20
0 20 40 60 80 100
Time (usec)

Figure 5.21. Measured inverter output voltage during reduced output test: (a) Voltage of
the first leg. (b) Voltage of the second leg. (c) AC inverter voltage.

Another test is performed to check the inverter behavior under different operating

frequencies. The inverter is tested with 180o phase-shift at 20 kHz and 40 kHz and the

results are indicated in Figure 5.22(a) and (b), respectively. As it can be noticed, a clean

full square wave is generated from the inverter with different operating frequency. Also,

the inverter is tested at 40 kHz with 90o phase shift between the inverter legs, as shown in

Figure 5.22(c).

106
 Voltage (V) Voltage (V) Voltage (V)
50
0 (a)
-50
0 50 100 150 200

20
0 (b)
-20
0 20 40 60 80 100

20
0 (c)
-20
0 20 40 60 80 100
Time (usec)

Figure 5.22. Measured inverter output voltage at different frequencies. (a) Full-square
voltage at 20kHz. (b) Full-square voltage at 40kHz. (c) Quasi-square voltage at 40kHz.

5.6.5 The Entire BIWPTS’s Performance

The VSHBCs, power pads and LCL compensation networks are linked together to

form a LCL BIWPTS, as shown in Figure 5.23 [171]. The system consists of a DC power

supply for DC-bus emulation; a 21 Ah, 53 V Polymer Li-ion battery pack to emulate EV

behavior; two HF VSHBCs; two identical circular pads with Ferrite cores\; two parallel

compensation capacitors; and two L-filters to form the LCL compensation topology. The

same system is simulated in MATLAB Simulink, as depicted in Figure 5.24. The two

model are analyzed and compared in this section. The system is designed to resonate at 40

kHz and the air-gap length between the primary and secondary coils is 150 mm, which is

107
the common case in EV situation. The design parameters of the LCL BIWPTS are indicated

in Table 5.3.

Figure 5.23. Experimental setup of LCL BIWPTS.

Figure 5.24. Simulink model of LCL BIWPTS.

108
 Table 5.3. Design Parameters of LCL BIWPTS.

Parameter Value Parameter Value Parameter Value

Lpi 26 µH Lsi 25.5 µH Cp 0.64 µF

Rpi 32 mΩ Rsi 31 mΩ Cs 0.66 µF

Lpc 25 µH Lsc 24 µH k 0.25

Rpc 30 mΩ Rsc 29 mΩ f 40 kHz

Ferrite Si Power Stranded copper

Core MOSFET Wires
I93/28/16 IXFB110N60P3 H07V-K
Capacitor Ceramic Driver FAN7391 Sensors Hall effect

Battery Li-ion Polymer (PL-1055275-14STM) Air-gap 150 mm

5.6.5.1 Full Supply Capacity Test

In this section, the system is analyzed during the full operation of supply capacity. In

this mode, the phase-shift between the two inverters’ legs (α and 𝛽) are set to 180o, to

provide the maximum available power on the supply. The BIWPTS is tested during G2V

(charging) mode of operation, as given in Figure 5.25. In this test, the phase-shift between

the two inverters’ voltages (δ) is set to -90o, to allow the power transfer from the DC-bus

to the EV’s battery. The figure shows the voltage and current waveforms in the primary

and secondary circuits. Figure 5.25(a) and (b) describe the primary and secondary

inverters’ variables, respectively. It can be noticed that, the two inverters generate full

square voltages with 90o phase-shift (Vsi lags Vpi). Also, it can be observed that, Vpi and Ipi

are almost in phase, however Vsi and Isi are anti-phase, which means that the system is

109
hitting the resonance and the primary and secondary power-flow are in opposite direction

(one side is transmitting the energy and the other side is receiving it). The primary and

secondary coils’ currents and voltages are described in Figure 5.25(c) and (d), respectively.

The inverters’ currents (Ipi and Isi) are not pure sinusoidal, but the coils’ currents (Ipc and

Isc) are almost pure sinusoidal, due to the filtering effect of the circuit impedance. Good

agreement can be observed between the simulation and experimental results.

V V I I
Exp. Sim. Exp. Sim.

50
10Ipi
Vpi

0 (a)
-50
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
50
10Isi
Vsi

0
(b)
-50
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
100
50
5Ipc
pc

0
V

-50 (c)

0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1

50
5Isc
Vsc

0 (d)
-50
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
Time (msec)

Figure 5.25. Experimental and simulated BIWPTS’s performance under full supply voltage
for G2V operation (α = β =180o, δ = -90o, Vdc=60 V). (a) Primary inverter variables. (b)
Secondary inverter variables. (c) Primary coil variables. (d) Secondary coil variables.

110
 The BIWPTS is analyzed during V2G (discharging) operation. In order to achieve

this mode, the secondary inverter’s voltage is set to lead the primary one by 90o phase-shift

(δ=90o). The parameters α and 𝛽 are kept fixed at 180o for supplying the full system

capacity. The discharging operation results are presented in Figure 5.26. As it can be

noticed, the secondary voltage leads the primary one by 90o, which allows the power to

flow from the EV to the DC-bus. In this case, Vpi and Ipi are anti-phase and Vsi and Isi are

in-phase, which means that the system is still working at UPF but with the reverse power

flow direction.

VExp. VSim. I Exp. I Sim.

50
10Ipi
Vpi

0 (a)
-50
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
50
10Isi
Vsi

0 (b)
-50
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
100
50
5Ipc
pc

0 (c)
V

-50
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
50
5Isc
Vsc

0 (d)
-50
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
Time (msec)
Figure 5.26. Experimental and simulated BIWPTS’s performance under full supply voltage
for V2G operation (α = β =180o, δ = 90o, Vdc=60 V). (a) Primary inverter variables. (b)
Secondary inverter variables. (c) Primary coil variables. (d) Secondary coil variables.

111
 5.6.5.2 Reduced Supply Capacity Test

To verify the power level adjustment capability, the system is analyzed during the

charging (G2V) operation with quasi-square voltage waveform. This situation is achieved

by setting the phase-shift between the inverter’s legs as α=𝛽=90o, and δ=-90o. The system

performance under these conditions is described in Figure 5.27. As it can be observed, Vpi

and Vsi waveforms exhibit the zero level, which causes the reduction of the transferred

currents and power, in consequence.

V V I I
Exp. Sim. Exp. Sim.

50
10Ipi
Vpi

0 (a)
-50
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
50
10Isi
Vsi

0 (b)
-50
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
100
50
5Ipc
Vpc

0 (c)
-50
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
50
5Isc
Vsc

0 (d)
-50
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
Time (msec)
Figure 5.27. Experimental and simulated BIWPTS performance under reduced supply
voltage for G2V operation (α = β =90o, δ = -90o, Vdc=60 V). (a) Primary inverter variables.
(b) Secondary inverter variables. (c) Primary coil variables. (d) Secondary coil variables.

112
 For a more clear comparison between the simulation and experimental results, the

normalized root mean square error (NRMSE) is evaluated, based on Equation (5-7), for the

presented cases and the results are indicated in Table 5.4. Most of the predicted variables

show less than 5% errors. The error in Ipi and Isi estimation is a little high due to the

existence of a DC-current offset in the measurements, which results from the very small

asymmetry in the inverter outputs.

2 2
𝑁𝑀𝑆𝐸 = √∑𝑚 𝑚
𝑖=1(𝑓(𝑡𝑖 ) − 𝑔(𝑡𝑖 )) ⁄∑𝑖=1(𝑓(𝑡𝑖 )) (5-7)

where, f(xi) and g(xi) are the value of the variable at sampling index (i), which are calculated

from experimental and simulation, respectively.

Table 5.4. NRMSE between Simulation and Experimental Results.

Mode Ipi Isi Ipc Isc Vpi Vsi Vpc Vsc

Full G2V 8.5 9 0.57 0.73 1.3 1.3 0.36 0.32

Full V2G 3.9 4.9 0.75 0.61 1.8 1.4 0.23 0.24

Reduced G2V 11.7 10 2 0.66 1 1.3 0.67 0.07

5.6.5.3 Practical Considerations during Testing the BIWPTS

The experimental tests and measurements of the BIWPTS are challenge and need to

be done carefully. Some practical issues are examined during testing the LCL BIWPTS

prototype, which are concluded below:

113
- The HF inverters’ switches and driver circuits need to be well designed to supply

symmetrical voltage waveform. Any small asymmetry in the voltage generates a large

DC current component, due to the very small DC impedance at resonance. This

asymmetry may be due to a small difference in dead-time of the inverter switches. This

problem does not exist, if there is a series capacitor in the system, such as LC-series or

CLCL compensation topology, since the series capacitor behaves as a high-pass filter.

- The polarities of the power pad coils must be well defined and connected correctly with

the inverters’ outputs, to provide the proper power-flow direction, based on the assigned

phase-shift between the two inverters’ voltages (δ).

- There is another design problem related to the compensation network elements due to

the errors in parameters’ estimation and the parasitic elements of both converters and

power pads. This problem was addressed in the prototype by initially evaluating the

compensation elements based on the resonant frequency and the estimated pads’ self-

inductances, as given in Equation (4-2). Then, these parameters were corrected by trial

and error, until the system shows pure resistive behavior and hits the resonant condition.

- The measuring sensors need to be carefully chosen with wide bandwidth to handle the

HF signals.

Conclusion

This chapter presented an effective design and implementation considerations for the

different components of the bidirectional inductive wireless power transfer system in EVs

applications. The design of HF voltage-source H-bridge converter is investigated in detail.

114
The details of the proposed design are presented including the parameters’ estimation,

components’ selection and PCB layout design. Moreover, two protection schemes were

introduced to the inverter circuit. Snubber circuits were designed for minimizing the

ringing effect. All the inverter components were embedded in one PCB using the

commercially available standard size (9"x6"), which results in size and weight reduction.

The proposed inverter design performance was tested under different operating and control

conditions. In addition, the analog implementation of the phase-shift control technique was

performed and tested. The circuit was able to supply the switching activities for two

synchronized inverters with the associated dead-times at different modulation conditions.

Practical design considerations for the magnetic power pads and the compensation

capacitors are addressed. Finally, two identical inverters were built and used in a LCL

bidirectional inductive wireless power transfer system, to achieve the two-way power-flow

operation between the EV and the power-grid. The entire system performance was

analyzed under different modes of operation (G2V and V2G), by simulation and

experimentally for verification purposes. The tests show very good correlation between the

simulation and the experimental results.

115
 Introduction

In a BIWPTS, the pads are loosely coupled with a large leakage inductance. A

compensation network is usually used to reduce the Volt-Ampere (VA) rating in the coil

and power supply. In early inductive charging designs, the compensation is set on primary

or secondary side only [172]. In the current IPT technology, where the coupling coefficient

is reduced to less than 0.3 for EV wireless charging, compensation at both the primary and

secondary side is recommended, to have a more flexible and advanced characteristic [82].

The simplest way to compensate the system inductance, is to add a capacitor at each side.

For a symmetrical BIWPTS, there are four basic compensation topologies, which are LC-

series, LC-parallel, LCL, and CLCL topology, as was discussed in chapter (4). Based on

the analytical models that were presented in chapter (4), an assessment analysis for the

steady-state performance of different topologies in BIWPTS for EV implementations is

investigated in this chapter. The analysis is presented for the three main compensation

configurations: LC-series, LC-parallel and LCL-topology. The study presents evident

criteria to pick among the different BIWPTS structures in the diverse applications. The

proposed analyses are verified using simulation and experimental tests. The results

demonstrate the ability of the proposed models to provide accurate estimation for

BIWPTS’s performance under various operating and control conditions. Also, the

evaluation analysis shows that LCL-topology is more appropriate for the bidirectional

116
operation in EV, due to the simple design and control requirements and being less sensitive

to the misalignment.

The contributions in this work are summarized as follows: 1) verifying the developed

theoretical models in chapter (4) by means of simulation and experimental tests, 2) utilizing

these models to assess the performance of the different topologies during the bidirectional

operation and present the merits and demerits of each structure, 3) achieving comparative

analysis for the BIWPTS’s performance parameters (power, power factor and efficiency)

among all topologies, under both full and light loading conditions, and 4) evaluating the

sensitivity of the different topologies against the misalignment in the system based on

theoretical models and measurements.

Assessment Criteria of Compensation Topologies in BIWPTS.

In unidirectional IPT systems, the primary side is connected to the grid and represents

the source of energy. The secondary side is coupled to the EV’s battery and behaves as a

load. Thus, in these systems, there is a specific source side and load side. In this case, the

choice of the compensation topology in the secondary side depends on various factors, such

as the output type (voltage/current) and the load. The choice of primary topology, on the

other hand, depends on other factors, such as the type of inverter used. In bidirectional

applications, the BIWPTS will be utilized to charge (G2V) and discharge (V2G) the EV’s

battery. During G2V operation, the primary side represents the source of energy, while the

secondary side acts as a load. In V2G service, the primary side behaves as a load, while the

secondary side supplies the required energy. Thus, in this case, there is neither a specific

source side nor load side. Hence, from the author’s point of view, the two sides need to be

117
treated equally. In addition, different considerations during the choice of the compensation

topology need to be taken into account. Due to these reasons, it is assumed that the

bidirectional system under consideration is symmetrical, in which the compensation

topologies in both sides are identical.

In general, as was reported in many works in the literature, the choice of the

compensation topology plays a very important role in the performance of IPT system.

Regardless of the mode of operation (V2G or G2V), the compensation topology in the

system helps to reduce the VA rating of power electronics converters, achieve UPF

operation (zero-phase-angle) condition, achieve soft switching [zero-voltage switching

(ZVS) or zero-current switching (ZCS)] in the power converters, enhance the power

transfer capability and increase the system efficiency [82], [89].

In BIWPTS, the compensation topologies need to be evaluated based on the

following requirements [139], [86], [82], [89], [146]:

- Providing two-way power-flow operation. The compensation topology needs to be

tested during both G2V and V2G operation. It should be able to facilitate a bidirectional

and controlled power-flow. In this case, the topology, which needs less complex

controllers, is preferred.

- Minimizing VA rating of power converters. In this case, two HF converters are adopted

in the system. Thus, the compensation structure on both sides needs to compensate the

reactive power requirements, such that the converters carry the active power only.

- Maximizing power transfer capability by compensating the inductive reactance.

118
- Enhancing the system efficiency. The ability of the compensation network to achieve

soft-switching and minimize the reactive power flow through the converters and feeder

leads to higher system efficiency.

- Offering more stability and less insensitivity to the variation of system parameters and

loading conditions. The resonant network must be able to realize constant current,

constant voltage or constant power under parameters’ variation (especially the mutual

inductance) and different loading conditions.

Performance Evaluation for Different Topologies.

The proposed mathematical models in chapter (4) are implemented and solved in

MATLAB environment. These models are based on the harmonic components and all the

parameters and variables are solved at each harmonic component and added together. The

results from these models were compared with Simulink models for the different

compensation configurations (Figure 4.3 and 5.24), for verification purposes. For fair

comparison and assessment among the various structures, the same design parameters,

given in Table 5.3, are considered in all of them and only the compensation topology varies.

The framework is designed to resonate at 40 kHz with 150 mm air-gap length between the

primary and secondary coils, which is the typical case in EV situation. The DC voltage

levels that are used in the Simulink models are Vdc=240 V and Vb=200 V. In addition,

power-flow assessment analysis for all topologies is presented for the same power-flow

level (1.5 kW). Each configuration is analyzed during both G2V and V2G operation,

considering the full and light loading conditions.

119
 6.3.1 Assessment of LC-Series Topology.

As it was mentioned before, the power-flow direction and the operation mode (either

V2G or G2V) is controlled by adjusting the sign of δ, while the power-flow magnitude is

controlled by adjusting α and/or β. Figure 6.1 describes the system performance with LC-

series topology under full charging (G2V) operation, which means that the inverters

transmit the full supply capacity. This mode is achieved by setting (α=β=180o and δ=90o).

The primary and secondary side variables are indicated in Figure 6.1(a) and 6.1(b),

respectively. The theoretical waveforms are plotted for the first 150 harmonics and they

show very good matching with the Simulink model results. As it can be noticed, the inverter

output voltages are full-square waves with amplitudes equal to the DC voltages.

Also, it can be seen from the figure that, Vpi and Ipi are in-phase, but Vsi and Isi are

anti-phase, which means that both the primary and secondary circuits are working almost

at UPF and the system is hitting the resonant frequency. Moreover, there is a phase-shift

between the primary and secondary voltages (Vpi lags Vsi by 90o), which allows the power

to flow from the grid to charge the EV. In addition, both inverters’ and coils’ currents are

equal and sinusoidal, with no DC current components due to the series capacitors, which

behave as high-pass filters. It is worth mentioning that the series resonance tank magnifies

the pad and capacitor voltages (Vpc and Vcp) much higher than the inverter voltages due to

the capacitive reactance, as was indicated in Equation (4-8). It can be seen in Figure 6.1(c)

and 6.1(d) that the peak of the coil voltage (Vpc) is almost equal to the peak of the capacitor

voltage (Vcp), and both are almost five times the peak of the inverter voltage (Vpi).

120
 VSim . VTheo. ISim . ITheo.
200

V pi(V)
Ipi(A)
0 (a)
-200
19.9 19.92 19.94 19.96 19.98 20
200
V si(V)
Isi(A)

0 (b)
-200
19.9 19.92 19.94 19.96 19.98 20
1000
V pc(V)

0
(c)
-1000
19.9 19.92 19.94 19.96 19.98 20
1000
V cp(V)

0 (d)
-1000
19.9 19.92 19.94 19.96 19.98 20
Time (msec)
Figure 6.1. LC-series topology performance under full supply capacity for G2V operation
(α=β=180o, δ=90o, Vdc=240 V and Vb=200 V). (a) Primary inverter variables. (b) Secondary
inverter variables. (c) Primary pad voltage. (d) Primary capacitor voltage.

Figure 6.2 describes the G2V operation with reduced power level, by adjusting α and

β to be 90o instead of 180o. A zero-voltage level appears in the inverter voltages to reduce

the RMS voltages, currents, and accordingly the power. The V2G operation under reduced

power level is depicted in Figure 6.3. In this case, the secondary voltage lag the primary

one by 90o, which allows the power to flow from the EV to the power grid. Also, Vpi and

Ipi are anti-phase and Vsi and Isi are in-phase, which means that the system is still working

at UPF, but with the reverse power flow direction.

121
 VSim . VTheo. ISim . ITheo.

200

V (V)
Ipi(A)
0 (a)

pi
-200
19.9 19.92 19.94 19.96 19.98 20

200
V (V)
Isi(A)
si 0 (b)

-200
19.9 19.92 19.94 19.96 19.98 20
1000
V (V)

0
pc

(c)

-1000
19.9 19.92 19.94 19.96 19.98 20
Time (msec)

Figure 6.2. LC-series topology performance under reduced supply capacity for charging
operation (G2V) (α=β=90o, δ=90o, Vdc=240 V and Vb=200 V). (a) Primary inverter
variables. (b) Secondary inverter variables. (c) Primary pad voltage.

VSim . VTheo. ISim . ITheo.

200
V (V)
Ipi(A)

0 (a)
pi

-200
19.9 19.92 19.94 19.96 19.98 20

200
V (V)
Isi(A)

0 (b)
si

-200
19.9 19.92 19.94 19.96 19.98 20
1000
V (V)

0 (c)
pc

-1000
19.9 19.92 19.94 19.96 19.98 20
Time (msec)

Figure 6.3 LC-series topology performance under reduced supply capacity for discharging
operation (V2G) (α=β=90o, δ=-90o, Vdc=240 V and Vb=200 V). (a) Primary inverter
variables. (b) Secondary inverter variables.

122
 From the presented results, it can be observed that in LC-series topology, huge

voltages appear across the power pads, which lead to large current-flow to achieve the

desired power-transfer. The inverter currents are sinusoidal, which facilitates the

implementation of soft-switching activities, such as ZVS or ZCS. In addition, the same

sinusoidal currents are passing in the wireless pad, which results in minimum radio

frequency interference on the nearby devices. However, LC-series structure exhibits

several limitations in EVs applications, as follows:

- The large pad current must be carried by the inverter. Therefore, the inverter needs to

be designed to handle this current, which is a challenge, especially in high-power and

HF applications.

- The wireless pad design requires many turns of winding with thick conductors to handle

the high-voltage and current operation.

- The compensation capacitors must be capable of supporting these high-voltages and

currents.

- The system experiences significant conduction (in power converters, passive elements

and feeders) and dielectric (capacitors) losses.

- Current controllers are necessary in this configuration to limit the pad currents in both

the system’s sides to match with the rated values.

- The high voltage on the pad may become very pronounced if the secondary circuit was

left open.

- It is very sensitive to misalignment.

Typically, this topology is used in low-power applications, and when the inverter

is located near the coupler, in order to minimize the feeder losses and voltage-drop.

123
 6.3.2 Assessment of LC-Parallel Topology

BIWPTS with LC-parallel configuration is analyzed in this section during G2V

operation using the theoretical and Simulink models, as given in Figure 6.4. In this

configuration, the capacitor is connected in parallel with the pad to supply the required

reactive power. In this case, the inverter provides only the real component that is required

to handle the system losses in the two circuits and the load requirements. The square-wave

voltages generated by the VSHBCs are rich in HF harmonics, as well as the desired

fundamental frequency. The direct connection of such source to a parallel resonant circuit

would cause excessive currents to flow at all higher harmonics of the drive frequency.

This is because the capacitor tank in the parallel resonant circuit would present a

progressively lower capacitive reactance to increasing frequencies. This results in large

current spikes at the switching transitions as the inverter tries to rapidly charge and

discharge the capacitor on rising and falling edges of the square-wave, as can be seen in

Figure 6.4(a), 6.4 (b) and 6.4 (c). These spikes are potentially very damaging to the VSHBC

and lead to large reactive power to be derived from the supply associated with poor PF

operation. The inclusion of L-filter between the inverter and capacitor negates this

problem, as is demonstrated in the next section. The pad current is almost sinusoidal and

does not show these spikes, which are absorbed by the parallel capacitor. In this test, the

theoretical model is solved for the first 150 harmonics orders and exhibits good correlation

with the Simulink model.

124
 VSim . VTheo. ISim . ITheo.

10*V si(kV) 10*V pi(kV)

2

Ipi(kA)
0 (a)
-2
19.9 19.92 19.94 19.96 19.98 20
2
Isi(kA)

0 (b)
-2
19.9 19.92 19.94 19.96 19.98 20
50
Ipi(A)

0 (c)
-50
19.9 19.92 19.94 19.96 19.98 20
50
Ipc(A)

0 (d)
-50
19.9 19.92 19.94 19.96 19.98 20
Time (msec)
Figure 6.4. LC-parallel topology performance under reduced supply capacity for G2V
operation (α=β=90o, δ=-90o, Vdc=240 V and Vb=200 V). (a) Primary inverter variables. (b)
Secondary inverter variables. (c) Zoomed primary inverter current. (d) Primary pad current.

6.3.3 Assessment of LCL Topology.

The LCL structure is an improved version of LC-parallel topology by adding an

impedance matching coil between the inverter and parallel resonant tank. This topology is

investigated during G2V operation using the proposed theoretical (for the first 15

harmonics) and Simulink models in this section. LCL-based BIWPTS’s performance for

full supply capacity operation is described in Figure 6.5. The figure shows all the primary

125
(Vpi, Ipi, Vpc and Ipc) and secondary (Vsi, Isi, Vsc and Isc) variables. As can be seen, the inverter

currents are limited to reasonable values and the spikes disappear after adding the L-filters.

Also, the coil currents become pure sinusoidal, leading to lower electromagnetic

interference. The inverter current is very small compared with the coil current, since it

represents only the active current component, while the reactive part is supplied from the

parallel capacitor. The peak inverter and capacitor voltages are almost equal to the DC

voltage levels, as the parallel resonance does not magnify the voltage like the series one.

V Sim. V Theo. ISim. ITheo.

10Ipi (A)

200
(V)

0 (a)
pi

-200
V

9.93 9.94 9.95 9.96 9.97 9.98 9.99 10

200
10Isi (A)
(V)

0 (b)
si

-200
V

9.93 9.94 9.95 9.96 9.97 9.98 9.99 10

5Ipc (A)

200
(V)

0 (c)
pc

-200
V

9.93 9.94 9.95 9.96 9.97 9.98 9.99 10

200
5Isc (A)
(V)

0 (d)
sc

-200
V

9.93 9.94 9.95 9.96 9.97 9.98 9.99 10

Time (msec)

Figure 6.5. LCL-topology performance under reduced supply capacity for charging
operation (G2V) (α=β=180o, δ=-90o, Vdc=240 V and Vb=200 V). (a) Primary inverter
variables. (b) Secondary inverter variables. (c) Primary coil variables. (c) Secondary coil
variables.

126
 6.3.4 Power-Flow Assessment for Different Configuration.

Performance evaluation parameters (power, efficiency and PF) for the three different

configurations during G2V operation are analyzed in this section. For fair comparative

analysis, the DC voltage levels (Vdc and Vb) for the three systems are adjusted, such that

they provide the same rated power (1.5 kW). These voltages are varied with a fixed ratio

for all topologies in order to deliver the same current distribution. Each configuration is

investigated at full and light (half) loading conditions. The full loading condition is

obtained by setting the control parameters (α and β) to be 180o, as indicated in Table 6.1.

The light loading condition is achieved by applying α=β=90o, as presented in Table 6.2.

The given efficiency calculations represent only the passive components’ efficiency and

do not consider the inverters’, feeders’ and supplies’ losses. The tables show that LC-series

topology provides the highest efficiency and PF in both the full and light load conditions,

due to the sinusoidal waveforms. LC-parallel exhibits very poor PF and the lowest

efficiency operation due to the large current spikes. LCL-topology gives higher PF and

efficiency than LC-parallel but less than LC-series, due to the non-sinusoidal inverter

currents, which increase the reactive power and the losses in the system.

Table 6.1. Power-Flow Performance for all Topologies under Full-Loading Conditions.

Topology Pp (kW) Ps (kW) %η PF

LC-series (Vdc=60 V and Vb=50 V) -1.536 1.479 96.33 0.999

LC-parallel (Vdc=236V and Vb=196 V) 1.532 -1.466 95.65 0.108

LCL (Vdc=240 V and Vb=200 V) 1.534 -1.473 96 0.9654

Evaluated at α=β=180o.

127
Table 6.2. Power-Flow Performance for all Topologies under Light-Loading Conditions.

Topology Pp (kW) Ps (kW) %η PF

LC-series (Vdc=60 V and Vb=50 V) -0.768 0.740 96.3 0.999

LC-parallel (Vdc=236V and Vb=196 V) 0.766 -0.733 95.6 0.108

LCL (Vdc=240 V and Vb=200 V) 0.767 -0.736 96 0.9654

Evaluated at α=β=90o.

Despite LCL-topology having two more passive components, it is more appropriate

for high-power applications (e.g. EVs) and for the situations where the inverters are

installed far from the pads, due to the following:

- The parallel capacitors behave as PF correction and supply the required reactive

(magnetizing) power.

- Only the real power component is driven from the supply, which leads to simple inverter

design and operation.

- The inverter currents are very small compared to the pad currents, which means low

conduction and feeder losses.

- The wireless pads and capacitors require simple design since they do not have to handle

high voltage operation.

- The system behaves as a current-source, therefore the bidirectional power-flow can be

easily achieved by controlling the inverter voltages.

- It is less sensitive to the misalignment than LC-series topology.

- No current spikes appear due to the filtration process.

128
 Experimental Evaluation for Different Topologies.

In this section, the BIWPTS prototype presented in Figure 5.24 with the design

parameters in Table 5.3 is utilized to evaluate the performance of the different topologies

based on experimental tests. In this prototype, the DC voltage levels (Vdc and Vb) are scaled

down (1/4 time the Simulink model values) to fit the available Li-ion battery module in the

laboratory. The used DC voltage levels are Vdc=60 V and Vb=50 V.

V Exp. V Theo. IExp. ITheo.

50
pi
pi

0
10I

(a)
V

-50
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
50
si
si

0
10I

(b)
V

-50
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
50
pc
pc

0 (c)
5I
V

-50
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
50
sc
sc

0 (d)
5I
V

-50
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
Time (msec)
Figure 6.6. Experimental and theoretical LCL-topology performance under full supply
capacity for G2V operation (α=β=180o, δ=-90o, Vdc=60 V, Vb=50 V). (a) Primary inverter
variables. (b) Secondary inverter variables. (c) Primary coil variables. (d) Secondary coil
variables.

129
 6.4.1 Experimental Validation of LCL Topology Model.

In this section, the LCL topology is tested and the instantaneous voltage and current

waveforms are recorded. These measured waveforms are compared with the theoretical

ones, estimated from the proposed analytical model, for verification purposes. The

response of LCL BIWPTS under the full supply capacity during G2V and V2G operation

is indicated in Figures 6.6 and 6.7, respectively. The figures present comparison between

experimental tests and theoretical model results. Good agreement is observed between the

two models.

VExp. VTheo. IExp. ITheo.

50
10Ipi
V pi

0 (a)
-50
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
50
10Isi
V si

0 (b)
-50
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
50
5Ipc
V pc

0 (c)
-50
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
50
5Isc
V sc

0
(d)
-50
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
Time (msec)
Figure 6.7. Experimental and theoretical LCL-topology performance under full supply
capacity for V2G operation (α=β=180o, δ=90o, Vdc=60 V, Vb=50 V). (a) Primary inverter
variables. (b) Secondary inverter variables. (c) Primary coil variables. (d) Secondary coil
variables.

130
 Moreover, the same prototype is analyzed to verify the effect of the phase-shift

parameters on the power-flow magnitude. In this case, the effect of changing α and 𝛽 on

the system power-flow is investigated. The system power is reduced by applying α=𝛽 =90o,

while δ = ±90o to control the power flow direction. The results of G2V operation with a

reduced power level are described in Figure 6.8. It can be observed that the two inverters’

voltages show zero-voltage level to reduce the RMS voltages, currents and, in

consequence, the power. This effect is clear in the systems’ currents and voltages. The

system performance under the V2G operation is shown in Figure 6.9. Also, the results show

good agreement between the theoretical and experimental data.

V Exp. V Theo. IExp. ITheo.

50
pi
pi

0
10I

(a)
V

-50
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
50
si
si

0
10I

(b)
V

-50
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
50
pc
pc

0 (c)
5I
V

-50
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
50
sc
sc

0 (d)
5I
V

-50
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
Time (msec)
Figure 6.8. Experimental and theoretical LCL-topology performance under reduced supply
capacity for G2V operation (α=β=90o, δ=-90o, Vdc=60 V, Vb=50 V). (a) Primary inverter
variables. (b) Secondary inverter variables. (c) Primary coil variables. (d) Secondary coil
variables.

131
 V V I I
Exp. Theo. Exp. Theo.
50

10Ipi
V pi 0 (a)
-50
0.04 0.06 0.08 0.1
50
10Isi
V si

0 (b)
-50
0.04 0.06 0.08 0.1
50
5Ipc
V pc

0 (c)
-50
0.04 0.06 0.08 0.1
50
5Isc
V sc

0 (d)
-50
0.04 0.06 0.08 0.1
Time (msec)
Figure 6.9. Experimental and theoretical LCL-topology performance under reduced supply
capacity for V2G operation (α=β=90o, δ=90o, Vdc=60 V, Vb=50 V). (a) Primary inverter
variables. (b) Secondary inverter variables. (c) Primary coil variables. (d) Secondary coil
variables.

In this analysis, two scales of BIWPTS are utilized. The large-scale is used in the

Simulink model, with Vdc=240 V, Vb=200 V. The small-scale is considered in the

prototype, with Vdc=60 V, Vb=50 V. In order to ensure the system scalability, the power-

flow is evaluated for both large and small-scale models during G2V operation, as presented

in Table 6.3. As can be noticed, both scales provide the same efficiency and PF. The ratio

between the power levels is about 16, which is the square of the scaling factor 4. Thus, it

132
can be concluded that the system is scalable, and small-scale prototypes can be utilized to

verify its performance.

Table 6.3. Large and Small-Scale Power-Flow Analysis.

Mode Pp (W) Ps (W) ɳ (%) PF (pu)

Large-scale G2V (Vdc=240 V & Vb=200 V) 1534 -1473 96 0.965

Small-scale G2V (Vdc=60 V & Vb=50 V) 95.855 -92 96 0.965

Evaluated at α=β=180o and δ=-90o.

6.4.2 Evaluating the Sensitivity to Misalignment for all Topologies.

Current IPT technology requires a perfect alignment between coils in order to keep

the transferred power within the desired limits. However, this condition is practically

impossible. Typically, different types of misalignment are expected between the two sides

during the practical implementation. Thus, the SAE J2954A international standard has

defined accepted limits and criteria for the different misalignments that the system may be

subjected to. The system must operate within a range of these misalignment to be able to

transfer the required power wirelessly. Two different categories of misalignments are

accurately defined in J2954A standard [149]. The first class is associated with the planar

offset in X, Y and Z directions. The second category is related to the angles between the

two sides. Three different angle misalignments are defined: Rotation, Roll and Yaw angles,

as indicated in Figure 6.10.

133
Figure 6.10. Angle misalignment in IPT system [149].

Various solutions are presented in the literature to minimize the misalignment issues,

such as installing electro-mechanic alignment systems. This system causes the charging

process to be slow and mechanically complex, or multiple primary systems (coil array

structures) are required [173]. Another method to mitigate the misalignment problems and

achieve fast charging process without the need for electro-mechanic positioning

mechanisms is to implement a very precise, complex, closed-loop frequency control [174]

that keeps the system operating at resonance, even with varying coupling factor. A cheap

and simple way to deal with the misalignment issues is to choose and design the

compensation network, such that the power-flow in the system becomes insensitive to the

misalignment [175]. Therefore, one important aspect to evaluate the performance of a

compensation network is to measure its sensitivity against the misalignment.

134
 Therefore, the performance of different topologies with the variation of the magnetic

coupling between the system’s sides, due to misalignment, is evaluated and compared in

this section. In this analysis, the system was operated to transfer the maximum power

during G2V mode (α=β=180o). All the investigated performance measurement parameters

are normalized with respect to their values at coupling factor (k) equal 0.2, as an

intermediate value. Only theoretical results are provided for LC-parallel topology, in order

not to damage the HF converters. The primary coil RMS current is measured and compared

with the theoretical (estimated) value under different coupling factors, as presented in

Figure 6.11. The figure shows that the coil current in LCL topology is independent from

the coupling factor. This proves that this structure is able to behave as a current-source

regardless of the misalignment. In contrast, in LC-series, the current is inversely

proportional with the coupling factor.

LC-series(est.)
2
Normalized I pc (pu)

LC-series(meas.)
LCL(est.)
1.5 LCL(meas.)
LC-parallel(est.)

0.1 0.15 0.2 0.25 0.3

coupling factor (k)

Figure 6.11. Measured and estimated normalized primary coil current under misalignment
for all topologies (G2V, α=β=180o).

135
 In Figure 6.12, the active and reactive primary power for all topologies are evaluated.

The measured powers are extracted from the experimental data using Fast-Fourier

transform (FFT) to find the magnitude of each harmonic component. The fundamental

component is used for active power estimation, while the higher-order harmonics are

considered for reactive power calculation. LC-parallel and LCL show linear and similar

increase in active power as k increases [Figure 6.12(a)]. However, LC series exhibits

significant reduction of power, as the coupling increases due to the sensitivity to the

variation of the reflected impedance. Also, Figure 6.12(b) indicates that the reactive power

of LCL topology is fixed regardless of the misalignment, unlike the other two

configurations.

LC-series(est.) LC-series(meas.) LCL(est.) LCL(meas.) LC-parallel(est.)

2 1.06
(a) (b)
Normalized P p (pu)

Normalized Q p (pu)

1.04
1.5
1.02

1 1

0.98

0.5
0.96
0.1 0.15 0.2 0.25 0.3 0.1 0.15 0.2 0.25 0.3
coupling factor (k) coupling factor (k)

Figure 6.12. Measured and estimated normalized power under misalignment for all
topologies (G2V, α=β=180o). (a) Active power. (b) Reactive power.

136
 Finally, the equivalent fundamental primary impedance for all topologies is

estimated based on the impedance model in Equation (4-6). This parameter is analyzed

versus the coupling factor in Figure 6.13. It can be observed that the equivalent impedance

of LCL topology is less sensitive to the misalignment, unlike that of LC-series and LC-

parallel, which nonlinearly increases and decreases, respectively as the coupling factor

increases. From Figure 6.13(b), both LC-series and LCL topology provide zero-impedance

angle (UPF), as was discussed before, while LC-parallel shows lagging inductive

impedance.

4
Norm. |Zp-eq | (pu)

LC-series
LCL
2 LC-parallel (a)

0
0.1 0.15 0.2 0.25 0.3

100
o
Imped. Angle

50 (b)

0
0.1 0.15 0.2 0.25 0.3
coupling factor (k)
Figure 6.13. Theoretical fundamental equivalent impedance under misalignment for all
topologies (G2V, α=β=180o). (a) Impedance magnitude. (b) Impedance angle.

137
 6.4.3 Summary for the Assessment Analysis.

The outcomes from the comparative analyses among the different topologies are

summarized in Table 6.4. The table presents the advantages and disadvantages of the

different topologies, in terms of many factors, such as passive components, control

complexity, power factor, etc. It can be noticed that LCL-structure is better in terms of

system’s design, control implementation, physical installation and the robustness against

misalignment, as is clarified below:

- As was discussed in section 6.3.1, in LC series configuration, the inverter should be

designed to handle both active and reactive current components, which represents huge

challenges in the semiconductor switches, especially in high-power and HF

applications. This item not only will represent a problem during the design, but also it

leads to significant conduction losses in the power converter and in the feeder, which

will reduce the overall system efficiency. Although the coupler efficiency in LC-series

is similar to that of LCL topology, the overall system efficiency, including the power

electronic converters and the feeders (which was not included in this study), will be less

in case of LC series.

- Regarding the control complexity, as was mentioned in section 6.3.3, LCL topology

behaves as a current-source, which is the desired operation for EV charging. In this case,

the power-flow can be easily controlled by adjusting the output voltage of the inverters.

However, in the case of the LC-series structure, the system behaves as a voltage-source

and current controllers are necessary, in addition to power-flow controllers, which

means more complex control.

- For the physical installation, in LCL topology, the inverter can be installed near to the

138
 power pad or far from it. This is because, in this case, the current and losses in the feeder

are small. On the other hand, in LC-series, it will be impractical to install the inverter

far from the power pad, since the feeder current and losses will be significant.

- Concerning the sensitivity to misalignment, LCL topology shows more robustness

against the misalignment, in contrast with LC-series network. This feature is very

important for IPT system in EV applications, in which the system is usually subjected

to the misalignment during the normal operation. The system may become completely

useless if it was too sensitive to the misalignment.

Table 6.4. Summary of the Assessment Analysis.

Topology LC-series LC-parallel LCL

No. of passive components 4 4 6

Bidirectional control
High ----- Low
Complexity
PF High Low Medium

Radio interference Low High Medium

Coupler efficiency High Low High

Design requirements High High Low

Sensitivity to misalignment High Low Low

close to close to or far

Coupler position close to inverter
inverter from inverter

Soft switching capability High Low Medium

DC blocking capacitor No need Need Need

139
 Thus, it can be concluded that LCL-topology is the most practical and appropriate

choice for bidirectional operation of EV. Even though LCL structure exhibits some

deficiencies in PF, radio interference and soft switching capability, all these problems can

be mitigated by adding DC-blocking capacitors with the proper design for both the

capacitors and filter inductors, which can be an open research point.

Conclusion

A comprehensive, steady-state performance analysis for various configurations of

bidirectional inductive power transfer system in EVs applications is presented in this

chapter. The analysis was based on the theoretical harmonics representation for each

topology that was presented in chapter (4). The proposed theoretical models were utilized

to study the control parameters’ effect on the steady-state system response and power flow.

Moreover, the BIWPTS’s performance evaluation parameters (active and reactive power,

efficiency and power factor) were estimated and compared. For validation purposes, a

BIWPTS model was built in Simulink/MATLAB for each configuration, using the same

design parameters, and its outcomes were compared with the theoretical models’ results.

The results demonstrate good correspondence between the theoretical and the simulated

models. In addition, the small-scale experimental prototype that was presented in chapter

(5) is tested and analyzed to verify the theoretical and simulation results. The proposed

analysis could predict the systems’ steady-state performance precisely as the correlation

between the simulation and experimental results manifest. The presented analysis

concludes that, among the different configurations, LCL-topology is more appropriate for

140
EV applications and in the situations where the inverter is a long way from the pad. It

shows current-source characteristics, power-factor correction capability, simple design and

control, high overall efficiency and power factor, and more robustness to misalignment.

141
 Introduction

Inductive wireless power transmission is a promising technology in the charging of

EVs. This technology is mainly based on the magnetic coupling between two or more coils.

The coupling coefficient (k) ranges from 0 to 1, based on the magnetic design and the

distance between the coils. In the conventional transformer, k is about 0.95 and it is called

strongly coupled technology, but in IPT systems k ranges from 0.01 to 0.4, and it is called

loosely coupled technology [176]. The electrical isolation in IPT systems provides safe,

reliable, maintenance free operation in harsh environment. For these features, IPT

technology has been approved to be convenient and reliable interface for charging and

discharging the EV’s battery, during long-term parking (stationary), short-term stops

(quasi-dynamic) or movement (dynamic) [97]. As was reported in chapter (6), among the

different topologies of BIWPTS, LCL-based system provides better performance,

especially for EVs applications. Precise model for such system can help the designers and

researchers to anticipate, optimize and evaluate the system’s behavior amid the

development. Thus, this chapter deals with developing accurate nonlinear dynamic model

for LCL BIWPTS in EVs applications.

According to the literature, several studies have been focused on BIWPTS modeling

analysis. In [142], [75] steady-state mathematical models for LCL BIWPTS were

developed. An analytical dynamic model based on state-space representation was

investigated in [143], [177] for BIWPTS with CLCL topology. In these works, the coils of

the wireless coupler were ideally modeled as linear inductors and the mutual coupling was

142
represented by the T-model, as in [75], or the reflected impedance, as in [142]. In these

models, the core losses and the nonlinear magnetic characteristics of ferrite materials of

the power pads were neglected. These approximations may lead to inaccurate prediction

and evaluation for the system performance under different control and operating

conditions. Considering such kind of features necessitates physics-based models for the

wireless coupler, such as FEMs. Several FEA for different structures of power pads in EV

applications were presented in the literature [84], [178]. In these works, FEMs were built

and pure sinusoidal currents were injected to emulate the system’s performance. In this

case, the effect of the nonlinearities due the power electronic converters and the dynamic

controllers on the system were neglected. For precise evaluation of BIWPTS, the impacts

of both the power electronics and the magnetic material characteristics need to be

considered in the same simulation environment.

Thus, this chapter presents a physics-based co-simulation platform for LCL BIWPTS

in EVs applications. The platform is established through the coupling between finite

element and circuit analysis. The power electronic converters and controllers are developed

in Simulink and the power pads are modeled in Magnet environment. The two parts are

linked together through the compatible Simulink Plug-in tool. In addition, a state-space

dynamic mathematical model for the same LCL BIWPTS is derived and implemented in

MATLAB environment. A 1.2 kW LCL BIWPTS is analyzed under different dynamics by

both models, and the results are compared. The effect of the nonlinearities and the magnetic

material characteristics on the system performance is assessed in terms of error and

harmonics analysis. The analysis considered both the full and the light loading operating

conditions in the system. The proposed co-simulation could provide accurate prediction for

143
the system dynamics, during both charging and discharging operation. The scheme is

generic and can be easily expanded to different pad structures, compensation networks and

inverter topologies.

The main contributions in this work are summarized below:

- A dynamic analytical model for LCL BIWPTS based on state-space representation is

derived, implemented and analyzed.

- A 2D-FEM for wireless coupler with DD pad structure is built considering the magnetic

characteristics of N87 ferrite material.

- Detailed analysis and explanation for the coupling between the power converters and

controller in Simulink and the 2D-FEM in Magnet is presented.

- The impact of the nonlinearities due to the magnetic material and the power electronics

on the BIWPTS’s performance based on error and harmonics analysis is investigated,

during different dynamics in the system.

- The system power flow under both light and full loading conditions is evaluated, during

G2V and V2G operations.

- The effectiveness of the platform is verified by means of analytical and experimental

results.

Analytical Dynamic Model of BIWPTS

Similar to the steady-state models, presented in chapter (4), the DC-bus is modeled

as a fixed DC source (Vdc), and the EV’s battery is represented by a fixed DC source (Vb),

in the nonlinear model. A dynamic mathematical model, based on state-space

representation, is presented in this section to predict the entire BIWPTS’s performance.

144
 7.2.1 HF VSHBC Dynamic Model

As was discussed before, two HF VSHBCs are utilized in the BIWPTS to supply the

wireless coupler with controlled square-wave voltages at the resonant frequency, as

indicated in Figure 7.1. These square-waves are controlled to achieve the required

bidirectional power-flow in the system. The phase-shift control technique, depicted in

Figure 4.4, is considered for this purpose. In this method, the controller adjusts the

magnitude and the phase angle of the two inverters’ voltages to manage the power flow-

magnitude and direction between the two sides of the system. The switching activities for

the primary inverter driven by the phase-shift control method are described in Figure 7.2.

The switching signals for the second leg (S13 and S14) are shifted by an angle α from the

first leg signals (S11 and S12). This parameter (α) varies between 0 to 180o to change the

RMS inverter’s output voltage (vpi) from zero to maximum, respectively. The switching

signals of the secondary inverter follow the same activities in Figure 7.1. There is a phase-

shift 𝛽 between the second leg signals (S23 and S24) and the first leg signals (S21 and S22).

Also, the four secondary signals are shifted by an angle δ from the first inverter switching

signals. Typically, the angle δ is used to manage the direction of the power-flow from the

power grid to the EV, and vice versa. The negative δ means performing G2V operation,

while positive δ provides V2G service. This parameter is theoretically set to ±90o to allow

maximum active power flow with almost UPF operation.

145
 Secondary inverter
Primary inverter
13 21 23
11
Lpi, Rpi Wireless coupler
Lsi, Rsi

Lpc M Lsc
DC Cp Cs EV’s
bus battery
(Vdc) (Vb)
14 24
12 22
2D S21-S24
S11-S14
FEM
Switching signals Switching signals
and drivers and drivers

Figure 7.1. Power electronics circuit diagram of LCL BIWPTS.

s1 s2 s3 s4
S11(t) Ts
Time
S12(t)

S13(t) α

S14(t)

ѱ(t)

vpi(t)
Vdc
Vdc

Figure 7.2. Switching activities and output voltage waveform of primary inverter.

146
 Depending on the switching actions in Figure 7.2, four switching states for the

primary inverter are stated in Table 7.1. The tri-state function [Ψp(t)] is evaluated from the

individual switching function using Equation (7-1) and indicated in Figure 7.2 and Table

7.1.

𝛹𝑝 (𝑡) = 𝑆11 (𝑡)𝑆14 (𝑡) − 𝑆12 (𝑡)𝑆13 (𝑡) (7-1)

The instantaneous output voltage of the primary inverter [vpi(t)] is expressed in terms

of the tri-state function and the DC-link voltage (Vdc), as in Equation (7-2).

𝑣𝑝𝑖 (𝑡) = 𝛹𝑝 (𝑡) × 𝑉𝑑𝑐 (7-2)

Table 7.1. Switching States of the Primary HF VSHVC.

State Period S11 S12 S13 S14 Ψp Vpi

S1 0 < t ≤ α/2πfs 1 0 0 1 1 Vdc

S2 α/2πfs < t ≤ Ts/2 1 0 1 0 0 0

S3 Ts/2 < t ≤ (Ts/2+α/2πfs) 0 1 1 0 -1 -Vdc

S4 (Ts/2+α/2πfs) < t ≤ Ts 0 1 0 1 0 0

For the secondary inverter, the switching states are the same as the primary ones, but

shifted in time by the angle δ, and the phase shift parameter is 𝛽 instead of α. Thus, the tri-

state function [Ψs(t)] and the instantaneous output voltage of the secondary inverter [vsi(t)]

are expressed, as in Equation (7-3).

147
 𝛹𝑠 (𝑡2 ) = 𝑆21 (𝑡2 )𝑆24 (𝑡2 ) − 𝑆22 (𝑡2 )𝑆23 (𝑡2 )
(7-3)
𝑣𝑠𝑖 (𝑡2 ) = 𝛹𝑠 (𝑡2 ) × 𝑉𝑏

where, 𝑡2 = 𝑡 + 𝛿 ⁄2𝜋𝑓𝑠 , fs is the switching frequency (fs=1/Ts), which is the same as the

resonant frequency (fr) in both sides.

7.2.2 Magnetic Coupler and Compensation Network Dynamic Model

In LCL BIWPTS structure, the compensation capacitor is connected in parallel with

each coil of the wireless coupler for PFC and reactive power compensation. Also, an L-

filter is introduced between the inverter and the capacitor for providing current-source

characteristics. The time-domain equivalent circuit of LCL-based symmetrical BIWPTS is

depicted in Figure 7.3. The HF inverters are represented by AC sources with square or

quasi-square waves. The system is designed for fixed frequency operation, and the two

sides operate at the same resonant frequency (ωr). The compensating capacitors are tuned

to resonate with the coil self-inductances based on Equation (4-2), to be independent from

the misalignment between the ground and the vehicle pads.

The filter inductances (Lpi and Lsi) are designed to match the pad self-inductances

(i.e. Lpi=Lpc and Lsi=Lsc) in order to provide fixed pad currents and to make the operating

frequency match the design resonant frequency, as was discussed in chapters (4) and (6).

The system dynamics are modeled based on the state-space representation, as given

in Equation (7-4).

ẋ = A x + B u (7-4)

where, x is the state vector and u is the control vector.

148
 ipi ipc isc isi
Lpi Rpi + M + Lsi Rsi
+
Lpc Lsc +
vpc vsc
- Cp Cs -
vpi Rpc Rsc vsi
- -

Figure 7.3. Dynamic electric equivalent circuit of symmetrical LCL BIWPTS.

These vectors, x and u, are stated for the LCL BIWPTS, in Equation (7-5).

x = [𝑖𝑝𝑖 𝑣𝑝𝑐 𝑖𝑝𝑐 𝑖𝑠𝑐 𝑣𝑠𝑐 𝑖𝑠𝑖 ]𝑇

(7-5)
𝑢 = [𝑣𝑝𝑖 𝑣𝑠𝑖 ]𝑇

where, the subscript p stands for the primary side variables and s identifies the secondary

side, 𝑖𝑝𝑖 and 𝑖𝑠𝑖 are the instantaneous inverter currents, 𝑖𝑝𝑐 and 𝑖𝑠𝑐 are the instantaneous

coil pad currents, and 𝑣𝑝𝑐 and 𝑣𝑠𝑐 are the instantaneous coil voltages.

Following the basic principles of circuit theory, the dynamic model is expressed by

six ODEs, as given in Equation (7-6).

1
𝑖𝑝𝑖̇ = (𝐿 ) (𝑣𝑝𝑖 − 𝑅𝑝𝑖 𝑖𝑝𝑖 − 𝑣𝑝𝑐 )
𝑝𝑖
1
𝑣𝑝𝑐̇ = (𝐶 ) (𝑖𝑝𝑖 − 𝑖𝑝𝑐 )
𝑝
𝐿 𝑅𝑝𝑐 𝐿𝑠𝑐
𝑖𝑝𝑐̇ = 𝑀𝛾 (𝑣𝑠𝑐 − 𝑅𝑠𝑐 𝑖𝑠𝑐 − ( 𝑀𝑠𝑐) 𝑣𝑝𝑐 + ( )𝑖𝑝𝑐 )
𝑀
(7-6)
𝑀 𝑅𝑝𝑐 𝑀
𝑖𝑠𝑐̇ = 𝐿𝑝𝑐 𝛾 (−𝑣𝑠𝑐 + 𝑅𝑠𝑐 𝑖𝑠𝑐 + (𝐿 ) 𝑣𝑝𝑐 − ( 𝐿 )𝑖𝑝𝑐 )
𝑝𝑐 𝑝𝑐
1
𝑣𝑆𝑐̇ = (𝐶 ) (𝑖𝑠𝑖 − 𝑖𝑠𝑐 )
𝑠
1
𝑖𝑠𝑖̇ = (𝐿 ) (𝑣𝑠𝑖 − 𝑅𝑠𝑖 𝑖𝑠𝑖 − 𝑣𝑠𝑐 )
𝑠𝑖

where, 𝛾 = 1/(𝑀2 − 𝐿𝑠𝑐 𝐿𝑝𝑐 ).

149
 According to Equations (7-4)-(7-6), the system matrix A and the input matrix B are

deduced, as in Equation (7-7).

−𝑅𝑝𝑖 −1
0 0 0 0
𝐿𝑝𝑖 𝐿𝑝𝑖 1
1 −1
0
𝐿𝑝𝑖
0 0 0 0
𝐶𝑝 𝐶𝑝 0 0
0 −𝛾𝐿𝑠𝑐 𝛾𝐿𝑠𝑐 𝑅𝑝𝑐 −𝛾𝑀𝑅𝑠𝑐 𝛾𝑀 0
A= ;B= 0 0 (7-7)
0 𝛾𝑀 −𝛾𝑀𝑅𝑝𝑐 𝛾𝐿𝑝𝑐 𝑅𝑠𝑐 −𝛾𝐿𝑝𝑐 0 0 0
−1 1 0 0
0 0 0 0 1
𝐶𝑠 𝐶𝑠
−1 −𝑅𝑠𝑖 [0 𝐿𝑠𝑖 ]
[ 0 0 0 0
𝐿𝑠𝑖 𝐿𝑠𝑖 ]

The system’s ODEs are programmed and solved in MATLAB script using the fourth

order Runge-Kutta (RK4) numerical technique, and utilized to predict the BIWPTS’s

dynamics.

Physics-Based Co-simulation Platform

A typical BIWPTS for EV connection consists of the power electronics converters,

passive compensation elements and power pads. The power pads are made of copper litz

wire to minimize the skin and proximity effects, and magnetic ferrite material to direct the

flux lines and reduce the magnetic path reluctance. In typical circuit models, the magnetic

coupler is modeled by self and mutual-inductances using T-model mutual representation

[75], or using the reflected impedance theory [142]. In this case, the parameters are

extracted or estimated at one operating point, and the nonlinearities due to the magnetic

material and the magnetic losses are not considered. Thus, relying on such approximate

models to simulate the system’s performance under different control and operating

conditions is not accurate and may lead to significant errors. In order to precisely simulate

150
the system’s performance, the magnetic characteristics need to be considered in the model.

Thus, combining the physics-based model for the power pads, such as FEM and the circuit

simulation is necessary. The main limitation for such coupling is the long computation

time. For minimizing this problem, compatible simulation tools are considered in this work.

More details about the FEM and the co-simulation are presented in this section.

7.3.1 Finite-Element Model in Magnet

Among the different pad structures presented in the literature, the polarized DD

design is considered in this work. It is simple in design and provides better coupling, greater

tolerance and cost effectiveness. Moreover, it offers a charge zone five times larger than

that possible with the typical circular pads for a similar material cost [84]. The typical

structure of one DD power pad is depicted in Figure 7.4.

Aluminum shield Coil

field
current
Ferrite sheet

Figure 7.4. Double-D power pad structure using sheet of ferrite core.

It comprises of two or more copper coils magnetically connected in series (flux from

one coil passes through the other) and electrically connected in parallel, to lower the

151
equivalent inductance seen by the power supply. The coils are placed on top of ferrite strips

or a sheet, to direct the flux streamlines from one coil to the other. Thus, there is no

magneto-motive force (MMF) produced that can drive the flux out of the back. This allows

aluminum shielding to be placed underneath with little impact on the coils’ quality factor,

creating a truly single-sided flux pad with a path height proportional to half of the pad

length [144], [176]. A 2D-FEM for the wireless coupler is developed in Magnet

environment. It consists of two identical DD power pads to satisfy the symmetrical

bidirectional operation, in which each side can behave as a source and a load at the same

time. Each pad consists of two Litz wire coils. The coils are attached to a ferrite sheet with

the commercial ferrite material N87. This material is the most commonly used in IPT

systems. It is composed of a combination of MnZn magnetic material. It offers good

magnetic characteristics, as shown in Figure 7.5, with mass density of 4850 kg/m3 [179].

These characteristics are considered in the developed 2D-FEM. The 2D-FEM for the two

identical DD power pads is shown in Figure 7.6, and the dimensions are presented in Table

7.2.

200
0.4 150
Pc (kW/m 3)

(a) (b)
0.3
B (T)

100
0.2
0.1 50

0 0
0 500 1000 0 0.1 0.2
H (A/m) Bpeak (T)

Figure 7.5. Characteristics of N87 ferrite material. (a) Magnetization characteristic

@10kHz and 25 oC. (b) Core losses characteristic @40kHz and 25 oC.

152
 Aluminum shield

e f Ferrite bar d

x x
b c
a

Current direction Copper coils

x x

Figure 7.6. 2D-FEM of one power pad of the wireless coupler.

Table 7.2. Dimensions of the Power Pad in 2D-FEM

Parameter Value Parameter Value

a 67 mm c 1 mm

b 7 mm d 1 mm

e 16 mm f 5 mm

g 100 mm

153
 The model was analyzed based on the 2D FEA for solving a time-varying field, using

the magnetic vector potential field formulation given in Equation (7-8) [180], [181].

𝜕𝐴
∇ × [𝜇 −1 (∇ × 𝐴)] = −𝜎 ( 𝜕𝑡 + ∇𝑉) (7-8)

where, μ is the magnetic permeability of the medium (H/m), and 𝜎 is the electric

conductivity of the medium (A/V.m), A is the magnetic vector potential (V.s/m), and V is

the scalar potential voltage function.

The nonlinearities that are investigated in this work are associated with two elements:

1) the nonlinear characteristics of the magnetic material, and 2) the nonlinearities in the

sources due to the power electronic components. The magnetic material nonlinearity is

represented in Equation (7-8) by μ. For the nonlinear magnetic materials (e.g. ferrite

material), the permeability μ depends on the local value of B, which changes nonlinearly

based on H, according to the B-H characteristic. Thus, iterative technique is required to

solve Equation (7-8) until the element permeability values have converged. The second

nonlinearity (associated to the sources) is characterized by the right-side of Equation (7-8),

which represents the current density vector of the source.

Dirichlet (Flux Tangential) boundary conditions are applied to the surrounding area

around the model. In this case, the entire outer boundary behaves as a flux line. This choice

is the best for this case, since it is equivalent to putting the model in a cavity of a material

with zero permeability, so that no flux can escape from the model. In addition, these

boundaries are taken sufficiently far away from the components of the model (≈10 times

the model dimensions), thus it represents a good approximation to the open boundaries.

Therefore, the model can provide accurate solution with minimum computational effort.

154
 Transient magnetic field analysis is performed in Magnet environment for the 2D-

FEM. The coil currents during this analysis are given in Figure 7.7. They are typically

sinusoidal, with a 90o phase-shift between them. Five time instants are defined in the

waveforms (t1, t2, t3, t4 and t5), and the magnetic flux distribution at these instants, when

the two power pads are perfectly aligned, are indicated in Figure 7.8. These positions

almost cover all the possibilities and are repeated periodically with the time.

50
i pc
i sc
Current (A)

0
t1 t2 t3 t4 t5

-50
0 0.01 0.02 0.03 0.04 0.05
Time (msec)

Figure 7.7. Instantaneous wireless pads currents waveforms.

155
 (a) (b)

(c)
(d)

(e) (f)

Figure 7.8. Magnetic flux density distribution of 2D-FEM. (a) At t1. (b) At t2. (c) At t3. (d)
At t4. (e) At t5. (f) Color map scale.

156
 7.3.2 FE-Simulink Co-simulation

For accurate evaluation of the magnetic material’s effect on the BIWPTS’s

performance, the 2D-FEM of the wireless coupler is coupled with the power converters’

simulation. For reasonable simulation time, two compatible tools are used in this platform,

as depicted in Figure 7.9. Simulink tool is used to simulate the compensation network, the

power electronic converters, and the controllers. Magnet software is considered to model

the power pads. The two models are coupled through the Simulink Plug-in tool. This tool

allows Simulink to control the sources of the coils in Magnet environment [145].

Figure 7.9. Co-simulation Simulink-Magnet Platform.

The details of the coupling between the circuit and finite element environments are

described in Figure 7.10. In Magnet environment, each coil is connected to a controlled-

voltage or current-source in the Circuit window. Voltage-source is preferred, as it results

in a more stable coupling for magnetic components and is less likely to yield unphysical

conditions [182]. In Simulink medium, each pad coil is denoted as a controlled current-

source. The voltages across the current-sources represent the control signals of the

157
controlled voltage-sources in Magnet. After analyzing the FEM, the coil currents are

calculated and used as control signals for the controlled current-sources in Simulink

environment, as indicated in Figure 7.10. The Simulink Plug-in tool has been designed to

operate properly when the Simulink step-size is fixed. It also provides the ability to set the

MagNet step-size to be different from that of Simulink. In this case, the step size of the FEA

can be much larger than that of the circuit analysis, to accelerate the simulation time.

i1 i2
Lpi, Rpi Lsi, Rsi
2D
+ FEM +
vpi v1 v2 vsi
Cp i1 i2 Cs
- + + -
- -
Simulink v1 v2 Simulink
Magnet

Figure 7.10. Coupling between Simulink and Magnet software.

Model Verification and Discussion

A 1.2kW LCL BIWPTS model (shown in Figure 7.1) is analyzed by both the

dynamic analytical model and the co-simulation platform. The framework is designed to

resonate at 40 kHz, with an air-gap length between the primary and secondary coils of 100

mm, and design parameters given in Table 7.3. The BIWPTS’s model comprises of two

DC supplies to emulate the DC-bus and the EV’s battery, two HF full-bridge inverters, two

identical DD power pads, two parallel compensation capacitors, and two L-filters. Each

power pad consists of two copper coils attached to a ferrite sheet of N87 magnetic material

158
from EPCOS. The analytical and co-simulation results during both G2V and V2G

operation are presented and discussed in this section.

Table 7.3. Design Parameters of the BIWPTS Model

Parameter Value Parameter Value Parameter Value

Lpi 26.1 µH Lsi 26.1µH Cp 0.61 µF

Rpi 38 mΩ Rsi 38 mΩ Cs 0.61 µF

Lpc 26.1µH Lsc 26.1 µH M 5.2 µH

Rpc 38 mΩ Rsc 38 mΩ F 40 kHz

Error Analysis

To show the nonlinearities in the system, selected dynamics are introduced to the

BIWPTS, such as starting, power-flow level variation and switching between the different

modes of operation (G2V and V2G). The system under these disturbances is analyzed in

time-domain, using both the co-simulation platform and the ODEs presented in section 7.2.

Both models are solved in MATLAB environment for 0.04 sec with 20 MHz sampling

frequency. The transient FEA is achieved at 2 MHz sampling frequency in Magnet

environment. The analyzed scenario, including the different dynamics, is indicated in

Figure 7.10. The BIWPTS starts with G2V operation to charge the EV at full power level

(α=β=180o and δ=-90o), as depicted in Figure 7.11(a). In this period, the two inverters

generate full-square wave output voltages, with 90o phase-shift (vsi lags vpi), as shown in

Figure 7.11(b) and 7.11(c).

159
 full full reduced reduced
200 G2V V2G V2G G2V

o 100
Angle
0  ,  (a)

-100
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
vpi-theo vpi-co-sim vsi-theo vsi-co-sim
400 (b)
v(t) (V)

200
0
-200
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
Time (s)
200 200 200

0 0 0

-200 -200 -200

0.00995 0.01 0.01005 0.0199 0.02 0.02 0.02995 0.03 0.03005
(c) (d) (e)

Figure 7.11. Introduced dynamics in BIWPTS (Vdc=240V and Vb=200V). (a) Phase shift
control parameters (α, β and δ). (b) Inverter output voltages. (c) Zoom for G2V to V2G
transition at full capacity. (d) Zoom for power level reduction in V2G mode. (e) Zoom for
V2G to G2V transition at reduced power level.

At 0.01 sec, the control parameter δ varies from -90o to 90o to discharge the EV and

realize V2G service. At this instant, the phase of vsi changes to lead vpi by 90o, while both

inverters still transfer the full power capacity [see Figure 7.11(c)]. The system keeps

working at V2G operation with full capacity until the control parameters α and β are

reduced from 180o to 90o at 0.02 sec. After 0.02 sec, the inverters generate quasi-square

waves with zero level to reduce the RMS voltages and the power flow, in consequence [see

160
Figure 7.11(d)]. Another mode switching action in applied at 0.03 sec to charge the EV

with a reduced power level [see Figure 7.11(e)].

Vtheo Itheo Vco-sim Ico-sim

300
10xI pi (A)
V pi (V)

200 (a)

100

0 0.01 0.02 0.03 0.04

300
5xI pc (A)
V pc (V)

250
(b)
200
150

0 0.01 0.02 0.03 0.04

Time (s)

Figure 7.12. RMS values of BIWPTS’s variables (Vdc=240V and Vb=200V). (a) Primary
inverter variables. (b) Primary coil variables.

The RMS value of the BIWPTS’s variables during the presented dynamics are

indicated in Figure 7.12. The figure compares between these variables from both co-

simulation and analytical model results. Large transients appear during the transition for

both power level variations and switching the mode of operation. The results show

significant mismatch between the analytical and FE co-simulation platform. The reasons for

these mismatches are the actual characteristics of the magnetic material, which are neglected

161
in the analytical model. The errors between the two results are evaluated and presented in

Figure 7.13. These errors are significant during the transient period and decrease at steady-

state region. Also, it can be noticed that the errors in the different modes of operation are

not the same. For example, ∆Ipi in V2G operation is slightly larger than that in G2V

operation. The reason for this behavior is the nonlinearities in the system.

6
 Ipi (A)

4 (a)
2
0
0 0.01 0.02 0.03 0.04
4
 Ipc (A)

(b)
2
0
0 0.01 0.02 0.03 0.04
30
 V pc (V)

20
10 (c)
0
-10
0 0.01 0.02 0.03 0.04
Time (s)

Figure 7.13. Error between analytical and FE co-simulation platform results. (a) Primary
inverter current. (b) Primary coil current. (c) Primary coil voltage.

The deviations in the system variables reflect errors in the power flow analysis. The

primary inverter power (Pp), during the same operating scenario from both the analytical

162
and co-simulation, is given in Figure 7.14(a). The error between the two results is depicted

in Figure 7.14(b). As can be seen, a large and non-homogeneous error is experienced

between the two data.

Ptheo
1
Pp (kW)

Pco-sim
0
-1 (a)

0.01 0.02 0.03 0.04

200
 Pp (W)

-200 (b)

0.01 0.02 0.03 0.04

Time (sec)

Figure 7.14. Active power flow. (a) Primary power (b) Primary power error.

Different performance is expected at different power levels (loading conditions) of

BIWPTS. In order to examine that, the same BIWPTS is analyzed at lower DC voltage

levels (Vdc and Vb are 60V and 50V instead of 240V and 200V, respectively). These voltage

levels allow only 1/16 of the full-power to be transferred. The NRMSE is evaluated, based

on Equation (7-9), for all the system’s variables, during both full and light loading

conditions, and presented in Table 7.4.

163
 2 2
𝑁𝑅𝑀𝑆𝐸 = √∑𝑚 𝑚
𝑖=1(𝑓(𝑡𝑖 ) − 𝑔(𝑡𝑖 )) ⁄∑𝑖=1(𝑓(𝑡𝑖 )) (7-9)

where, f(ti) and g(ti) are the value of the variable at sampling index (i), which are calculated

from the analytical and FEA co-simulation, respectively.

Table 7.4. %NRMSE at Full and Light Loading Conditions.

Mode ipi isi ipc isc vpc vsc pp ps

Full load (Vdc=240 V, Vb=200 V) 17.8 17.57 2.3 2.3 2.64 2.64 11.4 11.2

Light load (Vdc=60 V, Vb=50 V) 14.9 14.6 1.6 1.6 2.65 2.65 10.7 10.5

Harmonics Analysis

The nonlinearities in the arrangement due to the magnetic characteristics can be

identified by performing harmonics analysis in the system. A FFT study is implemented for

the primary inverter current (ipi), during both G2V and V2G operation. The inverter voltage

and current waveforms during G2V operation with full supply capacity are indicated in

Figure 7.15(a). As can be observed, ipi is in-phase with vpi, which means that the system is

hitting the resonance and working at UPF. The magnitude of FFT analysis for ipi from both

the analytical and finite element (FE) co-simulation models are presented in Figure 7.15(b).

Only odd harmonics appear in the current waveform due to the square-wave voltages. The

co-simulation results show large fundamental current and slight increase in the third

harmonic compared with the analytical results.

164
 vtheo itheo vco-sim ico-sim

200

10xi pi (A)
(a)
v pi (V)
0

-200
9.93 9.94 9.95 9.96 9.97 9.98 9.99
Time (msec)
6
co-simulation
|FFT| of ipi

4 theoritical
2 (b)

0
0 50 100 150 200 250 300
Frequency (kHz)

Figure 7.15. FFT analysis during full G2V operation (Vdc=240V, Vb=200V, α=β=180o and
δ=-90o). (a) Primary inverter variables. (b) FFT magnitude of primary inverter current.

For better evaluation, the total harmonic distortion (THD) for all the system’s

variables is estimated, and presented in Table 7.5. The table shows the THD at both full and

light loading conditions. In general, THDs in co-simulation results are smaller than that

from the analytical results due to the larger fundamental components in the co-simulation

data. The differences diminish in the light loading conditions, since the magnetic core

becomes less saturated and offers linear characteristics. The secondary inverter current (isi)

exhibits less THD than that of ipi. Also, it can be seen that THD of the coils’ variables (Ipc,

Isc, Vpc and Vsc) are very small, which means that these variables are almost sinusoidal due

to the filtration in the compensation network. Similar results are noticed during V2G

operation, as presented in Figure 7.16 and Table 7.6.

165
 Table 7.5. %THD during G2V Operation for Full and Light Loading

Mode ipi isi ipc isc vpc vsc

Full load 62.84 34.32 0.029 0.029 0.23 0.23

Theoretical
Light load 58.61 36.21 0.03 0.03 0.23 0.23

Full load 52.34 26.34 0.035 0.035 0.25 0.25

Co-simulation
Light load 52.28 26.44 0.035 0.035 0.25 0.25

v theo itheo v co-sim ico-sim

200 (a)
10xi pi (A)
v pi (V)

-200
19.94 19.96 19.98 20
Time (msec)
6
co-simulation
|FFT| of ipi

4 theoritical
2 (b)

0
0 50 100 150 200 250 300
Frequency (kHz)
Figure 7.16. FFT analysis during full V2G operation (Vdc=240V, Vb=200V, α=β=180o and
δ=90o). (a) Primary inverter variables. (b) FFT magnitude of primary inverter current.

166
 Table 7.6. %THD during V2G Operation for Full and Light Loading

Mode ipi isi ipc isc vpc vsc

Theoretical Full load 72.82 30.96 0.029 0.029 0.23 0.23

Light load 78.77 29.48 0.029 0.029 0.23 0.23

Co-simulation Full load 54.25 25.62 0.034 0.034 0.25 0.25

Light load 54.46 25.6 0.034 0.034 0.25 0.25

Conclusion

This chapter presents an accurate platform for predicting the performance of a

bidirectional inductive wireless power transfer system during G2V and V2G services. The

proposed co-simulation considered the nonlinearities in the system due to the power

electronic converters, the power flow controllers and the magnetic characteristics of the

power pads. The platform is developed using two compatible softwares based on circuits

and finite element analysis. The system is implemented and analyzed in MATLAB

environment. The results from the co-simulation are compared with the outcomes from a

six order state-space dynamic model for 1.2 kW LCL BIWPTS, with double-D pad

structure. The comparison shows the impact of the magnetic material characteristics in

terms of errors and harmonics. The analysis indicated the effectiveness of the proposed

platform, and its ability to be extended for different coupler designs, inverter topologies

and compensation configurations.

167
 Introduction

WPT is a technology, which could set the human free from the inconveniences of

wires. It is capable of transferring the electric power through a relatively large air-gap

without physical connection. WPT are intended to deliver power efficiently from a

stationary primary source to one or more movable secondary loads by magnetic coupling.

Bidirectional operation of this technology is exceptionally encouraging, particularly in EV

ancillary services. The proper magnetic design is one of the most difficult and critical

phases in developing a BIWPTS, especially for high-power (i.e. EVs) applications. Thus,

this chapter presents detailed analysis for the choice and design of the passive elements in

a BIWPTS. A clear methodology for achieving the best design for all the passive elements

in a symmetrical LCL BIWPTS is developed. The proposed strategy depends on the

electromagnetic field computations joined with the steady-state analytical model, presented

in chapter (4). A 3D-FEM for an 8 kW polarized DD power pad is developed and optimized

intuitively. The design is accomplished to meet with the WPT2 power level and Z2 class

in J2954A standard [149]. Moreover, the choice of the PFC capacitors and the impedance

matching coils was investigated. The best combination of the components’ parameters,

including the power pads, compensation capacitors and L-filters, was investigated.

Following the proposed design considerations, a LCL BIWPTS model is developed in

MATLAB Simulink, and its outcomes are compared with the theoretical model results.

Additionally, a small-scale experimental prototype was assembled, using the same design

168
specifications with reduced DC voltage levels. The models were tested during charging

(G2V) and discharging (V2G) operation, and good correlation is observed for all of them.

The Proposed Design Strategy

As was indicated in Figure 7.1, the LCL BIWPTS contains six passive components:

Lpi, Lsi, Cp, Cs, Lpc, and Lsc; in addition to the mutual inductance (M) between the primary

and secondary sides, which is directly related to the magnetic coupling factor (k). The

performance of a BIWPTS depends mainly on the values of these elements. They are

governing the system’s power transfer capability, efficiency and cost. Thus, these elements

must be decided and designed carefully, to accomplish the desired system’s operation. A

flow-chart for a clear design strategy to configure these parameters is depicted in Figure

8.1. The algorithm describes five different phases (I-V), which need to be followed, to

accomplish the BIWPTS’s design [178], [183].

The first phase in the design, I, is to decide the operating frequency (fr), the desired

power level (Pdes), the air-gap range and the compensation configuration. The SAE J2954A

standard defines the frequency range of 81.38 kHz to 90 kHz [149]. Also, four power levels

(WPT1, WPT2, WPT3 and WPT4) were established by the same standard to support 3.7,

7.7, 11.1 and 22 kVA, respectively. In addition, the air-gap between the two sides is

characterized by three different classes (Z1, Z2 and Z3) that cover most of the ranges in

the light-duty EVs. Thus, the operating frequency, power level and air-gap length can be

decided based on this standard.

169
 start

Choose desired fr & Pdes

Phase I Choose the compensation topology

Assume reasonable value

of ko for the desired air-gap

Define the pad design

parameters (Lpc & Lsc)
Phase II based on mathematical
model optimization

Achieve the pad

magnetic design using
Phase III FE analysis

NO
k ≥ ko?

YES

Estimate Cpc & Csc

Cpc=Csc=1/(ωr2Lpc)

Phase IV Choose Lpi & Lsi for

simusoidal inverter currents

NO
ηact ≥ ηdes?

YES
Phase V
Consider these
design parameters

end

Figure 8.1. Flow-chart for the proposed design methodology.

170
 Also, in this phase, the designer needs to choose the appropriate compensation

configuration, which is decided based on the assessment and comparative analyses

presented in Chapter (6) and published in [76]. According to the study, the most appropriate

network for EVs’ applications is the LCL-topology, which shows high efficiency, simple

design and control, and current-source characteristics. Due to these reasons, the LCL

compensation topology is considered in this work. The last step in phase I is to assume a

reasonable initial value for the coupling coefficient (ko), based on the provided ranges in

the J2954A standard. In this work, WPT2 power level (7.7 kVA), Z2 class (95-165 mm

air-gap) are chosen as design targets, since there are no specific characteristics for this

category and only proposals were presented in the J2954A standard. Also, the initial

desired coupling factor is assumed as ko=0.3; and the DC-bus and the EV’s battery voltage

are chosen as Vdc=420 V and Vb=350 V.

The second phase, II, of the design is to find the wireless power pad magnetic

parameters (Lpc and Lsc), which satisfy the desired power, frequency and magnetic coupling

requirements, previously defined in phase I. These parameters are decided based on the

developed steady-state mathematical model in chapter (4) for the LCL BIWPTS. This

model is utilized to predict the effect of the self-inductances of the power pads (Lpc and Lsc)

on the system’s performance. The BIWPTS’s performance is evaluated based on the active

power transfer capability, efficiency and power factor. The relations among these

evaluation parameters and the pads’ self-inductances are described in Figure 8.2. In this

figure, the system’s primary power (Pp), efficiency (η) and primary power factor (PFp) are

analyzed with respect to the self-inductance. In this analysis, it is assumed that the two

power pads are identical (Lpc = Lsc), due to the symmetry in the system. It can be noticed

171
that the inductance value that handles about 7.8 kW (WPT2), with 92.61% efficiency and

0.955 power factor is 19 μH. Thus, this value is considered in the design.

40
Pp (kW)

X: 19
20 Y: 7.819
(a)
0
20 40 60 80 100
100
 (%)

90 X: 19
Y: 92.61
80 (b)
70
20 40 60 80 100
PFp (pu)

X: 19
0.96 Y: 0.9551
(c)
0.95
20 40 60 80 100
Lpc=Lsc (uH)

Figure 8.2. Effect of the pad self-inductance on the BIWPTS’s performance (Vdc=420 V
and Vb=350 V). (a) Primary active power. (b) System’s efficiency. (c) Primary power
factor.

Once the self-inductances of the power pads are defined, the following step is to

achieve the magnetic design that is able to provide these values. Then, this design needs to

be optimized to fulfill the desired magnetic coupling requirement. This task is usually

achieved, using a 2D or 3D FEA. More details about the power pads’ magnetic design and

optimization are presented in the next section.

172
 Magnetic Design of the Power Pads

Several planar structures of the power pad for EVs applications are presented in the

literature, such as circular or rectangular [184], DD [176], bi-polar [185], [186], DDQ [176]

and tripolar [187]. The typical structure of all these configurations contains copper coils,

ferrite material and aluminum shield, as depicted in Figure 8.3. Among these different

configurations, only the circular/rectangular and the DD shapes are recommended by the

J2954A standard. By comparing between the circular and DD structure, the DD provides

better coupling, greater tolerance, greater charging zone, and cost effectiveness [84], [85],

[169], [176].

Figure 8.3. Different structures of power pads in EVs applications. (a) Circular. (b) Double-
D. (c) Bipolar. (d) Double-DQ.

173
 Because of the advantages of the polarized DD design, it is considered in this work.

As was discussed in chapter (7), each pad consists of two or more coils, electrically

connected in parallel, to reduce the equivalent inductance; aluminum shield and magnetic

ferrite cores. The ferrite cores can be arranged to form a solid sheet, as indicated in Figure

7.4, or multiple parallel bars, as depicted in Figure 8.3. Using the solid sheet guarantees

better magnetic coupling and shielding than those from the bars. However, it results in a

higher cost and weight for pads.

8.3.1 3D Finite Element Model of DD Power Pad

Due to the complex structure of the power pad, finding accurate analytical solutions

for electromagnetic field distribution may not be conceivable. However, distinctive

numerical analysis techniques can be utilized for electromagnetic field analysis. In this

study, static 3D-FEA was considered. The commercial tool ANSYS Maxwell software is

utilized for electromagnetic field calculation and magnetic parameters’ estimation. A 3D-

FEM is built for the DD power pad, as depicted in Figure 8.4. In this model, the bars’ core

option is considered to present the most cost-effective solution; and to be able to be built

with the available material in the laboratory. This model is optimized in order to achieve

the desired requirements, defined in phases I and II. For design simplicity, the standard

ferrite core (I93x28x16) of N87 magnetic material from EPCOS is utilized, without any

special cutting. In this structure, many different variables need to be defined in the design,

such as coil length (Lc), coil width (Wc), number of ferrite bars, number of I-cores in each

bar, spacing between bars (Df), shield dimensions (Ls, Ws, Ts and Gs), and coil side (CS),

174
which is a function of number of turns (N) and the distance between two adjacent turns

(pitch), as indicated in Figure 8.5.

Figure 8.4. 3D-FEM for DD wireless coupler based on ferrite bars.

Ws

Wc Gs

(a) (c)
Ts

Df

Ls
Lc CS

(b)

Figure 8.5. Different views of DD wireless coupler. (a) Front. (b) Top. (c) Side.

175
 8.3.2 3D Finite Element Analysis and Optimization of DD Power Pad

The developed model was analyzed based on the 3D FEA for solving the static

problem, using the static-magnetic vector potential partial differential equation given in

Equation (8-1). The magnetic field distribution, though a vertical section in the middle of

the model, while the currents in the two sides are equal to 23 A, is shown in Figure 8.6.

Also, the distribution of the magnetic flux density in the ferrite bars is depicted in Figure

8.7.

∇ × (𝜈(∇ × 𝐴̅)) = 𝐽̅𝑒 (8-1)

where, ν is the magnetic reluctivity (m/H), A is the magnetic vector potential (V.s/m), and

Je is the current density vector.

Figure 8.6. Magnetic field distribution across a vertical section in the middle of the 3D
FEM of the DD power pad.

176
Figure 8.7. Magnetic flus density distribution in the ferrite bars of the DD power pad model.

The 3D static FEA was performed to evaluate the design and obtain the magnetic

parameters (k and Lpc, Lsc), then these parameters were introduced inside the mathematical

model, to assess the BIWPTS’s power (Pp) and efficiency (η) [178]. Selected results for

optimizing the magnetic design are presented in Figures 8.8, 8.9 and 8.10. In these

analyses, the perfect alignment is assumed between the two power pads. In Figure 8.8, the

coil-width (Wc) is investigated at fixed ferrite bars (six bars with four I-cores/bar). In this

case, the number of turns was set to six turns in each coil. As can be noticed, for Wc that

nears the half of the ferrite bars’ length (186 mm), expanding Wc increases Lpc, k, Pp and

η, significantly. Greater increase in Wc demonstrates a small increase in Lpc, k and η, but

Pp experiences a maximum point at Wc = 260 mm. Increasing Wc over this value leads to

a drop in the power transfer. Thus, Wc = 260 mm is considered in this design and this value

is kept fixed in the flowing analysis.

177
 30 0.4
(a)
(uH)
25 0.3

k (pu)
pc
20 0.2
L

15 0.1
200 250 300 350

6.5 98
6 97
Pp (kW)

 (%)
5.5 96
(b)
5 95
4.5 94
200 250 300 350
coil width (mm)

Figure 8.8. Effect of coil-width on the evaluation parameters (Vdc=420 V and Vb=350 V).
(a) Primary self-inductance and coupling factor. (b) Primary power and efficiency.

The coil-width is kept fixed at 260 mm, and the spacing between the ferrite bars (Df)

is studied in Figure 8.9. The figure shows that a slight increase in Df increases Lpc, k and

η, but decreases Pp. Once Df becomes more than twice the bar width (2x28 mm), all the

parameters drop dramatically. The best Df is chosen at 46 mm, at which the model provides

the highest coupling factor.

178
 19 0.27

Lpc (uH)
18.5 0.26

k (pu)
18 (a) 0.25

17.5 0.24
40 45 50 55 60 65 70

7 96.6
P (kW)

6.9 96.4

 (%)
(b)
p

6.8 96.2

6.7 96
40 45 50 55 60 65 70
ferrite distance (mm)

Figure 8.9. Effect of Df on the evaluation parameters (Vdc=420 V and Vb=350 V). (a)
Primary self-inductance and coupling factor. (b) Primary power and efficiency.

Another variable [spacing between turns (pitch)] is optimized, as described in Figure

8.10. In this case, it is assumed that the pitch is fixed for all the coils in the DD power pad.

The figure demonstrates that expanding the pitch leads to a better coupling factor and

higher power transfer, but less system’s efficiency. Thus, the value of the pitch was decided

to achieve the required power level, with reasonable efficiency and coupling factor. This

parameter was set to 11 mm. Similarly, the entire pad’s design parameters are investigated

and the final parameters are presented in Table 8.1.

179
 40 0.4
(a)
(uH)

k (pu)
L
pc 20 0.3

0 0.2
0 5 10 15 20

20 98
(b)
P (kW)

 (%)
10 97
p

0 96
0 5 10 15 20
pitch (mm)

Figure 8.10. Effect the pitch on the evaluation parameters (Vdc=420 V and Vb=350 V). (a)
Primary self-inductance and coupling factor. (b) Primary power and efficiency.

Table 8.1. Final Design Parameters of the DD Power Pad.

Parameter value Parameter Value

Wc 260 mm Ls 700 mm

Lc 500 mm Ws 600 mm

Df 46 mm Ts 10 mm

CS 70 mm Strings 6

Turns 6 Bars 4

Pitch 11 mm

180
 8.3.3 Design of LCL Compensation Configuration

In the wake of completing the magnetic design and evaluation, the following phase,

IV in Figure 8.1, is to estimate the LCL compensation topology parameters (Lpi, Lsi, Cp and

Cs). For LCL topology, the capacitors are connected in parallel, and are tuned to resonate

with the self-inductances of the wireless pads, as was indicated in Equation (4-2). In this

case, the system behaves as a parallel-resonance circuit, with high equivalent impedance

at the resonant frequency to provide current-source characteristics.

Typically, the filters’ parameters (Lpi and Lsi) are assigned based on the power

transfer capabilities. However, the BIWPTS’s performance is very sensitive to the variation

of the filter inductance with respect to the pad self-inductance, as was depicted in Figure

4.12. For a 40 kHz resonant frequency, the operating frequency matches with 40 kHz only

when Lpi=Lpc and Lsi =Lsc. At these conditions, the system shows UPF operation (zero

impedance angle), as indicated in Figure 4.12(b) [102]. A small deviation of Lpi from Lpc

leads to deviation in the resonant frequency. In this case, the operating frequency needs to

be adjusted to match with the resonant frequency, which requires more complex

controllers. The last phase, V, before considering the design, is to check the system

efficiency, using the developed mathematical model. If it achieves the desired efficiency,

then the design will be considered, otherwise different magnetic design needs to be tried,

by repeating phases II to V.

Design Verification and Results

For confirming the adequacy of the proposed design methodology, two 8 kW

identical DD power pads were built, according to the parameters shown in Table 8.1. These

181
pads were developed using litz wire Type 2, 660/38 AWG 10, and the standard ferrite core

I93x28x16 of N87 magnetic material from EPCOS. The ferrite I-cores are attached together

to form the ferrite bars. Each pad contains six bars, with four cores per bar, as shown in

Figure 8.11. Also, it contains two litz wire coils, connected electrically in parallel and

magnetically in series. The final shape of the DD power pad is indicated in Figure 8.12.

Figure 8.11. Ferrite bars of N87 material from EPCOS for one DD power pad.

182
 Litz wire
Ferrite bar

I-core

Figure 8.12. Final shape of the developed 8 kW DD power pad.

The compensation networks are built using Metallized Film capacitors and hand-

made filters. The power pads, the compensation networks and the HF VSHBCs [described

in chapter (5)] are gathered, to form a prototype for an 8 kW LCL BIWPTS, as indicated

in Figure 8.13. It comprises of two identical DD power pads, two 40 kHz VSHBCs, a firing

board to generate synchronized switching signals for the converters using the PPM

technique, a 50 V Li-ion battery module to emulate the EV battery behavior, a

programmable MAGNA power supply to emulate the DC-bus, and sensors’ board with

183
LEM voltage and current transducers and oscilloscopes. The final BIWPTS’s design

parameters are described in Table 8.2.

Oscilloscope1 Li-ion Oscilloscope2 Programmable MAGNA

battery DC power supply

L-filter Capacitor
DC supplies
Sensors
Signal board
generator

Secondary DD pad (EV side)

Primary DD pad (grid side)

Secondary
inverter
Firing board Primary inverter

Figure 8.13. Test-bed for a symmetrical LCL BIWPTS based on the DD power pads.

Table 8.2. Design Parameters of LCL BIWPTS with DD pads structure.

Parameter Value Parameter Value Parameter Value

Lpi 19.8 µH Lsi 19.6 µH Cp 0.79 µF

Rpi 40 mΩ Rsi 40 mΩ Cs 0.8 µF

Lpc 20.1 µH Lsc 19.8 µH k 0.32

Rpc 60 mΩ Rsc 60 mΩ f 40 kHz

184
 In addition, these parameters are utilized inside the Simulink model (presented in

Figure 5.24), and the steady-state analytical model [presented in chapter (4)]. The two

models are analyzed and compared for 8 kW power level, during G2V (charging)

operation, as depicted in Figure 8.14. Very good agreement can be observed between the

analytical and simulated results.

VSim . VTheo. ISim . I Theo.

V sc, 5*Isc V pc, 5*Ipc V si, 10*Isi V pi, 10*Ipi

500
0
(a)
-500
19.93 19.94 19.95 19.96 19.97 19.98 19.99 20
500
0
(b)
-500
19.93 19.94 19.95 19.96 19.97 19.98 19.99 20
500
0
-500 (c)
19.93 19.94 19.95 19.96 19.97 19.98 19.99 20
500
0
-500 (d)
19.93 19.94 19.95 19.96 19.97 19.98 19.99 20
Time (msec)

Figure 8.14. Performance of an 8 kW LCL BIWPTS, during G2V operation (α=β=180o,

δ=-90o, Vdc=420V and Vb=350V), using Simulink and analytical models. (a) Primary
inverter variables. (b) Secondary inverter variables. (c) Primary coil variables. (d)
Secondary coil variables.

185
 Moreover, the DC voltage levels (Vdc and Vb) are scaled down (1/7 time the design

values, to fit the available Li-battery module in the laboratory), and the experimental

prototype is analyzed and compared with the analytical model’s results, during both G2V

and V2G operation, as indicated in Figure 8.15 and 8.16, respectively. Also, good

agreement can be observed between the two results. In order to check the system

scalability, the power-flow is evaluated for both the large and small-scale system, during

G2V operation, as presented in Table 8.3. As can be noticed, both scales provide the same

efficiency and power factor. The ratio between the power levels is about 49, which

represents the square of the scaling factor 7. In addition, the power, efficiency and power

factor results match with the desired values, defined by the proposed design strategy.

VExp. VTheo. IExp. ITheo.

50
10Ipi

0
Vpi

(a)
-50
0.04 0.06 0.08 0.1
50
10Isi

0
Vsi

-50 (b)
0.04 0.06 0.08 0.1
100
50
5Ipc
Vpc

0
-50 (c)
0.04 0.06 0.08 0.1
50
5Isc
Vsc

0 (d)
-50
0.04 0.06 0.08 0.1
Time (msec)

Figure 8.15. Performance of the small scale LCL BIWPTS, during G2V operation
(α=β=180o, δ=-90o, Vdc=60V and Vb=50V), using experimental and analytical models. (a)
Primary inverter variables. (b) Secondary inverter variables. (c) Primary coil variables. (d)
Secondary coil variables.

186
 VExp. VTheo. IExp. ITheo.
50

10Ipi
Vpi 0 (a)
-50
0.04 0.06 0.08 0.1
50
10Isi

0
Vsi

-50 (b)
0.04 0.06 0.08 0.1
100
50
5Ipc
Vpc

0 (c)
-50
0.04 0.06 0.08 0.1
50
5Isc
Vsc

0 (d)
-50
0.04 0.06 0.08 0.1
Time (msec)

Figure 8.16. Performance of the small scale LCL BIWPTS, during V2G operation
(α=β=180o, δ=90o, Vdc=60V and Vb=50V), using experimental and analytical models. (a)
Primary inverter variables. (b) Secondary inverter variables. (c) Primary coil variables. (d)
Secondary coil variables.

Table 8.3. The Power Flow of the Large and Small Scale Models

Mode Pp (W) Ps (W) η (%) PF (pu)

Large scale G2V 8050 7424 92.2 0.978

Small scale G2V 164 151 92.2 0.978

187
 Conclusion

In this chapter, an efficient design methodology for a symmetrical LCL BIWPTS

was presented. An 8-kW polarized DD pad magnetic design was investigated, and

intuitively optimized based on finite-element analysis, using ANSYS Maxwell tool. The

proposed procedure is utilized to design a LCL BIWPTS that satisfies the standard

requirements of WPT2 power level and Z2 class in the J2954A standard. This methodology

is generic, and can be utilized for different power levels, air-gap classes, resonant

frequencies and power pad structures. Also, the choice and design of the compensation

capacitors and the impedance matching parameters were investigated. The proposed design

technique was assessed based on simulation, analytical as well as experimental analysis

and results. The results confirm the effectiveness of the proposed strategy with good

agreement between the different models’ results.

188
 Introduction

The magnetic design of the power pads is crucial for the IPT system’s performance.

Several studies for investigating the design of the DD power pad structure in EVs

applications are presented in the literature [176], [178], [188]. In these works, the coil turns

were modeled as a rectangular block of copper (single-turn and single-layer), as was

discussed in chapter (8) and indicated in Figures 8.3 and 8.4. This way of modeling does

not represent the actual coil, which contains several turns with air-gaps among them. The

single-turn single-layer (STSL) model helps to reduce the computational time and effort,

during FEA. However, it introduces some errors in the analysis due to this approximation.

These errors are negligible, as long as the turns are close enough to each other. However,

if the separations between the turns are noticeable, this approximation will be incorrect and

may lead to significant errors. Moreover, in these presented studies, the authors picked up

some design variables, and analyzed them manually, within a specific range, as was

explained in section 8.3.2. This manual design optimization technique is acceptable in the

systems having few and independent design parameters. However, in complex situations,

like IPT systems, many design parameters need to configured, and these parameters are

related to each other. Thus, applying the manual optimization in IPT systems’ design is

very difficult, and requires a wide knowledge and experience, even though finding the

optimum solution is not guaranteed.

189
 Some automatic optimization algorithms for IPT systems’ design were presented in

the literature, based on Pareto multi-objective optimization [189]–[191]. These algorithms

were solved, using genetic algorithm, as in [191], and particle swarm, as in [190]. These

evolutionary techniques are very time consuming, especially when FEA becomes a part of

the optimization process. Moreover, the simplified coil model (STSL) was considered in

these works as well.

Different from what was presented in the literature, this chapter presents an automatic

multi-objective optimization for the design parameters of the DD power pad structure. A

detailed 2D FEM for the power pad, considering the separations among the turns, is

developed. The developed model is coupled with a fast and an efficient improved Tabu

search (ITS) algorithm, for optimization purposes. In the proposed algorithm, several

objectives are evaluated, such as magnetic coupling, misalignment (horizontal, vertical and

rotational), as well as the cost of the power pads. For verification purposes, the system

coupling performance and cost are investigated, using the optimum design parameters, and

compared with three other designs that are presented in the literature. The proposed

optimized design shows the best coupling performance with moderate cost. The results

proved the validity and advantages of the proposed design optimization methodology.

A Detailed 2D FEM of the DD Power Pad

In this study, the FE modeling and analysis are achieved, using ANSYS Maxwell

software for electromagnetic field calculation, and magnetic parameters’ estimation. A

detailed 2D quasi-static electromagnetic FEM is built and analyzed in this study. The coil

190
turns are exactly modeled, to accurately investigate the effect of the separations among

them, as shown in Figure 9.1.

Copper coil
P1 P2
Ts
Wc Dc

Tf Hs
Aluminum
shield
Wf
Ferrite sheet
Ws

Figure 9.1. A 2D-FEM of the DD power pad, showing the optimization variables.

Each turn is modeled as a stranded coil domain to emulate the litz wire performance,

which is commonly used in IPT systems. In this model, the separation among the turns

(pitch) in the middle sides is assigned to a variable P1, while the pitch in the outer sides is

associated to a variable P2. This way provides more freedom and flexibility for the search

algorithm to find the optimum design solution. In addition, the dimensions of the power

pad’s components (indicated in Figure 9.1) are assigned to different variables, to facilitate

their modification and optimization. The model was analyzed using the 2D FEA for solving

the static problem, based on the static magnetic vector potential partial differential

equation, given in Equation (8-1). The dimensions of the developed 2D FEM’s components

are indicated in Table 9.1. These dimensions are kept fixed during the optimization process,

and only the separations (P1 and P2) and the ferrite bars’ variables [width (Wf) and thickness

(Tf)] vary.

191
 Table 9.1. Dimensions of the 2D FEM for the DD pad structure.

Parameter Value Parameter Value

Coil width (Wc) 190 mm Shield thickness (Ts) 5 mm

Coil diameter (Dc) 5.6 mm Shield width (Ws) 500 mm

Turns per coil (N) 6 turns Shield edge (Hs) 50 mm

The Multi-objective Optimization Problem

Regardless of the compensation topology, the power transfer between the primary to

the secondary side of an IPT system is a function of the supply characteristics (frequency

and current), secondary pad magnetic parameters (self and mutual inductances) and load

characteristics (quality factor), as given in Equation (9-1) [85].

𝑃𝑜 = 𝜔𝐼12 (𝑀2 ⁄𝐿2 )𝑄2 (9-1)

where, ω is the frequency of the primary current, I1; M is the mutual inductance; L2 is the

secondary pad self-inductance; and Q2 is the secondary circuit quality factor.

Under the assumptions of 1) implementing parallel compensation at the secondary

circuit, which is the common case in EVs applications, 2) operating the system at fixed

resonant frequency in both the primary and secondary sides, and 3) considering identical

pads on both sides (L1=L2=L), the secondary quality factor (Q2) and mutual inductance

(M) can be written as in Equation (9-2).

𝑄2 = 𝑅𝐿 ⁄𝜔𝐿
(9-2)
𝑀 = √𝐿1 𝐿2 = 𝑘𝐿

where, RL is the load equivalent resistance.

192
 By substituting Equation (9-2) into (9-1), the output power at the secondary side can

be written, as in Equation (9-3).

𝑃𝑜 = 𝐼12 𝑘 2 𝑅𝐿 (9-3)

From Equation (9-3), it can be noted that the power transfer can be increased by

increasing the supply current (I12), the load equivalent resistance (RL) and the magnetic

coupling (k2). The term I12RL is limited by the power electronic converters’ capabilities and

the system’s losses and load. The coupling factor is related to the magnetic design of the

power pad and alignment condition between the two sides of the system. An IPT system

with a higher coupling factor means that the system is able to transfer more power

efficiently. Thus, the higher coupling factor, the better pad design. Generally speaking, the

coupling factor can be increased by extending the size and weight of materials used in the

power pad. However, this design will result in an expensive and impractical IPT system,

which is not acceptable, due to the limitations in EV’s manufacturing related to space,

weight and cost. Thus, the design optimization of the power pad exhibits a tradeoff between

coupling performance and cost. Therefore, in this work, both the coupling coefficient and

system’s cost are considered as objectives in the optimization problem. These objectives

are stated as functions of the geometry parameters given in Figure 9.1.

9.3.1 Optimization Objectives

In this study, the main concern is to investigate the effect of the separations among

the adjacent turns (P1 and P2) on the system performance, and to find the best combination

of these pitches and the dimensions of ferrite sheet [width (Wf) and thickness (Tf)] in the

193
DD pad structure. This combination must achieve the highest coupling performance and

the least cost. In this case, the power pad size and number of turns are kept fixed, with the

values given in Table 9.1.

9.3.1.1 Coupling Performance with Misalignment

One of the main objectives of the optimization problem is to maximize the mutual

coupling between the system’s sides. However, the coupling performance varies based on

the alignment between the two sides. The coupling factor is maximum, when the two sides

are perfectly aligned, and decreases with misalignment. The good pad design must be able

to achieve the best coupling, during both the perfect alignment and misalignments. Thus,

in this work, the average coupling factor (kave) is considered inside the cost function. This

factor kave represents the average coupling factor of four different positions for the coupler,

which cover the different possible misalignments in the system, as shown in Figure 9.2.

The first position [Figure 9.2(a)] denotes the perfect aligned situation. The second and third

positions [Figure 9.2(b) and (c)] represent the vertical and horizontal misalignment,

respectively. The fourth position investigates the rotational misalignment case [see Figure

9.2(d)]. For each combination of the optimization variables (P1, P2, Wf and Tf), the coupling

factor is evaluated at the four positions, using FEA. By knowing the individual factors, the

average value is estimated, using Equation (9-4), and considered as an optimization

objective. The misalignment parameters, in Figure 9.2 (g, v, h and θ), are set to match with

the maximum acceptable misalignments defined in the J2954A standard [149].

𝑘𝑎𝑣𝑒 = (𝑘𝑎 + 𝑘ℎ + 𝑘𝑣 + 𝑘𝑟 )⁄4 (9-4)

194
where, ka, kh, kv and kr represnt the coupling factor, during the perfect alignmnet, horizontal

misalignment, vertical misalignment and rotational misalignment, respectively.

g g

(a) (b)

h θ
g g

(c) (d)

Figure 9.2. Different positions of misalignments. (a) Aligned. (b) Vertical misalignment.
(c) Horizontal misalignment. (d) Rotational misalignment.

9.3.1.2 Material Cost of the DD Power Pad

The total material cost of the wireless coupler is represented by the cost of three main

parts: copper windings (Cc), ferrite material (Cf) and aluminum shield (Cs). The estimation

of each of these costs in terms of the pad’s dimensions is described in Equation (9-5).

𝐶𝑓 = 𝐿𝑓 × 𝑊𝑓 × 𝑇𝑓 × 𝑐𝑓
𝐶𝑠 = 𝐿𝑠 × 𝑊𝑠 × 𝑐𝑠ℎ (9-5)
𝐶𝑐 = 4[𝑁(𝐿𝑐 × 𝑊𝑐 ) − (𝑃1 + 3𝑃2 ) ∑𝑁
𝑚=1 𝑚 ] × 𝑐𝑐

195
where, N is the number of turns per coil; Wf, Lf and Tf are the dimensions of the ferrite bars;

Ws and Ls are the dimensions of the aluminum shield; Wc and Lc are the dimensions of the

copper coil; and cf, csh and cc are the cost coefficients of ferrite sheet, aluminum shield and

copper coil, respectively.

The dimensions of the pad’s components are indicated in Figure 9.1 and 9.3.

Although only the width dimensions are included in the 2D FEM, the entire components’

dimensions are required for the cost calculation. The width information is presented in

Table 9.1, and the length data are configured as ratios of them, according to the J2954A

standard. The length of each component is set to 76% of its width (i.e. Lf=0.76 Wf, Ls=0.76

Ws and Lc=0.76 x 2Wc).

Ws
Wc Coil

Wf
Lc
Ls Lf

Ferrite sheet

Aluminum shield

Figure 9.3. Top view of DD power pad with components’ dimensions.

196
 For each combination of the optimization variables, the three types of costs are

evaluated and added together to find the total cost of the power pad, as given in Equation

(9-6). For the identical pads’ case, the total material cost of the wireless coupler, which

represents one of the optimization objectives, is double each pad’s cost.

𝐶𝑡 = 𝐶𝑓 + 𝐶𝑠ℎ + 𝐶𝑐 (9-6)

9.3.2 Multi-objective Cost Function and Constraints

The basic objectives of the optimization algorithm are to minimize the coupler’s

material cost and maximize the magnetic coupling. The multi-objective optimization

function (MOF) is stated mathematically in Equation (9-7). For simplicity, the weighted

sum method is used to evaluate the two objectives, in which two different weights (w1 and

w2) are assigned to the two objectives.

𝑀𝑖𝑛 𝑀𝑂𝐹(𝑃1 , 𝑃2 , 𝑊𝑓 , 𝑇𝑓 ) = 𝑤1 𝐶𝑡 + 𝑤2 (1 − 𝑘𝑎𝑣𝑒 ) (9-7)

The optimization constraints are defined by the limits of each optimization variable,

as given in Equation (9-8). These limits are decided according to the physical structure

limitations and the designer experience.

𝑃1 𝑚𝑖𝑛 ≤ 𝑃1 ≤ 𝑃1 𝑚𝑎𝑥
𝑃2 𝑚𝑖𝑛 ≤ 𝑃2 ≤ 𝑃2 𝑚𝑎𝑥
𝐶𝑜𝑛𝑠𝑡𝑟𝑎𝑖𝑛𝑡𝑠 = 𝑊 (9-8)
𝑓 𝑚𝑖𝑛 ≤ 𝑊𝑓 ≤ 𝑊𝑓 𝑚𝑎𝑥

{ 𝑇𝑓 ≤ 𝑇𝑓 ≤ 𝑇𝑓
𝑚𝑖𝑛 𝑚𝑎𝑥

Solving of the Optimization Problem

After formulating the multiobjective optimization problem, including defining the

197
optimization variables, cost function and constraints, the next step is to solve this problem,

using one of the search algorithms. In this work, a modified version of the Tabu search

algorithm is considered, and linked with the developed 2D FEM, to find the optimum

design parameters. The details of the proposed search algorithm and the coupling with FEA

are presented in this section.

9.4.1 The Proposed Improved Tabu Search (ITS) Algorithm

An improved version of the Tabu search (ITS) algorithm is utilized to solve the

proposed optimization problem. The proposed algorithm is mainly based on the universal

Tabu search (UTS) approach for global optimization, with continuous variables in

electromagnetics [192]. There are two searching phases in the UTS: diversification and

intensification. The task of the diversification phase is to search the objective space widely,

while the function of the intensification phase is to locate the global optimal solution

precisely. In the ITS algorithm, the sampling points are generated with the Latin Hypercube

Sampling (LHS) method [193], which provides more uniform sampling in the objective

space than the random sampling used in the UTS technique. The advantages of the

proposed ITS method are: 1) the ability to jump out of the local optimal solutions, 2) high

convergence speed, and 3) simplicity of implementation and realization. The procedure of

the proposed ITS algorithm execution is described with the flowchart shown Figure 9.4,

and explained as follows:

Step 1: The algorithm generates the initial values of its parameters and defines the range

of the variables. It also states the iteration number; the maximum iteration number; the

iteration number for diversification and intensification, respectively.

198
 Initialized

Generate initial sampling

points with LHS

Calculate Fitness value

Sort and compare the fitness value

Update best_current and best_overall

Record elements of best_current and best_overall

Diversification phase: generate new sampling points based on best_current

Intensification phase: generate new sampling points based on best_overall

N
Converge?

END

Figure 9.4. Flowchart of the proposed ITS algorithm.

Step 2: It generates initial sampling points with LHS.

Step 3: It calculates the fitness value, using FEA, sorts and compares the fitness value to

199
get the best_current and best_overall. The best_current is the best objective function value

in the current iteration, and the best_overall is the best solution up till now.

Step 4: It generates new sampling points based on the rules in different phases.

Step 5: It checks the iteration number, if it meets the maximum iteration number, then it

goes to step 6; otherwise it goes to step 3.

Step 6: Stop and end the algorithm.

9.4.2 The Link between the ITS algorithm and FEA

The proposed ITS optimization algorithm is implemented in MATLAB script, and

coupled with the 2D-FEM of the inductive coupler in Maxwell environment. This link is

achieved using Visual Basic commands, as in Figure 9.5. The ITS algorithm generates the

populations, which contain information about the optimization variables (P1, P2, Wf and

Tf). These populations are passed to Maxwell environment, which contains four different

projects, I-IV. These projects represent the four different misalignment positions, shown

in Figure 9.2. The algorithm opens project I, and modifies its parameters based on the

current population. Then, it solves the project and estimates the coupling factor (ka). This

factor is saved in a text file, then the project is closed. The same steps are repeated for the

other three projects, to evaluate and save kv, kh and kr. The coupling factors are passed to

the MATLAB code, which evaluates the average coupling, total cost and the current MOF.

This value of MOF is passed to the ITS algorithm, to sort and compare the fitness values

to get the best_current and best_overall. These steps are repeated for each population until

the optimum solution is obtained, or the maximum number of iterations is reached.

200
 Maxwell environment
MATLAB
Project I environment

Open project I Wf, Tf, P1, P2

MATLAB
environment Modify its parameters

Run project I ka

Current Calculate ka
population
Save ka
ITS algorithm
in Figure 9.4 Wf Close project I
- Estimate kave
Tf
P1 Project II - Estimate Ct
P2 kv
- Estimate MOF

Project III kh

Project IV
kr

Current
fitness

Figure 9.5. Block diagram for the link between ITS algorithm in MATLAB and FEA in
Maxwell.

Results and Discussion

The proposed off-line multi-objective design optimization was achieved for an IPT

system with two identical DD power pads. The settings and results of the optimization

process are discussed in this section. Also, the coupling performance and cost of the

wireless coupler, considering the optimization results, are investigated.

201
 9.5.1 Optimization Results

The proposed ITS algorithm is implemented in MATLAB environment and linked

with the 2D-FEM of the inductive coupler, as discussed in section 9.4. The different

misalignment parameters (g, v, h and θ), shown in Figure 9.2, are chosen to represent the

maximum acceptable misalignments in the J2954A standard for the Z2 class, as indicated

in Table 9.2. The Z2 class covers an air-gap range between the two sides of the system of

95-165 mm. Thus, the avaerage of this range is considered as the normal vertical air-gap

(g=130 mm). Also, the table shows the cost coefficients that are utilized for material cost

calcualtion. These values represent the current cost in the market.

Table 9.2. The Misalignment Parameters and the Cost Coefficients during the
Optimization.

Parameter Value Parameter Value

G 130 mm cf 0.18 $/cm3

H 100 mm cc 2.04 $/m

V 35 mm csh 333.68 $/m2

Θ 6o -------- ---------

The boundaries of the optimization variables (constraints) are indicated in Table 9.3.

These limits are chosen, according to the physical limitations of the DD structure, and the

designer experience. The lower limits of the pitches (P1min and P2min) are set to zero, to

include the case, in which the windings are perfectly attached together without separations.

202
The upper limits are chosen to be twice the conductor diameter (P1max=P2max=2Dc), in order

to avoid the overlap between the coil-sides for the same coil. The boundaries of the width

of ferrite bars are decided to cover the whole range of the power pad width (2Wc). Thus,

Wfmin<<Wc and Wfmax>>2Wc, as given in Table 9.3. Finally, the limits of the ferrite bars’

thickness (Tfmin and Tfmax) are configured by trial and error.

Table 9.3. Boundaries of the Optimization Variables.

Parameter Value Parameter Value

P1min 0 mm Wfmin 100 mm

P1max 11.2 mm Wfmax 450 mm

P2min 0 mm Tfmin 3 mm

P2max 11.2 mm Tfmax 20 mm

The setting of the ITS algorithm, during the optimization process, are indicated in

Table 9.4. These parameters are chosen based on expeience, and trial and error. The

optimization process is repeated three times to confirm the optimum solution. The

progression of the multi-objective optimization algorithm is indicated in Figure 9.6. As can

be noticed, the algorithm contains two phases: phase I (diversification) and phase II

(intensification). In phase I, the best fitness decreases significantly, due to the wide search

feature in this stage. In phase II, the algorithm searches the global optimum accurately, thus

the reduction in the best fitness is very small. The final optimum design variables are

presented in Table 9.5.

203
 Table 9.4. Settings of the ITS Optimization Algorithm’s Parameters.

Parameter Value

Number of iteration in phase I 200

Number of iteration in phase II 100

Number of populations in phase I 8

Number of populations in phase II 5

Number of initial populations 20

Number of runs 3

1.19
Phase I Phase II
1.185
1.1573
1.18
1.1573
Best Fitness

1.175
1.1573
1.17

1.1573
1.165 200 250 300

1.16

1.155
0 50 100 150 200 250 300
Iterations

Figure 9.6. Progression of ITS optimization algorithm.

204
 Table 9.5. The Optimum Design Parameters.

Parameter Value Parameter Value

P1opt 8.09574409917262 mm Wfopt 248.292970599325 mm

P2opt 4.06555885383175 mm Tfopt 5.00004272074758 mm

9.5.2 Assessment of the IPT System’s Performance

The optimum values of the separations among the turns are introduced to the FEM

to evaluate the system’s performance. The optimum system performance is compared with

three other scenarios that exist in the literature. The power pad structure for the four

different scenarios is shown in Figure 9.7. In the Optimum scenario [Figure 9.7(a)], the

variables’ values in Table 9.5 are considered. In the Minimum scenario [Figure 9.7(b)], the

separations were set to zero, which represent the case in which the turns are attached

together without separations, as in [188]. For the Maximum one [Figure 9.7(c)], the pitches

were set to the maximum limits (P1=P2=11.2 mm), which was considered in [194]. The

Different scenario [Figure 9.7(d)] was proposed in [176], in which the separation in the

middle sides was set to maximum (P1=11.2 mm), while that for the outer sides was set to

zero (P2=0 mm).

Quasi-Static magnetic analysis is performed in Maxwell for each scenario, and the

magnetic flux distribution for all the cases are indicated in Figure 9.8. The figure shows

that the Optimum scenario provides higher magnetic flux link between the two pads than

the Maximum and Different scenarios. Also, the magnetic flux is directed perfectly

between the two pads, with minimum leakage, compared with the other cases.

205
 (a)

(b)

(c)

(d)

Figure 9.7. Four different design scenarios. (a) Optimum (P1=P1opt, P2=P2opt). (b) Minimum
(P1=P2=0). (c) Maximum (P1=P2=11.2 mm). (d) Different (P1=11.2mm, P2=0).

In addition, in the Optimum case, the flux density in the magnetic ferrite material is

better distributed, with less saturation effect, than the Minimum case. The magnetic flux in

the Minimum and Different scenarios shows overlap with the aluminum shield. This

overlap badly affects the coupling performance, and increases the sensitivity of the

magnetic parameters to the variation of the air-gap and misalignments in the system. On

the other hand, no overlap can be noticed in the Optimum case.

206
 (a)

(b)

(c)

(d)

Figure 9.8. Magnetic flux distribution for all scenarios (g=130 mm and v=h=θ=0). (a)
Optimum. (b) Minimum. (c) Maximum. (d) Different.

207
 For better performance and design evaluation, the magnetic coupling factor and the

coupler’s cost are numerically estimated, and compared for all scenarios. The coupling

factor under different vertical air-gaps (g) is studied in Figure 9.9. Also, the cost of two

identical power pads in all scenarios is evaluated, and compared in Table 9.6.

0.4
Optimum
Coupling factor, k (pu)

Minimum
0.35 Maximum
Different
0.3

0.25

0.2
100 110 120 130 140 150 160
Air gap, g (mm)

Figure 9.9. Coupling factor vs. air gap length with perfect alignment (h=θ=0).

As can be noticed, the attached turns with zero separations show the worst coupling

performance, and the highest cost. Applying a fixed maximum separation among the turns

provides the least cost, and better coupling than the Minimum scenario. Considering

different pitches in the system (Optimum and Different) shows the best coupling

performance, with moderate cost. The Optimum design provides a higher coupling factor

208
and less cost than the Different one. The difference in the coupling factor between the

Optimum and Different scenarios is large for small air-gap, and diminishes with increasing

the air-gap. This nonlinear behavior is due to the effect of the aluminum shield with the

variation of the air-gap. This impact is significant in the Different scenario [Figure 9.8(d)],

and almost disappears in the Optimum case [Figure 9.8(a)].

Table 9.6. Cost of the Wireless Coupler (Two Pads) in all Scenarios.

Scenario Cost ($) Scenario Cost ($)

Optimum 384 Maximum 375.6

Minimum 391 Different 387.1

Finally, the coupling factor under different horizontal and rotational misalignments

is investigated in Figure 9.10 and Figure 9.11, respectively. Similar behavior to that in

Figure 9.9 is noticed for all scenarios. For large horizontal misalignment (h>75 mm), the

Different and Optimum scenarios show very similar coupling performance, due to the wide

spread of the middle turns, which provides long flux pipe. The Different case shows slight

improvement in the coupling factor, higher than that from the Optimum one, at very large

h (>100 mm). However, at this level of misalignment, which exceeds the maximum

acceptable limits, it will be impractical and inefficient to operate the system. In addition,

still the Optimum one provides less cost than the Different case.

209
 0.35
Optimum
Minimum

Coupling factor, k (pu)

0.3
Maximum
Different
0.25

0.2

0.15

0.1
0 20 40 60 80 100
Horizontal misalignment, h (mm)

Figure 9.10. Coupling factor vs. horizontal misalignment with g=130 mm and θ=0.

Optimum
Minimum
Coupling factor, k (pu)

0.3 Maximum
Different

0.25

0.2

0.15
0 1 2 3 4 5 6
Rotational misalignment,  (mm)

Figure 9.11. Coupling factor vs. rotational misalignment with g=130 mm and h=0.

210
 Conclusion

This chapter presents a fast and an efficient multi-objective optimization technique,

for the magnetic design of the power pads in IPT systems, for EV applications. The

algorithm is based on an improved Tabu search technique, with the link of a 2D static finite-

element analysis. The study investigates the effect of the separations among the turns, and

the ferrite martial dimensions on the IPT system’s performance and cost. The results show

that using different separations for the middle and the outer coils’ sides leads to better

coupling performance and less cost. Increasing the separation on the middle sides to be

larger than that on the outer sides provides better performance. Minimizing the separation

on the outer sides does not provide the best coupling, as was claimed in the literature. The

outcomes of the proposed design optimization provides the best coupling performance

under different misalignments in the system, and shows moderate material cost.

211
 Introduction

Accomplishing effective G2V and G2V benefits, through wireless power interface

requires accurate and quick control reaction. Power-flow controllers are fundamental in the

BIWPTSs to manage the magnitude and direction of the power-flow between the EV and

power grid. Thus, this chapter proposes a new approach of active power-flow control for a

LCL BIWPTS in EVs’ ancillary services. The controller allows an EV to autonomously

charge and discharge its battery wirelessly, during long term parking (stationary) and/or

the transient stops (quasi-dynamic). According to the literature, power-flow controllers can

be grouped in terms of the location (primary, secondary or dual) [88], the HF resonant

inverter topology [195], and the modulation technique, such as pulse-width [161], [162],

pulse-phase [163], [164], pulse-frequency modulation [163], [165], [166] and a mix

between them [163]. Diverse sorts of control hypothesis, such as classical PI [165], [166],

proportional-integral-derivative (PID) [196], and FLC [162] have been used to drive the

above modulation variables. Classic PI and PID controllers encounter large settling time,

overshoots, and oscillations, and may not resist the uncertainties and disturbances. Fuzzy

controllers are more robust, however they show high computational processing, noisy

outcome and complexity in the design [132]. In addition, the design of these controllers

should be versatile with variation of the system’s parameters, which depend on the relative

position of the power pads. Model-based predictive active power-flow (PAPF) controllers

212
have all the earmarks of being promising in these applications, in terms of simplicity and

accuracy.

This chapter presents a new two-layer PAPF controller for LCL BIWPTS in EVs’

applications. The proposed controller has the ability to manage the bidirectional power-

flow between the EV and surrounding infrastructures, such as power grid, home micro-

grid, building micro-grid, road or another vehicle. The main contributions in the proposed

controller are:

- It consists of two levels of control:

o The first is responsible for communicating with the surrounding infrastructures and

gathering information from the driver, charging station, power grid and battery

management system (BMS). Then, based on these information, it takes a decision to

charge, discharge or abstain, and estimates the rate of charge or discharge.

o The second control layer receives the reference signal from the first one, and predicts

the control parameters for the two HF resonant converters (in the primary and

secondary sides) to achieve the desired power-flow.

- It offers intelligent autonomous performance that satisfies both the EV’s owner and

power grid operator.

- It guarantees the best driving performance with a minimum energy consumption cost

and minimum side effects on the power grid.

- It estimates EV’s psychological price, as a function of its battery state of charge (SOC),

and compares it with the energy price to decide whether to charge, discharge or abstain,

and how much the charging or discharging rate.

213
- It exists on the vehicle side, to allow the autonomous operation, so the EV can be

charged and discharged, independently, without the need of a centralized controller on

houses, parking garage, public parking or traffic signal.

- The proposed controller is adaptively estimating the system’s parameters to consider

the misalignment and the environmental conditions’ impacts on the system’s

performance. The parameter estimation is achieved using only one voltage sensor.

- The second layer control is designed based on an accurate analytical modeling for the

LCL BIWPTS. It provides very fast and stable response, during both the transient and

steady-state operation, in stationary and quasi-dynamic wireless power transfer

operation.

Detailed Power-Flow Modeling

The proposed controller is mainly depends on the analytical modeling for the LCL

BIWPTS, which provides rigorous prediction for the system power-flow. This precise

prediction helps to achieve the proper control design. Thus, this section presents accurate

mathematical representations for the active and reactive power-flow in the BIWPTS,

during EVs’ charging and discharging in V2G operation. Four different power models with

different levels of accuracy are developed. The developed formulas consider the design

and control parameters’ effect on the BIWPTS’s performance. Moreover, the impact of the

losses and harmonics in the system, due to the HF resonant converters, on the power-flow

performance is considered. In addition, practical limitations for the two-way power-flow

are inspected, during different modes of operation. The conditions for achieving maximum

214
active and reactive power, and unity and zero power factor operation are investigated and

verified.

10.2.1 Analytical Models of Active and Reactive Power-Flow

The proposed power-flow models are mainly depend on the steady-state

mathematical model, which is developed for LCL BIWPTS and presented in sections 4.3

and 4.4. Based on this model, four different active and reactive power-flow equations with

different levels of precision are developed, analyzed and compared [197]. According to the

assumption that most of the power-flow controllers exist on the vehicle side, all the power-

flow analyses are achieved for the secondary side. However, the system is symmetrical and

both sides have similar performance.

10.2.1.1 Total Power-Flow Model (PAPM and QRPM)

This model represents the fundamental and harmonics power components. It is

developed using the exact Vsi and Isi formulas in Equations (4-1) and (4-4), respectively,

and substituting in Equation (10-1) for secondary power estimation.

𝑃𝐴𝑃𝑀 = ∑𝑛=1,3,5,… 𝑅𝑒𝑎𝑙(𝑉𝑠𝑖_𝑛 𝐼𝑠𝑖_𝑛 ∗ )

(10-1)
𝑄𝑅𝑃𝑀 = ∑𝑛=1,3,5,… 𝐼𝑚𝑎𝑔(𝑉𝑠𝑖_𝑛 𝐼𝑠𝑖_𝑛 ∗ )

For implementing this model, the power is estimated at each harmonic component

and added together to represent the total power. This model provides accurate estimation

for the system’s power-flow, but it needs iterative solution, and it is not appropriate for a

fast control design. Thus, simpler models are developed rely on some reasonable

assumptions.

215
 10.2.1.2 Fundamental Power-Flow Model (PFAPM and QFRPM)

This model predicts the fundamental power component only. It is derived by

neglecting the harmonic components of Isi and Vsi, to find a simple formula for the

fundamental secondary inverter current (Isi_1), as in Equation (10-2).

𝑗𝑀 𝛾 𝐿𝑝𝑖 𝐿𝑠𝑖 𝐶𝑠 𝑅𝑠𝑖 +𝑗𝛾𝜔𝑟 𝑀2 𝑅𝑠𝑖 2 𝐶𝑝 2

𝐼𝑠𝑖_1 = 𝜔 𝑉𝑝𝑖_1 + 𝑉𝑠𝑖_1 (10-2)
𝑟 𝐿𝑝𝑖 𝐿𝑠𝑖 𝐿𝑝𝑖 𝐿𝑠𝑖 (𝐿𝑠𝑐 +𝐶𝑠 𝑅𝑠𝑖 2 )

where, γ is a real number, which is very close to unity, and is given as 𝛾 =

𝐿𝑝𝑖 𝐿𝑠𝑖
(𝐿𝑝𝑖 +𝐶𝑝 𝑅𝑝𝑐 𝑅𝑝𝑖 )(𝐿𝑠𝑖 +𝐶𝑠 𝑅𝑠𝑐 𝑅𝑠𝑖 )+𝜔𝑟2 𝑀2 𝐶𝑝 𝐶𝑠 𝑅𝑝𝑖 𝑅𝑠𝑖
; Vsi_1 and Vpi_1 are the fundamental components

of Vsi and Vpi, respectively.

Therefore, exact formulas for calculating the fundamental secondary powers are

obtained, by substituting Equation (10-2) into Equation (10-1), as given in Equation (10-

3). The active and reactive power are represented by two terms: the first term characterizes

the bulk power transfer, and the second one refers to the power losses. The losses’ term is

effective for active power-flow, since it is proportional with Rsi, however, it is negligible

for reactive power-flow, which depends on Rsi2, where Rsi is typically small (<0.5 Ω).

8 𝑀𝛾𝑉𝑑𝑐 𝑉𝑏 𝛼 𝛽 𝐶𝑠 𝑅𝑠𝑖 𝑉𝑏 2 𝛽
𝑃𝐹𝐴𝑃𝑀 = (𝜋2 ) [ 𝜔 𝑠𝑖𝑛 ( 2 ) 𝑠𝑖𝑛 ( 2 ) 𝑠𝑖𝑛(𝛿) + 𝐿 2 𝑠𝑖𝑛2 ( 2 )]
𝑟 𝐿𝑝𝑖 𝐿𝑠𝑖 𝑠𝑐 +𝐶𝑠 𝑅𝑠𝑖
𝛾𝜔𝑟 𝑀2 𝑅𝑝𝑖 2 𝐶𝑝 2 𝑉𝑏 2
(10-3)
−8 𝑀𝛾𝑉𝑑𝑐 𝑉𝑏 𝛼 𝛽 2 𝛽
𝑄𝐹𝑅𝑃𝑀 = ( 𝜋2 ) [ 𝜔 𝑠𝑖𝑛 ( 2 ) 𝑠𝑖𝑛 ( 2 ) 𝑐𝑜𝑠(𝛿) + 𝐿 2 ) 𝑠𝑖𝑛 ( 2 )]
𝑟 𝐿𝑝𝑖 𝐿𝑠𝑖 𝑝𝑖 𝐿𝑠𝑖 (𝐿𝑝𝑐 +𝐶𝑝 𝑅𝑝𝑖

10.2.1.3 Approximate Fundamental Power-Flow Model (PAFAPM and QAFRPM)

It provides simple formulas for the fundamental power, by neglecting the losses

terms in Equation (10-3). This will end up with the simple power-flow model in Equation

(10-4), which matches with the presented one in Equation (4-17), for the primary power.

216
 8 𝑀𝛾 α β
𝑃𝐴𝐹𝐴𝑃𝑀 = (𝜋2 ) [𝜔 𝑉𝑑𝑐 𝑉𝑏 sin (2 ) sin (2) sin(δ)]
𝑟 𝐿𝑝𝑖 𝐿𝑠𝑖
(10-4)
8 −𝑀𝛾 α β
𝑄𝐴𝐹𝑅𝑃𝑀 = (𝜋2) [𝜔 𝑉𝑑𝑐 𝑉𝑏 sin (2 ) sin (2) cos(δ)]
𝑟 𝐿𝑝𝑖 𝐿𝑠𝑖

By analyzing the abovementioned three models, we could get accurate and fast

prediction for the active power-flow, as is discussed in section 10.2.2, however, for reactive

power prediction, better model needs to be developed, as is presented in the next section.

10.2.1.4 Approximate Power-Flow Model (QARPM)

An accurate reactive power model can be developed by considering the effect of the

third and the fifth harmonic components on the reactive power analysis. A simple equation

for the reactive power-flow at each harmonic component is derived in Equation (10-5).

This formula is evaluated for the third and fifth harmonics, and added to the approximate

fundamental reactive power model in Equation (10-4). This results in an accurate and

explicit model for the reactive power-flow, as given in Equation (10-6).

8 𝑉 2 1 𝑛𝛽
𝑄𝑠_ℎ𝑎𝑟𝑚 = (𝜋2 ) 𝜔 𝑏𝐿 3
𝑠𝑖𝑛2 ( 2 ) (10-5)
𝑟 𝑠𝑖 𝑛

−𝑀𝛾𝑉𝑑𝑐 𝑉𝑏 𝛼 𝛽
𝑠𝑖𝑛 ( 2 ) 𝑠𝑖𝑛 ( 2 ) 𝑐𝑜𝑠(𝛿)
8 𝜔𝑟 𝐿𝑝𝑖 𝐿𝑠𝑖
𝑄𝐴𝑅𝑃𝑀 = (𝜋2 ) [ ] (10-6)
𝑉 2 1 3𝛽 1 5𝛽
+ 𝜔 𝑏𝐿 (33 𝑠𝑖𝑛2 ( 2 ) + 53 𝑠𝑖𝑛2 ( 2 ))
𝑟 𝑠𝑖

10.2.2 Models Verification and Comparison

The performance of the above developed models is investigated and compared in this

section. In this analysis, the active and reactive power are analyzed with respect to the

217
variation of the control parameter δ, which is responsible for managing the power-flow in

the BIWPTS. The powers are estimated in per-unit, considering their peak values (Ps_max

and Qs_max) as bases.

Figure 10.1 shows the secondary active power using the first three models (PAPM,

PFAPM and PAFAPM). As can be noticed, the fundamental active power model (PFAPM)

provides very accurate prediction with 0.2% normalized mean square error (NMSE), with

respect to the total power model (PAPM). However, the approximate fundamental model

(PAFAPM) shows 2.44% NMSE. Moreover, the fundamental model is simple and explicit,

thus it does not require iterative solution. Thus, this model is considered for active power-

flow control design analysis.

1.5 PAPM PFAPM PAFAPM

1.19
1 1.17
1.15
0.5 V2G mode
Ps (pu)

90
0
G2V mode -1.13
-0.5
-1.15
-1
-1.17
-90
-1.5
-150 -100 -50 0 50 100 150
 (degree)

Figure 10.1. Comparison between the different active power models (α = β = 180o, and
Ps_max is the base).

218
 The performance of the reactive power models is described in Figure 10.2. The

fundamental (QFRPM) and approximate fundamental (QAFRPM) models show very similar

and ideal symmetrical operation, due to neglecting the harmonics effect. However, they

provide poor reactive power prediction, with more than 30% NMSE with respect to the

total power model (QRPM). This large mismatch makes them inappropriate for a precise

reactive power-flow analysis. On the other hand, it can be observed that the approximate

reactive power model (QARPM) is able to predict the system’s reactive power accurately,

with 6.22% NMSE. Thus, this model is more applicable for reactive power-flow analysis.

QRPM QARPM QFRPM QAFRPM

1
-0.6
0.5
Qs (pu)

-0.8
0

-0.5

-1
-150 -100 -50 0 50 100 150
 (degree)

Figure 10.2. Comparison between the different reactive power models (α = β = 180o and
Qs_max is the base).

For better numerical evaluation, the powers are estimated by the four presented

models, and compared with measured values in Table 10.1, for the LCL BIWPTS shown

219
in Figure 5.3, with the parameters in Table 5.23 at lower air-gap. The power-flow analysis

is achieved, during G2V and V2G operations with full and light-loading conditions. The

measured power is extracted from the recorded voltage and current waveforms, using FFT

to get the magnitude of each harmonic component. The results indicate that the models

PFAPM and QARPM provide the most accurate prediction after the total models, with respect

to the measured power. In addition, they are represented in simple and explicit formulas,

which do not require iterative solution.

Table 10.1. Numerical Power-Flow Evaluation using the Different Models.

Mode PAPM QRPM PFAPM QFRPM PAFAPM QAFRPM PFAPM QARPM PMEAS QMEAS

Full G2V -102.5 21.3 -102.7 0 -104.5 0 -102.7 17 -99.7 30.4

Full V2G 106.4 21.3 106.3 0 104.5 0 106.3 17 110.5 31.1

Light G2V -51.2 10.6 -51.3 0 -52.2 0 -51.3 8.5 -49.4 13.6

Light V2G 53.2 10.6 53.1 0 52.2 0 53.1 8.5 57.3 14.8

Active powers are in Watt and reactive powers are in VAR. Full means α=𝛽 =180o and light means α=𝛽
=90o.

Power-Flow Criteria

In G2V and V2G concept the EV will not only inject active power to the grid, but

also can provide reactive power support. The change among the different modes of

operation (inject P, inject Q, absorb P, and absorb Q) can be achieved, by adjusting the

control parameter (δ), as indicated in Figure 10.3. Considering the fundamental power only,

220
the power-flow performance is symmetrical. However, due to the losses and the injected

harmonics from the HF converters, the system shows asymmetrical operation, as indicated

in Figures 10.1 and 10.2. The harmonic contents increase the absorbed reactive power from

the grid, during G2V operation, and limit the ability of the EV to inject Q to the grid, during

V2G mode [197]. This impact can be noticed by observing the maximum values of Qs in

Figure 10.3 (1 and 0.6947 pu at G2V and V2G operation, respectively).

X: 90 X: 180
x 10
-3 Y: 1 Y: 1
1 
20 P0 Ps0
0.8 10
absorbing
0.6 0 Q from
Qs0 grid
0.4 Ps -1 -0.5 0 (G2V)
0.2 Qs
Ps & Qs (pu)

0
-0.2

Q0
-0.4 suppling
Q to
-0.6 grid
X: -90 (V2G)
-0.8 Y: -0.9688
X: 0
Y: -0.6947
-1
absorbing P suppling P to
from grid (G2V) grid (V2G)
-150 -100 -50 0 50 100 150
 (degree)

Figure 10.3. BIWPTS’s active and reactive power-flow criteria.

221
 On the other hand, the system’s losses introduce small asymmetric effect on the

active power-flow (1 and 0.9688 pu at V2G and G2V operation, respectively). This

asymmetry performance results in some constraints and limitations for the system power-

flow, which need to be considered, during the control system design. These limitations are

stated mathematically by four novel formulas for power-flow criteria. These formulas

describe the conditions for achieving maximum active and reactive power, and unity and

zero power factor system operation [197].

10.3.1 Active Power-Flow Criteria

During the symmetrical operation by using the approximate fundamental power

model (PAFPM), the active power is zero, when δ equals zero. However, in the real operation,

the system withdraws a small amount of active power (Ps0) to supply the losses, as

described in Figure 10.3. This amount of power is stated mathematically in Equation (10-

7). In order to force the secondary active power to be zero, the angle δ must start from a

negative value (δp0), which can be calculated from Equation (10-8).

8 𝐶𝑠 𝑅𝑠𝑖 𝑉𝑏 2 𝛽
𝑃𝑠0 = (𝜋2 ) 𝐿 2 𝑠𝑖𝑛2 ( 2 ) (10-7)
𝑠𝑐 +𝐶𝑠 𝑅𝑠𝑖

β
𝜔𝑟 𝐿𝑝𝑖 𝐿𝑠𝑖 Cs Rsi Vb sin( )
−1 2
𝛿𝑝0 = − sin ( α ) (10-8)
MγVdc (Lsc +Cs Rsi 2 ) sin( )
2

10.3.2 Reactive Power-Flow Criteria

Due to the harmonics effect, the system consumes more reactive power than that is

able to inject. This effect makes the actual system deviates from the fact that at δ=±90o, the

222
reactive power is zero, as indicated in Figure 10.3. This deviation of reactive power (Qs0)

is stated in Equation (10-9). The value of δ that achieves zero reactive power (unity power

factor) operation (δQ0) can be calculated from Equation (10-10).

𝑉 2 3𝛽 5𝛽
𝑄𝑠0 = 𝜔 𝑏𝐿 (0.03 𝑠𝑖𝑛2 ( 2 ) + 0.0065 𝑠𝑖𝑛2 ( 2 )) (10-9)
𝑟 𝑠𝑖

3𝛽 5𝛽
𝐿𝑝𝑖 𝑉𝑏 [0.037 𝑠𝑖𝑛2 ()+0.008 𝑠𝑖𝑛2 ( )]
𝛿𝑄0 = cos −1 ( 2
𝛼 𝛽
2
) (10-10)
𝑀𝛾𝑉𝑑𝑐 𝑠𝑖𝑛( ) 𝑠𝑖𝑛( )
2 2

10.3.3 Verification for the Power-Flow Criteria

It can be noticed that the power-flow criteria formulas (10-7)-(10-10) are functions

of the design and control parameters. For the system parameters presented in Table 5.23,

these criteria are evaluated at different control conditions (α and 𝛽), as shown in Figure

10.4. Figure 10.4(a) shows the active power at δ=0 or 180o (Ps0), this value does not depend

on α, and increases gradually as increasing 𝛽, showing the worst (maximum) at 𝛽 = 180o.

Figure 10.4(b) describes the criterion δQ0, which rises as increasing 𝛽, and decreases as α

increases. The reactive power at maximum active power operation (Qs0) is given in Figure

10.4(c). It shows fluctuating performance against 𝛽, due to the strong coupling between

the shape of the inverter voltage (which is controlled by 𝛽) and the harmonic contents in

the system. The maximum value of Qs0 occurs also at 𝛽 = 180o. The criterion of UPF

operation (δP0) is given in Figure 10.4(d). At low power level (𝛽 < 80o), it shows large

deviation from 90o, but at 𝛽 > 80o, δQ0 becomes closer to 90o, and the worst value appears

at 𝛽 = 180o. This deviation is getting worse as α decreases. Based on these analysis, the

223
worst criteria almost appear at the full capacity operation (α=𝛽 =180o), which can be

considered as design limitations. The worst values of these criteria are estimated for the

LCL BIWPTS and presented in Table 10.2.

o o o
 =180  =90  =45
2
(W)

1 (a)
s0
P

0
0 20 40 60 80 100 120 140 160 180
0
-1
 P0
o

(b)
-2
-3
0 20 40 60 80 100 120 140 160 180
20
Qs0 (VAR)

10 (c)

0
0 20 40 60 80 100 120 140 160 180
100
 Q0

50
o

(d)

0
0 20 40 60 80 100 120 140 160 180
 (degree)

Figure 10.4. Power flow criteria with different control parameters (Vdc=60 V).

Table 10.2. Worst Power-Flow Criteria (α=𝛽=180o) for LCL BIWPTS.

Criterion Value Criterion Value

Ps0 1.8 W Qs0 16.94 VAR

δP0 -0.985o δQ0 80.67o

224
 By applying the criteria presented in Table 10.2 on the system, four different power-

flow modes of operation are obtained: 1) maximum active power (MAP), 2) unity power

factor (UPF), 3) maximum reactive power (MRP) and 4) zero power factor operation

(ZPF). The four modes are investigated and compared in Table 10.3.

Table 10.3. Different Modes of Power-Flow (α=𝛽=180o) in BIWPTS.

Mode Direction PAPM PFAPM QRPM QARPM

G2V (δ=-90o) -102.5 -102.7 21.3 17

(1) MAP
V2G (δ=90o) 106.4 106.3 21.3 17

G2V (δ=-δQ0=-80.67o) -101.1 -101.3 4.4 0

(2) UPF
V2G (δ= δQ0=80.67o) 105 104.9 4.4 0

V2G (δ=0o) 1.9 1.8 -83.3 -87.6

(3) MRP
G2V (δ=180o) 1.9 1.8 126 121.4

V2G (δ= δP0=-0.985o) 0.14 0 -83.3 -87.5

(4) ZPF
G2V (δ=180o- δQ0=180.985o) 0.14 0 126 121.3

In MAP mode, the system shows the maximum active power-flow in both G2V and

V2G operation (-102.7 and 106.3 W, respectively), but consumes a small amount of

reactive power due to the harmonic contents (17 VAR), which means that the power factor

is less than unity. By applying δQ0 instead of 90o (UPF mode), the active power slightly

drops and the reactive power almost vanishes, which means that the UPF operation is

225
realized. In the MRP mode, the system supplies or absorbs the maximum reactive power,

with absorbing a tiny amount of active power (1.8 W) to feed the losses. This small amount

can be avoided by applying δP0 criterion, to provide zero power factor operation [ZPF

mode]. Moreover, by comparing the peak power value, during V2G and G2V operation,

the asymmetry operation can be observed. The choice between the four modes of operation

depends on the applications and the operator’s priority. For example, if the operator is

interested in the magnitude of the power only, thus modes (1) and (3) are the best choice.

However, if the interest was for the system’s power factor, then modes (2) and (4) are

preferred. Anyway, regardless of the mode of operation, these limitations must be

considered, while designing the power-flow control system, as discussed in the next

sections.

Proposed Two-Layer Predictive Active Power-Flow Controller

The proposed controller is located on the vehicle side to manage the magnitude and

direction of the power-flow between the EV and power grid. The power-flow control is

achieved by implementing the phase shift modulation technique in both the primary and

the secondary sides. The proposed predictive controller is responsible for generating the

control parameters (α, 𝛽 and δ), to accomplish the desired power-flow magnitude and

direction. It consists of two control layers, as indicated in Figure 10.5. The first control

layer is responsible for gathering data from the BMS, EV’s driver/operator desire, charging

station and power grid. Then, it decides the mode of operation [discharge (V2G), charge

(G2V) or abstain], and estimates the required rate of discharge or charge.

226
 Primary inverter Secondary inverter BMS

11 13 21 23 s2
Wireless coupler
Lpi Ipc M Isc s1 Lsi
+ I + + + Vb
pi
Cp Isi Vsi
Vpi Vpc Vsc
Cs
Vdc - - - -
14 Lpc Lsc 22 24
12

S11-S14 S21-S24
primary secondary driver
driver
S21-S24
S11-S14
secondary PPM
primary
β δ
PPM
Second layer
α PAPF control
communication α Pref
gateway Wireless Es
First layer BMS data
communication
owner’s control
desire
retail price (Ŕ) α station data
communication gateway

Figure 10.5. The proposed two-layer predictive active power-flow control.

The outcomes from this layer are passed to the second layer PAPF control. This

controller predicts the phase-shift modulation parameters (α, 𝛽 and δ), for both the primary

and secondary sides, which are responsible for generating the switching activities of the

two resonant converters (primary and secondary). The generated signals are boosted

through the driving circuits to the gate’s voltage and power level of the power

semiconductor switch.

10.4.1 Control Parameters’ Design

The proposed two-layer control is designed based on the power-flow models and

227
criteria presented in sections 10.2 and 10.3. These analytical models are utilized to predict

the control parameters (α, 𝛽 and δ), which are responsible for achieving the desired power-

flow. The parameter (δ) is utilized to provide the wanted power-flow direction and mode

of operation (V2G or G2V). A negative δ provides G2V (charging) operation, but a positive

δ allows V2G (discharging) service. The value of δ is decided according to the operator’s

priority, whether it is the power magnitude or power factor. Keeping δ=±90o will achieve

maximum active power-flow (MAP) operation, while putting δ=±δQ0 [given in Equation

(10-10)] will accomplish UPF operation, as discussed in section 10.3.

The other two control parameters (α and 𝛽) are decided to provide the desired rate of

charge or discharge (active power magnitude). These parameters are predicted based on

PFAPM in Equation (10-3). By assuming that both the HF converters will be driven by the

same parameter (i.e α=𝛽) to control the active power-flow magnitude, the two parameters

are estimated using Equation (10-11) [198].

𝛼 = 𝛽 = cos −1[1 − (2𝑃𝑟𝑒𝑓 ⁄𝑃𝑚𝑎𝑥 )]

8 𝑀𝛾𝑉𝑑𝑐 𝑉𝑏 8 𝐶𝑠 𝑅𝑠𝑖 𝑉𝑏 2 (10-11)
𝑃𝑚𝑎𝑥 = (𝜋2 ) 𝜔 𝑠𝑖𝑛(𝛿) + (𝜋2 ) 𝐿 2
𝑟 𝐿𝑝𝑖 𝐿𝑠𝑖 𝑠𝑐 +𝐶𝑠 𝑅𝑠𝑖

where, Pref is the desired active power, which is estimated during the first control layer and

passed to the second one.

10.4.2 Online Mutual Inductance Estimation

As indicated in the previous section, the control parameters are functions of the

system’s design parameters. For the controller to be able to achieve the proper prediction,

the system’s parameters need to be known. The LCL BIWPTS under consideration is

228
designed for fixed resonant frequency (ωr) operation, on both the primary and secondary

sides. Also, the compensation capacitors (Cp, Cs) are tuned to resonate with the self-

inductance of the coils (Lpc, Lsc), which are identified after the design stage, and are

relatively fixed [184]. Therefore, the compensation capacitors are fixed as well. Under

these constraints, only the mutual inductance (M) varies in light of the vehicle position and

the misalignment between the power pads. Thus, this parameter needs to be adaptively

estimated, before applying the control algorithm. In this work, a modified online mutual

inductance estimation technique has been proposed. It depends on utilizing the developed

analytical model besides one direct measurement on the vehicle side. A known value for

the primary control parameter (α) is applied to the primary converter, while the secondary

circuit is left open, by opening the ideal switches S1 and S2 in Figure 10.5. The open circuit

secondary voltage (Es) is measured, and the primary coil current at open circuit secondary

(Ipc_OCS) is calculated, using Equation (10-12). Using these two recognized values, the

mutual inductance is estimated based on Equation (10-13) [199].

𝐼𝑝𝑐_𝑂𝐶𝑆 = −𝑗𝑉𝑝𝑖_1 ⁄[𝜔𝑟 (𝐿𝑝𝑐 + 𝐶𝑝 𝑅𝑝𝑐 𝑅𝑝𝑖 )] (10-12)

𝑀 = 𝐸𝑠 ⁄(𝜔𝑟 𝐼𝑝𝑐_𝑂𝐶𝑆 ) (10-13)

10.4.3 First Control Layer

As depicted in Figure 10.5, the first control layer receives the BMS’s data, EV’s

driver desire and the open circuit voltage (Es) measurement, directly from the vehicle side.

Also, it receives information about the grid retail price (Ŕ) and the charging station, through

the wireless communication link. Based on these data, it generates the reference active

229
power (Pref) to be passed to the second control layer (PAPF). The charging and discharging

decision and rate is defined, by comparing the EV’s psychological price (R) with the grid

retail price of the wireless charging (Ŕ) [200]. The EV’s psychological price depends on

the battery’s SOC, and is defined in this work using the first-order polynomial cost

function, as indicated in Figure 10.6(a). The EV’s psychological price, in terms of the

battery SOC, is stated mathematically in Equation (10-14).

𝑅𝑚𝑖𝑛 − 𝑅𝑚𝑎𝑥 𝑅𝑚𝑎𝑥 − 𝑅𝑚𝑖𝑛
𝑅= 𝑆𝑂𝐶 + 𝑅𝑚𝑎𝑥 + 𝑆𝑂𝐶𝑚𝑖𝑛 (10-14)
𝑆𝑂𝐶𝑚𝑎𝑥 −𝑆𝑂𝐶𝑚𝑖𝑛 𝑆𝑂𝐶𝑚𝑎𝑥 −𝑆𝑂𝐶𝑚𝑖𝑛

where, SOCmin and SOCmax are the limits of the linear operating region of the EV’s battery,

which is typically 20-80%; Rmin and Rmax are the limits of EV psychological price, which

are chosen to cover the range of variation of the wireless charger retail price changes (Ŕ),

as depicted in Figure 10.6(a). The retail price for wireless charging of a road is available

for the EV’s drivers based on the current traffic information systems of Intelligent

Transportation Systems (ITSs) [200].

The charging and discharging benefit is evaluated, as ∆R=Ŕ-R. The reference power-

flow direction and rate (Pref) is assessed in terms of ∆R, as described in Figure 10.6(b) and

Equation (10-15).

𝑃𝑚𝑎𝑥,𝑑 ∆𝑅 ≥ ∆𝑅𝑚𝑎𝑥,𝑑
𝑚𝑑 ∆𝑅 + 𝑘𝑑 0 < ∆𝑅 < ∆𝑅𝑚𝑎𝑥,𝑑
𝑃𝑟𝑒𝑓 = (10-15)
𝑚𝑐 ∆𝑅 + 𝑘𝑐 ∆𝑅𝑚𝑎𝑥,𝑐 < ∆𝑅 ≤ 0
{ 𝑃𝑚𝑎𝑥,𝑐 ∆𝑅 ≤ ∆𝑅𝑚𝑎𝑥,𝑐

𝑃𝑚𝑎𝑥,𝑑 −𝑃𝑚𝑖𝑛,𝑑 𝑃𝑚𝑎𝑥,𝑐 −𝑃𝑚𝑖𝑛,𝑐

where, 𝑚𝑑 = , 𝑘𝑑 = 𝑃𝑚𝑖𝑛,𝑑 , 𝑚𝑐 = , 𝑘𝑐 = 𝑃𝑚𝑖𝑛,𝑐 .
∆𝑅𝑚𝑎𝑥,𝑑 ∆𝑅𝑚𝑎𝑥,𝑐

Pmax,c and Pmax,d are the maximum rate of charge and discharge, respectively, which

are estimated from PFAPM in Equation (10-11), by applying α=𝛽=180o. Pmin,c and Pmin,d are

230
the minimum rate of charge and discharge, respectively, which are defined based on the

EV’s battery manufactures and charging system’s limits.

charge discharge
Rmax R
∆Rmax,c Ŕ
(a)
∆Rmax,d
Rmin SOC
0 SOCmin SOCmax 100%
Pref

Pmax,d
md (b)
Pmin,d
∆Rmax,c 0
Pmin,c ∆Rmax,d ∆R

mc

Pmax,c

Figure 10.6. First layer control principle. (a) EV psychological price and grid retail price.
(b) Reference power vs. cost benefits.

Over SOCmax, the benefit is maximum (∆Rmax,d=Ŕ-Rmin), and the EV will discharge

at the maximum limit (Pmax,d), in order to avoid overcharging situation. Below SOCmin, the

EV will keep charging at the maximum capacity (Pmax,c), to help the vehicle to reach its

231
destination. In this case, the benefit is maximum with negative sign (∆Rmax,c=Ŕ-

Rmmax).When the two prices (Ŕ and R) match, i.e ∆R=0, the EV will charge at the minimum

limit (Pmin,c). When ∆R lies between 0 and ∆Rmax,d, the EV will discharge following a linear

profile with a slop md. When ∆R ranges between 0 and ∆Rmax,c, the EV will charge through

another linear profile with a slop mc.

If the EV’s driver has some restrictions, during the integration with the power grid,

the controller allows him/her to introduce these limitations, in terms of the desired final

SOC (SOCf), integration (connection) time (tc) and required operation (Mode). Then, it

calculates the required rate (Pres) to satisfy these limitations using Equation (10-16).

𝑆𝑂𝐶𝑓 . 𝑄𝑛
𝑃𝑟𝑒𝑠 = ∓ (10-16)
𝜂𝑏 . 𝑡𝑐

where, Qn is the nominal battery capacity, and ηb is the charging and discharging efficiency

[201], [202].

The proposed first control layer provides five different modes of operation:

Mode I means abstain (no interaction);

Mode II provides restricted operation by manually entering Mode, SOCf and tc;

Mode III allows charge only service, without considering the grid price;

Mode IV enables charge only, with considering the grid price; and

Mode V activates the interaction operation, in which the system will be able to charge and

discharge automatically, based on the grid price and EV’s psychological price.

The sequence of the algorithm implementation linked with the available wireless

communication system is indicated in Figure 10.7 and described as follows:

232
 start
Mode I
NO
Pref=0 interact?

A YES
Mode II
YES
restricted?

NO Mode V
NO NO enter
source YES calculate Mode, SOCf, tc
Pref=0 available? bidirectional? ∆R= Ŕ-R
A YES YES NO
NO source YES SOC<SOCf? Mode=charge? SOC>SOCf?
YES
available? ∆R≤0? NO NO
Pref=0 SOC≥100%?
YES YES
NO
A NO Pref=0 YES
NO NO source NO
calculate grid needs? grid needs?
available?
R from (10-14) A Pref=0 YES Pref=0 YES YES Pref=0
estimate YES YES
Mode III Mode IV A M & calculate δQ0 A SOC≥SOCmax? SOC≤SOCmin?
consider A
NO grid price? YES
NO Pref=0 Pref=0 NO
set set set δ=δQ0 for UPF
∆R= R ∆R= Ŕ-R estimate estimate
or δ=90o for MAP M & calculate δQ0 A A M & calculate δQ0
operation
YES NO
Pref=0 ∆R>0? estimate set δ=-δQ0 for UPF set δ=δQ0 for UPF
Pmax,d from (10-11) or δ=-90o for MAP or δ=90o for MAP
estimate & ∆Rmax,d
A operation operation
M & calculate δQ0 .
YES NO
∆R ≥ ∆Rmax,d? estimate
set δ=-δQ0 for UPF Pmax from (10-11),
estimate Pres from (10-16).
or δ=-90o for MAP assign Pref from
operation Pref=Pmax,d (10-15)
YES NO
estimate ǀPresǀ≥ǀPmaxǀ?
Pmax,c from (10-11)
& ∆Rmax,c assign assign
Pref=Pmax Pref=Pres
YES NO
∆R ≥ ∆Rmax,c?
A
estimate
assign Pref from
Pref=Pmax,c (10-15)

A
update Pref & M

Figure 10.7. The proposed first layer control algorithm.

- A vehicle enters the coverage area of the wireless communication system, and receives

a notification that there is a WPT service available in the area.

- If the vehicle wants to interact, it will send a request of service to the grid side unit,

otherwise it will ignore the notification.

- Based on the driver’s desire, EV battery SOC, grid price and the BIWPTS’s

specifications, the first control layer checks the grid resources availability, estimates the

mutual inductance, decides the mode of operation (among modes I to V), and calculates

Pref, depending on Figures 10.6 and 10.7, and using Equations (10-12)-(10-16).

233
- The first layer updates the variables M and Pref, and passes them to the second control

layer (PAPF), which predicts the phase-shift parameters (α, 𝛽 and δ), as is explained in

the next section.

10.4.4 Second Control Layer

This control layer receives the reference power from the first stage, and predicts the

modulation parameters (α, 𝛽 and δ) that direct the system to transfer this power. The

operation procedure of this control layer is explained in Figure 10.8. The control gets M,

Pmax and Pref, from the first stage, and checks the sign of Pref. If Pref=0, then it keeps the

control parameters as zero. In this case, there is no power transfer, which can be due to

many reasons, such as limited resources, the grid does not need extra power, the driver

does not want to sell energy, while the EV fully charged or ∆R>0, or the driver does not

want to interact with the grid at all [199]. On the other hand, if Pref is positive, the controller

sets δ=-δQ0 or -90o, to charge the battery; and if it is negative, the controller assigns δ=δQ0

or 90o, for discharging operation. Finally, the controller checks the charger limit (Pmax), if

Pref is exceeding of this limit, then α, 𝛽 will be set to the maximum, i.e α=𝛽=180o,

otherwise they will be estimated, using Equation (10-11). Finally, the controller updates

the new parameters, and passes δ and 𝛽 to the secondary PPM system, and sends α to the

primary PPM, through the wireless communication, as indicated in Figure 10.5. The PPM

generates the switching signals for both the primary and secondary inverters, which are

boosted through the driving circuits to the gate voltage and power level of the

semiconductor switch.

234
 start

get Pref, Pmax & M

from 1st control layer
Abstain
YES
set α=β=δ=0o Pref = 0?
NO
end
YES NO
Discharge Pref > 0? Charge

set δ=-δQ0 for UPF set δ=δQ0 for UPF

or δ=-90o for MAP or δ=90o for MAP
operation operation

YES NO
ǀPref ǀ≥ǀPmaxǀ?

predict α & β
set α=β=180o
using (10-11)

update α, β, δ

Figure 10.8. The proposed second control layer algorithm.

Control Implementation and Results

The prototype for LCL BIWPTS, shown in Figure 8.13 with the parameters in Table

8.2, is driven by the developed controller and analyzed by means of simulation and

experimental tests. The proposed control is implemented using Cmod S6 FPGA board,

shown in Figure 10.9. The implementation of the PPM inside the FPGA board for both

inverters is depicted in Figure 10.10. The figure shows that the dead-time between the

switches in the same legs are digitally implemented inside the board, using logic AND

gates. The signals from the FPGA are boosted up to the MOSFETs’ voltage and power

level, using FAN7391 ICs. The proposed PAPF controller is implemented and its

performance is compared with a PI controller, for verification purposes.

235
Figure 10.9. Cmod S6 FPGA board.

0-5 V 0-15 V
+ sin >0 Z-Δ s11 s11
α +
Two-layer controller

s12 FAN7391 s12

<0 Z-Δ To the
primary
sin >0 Z-Δ s13 s13 inverter
ωt
FAN7391
<0 Z-Δ s14 s14
Phase-pulse modulation
(PPM) Dead-time Driver
β 0-5 V 0-15 V
+
+
+
sin >0 Z -Δ s21 s21
+ Phase
δ
shift s22 FAN7391 s22
<0 Z-Δ To the
secondary
sin >0 Z-Δ s23 s23 inverter
ωt
FAN7391
<0 Z-Δ s24 s24
Phase-pulse modulation
(PPM) Dead-time Driver

Figure 10.10. Implementation of FPGA-based PPM including the dead-time and the driver.

236
 14

I (A) & V (V)

12
(a)
10
Ipc-OCS meas. Ipc-OCS est. Es meas.
8
0 0.005 0.01 0.015 0.02 0.025 0.03

5.5
5
M (uH)

4.5 (b)
measured estimated
4
3.5
0 0.005 0.01 0.015 0.02 0.025 0.03
Time (sec)

Figure 10.11. Mutual inductance estimation performance (Vdc=60V). (a) Es and Ipc_OCS. (b)
Mutual inductance.

10.5.1 Testing the Proposed Mutual Inductance Estimation

The performance of the proposed mutual inductance estimation procedure is

described in Figure 10.11. The parameter α is set to 180o, and the open circuit secondary

voltage (Es) is measured, as indicated by the green dashed line in Figure 10.11(a). The

primary coil current is calculated, using Equation (10-12), and also measured for

comparison purpose. As can be seen in Figure 10.11(a), the estimated and the measured

values show good agreement. Depend on Equation (10-13), the mutual inductance is

estimated and compared with the real actual value in Figure 10.11(b). It can be noted that

the proposed technique could reach to the correct value of M accurately, in less than 5

msec, which proves the real-time implementation of this technique.

237
 10.5.2 Verifying the Proposed Controller Performance

The proposed PAPF control is enforced and tested in this section. The LCL

BIWPTS’s performance, during the Abstain and Discharging operation, is described in

Figure 10.12. The system starts with the Abstain mode, and switches to the Discharge mode

at 0.04 sec. During the Abstain operation, all the control parameters were set to zero, and

no power transfer occurs. In the wake of applying the Discharge, δ was changed in

accordance with 90o to accomplish V2G service, while α and β were adjusted to match with

the desired power (Pref), as shown in Figure 10.12(a) and 10.12(b). At 0.08 sec, Pref exceeds

the power limits, accordingly the controller conforms α and β to 180o to provide the

maximum available power. Then, at 0.12 sec, the required power diminished, and the

control framework takes after the new value. During these transitions of power-flow, the

variation of Vsi can be seen in Figure 10.12(c) and 10.12(d).

Abstain Discharge
200
  =
o
Angle

100 (a)

0
0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
150
P (W)

100 (b)
s

50 actual reference
0
0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
50
Vsi(V)

0 (c)

-50
0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
Time (sec) (d)
50 50 50
0 0 0
-50 -50 -50
0.04 0.04 0.0802 0.1201 0.1203

Figure 10.12. PAPF control performance during Abstain and Discharge modes. (a) Control
parameters. (b) Secondary power. (c) Secondary voltage. (d) Zoomed secondary voltage.

238
 The performance of the proposed controller is compared with the classical

proportional integral (PI) control, as delineated in Figure 10.13. The figure demonstrates

the reference active power, and the actual system power-flow due to both the proposed

PAPF and PI controllers. During the Abstain mode, the PI controller couldn’t keep the

power at zero, which prompts to unwanted inverters’ switching. Additionally, it shows

noisy performance, during both the transient and steady-state response. Also, a large delay

is experienced in the PI transient response (at 0.12 sec). The proposed controller shows

more precise and faster tracking performance for the reference power.

160
140

120

100
P (W)

80
s

PI
60
reference
40 PAPF
20

0
0.02 0.04 0.06 0.08 0.1 0.12 0.14
Time (sec)

Figure 10.13. Comparison between the performance of PI and the proposed PAPF control
during Abstain and Discharge modes.

239
 The behavior of the proposed controller, during the transition from Discharge to

Charge modes, is examined in Figure 10.14. In this test, the controller changes the

framework from discharging (V2G) to charging (G2V) operation at 0.06 sec, by adjusting

δ from 90o to -90o [see Figure 10.14(a)]. As a result of δ variation, the power-flow is reversed

from positive to negative, following the new reference value. The system remains working

in the Charging mode the rest of the time with various power levels. The inverter secondary

voltage variations, due to the control actions, are indicated in Figure 10.14(c) and 10.14(d).

Discharge Charge
200
Angleo

100 (a)
  =
0
-100
0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
50
0 actual reference
P (W)

-50 (b)
s

-100
-150
0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
50
Vsi(V)

0 (c)
-50
0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
Time (sec)
(d)
50 50 50
0 0 0
-50 -50 -50
0.06 0.06 0.06 0.09 0.1 0.11 0.1399 0.14 0.14

Figure 10.14. PAPF control performance during Discharge and Charge modes. (a) Control
parameters. (b) Secondary power. (c) Secondary voltage. (d) Zoomed secondary voltage.

240
 During the entire period of operation, the controller succeeded to follow the reference

power value, and the system was capable of transferring the required power. A similar test

is applied, while utilizing the PI controller, and compared with the proposed PAPF control,

as indicated in Figure 10.15. The proposed controller demonstrates fast transient response,

with accurate steady-state attitude, compared with the PI controller.

50

0
Ps(W)

-50 PI reference PAPF

-100

-150

0.04 0.06 0.08 0.1 0.12 0.14 0.16

Time (sec)

Figure 10.15. Comparison between the performance of PI and the proposed PAPF control
during Discharge and Charge modes.

Conclusion

A two-layer vehicle-side predictive active power-flow controller for the LCL

BIWPTS in V2G applications is presented in this chapter. The controller is suitable for

241
stationary and quasi-dynamic interaction between the EV and power grid. It allows the EV’s

driver to choose among five different modes of operation. Also, it considers the grid retail

price to achieve the maximum profits for the EVs’ owners. The proposed controller is based

on the system analytical modeling, and it has the ability to predict the modulation parameters

for both the primary and secondary inverters, to achieve the desired active power-flow.

Moreover, it considers the misalignment in WPT system, by adaptively estimating the

mutual inductance before applying the proposed algorithm. The practical implementation

of the proposed controller evidenced its performance’s accuracy and fast response, during

both the transient and steady-state operation, compared to the available controller in the

literature.

242
 Introduction

Wireless charging methods present a new revolution in the EV industry [203]. Unlike

conventional plug-in EVs tethered to a charger, no power connection is needed. Two

different visions for implementing wireless charging in EVs’ applications are intensively

investigated in the literature: stationary and dynamic. In both modes, a primary coil is

buried in the ground underneath the vehicle, where in stationary mode, the energy is

transferred to a parked vehicle using a single coil, while in dynamic mode, the energy is

transferred to a moving vehicle using a series of coils. The removal of cables, autonomy

for the driver, and relatively low maintenance has improved the practicality of this

technology. Stationary system topologies were demonstrated as early as the General

Motors EV1 in 1998 through the implementation of IPT [204]. In [83], a review is

presented noting that current stationary applications are beginning to reach their maturity,

as the SAE has already established industry-wide specification guidelines in the SAE

J2954A standard [149]. Current research has focused on the optimization and design of

coupled coils, which have a profound impact on the system’s efficiency. Despite the ability

to transfer energy more efficiently using stationary WPT since it offers better alignment

between pads, it is still limited by its requirement for the EV to park in a specific position.

As wireless charging technology continues to grow, dynamic (online) WPT (DWPT)

charging has been introduced accompanied by a number of new advantages [205]. An

online EV has already been tested at the Korea Advanced Institute of Science and

243
Technology demonstrating a commercially available DWPT charging system for busses

[206]. A review of DWPT technologies was conducted in [82], identifying that it could

provide the foundation for mass market penetration of EVs regardless of the battery

technology. In 2015, a comprehensive study was conducted evaluating the feasibility of

installing DWPT charging lanes on UK roadways [207]. Through the introduction of WPT

in EVs, concerns associated with traditional plug-in EVs including charging duration and

range limitations could be alleviated.

DWPT is not limited to benefiting the EV’s driver, but can also provide enhanced

V2G services as well. In [73], a bi-directional WPT (BWPT) V2G scheme was proposed

including the potential inclusion of mobile EVs. The inclusion of DWPT added flexibility

to V2G schemes but despite a number of advantages, several high level challenges exist in

its deployment where cost becomes the center of attention [208]. Charging EVs consumes

a large amount of energy and with the inclusion of mobile loads at this magnitude, the

economic toll is unclear, which has incited feasibility studies on this topic [209]. In [200],

the authors quantified some of these issues through the development of an analytical model,

predicting variations in the locational marginal price of energy. However, massive initial

cost associated with constructing a robust embedded coil network along the road was a

serious concern. Furthermore, reliable, high speed bi-directional communications are

required to handle the control. However, this is the most practical solution to support

highway driving.

For city driving, an intermediate solution could provide a balance between the

infrastructure costs while still enabling a majority of the same advantages dynamic

charging has to offer. For city driving, an alternative, such as a quasi (semi)-DWPT

244
(QDWPT) system, could provide a balance between the advantages seen in the stationary

versus a dynamic system. Using QDWPT, an EV could charge during transient stops on

the city roads, such as bus stops for electric buses and traffic signals for EVs. Little work

has been presented in the literature that analyzes the QDWPT concept. In [97], the effect

of implementing QDWPT over one driving cycle with different profiles was briefly

studied. To the author’s knowledge, detailed modeling with comprehensive performance

and feasibility analysis based on the J2954A standardized power levels have not yet been

studied for a QDWPT system.

In this chapter, the feasibility of implementing the QDWPT system at traffic signals

is explored. The comprehensive charging and EV system architecture has been modeled

and exercised, through the integration of multiple driving scenarios. The novelties of this

work are as follows: 1) modeling analysis has been conducted for the entire wirelessly

connected EV, including BIWPTS, EV’s battery, and traction system with experimental

verification, 2) feasibility assessments in implementing a QDWPT system at traffic signals

is investigated under three distinct WPT scenarios: fixed power charging, variable power

charging and fixed power charging and discharging , 3) a comparative analysis over the

maximum driving range and duration per cycle has been examined, 4) quantification of the

additional distance gained over all charging scenarios for each consumed kWh is

calculated, and 5) the effect of coil misalignment in the WPT system over the driving

performance is investigated.

Quasi-Dynamic BIWPTS Description

The implementation of a QDWPT system at traffic signals could provide a promising

245
solution for EV charging. This concept can not only take advantage of dynamic charging

features, but would also significantly reduce the cost of infrastructure, simplify control

complexity, and potentially increase the transfer efficiency by enabling better alignment.

As shown in Figure 11.1, the primary wireless string of pads are placed beneath the

pavement in each travel lane at each direction of the intersection and are depicted in blue.

Each pad can be driven by an independent power converter or one converter can be utilized

to drive a few of them, while controlling the current in each pad. Thus, the primary pads

can be selectively excited based on the EV’s position, such that the energized pads are

covered by the vehicle [82]. To determine the optimal number of WPT coils to support

each lane, a traffic-flow analysis can be conducted to define the minimum coverage

distance. In this system, it is assumed that over the course of a full traffic light cycle across

all directions, wireless coils are available for all stopped traffic (under a red signal). While

stopped, authentication with a charging controller can activate either V2G or G2V services.

Conversely, the wireless coils fixed in directions with a green signal will have their coils

de-energized. Apart from the upfront installation cost to the system, BIWPTSs’ activation

can be coordinated with existing traffic light controllers and installed in combination with

common inductive loops used for traffic detection [210].

The integration between EVs and the surrounding infrastructures at traffic signal can

be represented by the DC-bus configuration shown in Figure 1.1. In this structure, the AC

grid connection is established using a bi-directional grid-tied converter responsible for

regulating a common DC-bus voltage. Each EV at the traffic signal is then connected to

the DC-bus through its own BIWPTS to facilitate charging or discharging operations. As

an added benefit, the introduction of a localized DC-bus at the traffic signal would help

246
ease the integration of RESs, such as PV generations, as shown in Figures 1.1 and 11.1. In

this case, the PV panels can be distributed along the road sides, and coupled with the DC-

bus through a unidirectional DC-DC converter. In addition, external DC loads in and

around the traffic signal area can be connected to the same DC-bus through another DC-

DC converter. Through the inclusion of renewable energy, the bulk load demand can

reduce grid stress when EVs are in G2V mode, while in V2G mode, the EV battery can aid

in smoothening the PV generation.

Figure 11.1. The proposed implementation of Quasi-dynamic WPT system at traffic signal.

247
 Modeling of Wirelessly Connected Light-Duty EV

A block diagram of the wirelessly connected EV configuration is indicated in Figure

11.2 [92]. The structure consists of three main parts: a BIWPTS, EV’s battery and EV’s

drive system. The system has two isolated sides: the grid and vehicle. The two sides are

talking to each other through radio communication. The power-flow between these sides

is managed by the secondary controller. The modeling of each part in the system is

described in this section.

Primary (grid) Secondary (vehicle)

side Wireless coupler with tuning capacitors side
DC bus Mechanical system
HF inverter HF inverter EV battery EV drive system

M PE Pm=τm ωm
Motor

ηm=f(τm, ωm)
Primary Secondary BMS data
controller controller

Radio
communications BIWPTS

Figure 11.2. Block diagram of a wirelessly connected EV through the BIWPTS.

11.3.1 Lithium-Ion EV’s Battery Model

Present-day EVs are subjected to heavy power and energy demands, which are not

limited to high transient discharge currents, but also sporadic regenerative braking charge

currents. Furthermore, with the inclusion of QDWPT, a battery model accounting for

multiple time-constants is needed to accurately depict the dynamic response of the battery

248
system. Simulation of the EV requires an advanced battery model capable of depicting

precise SOC, I-V characteristics, and accurate dynamic behavior. In this study, a 21 Ah

lithium-ion base module has been utilized to represent the EV’s battery pack, as depicted

in Figure 11.3(a). This module contains 14 PL8048168 cells in series at a nominal voltage

of 51.8 V, as shown in Figure 11.3(b) [211]. In this section, a model is developed for

simulation purposes, where seven modules are placed in series and three in parallel to reach

a standard EV pack voltage and capacity of 362.6 V and 63 Ah, respectively.

(a) (b)

Figure 11.3. Lithium-ion test battery. (a) Battery Module. (b) PL8048168 battery cell.

In [212], a procedure for obtaining a 3rd order dynamic battery model for EV

simulations was presented, where a series of standardized charge and discharge current

pulses are administered throughout the full SOC range. High charge and discharge C-rates

are utilized to push the battery out of equilibrium highlighting its dynamic response. To

differentiate between the voltage recovery governed by multiple time constants, test pulses

249
are administered over second, minute, and hour ranges. An analysis of the voltage deviation

and an exponential curve fit of the voltage recovery extracts the ohmic resistance 𝑅0 and

the associated impulse response components. The driving profile and quasi-dynamic

wireless charging simulations fall within the second and minute response periods, thus the

hour time constant was omitted, reducing the model to 2nd order, as shown in Figure 11.4.

To obtain the open circuit voltage (OCV) 𝑉𝑜𝑐 , long rest periods were featured between each

SOC step, during the charging and discharging cycles. Following a comprehensive test,

final measurements for each component were plotted over the full SOC range and curve

fitted, as shown in Figure 11.5.

Rsec Rmin
R0 Ib
Calculate
+ PE Electric Motor &
Ib,ref X Regenerative
Csec Cmin
Vb Braking Power
Voc - ÷ Vb v(t)
Driving Profile
30

Vb
Speed (mph)

20

Calculate Voc, R0, Csec, Rsec, Cmin, Rmin 10

Ib 0
0 100 200 300
Duration (s)
400 500 600

Figure 11.4. 2nd order dynamic battery equivalent circuit model with EV traction system.

The OCV (𝑉𝑜𝑐 ) and both minutes (𝑅𝑚𝑖𝑛 and 𝐶𝑚𝑖𝑛 ) and seconds (𝑅𝑠𝑒𝑐 and 𝐶𝑠𝑒𝑐 )

impulse response components produce different values dependent on whether the battery

is in charging or discharging mode. The charging experimental result and best curve fits

250
are shown in blue and discharging in red. 𝑅0 does not significantly vary between charging

and discharging, thus the dataset was combined. The closest fit is consistent with a 4th order

polynomial given in Equation (11-1).

𝑅0 (𝑆𝑂𝐶) = 𝑎1 𝑆𝑂𝐶 4 + 𝑎2 𝑆𝑂𝐶 3 + 𝑎3 𝑆𝑂𝐶 2 + 𝑎4 𝑆𝑂𝐶 + 𝑎5 (11-1)

where, a1-a5 are the R0 equation’s coefficients, which are the same for both charging and

discharging modes.

Fit (ch.) Exp. (ch.) Fit (disch.) Exp. (disch.)

60
0.24 Fit
55 (a)
(V)

R ( ) (b)
Exp.
oc

0
V

50 0.22

0 50 100 0 50 100
0.03
40 (d)
( )

(F)

(c)
0.02
sec
sec

20
C
R

0.01
0 50 100 0 50 100

2000
0.12
( )

(f)
(F)

(e)
0.1 1500
min
min

0.08
C
R

0.06 1000
0 50 100 0 50 100
SoC (%) SoC (%)

Figure 11.5. Curve fitted battery model coefficients. (a) OCV (Voc). (b) R0. (c) Rsec. (d)
Csec. (e) Rmin. (f) Cmin.

251
 For the OCV, the SOC to voltage relationship complexity warrants the need of a 5 th

order polynomial, as given in Equation (11-2).

𝑉𝑜𝑐 (𝑆𝑂𝐶) = 𝑎6 𝑆𝑂𝐶 5 + 𝑎7 𝑆𝑂𝐶 4 + 𝑎8 𝑆𝑂𝐶 3 + 𝑎9 𝑆𝑂𝐶 2 + 𝑎10 𝑆𝑂𝐶 + 𝑎11 (11-2)

where, 𝑎6 -𝑎11 represent the Voc’s equation’s coefficients, which have different values in

both the charging and discharging operation.

The curve fits of the remaining impulse response parameters over the second and

minute time spans are accomplished through the formulas given in Equations (11-3)-(11-

6).

𝑅𝑠𝑒𝑐 (𝑆𝑂𝐶) = 𝑎12 𝑆𝑂𝐶 3 + 𝑎13 𝑆𝑂𝐶 2 + 𝑎14 𝑆𝑂𝐶 + 𝑎15 (11-3)

𝐶sec (𝑆𝑂𝐶) = 𝑎16 𝑆𝑂𝐶 3 + 𝑎17 𝑆𝑂𝐶 2 + 𝑎18 𝑆𝑂𝐶 + 𝑎19 (11-4)

𝑅min (𝑆𝑂𝐶) = 𝑎20 𝑆𝑂𝐶 3 + 𝑎21 𝑆𝑂𝐶 2 + 𝑎22 𝑆𝑂𝐶 + 𝑎23 (11-5)

𝐶min (𝑆𝑂𝐶) = 𝑎24 𝑒 𝑎25 𝑆𝑂𝐶 + 𝑎26 𝑒 𝑎27 𝑆𝑂𝐶 (11-6)

where, 𝑎12 -𝑎27 represent the parameters’ coefficients, which are different between

charging and discharging operation.

The model’s coefficients (a1-a27) are extracted by curve fitting the measured data,

during both the charging and discharging operation, as indicated in Figure 11.5. The

extracted coefficients for the chosen li-ion battery module are presented in Table 11.1. The

battery module’s model shown in Figure 11.4, is implemented in MATLAB Simulink,

where dynamic values for each of the components are passed to variable resistances,

capacitances, and a controlled voltage-source. The SOC input to each is calculated through

252
the combination of the traditional current integration technique and an adjustment factor

for the discharge rate 𝑓𝑟𝑎𝑡𝑒 , as described in Equation (11-7). This factor accounts for the

adjusted capacity, as a result of the internal losses observed at high discharge currents.

Table 11.1. Charging and Discharging Coefficients of the Battery Module’s Analytical
Model.

Coefficient Charge Discharge Coefficient Charge Discharge

𝑎1 0.29500 𝑎15 0.01895 0.02218

𝑎2 -0.5399 𝑎16 -69.94 63.22

𝑎3 0.35810 𝑎17 121 -113.6

𝑎4 -0.09226 𝑎18 -81.3 66

𝑎5 0.22350 𝑎19 40.12 15.64

𝑎6 78.99 70.98 𝑎20 0.3397 -0.1734

𝑎7 -237.5 -202.2 𝑎21 -0.4096 0.3648

𝑎8 266.3 218.2 𝑎22 0.1516 -0.1666

𝑎9 -133.2 -107.3 𝑎23 0.05377 0.09134

𝑎10 34.03 27.89 𝑎24 2.184x10-6 720.8

𝑎11 49.24 49.36 𝑎25 19.25 0.6925

𝑎12 0.01715 -0.04911 𝑎26 928.7 419.6

𝑎13 0.01964 0.09523 𝑎27 0.5496 -2.012

𝑎14 -0.0251 -0.05824 - - -

253
 𝑡 𝑓𝑟𝑎𝑡𝑒 [𝐼𝑏 (𝜏)]
𝑆𝑂𝐶(𝑡) = 𝑆𝑂𝐶(𝑡0 ) − ∫0 𝑑𝜏 (11-7)
3600𝐶𝑏

where, Cb is the nominal battery capacity (Ah).

The final terminal voltage at the battery (𝑉𝑏 ) is estimated as in Equation (11-8).

−𝑡 −𝑡
𝑉𝑏 = 𝑉𝑜𝑐 ± 𝐼𝑏 (𝑡) [𝑅0 + 𝑅𝑠𝑒𝑐 (1 − 𝑒 𝜏𝑠𝑒𝑐 ) + 𝑅𝑚𝑖𝑛 (1 − 𝑒 𝜏𝑚𝑖𝑛 )] (11-8)

where, 𝜏𝑠𝑒𝑐 = 𝑅𝑠𝑒𝑐 𝐶𝑠𝑒𝑐 , is the seconds time-constant, and 𝜏𝑚𝑖𝑛 = 𝑅𝑚𝑖𝑛 𝐶𝑚𝑖𝑛 , is the

minutes time-constant.

To verify the accuracy of the model, short (1.5 s) and long (6 min) charge and

discharge current pulses were applied and compared at four different SOC levels, where

the results are shown in Figure 11.6. The top plots depict the voltage response during and

following discharge current pulses of 0.75C, where the measured values at multiple SOC

levels are shown in black solid lines compared to the simulation in colored dotted lines.

The model reveals a close match to the experimental values, where the most variation is

observed when charging from a low SOC or discharging from a high SOC. In the case of

the EV, the practical SOC operating range is between 20-80% SOC to preserve the lifespan,

thus extremely low and high SOC levels are avoided in this simulation.

Using the single module model as a reference, the final configuration is expanded to

the EV testing level placing seven models in series and three in parallel. Two versions were

developed for simulation: one for conducting in real-time and a second time-scaled version

(0.001x) to assist in reducing processing times for longer driving simulations. The time-

scaled version includes a scaling of the capacitance to scale the time constants 𝜏𝑠𝑒𝑐 , 𝜏𝑚𝑖𝑛 ,

as well as a revised time base for the SOC calculation.

254
 56
56 (a) 90% SOC 90% SOC

54 67% SOC
54 67% SOC
Vb (V)

46% SOC 45% SOC

52 52
24% SOC 17% SOC

50 50 (b)

48 48
0 500 1000 0 50 100 150 200
58 77% SOC
56
84% SOC 61% SOC

54 47% SOC
56 65% SOC
Vb (V)

51% SOC
52 22% SOC
77%
54 (c) 61% S
22% SOC
50
47% SOC
(d)
48 22% S
52
0 500 1000 0 50 100 150 200
Time (s) Time (s)

Figure 11.6. Battery model verification tests. (a) Long discharge. (b) Short discharge. (c)
Long charge. (d) Short charge.

11.3.2 EV’s Drive System Model

In this section, the specifications of the EV under test are discussed, as well as the

selected driving cycles and powertrain model used to calculate the power exchange in the

EV’s battery system. The EV modeled in this study aligns to provisions published by the

US National Highway Traffic Safety Administration (NHTSA) classified as a medium

passenger car similar to the small-sized sedan with a curb weight of 1680 kg.

255
 To evaluate the strength of the proposed EV BIWPT architecture, the Federal Test

Procedure (FTP-72) dynamometer driving profile was selected for use in simulation. FTP-

72, sometimes referred to as LA4, represents a typical city driving test conditions [213]. A

7.5 mile commute is run over an approximately 22 minute period, at an average speed of

19.6 mi/h and is shown in Figure 11.7(a). The FTP-72 consists of two phases: 1) 505 sec

“cold start cycle” taking the vehicle up to a high speed (56.7 mi/h), and 2) an 867 sec

“transient” phase representing stop-and-go city driving. In the case of the EV, phase 1

presents the greatest challenge tests on the battery pack in terms of power and energy

output, as the high speed driving portion reduces the availability of quasi-dynamic WPT.

Phase 2 subjects the EV to frequent stop-and-go at traffic signals, where WPT will be

initiated. The FTP-72 speed profile is then passed to the EV powertrain model to produce

the power profile.

The EV under test is modeled to include both drive power applied to the motor, as

well as account for regenerative braking recovery power. First, the resistance force 𝐹𝑣 of

the EV at speed 𝑣(𝑡) can be calculated by the summation of aerodynamic drag, rolling

resistance, and grading resistance at angle 𝜗, as given in Equation (11-9) [214].

1 2
𝐹𝑣 (𝑣(𝑡)) = 𝜌 𝑐 𝐴 𝑣(𝑡) + 𝐹𝑅 (𝑣(𝑡)) + 𝑔𝑚𝑡 sin 𝜗 (11-9)
2 𝑎 𝐷 𝑓

where, 𝜌𝑎 , 𝑐𝐷 , 𝐴𝑓 , and 𝑚𝑡 represent the air density (1.205 kg/m³), drag coefficient (0.32),

frontal EV area (2.31 m²), and vehicle mass, respectively.

The rolling resistance function [𝐹𝑅 (𝑣(𝑡))] is derived from a fifth order polynomial

function, which can be found in detail with a complete list of its associated coefficients in

[214]. To calculate the motor torque and speed, the wheel resistance and dynamic torque

256
for acceleration are passed through the gearbox 𝐺𝑟 (6.45), as given in Equation (11-10).

𝑟𝑤ℎ 𝐹𝑣 (𝑡) 𝜃𝑣 𝑑𝑣(𝑡)

𝜏𝑚 (𝑡) = +𝐺
𝐺𝑟 𝑟 𝑟𝑤ℎ 𝑑𝑡
𝑣(𝑡)
(11-10)
𝜔𝑚 (𝑡) = 𝑟
𝑤ℎ 𝐺𝑟

where, 𝑟𝑤ℎ and 𝜃𝑣 represent the radius of the wheels (0.29 m) and total vehicle inertia (145

kg·m²), respectively.

The resulting bi-directional battery power (PE) (see Figure 11.2) is then calculated as

in Equation (11-11).

𝜏 𝜔 /𝜂 (𝑡), 𝜏𝑚 (𝑡) ≥ 0
𝑃𝐸 (𝑡) = { 𝑚 𝑚 𝑚 (11-11)
𝜏𝑚 𝜔𝑚 𝜂𝑚 (𝑡), otherwise

where, 0 < 𝜂𝑚 (𝑡) < 1 is the motor-inverter efficiency, which is a function of the motor

speed and torque interpolated from [214].

60
speed (mph)

40
(a)
20

0
0 250 500 750 1000 1250
20 Regenerated power
PE (kW)

0 (b)

-20
Motoring power
-40
0 250 500 750 1000 1250
Time (sec)

Figure 11.7. FTP-72 Driving profile. (a) Speed. (b) Electric power.

257
 The FTP-72 Driving profile is passed through the traction system’s model given in

Equations (11-9)-(11-11) to generate the resulting motoring and regenerated power profile

indicated in Figure 11.7(b). This power profile is divided by the battery voltage (Vb) to

estimate the reference battery current (Ib,ref), as depicted in Figure 11.4. The reference

battery current is utilized as a control signal to manage a controlled current-source that

emulates the EV’s traction system.

11.3.3 Model Verification

The Simulink model of an LCL BIWPTS presented in Figure 5.25 is utilized and

linked with the battery and the drive system models presented above, to form a

comprehensive simulation platform for the wirelessly connected EV shown in Figure 11.2.

The small-scale LCL BIWPTS prototype shown in Figure 5.23 is utilized to verify the

proposed simulation platform. The prototype contains one module of the 51.8 V 21 Ah

lithium-ion battery pack to emulate the EV’s behavior, which is the same module that is

considered in the model. The system is analyzed in both charging (G2V) and discharging

(V2G) operation and compared with the simulated model where the results are presented

in Figures 5.25-5.27.

The experimental and simulated models of the BIWPTS are designed and analyzed

for 40 kHz resonant frequency, which is different from the defined frequency in the SAE

J2954A standard (81.38-90 kHz) [149]. However, this study focused on the impact of

implementing a QDWPT system on the driving performance in terms of driving miles and

hours. This performance is affected only by the power level of charging and discharging.

As long as the WPT system was able to transfer the required power level, at any frequency,

258
these results will not change. A different resonant frequency for the same power level leads

to different design parameters. In order to clarify this point, a simulation-based analysis is

conducted. The analysis compares between the performances of two LCL BIWPTSs with

two different resonance frequencies: 40 kHz and 85 kHz. The two systems are designed to

supply the same power level (11.1 kW). As can be noticed in Table 11.2, the two systems

have different design parameters, in order to be able to transfer the same amount of power.

Table 11.2. Design Parameters of an 11.1 kW LCL BIWPTS at 40 and 85 kHz.

40 kHz design 85 kHz design

Parameter Value Parameter Value

Lpc=Lpi=Lsi=Lsc 25.5 µH Lpc=Lpi=Lsi=Lsc 12 µH

Rpc=Rpi=Rsi=Rsc 30 mΩ Rpc=Rpi=Rsi=Rsc 20 mΩ

Cp=Cs 0.62 µF Cp=Cs 0.292 µF

The higher frequency system (85 kHz) provides simpler passive components’ design

(lower pad self-inductances, filter inductances and compensation capacitors). The

performance of both designs (40 kHz and 85 kHz) is analyzed during G2V operation and

compared in Figures 11.8 and 11.9, respectively. As can be observed, the two systems show

very similar voltage and current waveforms. Thus, it can be concluded that using the 40

kHz system’s design to evaluate the simulation platform is applicable since the amount of

power-transfer is only matter.

259
 V I

10*Ipi(A)
500

V pi(V)
0
-500
19.94 19.96 19.98 20

10*Isi(A)
500
V si(V) 0
-500
19.94 19.96 19.98 20
5*Ipc(A)

500
V pc(V)

0
-500
19.94 19.96 19.98 20
5*Isc(A)

500
V sc(V)

0
-500
19.94 19.96 19.98 20
Time (msec)

Figure 11.8. BIWPTS’s performance at 40 kHz resonant frequency, during G2V operation
(α=β=180o, δ=-90o).

V I
10*Ipi(A)

500
V pi(V)

0
-500
19.94 19.96 19.98 20
10*Isi(A)

500
V si(V)

0
-500
19.94 19.96 19.98 20
5*Ipc(A)

500
V pc(V)

0
-500
19.94 19.96 19.98 20
5*Isc(A)

500
V sc(V)

0
-500
19.94 19.96 19.98 20
Time (msec)

Figure 11.9. BIWPTS’s performance at 85 kHz resonant frequency, during G2V operation
(α=β=180o, δ=-90o).

260
 Feasibility Analysis of Implementing QDWPT Systems at Traffic Signals

After modeling and verification of the small scale prototype of the wirelessly

connected EV, a large scale model has been used to evaluate the effectiveness of

implementing the QDWPT system at traffic signals. This implementation will allow EVs

to charge and discharge during the stop time. The feasibility of this project from the driving

performance and the consumed energy point of view is presented in this section. The FTP-

72 driving profile is considered in this analysis. Three different charging scenarios have

been investigated: fixed power charging, variable power charging, and fixed power

charging and discharging operation.

11.4.1 Fixed Power Charging Scenario

In this scenario, it is assumed that the EV charging will start automatically once the

vehicle stops at the traffic signal. The charging power is kept fixed by the power-flow

controller of the BIWPTS. The control system incrementally adjusts the battery current to

maintain the target power, while keeping the system power factor very close to unity due

to the resonance operation. Four different standard charging levels are analyzed:

WPT1=3.7 kVA, WPT2=7.7 kVA, WPT3=11.1 kVA, and WPT4=22 kVA, based on the

SAE J2954A international standard. The whole driving performance during the WPT2

charging level is indicated in Figure 11.10. As can be noticed, the charging operation is

initiated during the stop time of the vehicle only (zero speed period). In this level, the

charging power is about 7.7 kW [Figure 11.10(c)]. The behavior of the EV’s battery SOC

during the entire driving profile is indicated in Figure 11.10(d). The EV starts with 80%

initial SOC (SOCi), which decreases with driving. Due to the charging energy during the

261
stops, the reduction in the SOC is less than what is expected, as will be indicated later. The

EV’s battery current, including the motoring, regenerated and charging current is shown in

Figure 11.10(e). Both the charging and the regenerative currents are negative, while the

motoring current is positive.

A second test is analyzed for a fixed charging scenario with WPT4 (22 kW) power

transfer, and the results are indicated in Figure 11.11. The figure shows the speed profile,

battery voltage, WPT charger current, WPT charger power and SOC, over the entire FTP-

72 driving cycle. Also, a zoom of all fields, during a 25-second traffic intersection stop

occurring between 1005 and 1030 seconds, is depicted in the same figure. It can be noticed

that the battery voltage steadily rising, while the current is being incrementally reduced to

maintain a constant power charging level of 22 kW. Also, the figure shows that after a

driving cycle, the reduction in the SOC is less than the case of WPT2 level in Figure 11.10.
speed (mph)

60
40
20 (a)
0
0 200 400 600 800 1000 1200
20
P (kW)

0
-20 (b)
E

-40
0 200 400 600 800 1000 1200
10
Pc (kW)

5 (c)
0
0 200 400 600 800 1000 1200
SOC (%)

80
75 (d)
0 200 400 600 800 1000 1200
50
0
I (A)

-50 (e)
-100
b

-150
0 200 400 600 800 1000 1200
Time (sec)

Figure 11.10. Driving performance under fixed power charging WPT2 (SOCi =80%). (a)
Speed. (b) Vehicle power. (c) Charger power. (d) SOC. (e) Battery current.

262
 60

speed (mph)
40
20

speed (mph)
2
0 1
0 200 400 600 800 1000 1200 0
450 -1
1005 1010 1015 1020 1025 1030
Vb (V)

400 422
350

Vb (V)
421

300 420
0 200 400 600 800 1000 1200 419
1005 1010 1015 1020 1025 1030
60 52.6
Ic (A)

40

I (A)
52.4
20

c
52.2
0
0 200 400 600 800 1000 1200 1005 1010 1015 1020 1025 1030
22.2
30

P (kW)
Pc (kW)

20 22

c
10
21.8
0 1005 1010 1015 1020 1025 1030
0 200 400 600 800 1000 1200 76.5

SOC (%)
76
SOC (%)

80
1005 1010 1015 1020 1025 1030
Time (sec)
75
0 200 400 600 800 1000 1200
Time (sec)

Figure 11.11. Driving performance under fixed power charging WPT4 (SOCi =80%). (a)
Speed. (b) Battery voltage. (c) Charger current. (c) Charger power. (d) SOC.

Similar to the analyses in Figures 11.10 and 11.11, this test has been conducted for

the other standard charging levels, and the driving performance is compared with the case

while there is no QDWPT charging, as shown in Figure 11.12. The figure shows the EV’s

battery SOC through the driving period without and with implementing the WPT charger

at the traffic signal. It can be observed that through the utilization of QDWPT charging,

the driving range is extended as a function of the charging power level for the same stop

time. All charging levels exhibit a shallower reduction in the SOC, where WPT4 results in

only a 3% SOC reduction at the end of the driving cycle. WPT4 and higher levels appear

promising for these applications since the vehicle may recover the initial SOC by the end

of the driving profile.

263
 No charging
70 WPT1
WPT2
WPT3
68 WPT4
SOC (%)

66

64

62

0 200 400 600 800 1000 1200

Time (sec)

Figure 11.12. Driving performance at different charging levels (SOCi =70%).

11.4.2 Variable Power Charging Scenario

In this case, the charging power rate varies automatically based on the EV’s SOC. It

is desirable that an EV that has a lower SOC should be permitted to charge at a higher

charging rate, while as the SOC increases, the rate decreases linearly, as depicted in Figure

11.13. In this figure, 𝑃𝑚𝑖𝑛 and 𝑃𝑚𝑎𝑥 represent the minimum and maximum acceptable limits

of the charging rate, which are defined based on the charger limits. 𝑆𝑂𝐶𝑚𝑖𝑛 and 𝑆𝑂𝐶𝑚𝑎𝑥

denote the SOC range, in which the power is adjusted where below 𝑆𝑂𝐶𝑚𝑖𝑛 , a constant

264
𝑃𝑚𝑎𝑥 is supplied while above 𝑆𝑂𝐶𝑚𝑎𝑥 , charging is deactivated. This behavior would help

aid in limiting utility grid stress, while simultaneously prioritizing charging amongst EVs

and preserving their battery lifetimes, by greatly reducing the charging rate when the

battery approaches the maximum SOC level [215].

Pc
Pmax

Pmin
SOC
0 SOCmin SOCmax 100%

Figure 11.13. Charging rate decision based on SOC.

In this study, we have considered a safe linear SOC operating range of 𝑆𝑂𝐶𝑚𝑖𝑛 =

30% to 𝑆𝑂𝐶𝑚𝑎𝑥 = 80%, where 𝑃𝑚𝑎𝑥 is set to the WPT4 level, and the minimum supported

charging level (𝑃𝑚𝑖𝑛 ) is set to 30% of 𝑃𝑚𝑎𝑥 . WPT4 is chosen since it has shown promising

performance in this application, as was indicated in Figure 11.12. The charging rate

adjustment is achieved by the secondary side controller shown in Figure 11.2. The driving

performance under this scenario is indicated in Figure 11.14. In this case, the FTP-72

265
profile is extended by multiple repetitions to cover the full operating SOC range (80-20%).

It can be noticed that the charging rate starts with 𝑃𝑚𝑖𝑛 at 80% SOC increasing linearly as

the SOC decreases. Once the SOC reaches 30%, the charger supplies 𝑃𝑚𝑎𝑥 for the

remaining period. By applying this algorithm, the driving time is extended by more than 5

hours.
speed (mph)

60
40
20 (a)

0
0 5000 10000 15000
30
Pc (kW)

20
10 (b)
0
0 5000 10000 15000
80
SOC (%)

60
(c)
40
20
0 5000 10000 15000
Time (sec)

Figure 11.14. Driving performance under variable power charging level (SOCi = SOCmax
=80%, SOCmin =30%). (a) Speed. (b) Charger power. (c) Battery SOC.

11.4.3 Fixed Power Charging and Discharging Scenario

This scenario is applied to identify the ability of the QDWPT system to achieve V2G

266
and G2V operation. In this case, the EV can charge and discharge during the stop time at

the traffic signal with a fixed level of WPT4. The charging and discharging decision is

made by comparing the EV’s psychological price (R) (the maximum acceptable price) with

the grid retail price of the wireless charger (R’), where R depends on the EV’s battery SOC,

as was presented in Figure 10.6(a) and Equation (10-14).

The wireless charger retail price (R’), should be more than that of the base load,

which is 4 cents/kWh [200]. Thus, R’ is set to change from 4.5 to 6.5 cents/kWh. The retail

price for wireless charging of a road is available for the EV owners based on the current

traffic information systems of Intelligent Transportation Systems (ITS) [200]. The

switching between charging and discharging operation is achieved by the vehicle-side

controller after the EV’s driver accepts. The controller compares the psychological price

with the retail price of the road. If R is higher than R’, the EV charges the battery; and if R

is less than R’, the EV discharges the battery. This procedure is applied using the same

SOC range as was established in the previous section. Over 80%, the EV will discharge in

order to avoid overcharging, and below 30% the EV will keep charging.

The driving performance during this scenario is indicated in Figure 11.15. These

results are obtained by assuming the retail price R’=5.5 cents/kWh, and the maximum and

the minimum accepted EV psychological price is 6.5 and 4.5 cents/kWh, respectively. It

can be noticed, at high SOC (≥55%), R> R’, thus the vehicle discharges and the SOC drops

dramatically [Figure 11.15(c)]. When the SOC goes below a certain value, R becomes less

than R’ and the BIWPTS starts to charge the battery. These results will vary dynamically

as the retail price changes.

267
 speed (mph)
60
40
20 (a)
0
0 2000 4000 6000 8000 10000 12000
Pc (kW)

20
0 (b)
-20
0 2000 4000 6000 8000 10000 12000
80
SOC (%)

60
40
(c)

0 2000 4000 6000 8000 10000 12000

Time (sec)

Figure 11.15. Driving performance under charging and discharging operation (WPT4,
SOCi =80%, Ƥ’=5.5 cents/kWh). (a) Speed. (b) Charger power. (c) SOC.

11.4.4 Driving Cycle Performance Evaluation

For a more clear driving performance assessment, the driving range for all the

different scenarios are evaluated and compared with the case where there is no charging at

all. This range is estimated based on two factors: full cycle time (FCT) [maximum

continuous driving time per battery cycle (80-20%)], and miles per cycle (mpc) (maximum

driving distance per cycle). Since these factors are functions of the initial SOC (SOCi), they

are estimated at different SOCi, as indicated in Figure 11.16. As can be shown,

268
implementing the QDWPT technology at the traffic signal extends the driving range (time

and distance) for the whole range of SOCi. Using the WPT4 fixed charging level, the

driving range is tripled compared to the no-charging case. The automatic variable charging

(WPTAC) and the automatic charging and discharging (WPTACD) scenarios show

significant improvement in the driving range as well. This analysis is achieved under the

assumption that the WPT system is working with perfect alignment between the two power

pads, and the system is able to transfer the full power capacity.

8 No charging
300
WPT1
(a) (b)
7 WPT2
WPT3 250
Mile per cycle (mile)
Full cycle time (hr)

6 WPT4
WPTAC
5 WPTACD
200

4 150
3
100
2
1 50

30 40 50 60 70 80 30 40 50 60 70 80
SOCi (%) SOCi (%)

Figure 11.16. Driving range for all different scenarios. (a) Full cycle time (FCT). (b)
Maximum miles per cycle (mpc).

Logically, increasing the charging power level will result in increasing the stored

energy in the battery for the same charging period, resulting in an improvement in the

driving range. However, at the same time, the energy consumption and cost will increase.

269
Thus, for fair evaluation among the different scenarios, a performance index is introduced

in this work. A benefit measure factor has been deduced (b.f) for all scenarios. This factor

evaluates the extra mpc that the wireless charging adds to the no-charging case over the

average consumed energy per cycle (kWh), as given in Equation (11-12).

𝑚𝑝𝑐ǀ𝑤𝑖𝑡ℎ 𝑐ℎ𝑎𝑟𝑔𝑒𝑟 −𝑚𝑝𝑐ǀ𝑤𝑖𝑡ℎ𝑜𝑢𝑡 𝑐ℎ𝑎𝑟𝑔𝑒𝑟

𝑏. 𝑓 = (miles/kWh) (11-12)
𝐴𝑣𝑒𝑟𝑎𝑔𝑒 𝐶𝑜𝑛𝑠𝑢𝑚𝑒𝑑 𝐸𝑛𝑒𝑟𝑔𝑦 𝑝𝑒𝑟 𝑐𝑦𝑐𝑙𝑒

This factor is estimated for all scenarios and compared at different SOCi, as presented

in Table 11.3.

Table 11.3. Benefit Factor Evaluation for all Scenarios (miles/kWh).

80% 70% 60% 50% 40% 30% Average

Scenario
SOCi SOCi SOCi SOCi SOCi SOCi (mile/kWh)

WPT1 6.5583 7.5684 9.7528 7.4983 3.4248 1.2626 90.0.6

WPT2 7.1885 7.2134 7.1898 7.3160 5.7584 1.3605 6.0044

WPT3 6.8806 6.8902 6.9911 7.7204 6.4914 3.1659 6.3566

WPT4 7.1200 6.8886 6.8183 6.6669 6.6056 6.5272 6.7711

WPTAC 6.7995 6.7651 7.0121 6.6713 6.7812 6.4574 6.7478

WPTACD 6.1799 6.4199 6.6774 6.6357 6.5294 6.4116 6.4755

For WPT1, WPT2 and WPT3, b.f changes dramatically with different SOCi. Smaller

benefits are observed at low SOCi, and high benefits at high SOCi. On the other hand,

270
WPT4, WPTAC, and WPTACD show flat and high benefits over the entire SOCi range.

By calculating the average b.f, these three scenarios provide the highest benefits, regardless

of SOCi. Notice the benefit factor in the case of WPTACD is dynamic based on the retail

price, which is 5.5 cents/kWh in this case. These results can be supplied to the power-flow

controller to decide the most beneficial scenario based on the initial SOC.

Impact of Misalignment on the Feasibility Analysis

One of the main practical limitations of the WPT system is the alignment between

the two sides, which is significantly dependent on the primary pad structure. Based on the

literature [82], there are two main visions for implementing the primary pad in dynamic

WPT systems. The first vision is by burying a long track coupler in the road, in which an

EV is running along with the track and continues power is transferred, as shown in Figure

11.17 [216]. The problem of the long track design is that the pickup coil only covers a

small portion of the track, which makes the coupling coefficient very small. Poor coupling

generates efficiency and EMI concerns. In addition, this structure shows serious issues

related to the safety of the creatures crossing the road, while the primary circuit is being

energized.

The second vision is to implement a string of pads near to each other similar to the

one utilized in the stationary charging, as depicted in Figure 11.18 [217]. Each pad can be

driven by an independent power converter, or one converter can be utilized to drive a

number of pads, with controlling the current in each pad. Thus, the primary pads can be

selectively excited without a high-frequency common current. Also, the energized primary

pads are covered by the vehicle, thus safer operation is experienced. The efficiency and

271
EMI performance can be as good as that in a stationary charging application, however, with

a tradeoff in higher cost.

Figure 11.17. Long track primary coupler.

Figure 11.18. String of pads primary coupler.

272
 The author believes that the second vision is more appropriate for the quasi-dynamic

WPT system, since it can handle both the stationary and dynamic operation. By considering

the second vision, it can be concluded that different sizes and geometries of the vehicle

should not introduce significant misalignment effect. The expected misalignments will

result from the stop position, which can be minimized by implementing an automatic

guiding system.

Therefore, during the transient stops, it is expected that the EV will experience some

misalignments, during charging and discharging operation, resulting in a considerable

deviation from the ideal case that is assumed in the previous feasibility analysis. This

phenomenon has been considered and presented in this section. The misalignment range is

defined based on the J2954A standard, in which the accepted range of variation of coupling

factor in the Z2 class is defined as 0.082 to 0.215 [149]. A reflection of this range on the

power transfer is evaluated using the system model to be from approximately 40% to 100%

of the associated power level capacity. Therefore, the misalignment is expected to vary

randomly within this range.

The driving performance considering the misalignment is investigated during fixed

power charging at the WPT4 level in Figure 11.19. This figure shows the charging power

and the battery SOC over the whole driving profile for three cases: (1) Aligned, when the

two sides are perfectly aligned and Pc=22kW, (2) Worst Misalignment, when the EV

experiences the maximum misalignment and Pc=0.4x22kW, and (3) Random

Misalignment, in which the degree of misalignment is haphazardly assigned at each stop,

based on a uniform distribution random variable. As can be noticed, the misalignment

negatively affects the driving performance. For clear evaluation of this effect, the FTC,

273
mbc and b.f are estimated for all previously mentioned scenarios at SOCi=80%, as

presented in Table 11.4. Considering the misalignment, WPT1 is revealed to be impractical

in terms of performance for FTC and mbc, since it gives similar results to the case with no

charging. With WPT4, a significant drop is experienced in the driving range, however, it

still provides almost double the case without charging. With implementing an automatic

guiding system, it is expected that the driving performance to be very similar to the perfect

alignment case.

4
x 10
3

2
P (kW)

(a)
c

0
0 200 400 600 800 1000 1200
Aligned
Worst Misalignment
80 Random Misalignment
No charging
SOC (%)

(b)
75

70
0 200 400 600 800 1000 1200
Time (sec)

Figure 11.19. Driving performance considering a misalignment (WPT4, SOCi =80%). (a)
Charging power. (b) Battery SOC.

274
 Table 11.4. Driving Performance Considering Misalignment.

Aligned Worst Misalignment Random Misalignment

Scenario FCT mpc FCT mpc FCT mpc

b.f b.f b.f
(hr) (miles) (hr) (miles) (hr) (miles)

WPT1 2.82 110.5 6.56 2.71 106.2 6.14 2.72 106.7 6.44

WPT2 3.33 130.5 7.18 2.77 108.64 6.21 3.1 121.14 7.81

WPT3 3.87 151.84 6.88 3.01 117.88 7.73 3.375 132.25 7.28

WPT4 7.67 300.4 7.12 3.492 136.8 6.814 4.99 195.37 7.01

WPTAC 5.375 210.6 6.7995 3.164 124 6.78 3.994 156.5 6.79

WPTACD 4.98 195.3 6.18 2.872 112.54 6.04 3.48 136.3 6.13

Conclusion

This chapter proposed, modeled, and analyzed the feasibility of installing the qusi-

dynamic WPT systems at traffic signals to improve the range and driving time constraints

present in the EV. A precise model for wirelessly connected EV, including BIWPTS, a

lithium-ion battery array and drive system was developed and verified by means of

simulation and experimentally. Three different scenarios of QDWPT system

implementation were evaluated based on the FTP-72 city driving profile. Under the first

scenario, the EV charges at fixed power levels during stops at the traffic intersections. As

expected, driving range is increased as the charging rate increases until reaching WPT4,

where energy recovery from charging is close to that of which was consumed by the electric

275
motor. In the second scenario, a variable power charging profile is applied to reduce the

grid stress, while dramatically extending the driving range. In the final scenario, V2G and

G2V operation was tested considering the grid retail price and EV’s psychological price.

The performance of these scenarios were evaluated based on the continuous driving time

per battery cycle and miles per cycle. A benefit measure factor was introduced to quantify

the tradeoff between energy consumption and extra miles added, due to QDWPT charging.

At fixed charging rates below WPT4, the benefit measure factor changed dramatically

when driving started at different initial SOC values, revealing that starting at a low SOC

yielded less benefit than starting at a high SOC. However, at WPT4, WPTAC and

WPTACD, high benefits were revealed regardless of the initial SOC. Finally, the effect of

coil misalignment during WPT system operation was investigated for all scenarios

questioning the feasibility of WPT1 charging given a misalignment. This study concluded

that QDWPT at traffic signals could provide a promising solution to dramatically extend

the driving range of EVs, while increasing the operating time between traditional charging

cycles, especially with high power levels.

276
 This dissertation presented modeling, design and control analyses for wirelessly

connected electric vehicles with the power grid to accomplish grid-to-vehicle and vehicle-

to-grid services, in the presence of Photovoltaic generation. Developing accurate analytical

and physics-based models that are able to predict the system’s dynamics, during transient

and steady-state operation, utilizing these models to design and optimize the system’s

components, and then evaluating the entire system’s performance, is the general goal of

this research. How to choose the wireless power pads and optimize their design parameters;

how to choose, design and build the appropriate resonant converter topology; and how to

design and implement the appropriate controller that is able to manage the bidirectional

power-flow in the system, are the central questions that drive most of these research

activities.

In this work, the power conversion systems that integrate the EVs, PV generations

and power grid through a common DC-bus, to achieve G2V and V2G services, are

researched. First, the coupling between the PV power system and the DC-link is

investigated. An accurate nonlinear dynamic model for the grid-connected PV system is

developed and utilized to predict the system’s performance during the different dynamics.

Then, this model is linearized to obtain the small signal model for the same system. Based

on this model, clear design considerations for the maximum power point tracking algorithm

in the system are stated and applied for the different MPPT algorithms. In addition, an

advanced intelligent MPPT algorithm based on fuzzy logic is proposed and its design

parameters are optimized using both the developed small signal model and genetic

277
algorithm search technique. The proposed intelligent MPPT algorithm is simulated and

experimentally tested under fast variation of irradiance and temperature conditions, and

compared with the P&O and INC algorithms. The proposed controller shows faster

transient response, less steady state error and robust tracking performance.

Another part in the V2G system that is researched in this work is the bidirectional

wireless integration between the EV and DC-bus. In this topic, many aspects have been

explored, analyzing practical issues in the system. The themes that have been examined in

this area are summarized as follows:

- Developing accurate mathematical and physics-based models for the bidirectional

inductive wireless power transfer system in EVs applications. The models were able to

predict the system’s performance and power-flow, and novel practical limitations for

active and reactive power-flow during G2V and V2G operations are stated.

- Designing, building and testing experimental prototypes and computer based models for

the BIWPTS to verify the proposed theoretical analyses.

- Performing comparative and assessment analyses for the main compensation topologies

in the symmetrical BIWPTS: LC-series, LC-parallel and LCL topology. The BIWPTSs’

performance is evaluated by studying the active and reactive power-flow, efficiency,

power factor, physical installation, design and control complexity, and sensitivity to the

misalignment. The results demonstrate that LCL-topology is more appropriate for EVs

applications and in the situations where the inverter is a long way from the pad. This

topology shows current-source characteristics, power factor correction capability,

simple design and control, high overall efficiency and power factor, and more

robustness to misalignment.

278
- Building an accurate physics-based co-simulation platform for the LCL BIWPTS that

links the circuit model of the power electronic converters and controllers in Simulink,

with the finite-element model of the electro-magnetic coupler. The effectiveness of the

platform is clarified by investigating the nonlinearities in the system’s performance due

to the magnetic material and power electronics converters, in terms of error and

harmonics analysis.

- Proposing a clear design methodology for the different components of the LCL

BIWPTS in EVs applications. The proposed strategy depends on electromagnetic field

computations, joined with analytical models. A 3-D finite-element model for an 8-kW

polarized double-D power pad was modeled, optimized, and built. Moreover, the choice

of the power factor correction and the impedance matching elements was investigated.

- Accomplishing an automatic magnetic design multi-objective optimization for the

double-D power pad’s parameters. The design parameters are chosen, such that the

system shows the best coupling performance under different misalignments with

minimum cost. The optimization analysis is achieved by linking the efficient improved

Tabu search algorithm in the MATLAB environment with a 2D FEM in Maxwell. For

verification purposes, the system coupling performance and cost are investigated using

the optimum design parameters and compared with three other designs presented in the

literature. The proposed optimized design shows the best coupling performance with

moderate cost.

- Developing a new two-layer predictive power-flow controller to manage the

bidirectional power-flow between the EV and power grid, during the long-term parking

(stationary) and transient stops at traffic signals (quasi-dynamic). The controller is

279
 designed to be placed on the vehicle’s side to allow the autonomous charging and

discharging operation for the EV’s battery. It gives the capability to the EV’s operator

to choose among five different modes of operation, considering the grid’s retail price and

EV’s battery SOC. The proposed controller is able to predict the control parameters for

both the primary and secondary converters based on the system’s analytical model.

Moreover, it considers the misalignment in the system, by adaptively estimating the

mutual inductance before applying the proposed algorithm. The practical

implementation of the proposed controller evidenced its performance’s accuracy and

fast response, during both the transient and steady-state operation, compared to the

available controller in the literature.

- Building a comprehensive simulation model for a light-duty wirelessly connected EV,

including BIWPTS, EV’s battery and the traction system. The model was utilized to

investigate the feasibility of implementing the quasi-dynamic wireless charging systems

at the traffic signals on the road, to charge the EV during the transient stops at the

intersections. This study demonstrated that implementing the quasi-dynamic wireless

charging systems at the traffic signals could provide a promising solution to

dramatically extend the driving range of EVs while increasing the operating time

between traditional charging cycles, especially with high power levels.

280
 The main focus of this dissertation is to present modeling, design and control

analyses for the power conversion systems between the EVs, PV generations and power-

grid, to accomplish G2V and V2G services. More focus was provided to the bidirectional

wireless power transfer system between each EV and the DC-bus. However, to further

improve the quality of the power conversion systems, the following recommendations are

stated:

- Utilizing the wide-band gap switching devices for building the HF resonance converters.

- Considering the switching frequency range recommended by the J2954A standard

(81.38-90 kHz) in the prototype design instead of 40 KHz.

- Performing EMI and EMC analysis for both the wireless coupler and the power

electronics converters.

- Investigating the design of the aluminum shield and its impact on the system’s

performance.

- Including the wireless data communication between the two sides of the BIWPTS and

exploring its impact on the autonomous controller’s performance.

- Modifying the physics-based co-simulation platform to represent the BIWPTS in

dynamic and quasi-dynamic WPT operation besides the stationary situation.

Following the presented study, several future research works can be performed, as

follows:

- Developing and optimizing new power pad structures for high-power inductive

281
 stationary, dynamic and quasi-dynamic charging systems for EVs.

- Proposing new arrangements for the compensation networks in the inductive charger

that provide interoperable and in-sensitive power-flow operation.

- Researching the high frequency resonant power converters for the wireless charger,

involving the wide-band gap switching devices, such as SiC and GaN.

- Developing intelligent and adaptive controllers for managing the bidirectional active

and reactive power-flow operation.

- Building an accurate simulation platform to test and evaluate the different

implementation visions of inductive power charging for EVs (stationary, dynamic and

quasi-dynamic).

- Developing autonomous and centralized high-level intelligent controllers that achieve

the integration among a fleet of EVs, RESs and power grid based on EV’s operator

desire, energy price, EV’s psychological price, grid voltage and frequency.

- Implementing the wireless data communication for the inductive charging system and

investigating the cyber and physical security issues in both power and data transfer.

- Investigating the rule of implementing different energy storage technologies, such as

Li-ion battery, super-capacitor and flywheel, in the EV’s power train, with the

wirelessly connected EVs.

282
[1] M. Yilmaz and P. T. Krein, “Review of the Impact of Vehicle-to-Grid Technologies
on Distribution Systems and Utility Interfaces,” IEEE Trans. Power Electron., vol.
28, no. 12, pp. 5673–5689, Dec. 2013.

[2] M. Duvall and E. Knipping, “Environmental assessment of plug-in hybrid electric

vehicles. Volume 1: Nationwide Greenhouse Gas Emissions,” EPRI/NRDC, Palo
Alto, CA, 1015325, 2007.

[3] C. Li, Y. Cao, Y. Kuang, and B. Zhou, Influences of Electric Vehicles on Power
System and Key Technologies of Vehicle-to-Grid. Springer, 2016.

[4] E. L. Karfopoulos, E. M. Voumvoulakis, and N. Hatziargyriou, “Steady-state and

dynamic impact analysis of the large scale integration of plug-in EV to the operation
of the autonomous power system of Crete Island,” in MedPower 2014, pp. 1–9.

[5] H. Xiao, Y. Huimei, W. Chen, and L. Hongjun, “A survey of influence of electrics

vehicle charging on power grid,” in 2014 9th IEEE Conference on Industrial
Electronics and Applications, 2014, pp. 121–126.

[6] N. Hartmann and E. D. Özdemir, “Impact of different utilization scenarios of electric

vehicles on the German grid in 2030,” J. Power Sources, vol. 196, no. 4, pp. 2311–
2318, Feb. 2011.

[7] S. Hadley, “Impact of Plug-in Hybrid Vehicles on the Electric Grid,” Oak Ridge
National Laboratory, Oak Ridge, TN, 2006.

[8] M. K. Meyers, K. Schneider, and R. Pratt, “Impacts Assessment of Plug-in Hybrid

Vehicles on Electric Utilities and Regional US Power Grids,” Pacific Northwest
National Laboratory, State of Washington, Technical Analysis, 2007.

[9] S. W. Hadley and A. Tsvetkova, “Potential Impacts of Plug-in Hybrid Electric

Vehicles on Regional Power Generation,” Oak Rige National Lab, Tennessee, 2008.

[10] A. H. Hajimiragha, C. A. Canizares, M. W. Fowler, S. Moazeni, and A. Elkamel, “A

Robust Optimization Approach for Planning the Transition to Plug-in Hybrid Electric
Vehicles,” IEEE Trans. Power Syst., vol. 26, no. 4, pp. 2264–2274, Nov. 2011.

[11] A. Hajimiragha, C. A. Canizares, M. W. Fowler, and A. Elkamel, “Optimal Transition

to Plug-In Hybrid Electric Vehicles in Ontario, Canada, Considering the Electricity-
Grid Limitations,” IEEE Trans. Ind. Electron., vol. 57, no. 2, pp. 690–701, Feb. 2010.

[12] P. Denholm and W. Short, “An Evaluation of Utility System Impacts and Benefits of
Optimally Dispatched Plug-in Hybrid Electric Vehicles,” National Renewable
Energy Laboratory, Colorado, 2007.

283
[13] A. Ipakchi and F. Albuyeh, “Grid of the future,” IEEE Power Energy Mag., vol. 7,
no. 2, pp. 52–62, Mar. 2009.

[14] J. C. Gomez and M. M. Morcos, “Impact of EV battery chargers on the power quality
of distribution systems,” IEEE Trans. Power Deliv., vol. 18, no. 3, pp. 975–981, Jul.
2003.

[15] A. M. Leite da Silva, I. P. Coutinho, A. C. Zambroni de Souza, R. B. Prada, and A.

M. Rei, “Voltage collapse risk assessment,” Electr. Power Syst. Res., vol. 54, no. 3,
pp. 221–227, Jun. 2000.

[16] A. Dubey and S. Santoso, “Electric Vehicle Charging on Residential Distribution

Systems: Impacts and Mitigations,” IEEE Access, vol. 3, pp. 1871–1893, 2015.

[17] S. Rahman and G. B. Shrestha, “An investigation into the impact of electric vehicle
load on the electric utility distribution system,” IEEE Trans. Power Deliv., vol. 8, no.
2, pp. 591–597, Apr. 1993.

[18] Y. Chen, A. Oudalov, and J. S. Wang, “Integration of electric vehicle charging system
into distribution network,” in 8th International Conference on Power Electronics -
ECCE Asia, 2011, pp. 593–598.

[19] J. Taylor, J. W. Smith, and R. Dugan, “Distribution modeling requirements for

integration of PV, PEV, and storage in a smart grid environment,” in 2011 IEEE
Power and Energy Society General Meeting, 2011, pp. 1–6.

[20] J. Taylor, A. Maitra, M. Alexander, D. Brooks, and M. Duvall, “Evaluation of the

impact of plug-in electric vehicle loading on distribution system operations,” in 2009
IEEE Power Energy Society General Meeting, 2009, pp. 1–6.

[21] S. Shao, M. Pipattanasomporn, and S. Rahman, “Challenges of PHEV penetration to

the residential distribution network,” in 2009 IEEE Power Energy Society General
Meeting, 2009, pp. 1–8.

[22] S. W. Hadley, “Evaluating the impact of Plug-in Hybrid Electric Vehicles on regional
electricity supplies,” in 2007 iREP Symposium - Bulk Power System Dynamics and
Control - VII. Revitalizing Operational Reliability, 2007, pp. 1–12.

[23] O. Warweg, F. Schaller, S. Ritter, and P. Bretschneider, “Technical and regulatory

challenges for the integration of electric vehicles into the energy system,” in 2011 2nd
IEEE PES International Conference and Exhibition on Innovative Smart Grid
Technologies, 2011, pp. 1–8.

[24] K. Clement-Nyns, E. Haesen, and J. Driesen, “The Impact of Charging Plug-In

Hybrid Electric Vehicles on a Residential Distribution Grid,” IEEE Trans. Power
Syst., vol. 25, no. 1, pp. 371–380, Feb. 2010.

284
[25] A. G. Boulanger, A. C. Chu, S. Maxx, and D. L. Waltz, “Vehicle Electrification:
Status and Issues,” Proc. IEEE, vol. 99, no. 6, pp. 1116–1138, Jun. 2011.

[26] A. Dubey, S. Santoso, and M. P. Cloud, “Understanding the effects of electric vehicle
charging on the distribution voltages,” in 2013 IEEE Power Energy Society General
Meeting, 2013, pp. 1–5.

[27] A. Dubey, S. Santoso, and M. P. Cloud, “Comparative analysis of effects of electric

vehicle loads on distribution system voltages,” in 2014 IEEE PES T D Conference
and Exposition, 2014, pp. 1–5.

[28] “Transportation Electrification: A Technology Overview,” Electric Power Research

Institute, Technical Results 1021334, Jul. 2011.

[29] C. Weiller, “Plug-in hybrid electric vehicle impacts on hourly electricity demand in
the United States,” Energy Policy, vol. 39, no. 6, pp. 3766–3778, Jun. 2011.

[30] J. Cook, C. Churchwell, and S. George, “Final Evaluation of San Diego Gas &
Electric Plug-In Electrical Vehicle TOU,” San Diego Gas & Electric, San Diego,
2014.

[31] S. L. Schey, J. Smart, and D. R. Scoffield, “A First Look at the Impact of Electric
Vehicle Charging on the Electric Grid in the EV Project,” vol. 5, Jan. 2012.

[32] S. Shao, T. Zhang, M. Pipattanasomporn, and S. Rahman, “Impact of TOU rates on

distribution load shapes in a smart grid with PHEV penetration,” in IEEE PES T D
2010, 2010, pp. 1–6.

[33] Y. Gao, C. Wang, Z. Wang, and H. Liang, “Research on time-of-use price applying
to electric vehicles charging,” in IEEE PES Innovative Smart Grid Technologies,
2012, pp. 1–6.

[34] A. Dubey, S. Santoso, M. P. Cloud, and M. Waclawiak, “Determining Time-of-Use

Schedules for Electric Vehicle Loads: A Practical Perspective,” IEEE Power Energy
Technol. Syst. J., vol. 2, no. 1, pp. 12–20, Mar. 2015.

[35] C. Gerkensmeyer, M. KintnerMeyer, and J. DeSteese, “Technical Challenges of Plug-

in Hybrid Electric Vehicles and Impacts to the US Power System: Distribution
System Analysis,” PNNL, Technical Report, Jan. 2010.

[36] S. Han, S. Han, and K. Sezaki, “Development of an Optimal Vehicle-to-Grid

Aggregator for Frequency Regulation,” IEEE Trans. Smart Grid, vol. 1, no. 1, pp.
65–72, Jun. 2010.

[37] S. Shao, M. Pipattanasomporn, and S. Rahman, “Demand Response as a Load

Shaping Tool in an Intelligent Grid With Electric Vehicles,” IEEE Trans. Smart Grid,
vol. 2, no. 4, pp. 624–631, Dec. 2011.

285
[38] Y. Cao et al., “An Optimized EV Charging Model Considering TOU Price and SOC
Curve,” IEEE Trans. Smart Grid, vol. 3, no. 1, pp. 388–393, Mar. 2012.

[39] Z. Li, Q. Guo, H. Sun, S. Xin, and J. Wang, “A New Real-Time Smart-Charging
Method Considering Expected Electric Vehicle Fleet Connections,” IEEE Trans.
Power Syst., vol. 29, no. 6, pp. 3114–3115, Nov. 2014.

[40] S. Deilami, A. S. Masoum, P. S. Moses, and M. A. S. Masoum, “Real-Time

Coordination of Plug-In Electric Vehicle Charging in Smart Grids to Minimize Power
Losses and Improve Voltage Profile,” IEEE Trans. Smart Grid, vol. 2, no. 3, pp. 456–
467, Sep. 2011.

[41] S. Shao, M. Pipattanasomporn, and S. Rahman, “Grid Integration of Electric Vehicles

and Demand Response With Customer Choice,” IEEE Trans. Smart Grid, vol. 3, no.
1, pp. 543–550, Mar. 2012.

[42] E. Sortomme, M. M. Hindi, S. D. J. MacPherson, and S. S. Venkata, “Coordinated

Charging of Plug-In Hybrid Electric Vehicles to Minimize Distribution System
Losses,” IEEE Trans. Smart Grid, vol. 2, no. 1, pp. 198–205, Mar. 2011.

[43] Y. He, B. Venkatesh, and L. Guan, “Optimal Scheduling for Charging and
Discharging of Electric Vehicles,” IEEE Trans. Smart Grid, vol. 3, no. 3, pp. 1095–
1105, Sep. 2012.

[44] C. Jin, J. Tang, and P. Ghosh, “Optimizing Electric Vehicle Charging: A Customer’s
Perspective,” IEEE Trans. Veh. Technol., vol. 62, no. 7, pp. 2919–2927, Sep. 2013.

[45] J. de Hoog, T. Alpcan, M. Brazil, D. A. Thomas, and I. Mareels, “Optimal Charging

of Electric Vehicles Taking Distribution Network Constraints Into Account,” IEEE
Trans. Power Syst., vol. 30, no. 1, pp. 365–375, Jan. 2015.

[46] K. Mets, T. Verschueren, W. Haerick, C. Develder, and F. D. Turck, “Optimizing

smart energy control strategies for plug-in hybrid electric vehicle charging,” in 2010
IEEE/IFIP Network Operations and Management Symposium Workshops, 2010, pp.
293–299.

[47] B. W. Zhou, T. Littler, and H. F. Wang, “The impact of vehicle-to-grid on electric

power systems: A review,” in 2nd IET Renewable Power Generation Conference
(RPG 2013), 2013, pp. 1–4.

[48] X. E. Yu, Y. Xue, S. Sirouspour, and A. Emadi, “Microgrid and transportation

electrification: A review,” in 2012 IEEE Transportation Electrification Conference
and Expo (ITEC), 2012, pp. 1–6.

[49] S. Habib and M. Kamran, “A novel vehicle-to-grid technology with constraint

analysis-a review,” in 2014 International Conference on Emerging Technologies
(ICET), 2014, pp. 69–74.

286
[50] G. Xiao, C. Li, Z. Yu, Y. Cao, and B. Fang, “Review of the impact of electric vehicles
participating in frequency regulation on power grid,” in 2013 Chinese Automation
Congress, 2013, pp. 75–80.

[51] Y. Jang and M. M. Jovanović, “Contactless electrical energy transmission system,”

US6301128 B1, 09-Oct-2001.

[52] S. B. Peterson, J. F. Whitacre, and J. Apt, “The economics of using plug-in hybrid
electric vehicle battery packs for grid storage,” J. Power Sources, vol. 195, no. 8, pp.
2377–2384, Apr. 2010.

[53] P. M. de Quevedo, G. Muñoz-Delgado, and J. Contreras, “Impact of Electric Vehicles

on the Expansion Planning of Distribution Systems considering Renewable Energy,
Storage and Charging Stations,” IEEE Trans. Smart Grid, vol. PP, no. 99, pp. 1–1,
2017.

[54] M. Moeini-Aghtaie, H. Farzin, M. Fotuhi-Firuzabad, and R. Amrollahi, “Generalized

Analytical Approach to Assess Reliability of Renewable-Based Energy Hubs,” IEEE
Trans. Power Syst., vol. 32, no. 1, pp. 368–377, Jan. 2017.

[55] J. Ugirumurera and Z. J. Haas, “Optimal Capacity Sizing for Completely Green
Charging Systems for Electric Vehicles,” IEEE Trans. Transp. Electrification, vol. 3,
no. 3, pp. 565–577, Sep. 2017.

[56] H. N. T. Nguyen, C. Zhang, and J. Zhang, “Dynamic Demand Control of Electric

Vehicles to Support Power Grid With High Penetration Level of Renewable Energy,”
IEEE Trans. Transp. Electrification, vol. 2, no. 1, pp. 66–75, Mar. 2016.

[57] R. Wang, P. Wang, and G. Xiao, “Two-Stage Mechanism for Massive Electric
Vehicle Charging Involving Renewable Energy,” IEEE Trans. Veh. Technol., vol. 65,
no. 6, pp. 4159–4171, Jun. 2016.

[58] A. Y. S. Lam, K. C. Leung, and V. O. K. Li, “Capacity Estimation for Vehicle-to-

Grid Frequency Regulation Services With Smart Charging Mechanism,” IEEE Trans.
Smart Grid, vol. 7, no. 1, pp. 156–166, Jan. 2016.

[59] E. L. Karfopoulos and N. D. Hatziargyriou, “Distributed Coordination of Electric

Vehicles Providing V2G Services,” IEEE Trans. Power Syst., vol. 31, no. 1, pp. 329–
338, Jan. 2016.

[60] S. Sahoo, S. Prakash, and S. Mishra, “Handshaking V2G strategy for grid connected
PV assisted charging station,” IET Renew. Power Gener., vol. 11, no. 11, pp. 1410–
1417, 2017.

[61] N. Rahbari-Asr, M. Y. Chow, J. Chen, and R. Deng, “Distributed Real-Time Pricing

Control for Large-Scale Unidirectional V2G With Multiple Energy Suppliers,” IEEE
Trans. Ind. Inform., vol. 12, no. 5, pp. 1953–1962, Oct. 2016.

287
[62] R. Hara, H. Kita, T. Tanabe, H. Sugihara, A. Kuwayama, and S. Miwa, “Testing the
technologies,” IEEE Power Energy Mag., vol. 7, no. 3, pp. 77–85, May 2009.

[63] A. Woyte, V. V. Thong, R. Belmans, and J. Nijs, “Voltage fluctuations on distribution

level introduced by photovoltaic systems,” IEEE Trans. Energy Convers., vol. 21, no.
1, pp. 202–209, Mar. 2006.

[64] J. A. Short, D. G. Infield, and L. L. Freris, “Stabilization of Grid Frequency Through

Dynamic Demand Control,” IEEE Trans. Power Syst., vol. 22, no. 3, pp. 1284–1293,
Aug. 2007.

[65] U. C. Chukwu and S. M. Mahajan, “V2G Parking Lot With PV Rooftop for Capacity
Enhancement of a Distribution System,” IEEE Trans. Sustain. Energy, vol. 5, no. 1,
pp. 119–127, Jan. 2014.

[66] M. J. E. Alam, K. M. Muttaqi, and D. Sutanto, “Effective Utilization of Available

PEV Battery Capacity for Mitigation of Solar PV Impact and Grid Support With
Integrated V2G Functionality,” IEEE Trans. Smart Grid, vol. 7, no. 3, pp. 1562–1571,
May 2016.

[67] M. Falahi, H. M. Chou, M. Ehsani, L. Xie, and K. L. Butler-Purry, “Potential Power

Quality Benefits of Electric Vehicles,” IEEE Trans. Sustain. Energy, vol. 4, no. 4, pp.
1016–1023, Oct. 2013.

[68] M. E. Ropp, “Similarities between vehicle-to-grid interfaces and photovoltaic

systems,” in 2009 IEEE Vehicle Power and Propulsion Conference, 2009, pp. 1221–
1225.

[69] N. K. Kandasamy, K. Kandasamy, and K. J. Tseng, “Loss-of-life investigation of EV

batteries used as smart energy storage for commercial building-based solar
photovoltaic systems,” IET Electr. Syst. Transp., vol. 7, no. 3, pp. 223–229, 2017.

[70] K. W. Hu and C. M. Liaw, “Incorporated Operation Control of DC Microgrid and

Electric Vehicle,” IEEE Trans. Ind. Electron., vol. 63, no. 1, pp. 202–215, Jan. 2016.

[71] A. A. A. Elgammal and A. M. Sharaf, “Self-regulating particle swarm optimised

controller for (photovoltaic-fuel cell) battery charging of hybrid electric vehicles,”
IET Electr. Syst. Transp., vol. 2, no. 2, pp. 77–89, Jun. 2012.

[72] J. Tomić and W. Kempton, “Using fleets of electric-drive vehicles for grid support,”
J. Power Sources, vol. 168, no. 2, pp. 459–468, Jun. 2007.

[73] P. Javanbakht, S. Mohagheghi, B. Parkhideh, S. Dutta, R. Chattopadhyay, and S.

Bhattacharya, “Vehicle-to-grid scheme based on inductive power transfer for
advanced distribution automation,” in 2013 IEEE Energy Conversion Congress and
Exposition, 2013, pp. 3250–3257.

288
[74] M. Ş. Kuran, A. C. Viana, L. Iannone, D. Kofman, G. Mermoud, and J. P. Vasseur,
“A Smart Parking Lot Management System for Scheduling the Recharging of Electric
Vehicles,” IEEE Trans. Smart Grid, vol. 6, no. 6, pp. 2942–2953, Nov. 2015.

[75] A. Mohamed, A. Berzoy, and O. Mohammed, “Experimental Validation of

Comprehensive Steady-state Analytical Model of Bidirectional WPT System in EVs
Applications,” IEEE Trans. Veh. Technol., vol. PP, no. 99, pp. 1–1, 2016.

[76] A. A. S. Mohamed, A. Berzoy, F. G. N. de Almeida, and O. Mohammed, “Steady-

state performance assessment of different compensation topologies in two-way IWPT
system for EV ancillary services,” in 2016 IEEE Industry Applications Society Annual
Meeting, 2016, pp. 1–8.

[77] G. Lempidis, Y. Zhang, M. Jung, R. Marklein, S. Sotiriou, and Y. Ma, “Wired and
wireless charging of electric vehicles: A system approach,” in Electric Drives
Production Conference (EDPC), 2014 4th International, 2014, pp. 1–7.

[78] A. A. S. Mohamed, A. Berzoy, F. G. N. de Almeida, and O. Mohammed, “Modeling

and Assessment Analysis of Various Compensation Topologies in Bidirectional
IWPT System for EV Applications,” IEEE Trans. Ind. Appl., vol. 53, no. 5, pp. 4973–
4984, Sep. 2017.

[79] A. A. S. Mohamed and O. Mohammed, “Physics-Based Co-simulation Platform with

Analytical and Experimental Verification for Bidirectional IPT System in EV
Applications,” IEEE Trans. Veh. Technol., vol. P.P, no. P.P, 2017.

[80] N. Tesla, “System of transmission of electrical energy.,” US645576 A, 20-Mar-1900.

[81] A. Kurs, A. Karalis, R. Moffatt, J. D. Joannopoulos, P. Fisher, and M. Soljačić,

“Wireless Power Transfer via Strongly Coupled Magnetic Resonances,” Science, vol.
317, no. 5834, pp. 83–86, Jul. 2007.

[82] S. Li and C. C. Mi, “Wireless Power Transfer for Electric Vehicle Applications,”
IEEE J. Emerg. Sel. Top. Power Electron., vol. 3, no. 1, pp. 4–17, Mar. 2015.

[83] S. Lukic and Z. Pantic, “Cutting the Cord: Static and Dynamic Inductive Wireless
Charging of Electric Vehicles,” IEEE Electrification Mag., vol. 1, no. 1, pp. 57–64,
Sep. 2013.

[84] G. A. Covic and J. T. Boys, “Modern Trends in Inductive Power Transfer for
Transportation Applications,” IEEE J. Emerg. Sel. Top. Power Electron., vol. 1, no.
1, pp. 28–41, Mar. 2013.

[85] G. A. Covic and J. T. Boys, “Inductive Power Transfer,” Proc. IEEE, vol. 101, no. 6,
pp. 1276–1289, Jun. 2013.

289
[86] B. Esteban, M. Sid-Ahmed, and N. C. Kar, “A Comparative Study of Power Supply
Architectures in Wireless EV Charging Systems,” IEEE Trans. Power Electron., vol.
30, no. 11, pp. 6408–6422, Nov. 2015.

[87] M. Kamar Uddin, G. Ramasamy, S. Mekhilef, K. Ramar, and Y.-C. Lau, “A review
on high frequency resonant inverter technologies for wireless power transfer using
magnetic resonance coupling,” in 2014 IEEE Conference on Energy Conversion
(CENCON), 2014, pp. 412–417.

[88] R. Bosshard, U. Badstubner, J. W. Kolar, and I. Stevanovic, “Comparative evaluation

of control methods for Inductive Power Transfer,” in 2012 International Conference
on Renewable Energy Research and Applications (ICRERA), 2012, pp. 1–6.

[89] W. Zhang and C. C. Mi, “Compensation Topologies of High-Power Wireless Power

Transfer Systems,” IEEE Trans. Veh. Technol., vol. 65, no. 6, pp. 4768–4778, Jun.
2016.

[90] F. Musavi and W. Eberle, “Overview of wireless power transfer technologies for
electric vehicle battery charging,” IET Power Electron., vol. 7, no. 1, pp. 60–66, Jan.
2014.

[91] T. W. Ching and Y. S. Wong, “Review of wireless charging technologies for electric
vehicles,” in 2013 5th International Conference on Power Electronics Systems and
Applications (PESA), 2013, pp. 1–4.

[92] A. A. S. Mohamed, C. R. Lashway, and O. Mohammed, “Modeling and Feasibility

Analysis of Quasi-Dynamic WPT System for EV Applications,” IEEE Trans. Transp.
Electrification, vol. 3, no. 2, pp. 343–353, Jun. 2017.

[93] Inductive Powering - Basic Theory and Application | Koenraad van Schuylenbergh |
Springer.

[94] B. I. Ashraf, Wireless Power Transfer: Fundamentals and Technologies, 1st edition.
S.l.: Auris Reference, 2017.

[95] K. Y. Kim, Wireless Power Transfer: Principles and Engineering Explorations.

Intech, 2012.

[96] A. G. Batres, A. Moghe, and J. Taiber, “A communication architecture for wireless

power transfer services based on DSRC technology,” in 2016 IEEE Transportation
Electrification Conference and Expo (ITEC), 2016, pp. 1–8.

[97] S. Chopra and P. Bauer, “Driving Range Extension of EV With On-Road Contactless
Power Transfer #x2014;A Case Study,” IEEE Trans. Ind. Electron., vol. 60, no. 1,
pp. 329–338, Jan. 2013.

290
[98] A. K. Swain, S. Devarakonda, and U. K. Madawala, “Modeling, Sensitivity Analysis,
and Controller Synthesis of Multipickup Bidirectional Inductive Power Transfer
Systems,” IEEE Trans. Ind. Inform., vol. 10, no. 2, pp. 1372–1380, May 2014.

[99] M. Kabalo, F. Berthold, B. Blunier, D. Bouquain, S. Williamson, and A. Miraoui,

“Efficiency comparison of wire and wireless battery charging: Based on connection
probability analysis,” in 2014 IEEE Transportation Electrification Conference and
Expo (ITEC), 2014, pp. 1–6.

[100] X. Huang, H. Qiang, Z. Huang, Y. Sun, and J. Li, “The Interaction Research of
Smart Grid and EV Based Wireless Charging,” in 2013 IEEE Vehicle Power and
Propulsion Conference (VPPC), 2013, pp. 1–5.

[101] A. A. S. Mohamed, A. Berzoy, and O. Mohammed, “Power Flow Modeling of

Wireless Power Transfer for EVs Charging and Discharging in V2G Applications,”
in 2015 IEEE Vehicle Power and Propulsion Conference (VPPC), 2015, pp. 1–6.

[102] A. A. S. Mohamed, F. G. N. de Almeida, and O. Mohammed, “Harmonics-based

steady-state mathematical model of bidirectional inductive wireless power transfer
system in V2G applications,” in 2016 IEEE Transportation Electrification
Conference and Expo (ITEC), 2016, pp. 1–6.

[103] Y. Mahmoud and E. F. El-Saadany, “A Photovoltaic Model With Reduced

Computational Time,” IEEE Trans. Ind. Electron., vol. 62, no. 6, pp. 3534–3544, Jun.
2015.

[104] Y. A. Mahmoud, W. Xiao, and H. H. Zeineldin, “A Parameterization Approach for

Enhancing PV Model Accuracy,” IEEE Trans. Ind. Electron., vol. 60, no. 12, pp.
5708–5716, Dec. 2013.

[105] L. H. I. Lim, Z. Ye, J. Ye, D. Yang, and H. Du, “A Linear Identification of Diode
Models from Single #x2013; Characteristics of PV Panels,” IEEE Trans. Ind.
Electron., vol. 62, no. 7, pp. 4181–4193, Jul. 2015.

[106] M. A. Farahat, H. M. B. Metwally, and A. Abd-Elfatah Mohamed, “Optimal choice

and design of different topologies of DC–DC converter used in PV systems, at
different climatic conditions in Egypt,” Renew. Energy, vol. 43, pp. 393–402, Jul.
2012.

[107] R. Middlebrook and S. Cuk, “A general unified approach to modelling switching-

converter power stages,” in 1976 IEEE Power Electronics Specialists Conference,
1976, pp. 18–34.

[108] A. A. S. Mohamed, A. Berzoy, and O. A. Mohammed, “Design and Hardware

Implementation of FL-MPPT Control of PV Systems Based on GA and Small-Signal
Analysis,” IEEE Trans. Sustain. Energy, vol. 8, no. 1, pp. 279–290, Jan. 2017.

291
[109] X. Li, Y. Li, J. E. Seem, and P. Lei, “Detection of Internal Resistance Change for
Photovoltaic Arrays Using Extremum-Seeking Control MPPT Signals,” IEEE Trans.
Control Syst. Technol., vol. 24, no. 1, pp. 325–333, Jan. 2016.

[110] A. Ahmed, L. Ran, S. Moon, and J.-H. Park, “A Fast PV Power Tracking Control
Algorithm With Reduced Power Mode,” IEEE Trans. Energy Convers., vol. 28, no.
3, pp. 565–575, Sep. 2013.

[111] Y. Mahmoud and E. F. El-Saadany, “A Photovoltaic Model With Reduced

Computational Time,” IEEE Trans. Ind. Electron., vol. 62, no. 6, pp. 3534–3544, Jun.
2015.

[112] Y. A. Mahmoud, W. Xiao, and H. H. Zeineldin, “A Parameterization Approach for

Enhancing PV Model Accuracy,” IEEE Trans. Ind. Electron., vol. 60, no. 12, pp.
5708–5716, Dec. 2013.

[113] G. M. Masters, Renewable and Efficient Electric Power Systems, 2 edition.

Hoboken, New Jersey: Wiley, 2013.

[114] A. Mohamed, H. Metwally, and M. Farahat, Photovoltaic Water Pumping Systems.

LAP LAMBERT Academic Publishing, 2012.

[115] M. Fazzolari, R. Alcala, Y. Nojima, H. Ishibuchi, and F. Herrera, “A Review of the

Application of Multiobjective Evolutionary Fuzzy Systems: Current Status and
Further Directions,” IEEE Trans. Fuzzy Syst., vol. 21, no. 1, pp. 45–65, Feb. 2013.

[116] A. A. Ramadan, M. T. Faheem, A. F. Amer, and S. A. Sallam, “New genetic-based

design of a Pi-like fuzzy logic speed controlter for an induction motor,” in Second
IEEE International Conference on Computational Cybernetics, 2004. ICCC 2004,
2004, pp. 25–32.

[117] “bp 4175t pv module datasheet - Google Search.” [Online]. Available:

https://www.google.com/?gws_rd=ssl#q=bp+4175t+pv+module+datasheet.
[Accessed: 20-Oct-2015].

[118] B. Subudhi and R. Pradhan, “A Comparative Study on Maximum Power Point

Tracking Techniques for Photovoltaic Power Systems,” IEEE Trans. Sustain. Energy,
vol. 4, no. 1, pp. 89–98, Jan. 2013.

[119] M. A. G. de Brito, L. Galotto, L. P. Sampaio, G. de Azevedo e Melo, and C. A.

Canesin, “Evaluation of the Main MPPT Techniques for Photovoltaic Applications,”
IEEE Trans. Ind. Electron., vol. 60, no. 3, pp. 1156–1167, Mar. 2013.

[120] A. A. S. Mohamed, A. Berzoy, and O. Mohammed, “Optimized-fuzzy MPPT

controller using GA for stand-alone photovoltaic water pumping system,” in IECON
2014 - 40th Annual Conference of the IEEE Industrial Electronics Society, 2014, pp.
2213–2218.

292
[121] N. Femia, G. Petrone, G. Spagnuolo, and M. Vitelli, “Optimization of perturb and
observe maximum power point tracking method,” IEEE Trans. Power Electron., vol.
20, no. 4, pp. 963–973, Jul. 2005.

[122] N. Femia, G. Petrone, G. Spagnuolo, and M. Vitelli, “A Technique for Improving

P #x00026;O MPPT Performances of Double-Stage Grid-Connected Photovoltaic
Systems,” IEEE Trans. Ind. Electron., vol. 56, no. 11, pp. 4473–4482, Nov. 2009.

[123] M. A. Elgendy, B. Zahawi, and D. J. Atkinson, “Assessment of Perturb and Observe

MPPT Algorithm Implementation Techniques for PV Pumping Applications,” IEEE
Trans. Sustain. Energy, vol. 3, no. 1, pp. 21–33, Jan. 2012.

[124] A. A. S. Mohamed, A. Berzoy, and O. Mohammed, “Control Parameters

Optimization for PM DC Motor in Photovoltaic Applications,” presented at the IEEE
International Electric Machines and Drives Conference (IEMDC) 2015, 2015.

[125] S. K. Kollimalla and M. K. Mishra, “Variable Perturbation Size Adaptive P amp;O

MPPT Algorithm for Sudden Changes in Irradiance,” IEEE Trans. Sustain. Energy,
vol. 5, no. 3, pp. 718–728, Jul. 2014.

[126] Y. Jiang, J. A. A. Qahouq, and T. A. Haskew, “Adaptive Step Size With Adaptive-
Perturbation-Frequency Digital MPPT Controller for a Single-Sensor Photovoltaic
Solar System,” IEEE Trans. Power Electron., vol. 28, no. 7, pp. 3195–3205, Jul.
2013.

[127] A. K. Abdelsalam, A. M. Massoud, S. Ahmed, and P. Enjeti, “High-Performance

Adaptive Perturb and Observe MPPT Technique for Photovoltaic-Based Microgrids,”
IEEE Trans. Power Electron., vol. 26, no. 4, pp. 1010–1021, Apr. 2011.

[128] A. Berzoy, E. Baethge, J. Restrepo, and J. Viola, “Fuzzy Control System for
Maximum Power Point Tracking in Solar Panels Based on DC-DC Converter PI
Current Control,” in Andean Region International Conference (ANDESCON), 2012
VI, 2012, pp. 119–122.

[129] T. L. Kottas, Y. S. Boutalis, and A. D. Karlis, “New maximum power point tracker
for PV arrays using fuzzy controller in close cooperation with fuzzy cognitive
networks,” IEEE Trans. Energy Convers., vol. 21, no. 3, pp. 793–803, Sep. 2006.

[130] A. El Khateb, N. Abd Rahim, J. Selvaraj, and M. N. Uddin, “Fuzzy-Logic-

Controller-Based SEPIC Converter for Maximum Power Point Tracking,” IEEE
Trans. Ind. Appl., vol. 50, no. 4, pp. 2349–2358, Jul. 2014.

[131] M. A. A. Mohd Zainuri, M. A. Mohd Radzi, A. C. Soh, and N. A. Rahim,

“Development of adaptive perturb and observe-fuzzy control maximum power point
tracking for photovoltaic boost dc-dc converter,” IET Renew. Power Gener., vol. 8,
no. 2, pp. 183–194, Mar. 2014.

293
[132] G. J. Klir and B. Yuan, Fuzzy Sets and Fuzzy Logic: Theory and Applications, 1st
edition. Upper Saddle River, N.J: Prentice Hall, 1995.

[133] G. J. Klir and B. Yuan, Fuzzy Sets and Fuzzy Logic: Theory and Applications, 1st
edition. Upper Saddle River, N.J: Prentice Hall, 1995.

[134] T. D. Vrionis, X. I. Koutiva, and N. A. Vovos, “A Genetic Algorithm-Based Low

Voltage Ride-Through Control Strategy for Grid Connected Doubly Fed Induction
Wind Generators,” IEEE Trans. Power Syst., vol. 29, no. 3, pp. 1325–1334, May
2014.

[135] D. Sera, L. Mathe, T. Kerekes, S. V. Spataru, and R. Teodorescu, “On the Perturb-
and-Observe and Incremental Conductance MPPT Methods for PV Systems,” IEEE
J. Photovolt., vol. 3, no. 3, pp. 1070–1078, Jul. 2013.

[136] H. Liu, H. Ning, Y. Zhang, and L. T. Yang, “Aggregated-Proofs Based Privacy-

Preserving Authentication for V2G Networks in the Smart Grid,” IEEE Trans. Smart
Grid, vol. 3, no. 4, pp. 1722–1733, Dec. 2012.

[137] U. K. Madawala and D. J. Thrimawithana, “Current sourced bi-directional

inductive power transfer system,” IET Power Electron., vol. 4, no. 4, pp. 471–480,
Apr. 2011.

[138] J. G. Meins and J. D. Sinsley, “Method and apparatus for supplying contactless
power,” US6515878 B1, 04-Feb-2003.

[139] U. K. Madawala and D. J. Thrimawithana, “A Bidirectional Inductive Power

Interface for Electric Vehicles in V2G Systems,” IEEE Trans. Ind. Electron., vol. 58,
no. 10, pp. 4789–4796, Oct. 2011.

[140] D. J. Thrimawithana and U. K. Madawala, “A contactless bi-directional power

interface for plug-in hybrid vehicles,” in IEEE Vehicle Power and Propulsion
Conference, 2009. VPPC ’09, 2009, pp. 396–401.

[141] C.-S. Wang, G. A. Covic, and O. H. Stielau, “Investigating an LCL load resonant
inverter for inductive power transfer applications,” IEEE Trans. Power Electron., vol.
19, no. 4, pp. 995–1002, Jul. 2004.

[142] D. J. Thrimawithana and U. K. Madawala, “A Generalized Steady-State Model for

Bidirectional IPT Systems,” IEEE Trans. Power Electron., vol. 28, no. 10, pp. 4681–
4689, Oct. 2013.

[143] A. K. Swain, M. J. Neath, U. K. Madawala, and D. J. Thrimawithana, “A Dynamic

Multivariable State-Space Model for Bidirectional Inductive Power Transfer
Systems,” IEEE Trans. Power Electron., vol. 27, no. 11, pp. 4772–4780, Nov. 2012.

294
[144] A. A. S. Mohamed, A. Berzoy, and O. A. Mohammed, “Physics-based FE model
and analytical verification of bi-directional inductive wireless power transfer system,”
in 2016 IEEE/ACES International Conf. on Wireless Information Techn. and Systems
(ICWITS) and Applied Computational Electromagnetics (ACES), 2016, pp. 1–2.

[145] A. A. S. Mohamed, A. Berzoy, and O. A. Mohammed, “Physics-Based Co-

Simulation Platform for EMC Analysis of Two-Way Inductive WPT System in EV
Applications,” in 2016 IEEE Vehicle Power and Propulsion Conference (VPPC),
2016, pp. 1–6.

[146] L. Zhao, D. J. Thrimawithana, and U. K. Madawala, “A comparison of LCL and

LC bi-directional inductive power transfer systems,” in Electronics and Application
Conference and Exposition (PEAC), 2014 International, 2014, pp. 766–771.

[147] A. A. S. Mohamed, A. Berzoy, F. G. N. N. de Almeida, and O. Mohammed,

“Steady-State Performance Assessment of Different Compensation Topologies in
Two-way IWPT System for EV Ancillary Services,” IEEE Trans. Ind. Appl., vol. PP,
no. 99, pp. 1–1, 2017.

[148] L. Tan, B. Wu, and S. Rivera, “A bipolar-DC-bus EV fast charging station with
intrinsic DC-bus voltages equalization and minimized voltage ripples,” in IECON
2015 - 41st Annual Conf. of the IEEE Indus. Electro. Soc., 2015, pp. 002190–002195.

[149] “J2954A (WIP) Wireless Power Transfer for Light-Duty Plug-In/ Electric Vehicles
and Alignment Methodology - SAE International.” [Online]. Available:
http://standards.sae.org/wip/j2954/. [Accessed: 10-Jul-2016].

[150] W. Zhang, S.-C. Wong, C. K. Tse, and Q. Chen, “Design for Efficiency
Optimization and Voltage Controllability of Series #x2013;Series Compensated
Inductive Power Transfer Systems,” IEEE Trans. Power Electron., vol. 29, no. 1, pp.
191–200, Jan. 2014.

[151] W. Zhang, S.-C. Wong, C. K. Tse, and Q. Chen, “Design for Efficiency
Optimization and Voltage Controllability of Series #x2013;Series Compensated
Inductive Power Transfer Systems,” IEEE Trans. Power Electron., vol. 29, no. 1, pp.
191–200, Jan. 2014.

[152] V. PALTANEA, G. PALTANEA, H. GAVRILĂ, and G. IONESCU,

“MAGNETIC PROPERTIES OF N87 MnZn (EPCOS TYPE) FERRITE,” UPB Sci
Bull, vol. 75, no. 3, 2013.

[153] K. L. Kaiser, Electromagnetic Compatibility Handbook. CRC Press, 2004.

[154] J. T. Boys, G. A. J. Elliott, and G. A. Covic, “An Appropriate Magnetic Coupling

Co-Efficient for the Design and Comparison of ICPT Pickups,” IEEE Trans. Power
Electron., vol. 22, no. 1, pp. 333–335, Jan. 2007.

295
[155] H. Yoo, E. Shim, J. Kang, G. Choi, C. Lee, and B. Bang, “100kHz IGBT inverter
use of LCL topology for high power induction heating,” in 2011 IEEE 8th
International Conference on Power Electronics and ECCE Asia (ICPE ECCE), 2011,
pp. 1572–1575.

[156] S. Dieckerhoff, M. J. Ruan, and R. W. D. Doncker, “Design of an IGBT-based

LCL-resonant inverter for high-frequency induction heating,” in Conference Record
of the 1999 IEEE Industry Applications Conference, 1999. Thirty-Fourth IAS Annual
Meeting, 1999, vol. 3, pp. 2039–2045 vol.3.

[157] “IXFB110N60P3 - IXYS SEMICONDUCTOR - Power MOSFET, N Channel, 110

A, 600 V, 0.056 ohm, 10 V, 5 V | Farnell element14 Israel.” [Online]. Available:
http://il.farnell.com/ixys-semiconductor/ixfb110n60p3/mosfet-n-ch-600v-110a-to-
264/dp/2429703. [Accessed: 14-Feb-2016].

[158] “FAN7391 PDF Datasheet 625V,” Shell. [Online]. Available:

https://www.fairchildsemi.com/products/power-management/mosfet-igbt-gate-
drivers/high-side-drivers/FAN7391.html. [Accessed: 14-Feb-2016].

[159] “AN-6076 Design and Application Guide of Bootstrap Circuit for High-Voltage
Gate-DriveIC-AN-6076.pdf.”[Online]. Available: https://www.fairchildsemi.com/
application-notes/AN/AN-6076.pdf. [Accessed: 14-Feb-2016].

[160] “Ringing Reduction Techniques for NexFET High Performance MOSFETs -

slpa010.pdf.” [Online]. Available: http://www.ti.com/lit/an/slpa010/slpa010.pdf.
[Accessed: 14-Feb-2016].

[161] J.-Y. Lee and B.-M. Han, “A Bidirectional Wireless Power Transfer EV Charger
Using Self-Resonant PWM,” IEEE Trans. Power Electron., vol. 30, no. 4, pp. 1784–
1787, Apr. 2015.

[162] K. Thirugnanam, T. P. E. R. Joy, M. Singh, and P. Kumar, “Modeling and Control

of Contactless Based Smart Charging Station in V2G Scenario,” IEEE Trans. Smart
Grid, vol. 5, no. 1, pp. 337–348, Jan. 2014.

[163] M. Bojarski, K. K. Kutty, D. Czarkowski, and F. De Leon, “Multiphase resonant

inverters for bidirectional wireless power transfer,” in Electric Vehicle Conference
(IEVC), 2014 IEEE International, 2014, pp. 1–7.

[164] L. Zhao, D. J. Thrimawithana, and U. K. Madawala, “A hybrid bi-directional IPT

system with improved spatial tolerance,” in Future Energy Electronics Conference
(IFEEC), 2015 IEEE 2nd International, 2015, pp. 1–6.

[165] M. J. Neath, U. K. Madawala, and D. J. Thrimawithana, “A new controller for bi-

directional inductive power transfer systems,” in 2011 IEEE International Symposium
on Industrial Electronics (ISIE), 2011, pp. 1951–1956.

296
[166] U. K. Madawala, M. Neath, and D. J. Thrimawithana, “A Power #x2013;Frequency
Controller for Bidirectional Inductive Power Transfer Systems,” IEEE Trans. Ind.
Electron., vol. 60, no. 1, pp. 310–317, Jan. 2013.

[167] O. C. Onar, S. L. Campbell, L. E. Seiber, C. P. White, and M. Chinthavali, “A high-

power wireless charging system development and integration for a Toyota RAV4
electric vehicle,” in 2016 IEEE Transportation Electrification Conference and Expo
(ITEC), 2016, pp. 1–8.

[168] “Guidelines for limiting exposure to time-varying electric, magnetic, and

electromagnetic fields (up to 300 GHz). International Commission on Non-Ionizing
Radiation Protection,” Health Phys., vol. 74, no. 4, pp. 494–522, Apr. 1998.

[169] S. Kim, G. Covic, and J. Boys, “Tripolar Pad for Inductive Power Transfer Systems
for EV Charging,” IEEE Trans. Power Electron., vol. PP, no. 99, pp. 1–1, 2016.

[170] “CHE Series Metallized Film Capacitors On High Energy Corp.” [Online].
Available: http://highenergycorp.thomasnet-navigator.com/viewitems/metallized-
film-capacitors/che-series-metallized-film-capacitors. [Accessed: 02-Jul-2017].

[171] A. A. S. Mohamed, D. Allen, T. Youssef, and O. Mohammed, “Optimal design of

high frequency H-bridge inverter for wireless power transfer systems in EV
applications,” in 2016 IEEE 16th International Conference on Environment and
Electrical Engineering (EEEIC), 2016, pp. 1–6.

[172] N. H. Kutkut and K. W. Klontz, “Design considerations for power converters

supplying the SAE J-1773 electric vehicle inductive coupler,” in Proceedings of
APEC 97 - Applied Power Electronics Conference, 1997, vol. 2, pp. 841–847 vol.2.

[173] J. Achterberg, E. A. Lomonova, and J. de Boeij, “Coil Array Structures Compared

for Contactless Battery Charging Platform,” IEEE Trans. Magn., vol. 44, no. 5, pp.
617–622, May 2008.

[174] A. P. Sample, D. T. Meyer, and J. R. Smith, “Analysis, Experimental Results, and

Range Adaptation of Magnetically Coupled Resonators for Wireless Power
Transfer,” IEEE Trans. Ind. Electron., vol. 58, no. 2, pp. 544–554, Feb. 2011.

[175] J. L. Villa, J. Sallan, J. F. S. Osorio, and A. Llombart, “High-Misalignment Tolerant

Compensation Topology For ICPT Systems,” IEEE Trans. Ind. Electron., vol. 59, no.
2, pp. 945–951, Feb. 2012.

[176] M. Budhia, J. T. Boys, G. A. Covic, and C.-Y. Huang, “Development of a Single-

Sided Flux Magnetic Coupler for Electric Vehicle IPT Charging Systems,” IEEE
Trans. Ind. Electron., vol. 60, no. 1, pp. 318–328, Jan. 2013.

[177] B. Liangu, T. Houjun, and L. Chao, “Bi-directional Contactless Inductive Power

Transfer System Modeling and verifying,” Przegląd Elektrot., vol. R. 89, nr 1b, 2013.

297
[178] A. A. S. Mohamed, A. A. Marim, and O. A. Mohammed, “Magnetic Design
Considerations of Bidirectional Inductive Wireless Power Transfer System for EV
Applications,” IEEE Trans. Magn., vol. 53, no. 6, pp. 1–5, Jun. 2017.

[179] “Ferrites and Accessories - Data Book 2013 - ferrites-and-accessories-db-

130501.pdf.” [Online]. Available:
https://en.tdk.eu/download/519704/069c210d0363d7b4682d9ff22c2ba503/ferrites-
and-accessories-db-130501.pdf. [Accessed: 05-Mar-2017].

[180] D. K. Cheng, Field and Wave Electromagnetics, 2nd edition. Reading, Mass:
Addison-Wesley, 1989.

[181] “MagNet.” [Online]. Available:

https://www.infolytica.com/secured/customer/elite/livedocs/MagNet/. [Accessed:
18-Aug-2017].

[182] “Simulink.” [Online]. Available:

http://www.infolytica.com/secured/customer/elite/livedocs/MagNet/#07-
Circuits/SpecializedTopics/Simulink.htm?Highlight=simulink. [Accessed: 05-Mar-
2017].

[183] A. A. S. Mohamed, A. Berzoy, and O. Mohammad, “Magnetic design

considerations of bidirectional inductive wireless power transfer system for EV
applications,” in 2016 IEEE Conference on Electromagnetic Field Computation
(CEFC), 2016, pp. 1–1.

[184] M. Budhia, G. A. Covic, and J. T. Boys, “Design and Optimization of Circular

Magnetic Structures for Lumped Inductive Power Transfer Systems,” IEEE Trans.
Power Electron., vol. 26, no. 11, pp. 3096–3108, Nov. 2011.

[185] G. A. Covic, M. L. G. Kissin, D. Kacprzak, N. Clausen, and H. Hao, “A bipolar

primary pad topology for EV stationary charging and highway power by inductive
coupling,” in 2011 IEEE Energy Conversion Congress and Exposition, 2011, pp.
1832–1838.

[186] A. Zaheer, D. Kacprzak, and G. A. Covic, “A bipolar receiver pad in a lumped IPT
system for electric vehicle charging applications,” in 2012 IEEE Energy Conversion
Congress and Exposition (ECCE), 2012, pp. 283–290.

[187] S. Kim, G. A. Covic, and J. T. Boys, “Tripolar Pad for Inductive Power Transfer
Systems for EV Charging,” IEEE Trans. Power Electron., vol. 32, no. 7, pp. 5045–
5057, Jul. 2017.

[188] M. Budhia, G. A. Covic, J. T. Boys, and C.-Y. Huang, “Development and

evaluation of single sided flux couplers for contactless electric vehicle charging,” in
2011 IEEE Energy Conversion Congress and Exposition (ECCE), 2011, pp. 614–621.

298
[189] R. Bosshard, J. W. Kolar, J. Mühlethaler, I. Stevanović, B. Wunsch, and F. Canales,
“Modeling and $eta $ - $alpha $ -Pareto Optimization of Inductive Power Transfer
Coils for Electric Vehicles,” IEEE J. Emerg. Sel. Top. Power Electron., vol. 3, no. 1,
pp. 50–64, Mar. 2015.

[190] S. Bandyopadhyay, V. Prasanth, P. Bauer, and J. A. Ferreira, “Multi-objective

optimisation of a 1-kW wireless IPT systems for charging of electric vehicles,” in
2016 IEEE Transportation Electrification Conf. and Expo (ITEC), 2016, pp. 1–7.

[191] M. Moghaddami, A. Anzalchi, A. Moghadasi, and A. Sarwat, “Pareto optimization

of circular power pads for contactless electric vehicle battery charger,” in 2016 IEEE
Industry Applications Society Annual Meeting, 2016, pp. 1–6.

[192] Y. Shiyou, N. Guangzheng, L. Yan, T. Baoxia, and L. Ronglin, “A universal tabu

search algorithm for global optimization of multimodal functions with continuous
variables in electromagnetics,” IEEE Trans. Magn., vol. 34, no. 5, pp. 2901–2904,
Sep. 1998.

[193] K. R. Davey, “Latin Hypercube Sampling and Pattern Search in Magnetic Field
Optimization Problems,” IEEE Trans. Magn., vol. 44, no. 6, pp. 974–977, Jun. 2008.

[194] A. A. S. Mohamed, A. A. Marim, and O. A. Mohammed, “Magnetic Design

Considerations of Bidirectional Inductive Wireless Power Transfer System for EV
Applications,” IEEE Trans. Magn., vol. 53, no. 6, pp. 1–5, Jun. 2017.

[195] F. Turki and U. Reker, “Further design approaches of the standardization: Inductive
charging of electric vehicles,” in Electric Drives Production Conference (EDPC),
2012 2nd International, 2012, pp. 1–6.

[196] M. J. Neath, A. K. Swain, U. K. Madawala, and D. J. Thrimawithana, “An Optimal

PID Controller for a Bidirectional Inductive Power Transfer System Using
Multiobjective Genetic Algorithm,” IEEE Trans. Power Electron., vol. 29, no. 3, pp.
1523–1531, Mar. 2014.

[197] A. A. S. Mohamed, A. Berzoy, and O. Mohammed, “Experimental Validation of

Comprehensive Steady-state Analytical Model of Bidirectional WPT System in EVs
Applications,” IEEE Trans. Veh. Technol.

[198] A. A. S. Mohamed, A. Berzoy, and O. Mohammed, “Predictive active power-flow

control of two-way wireless power transfer system in V2G services,” in 2016 IEEE
2nd Annual Southern Power Electronics Conference (SPEC), 2016, pp. 1–6.

[199] A. A. S. Mohamed, T. Youssef, and O. Mohammed, “Vehicle side predictive

power-flow control of bidirectional WPT system for EV ancillary services,” in 2017
IEEE Applied Power Electronics Conference and Exposition (APEC), 2017, pp.
3211–3217.

299
[200] C. H. Ou, H. Liang, and W. Zhuang, “Investigating Wireless Charging and Mobility
of Electric Vehicles on Electricity Market,” IEEE Trans. Ind. Electron., vol. 62, no.
5, pp. 3123–3133, May 2015.

[201] R. Xiong, H. He, F. Sun, and K. Zhao, “Evaluation on State of Charge Estimation
of Batteries With Adaptive Extended Kalman Filter by Experiment Approach,” IEEE
Trans. Veh. Technol., vol. 62, no. 1, pp. 108–117, Jan. 2013.

[202] C. R. Lashway and O. A. Mohammed, “Adaptive Battery Management and

Parameter Estimation Through Physics-Based Modeling and Experimental
Verification,” IEEE Trans. Transp. Electrification, vol. 2, no. 4, pp. 454–464, Dec.
2016.

[203] F. Musavi and W. Eberle, “Overview of wireless power transfer technologies for
electric vehicle battery charging,” IET Power Electron., vol. 7, no. 1, pp. 60–66, Jan.
2014.

[204] J. G. Hayes, “Battery charging systems for electric vehicles,” in 1998/262), IEE
Colloquium on Electric Vehicles - A Technology Roadmap for the Future (Digest No,
1998, p. 4/1-4/8.

[205] S. Jeong, Y. J. Jang, and D. Kum, “Economic Analysis of the Dynamic Charging
Electric Vehicle,” IEEE Trans. Power Electron., vol. 30, no. 11, pp. 6368–6377, Nov.
2015.

[206] “Korean Electric Buses Go Wireless - Korea Real Time - WSJ.” [Online].
Available: http://blogs.wsj.com/korearealtime/2013/08/08/korean-electric-buses-go-
wireless/. [Accessed: 29-Jul-2016].

[207] “Feasibility study — powering electric vehicles on England’s major roads,” 01-
Jan-1970. [Online]. Available:
http://www.highways.gov.uk/knowledge/publications/1902/. [Accessed: 18-Jan-
2017].

[208] J. M. Miller, P. T. Jones, J. M. Li, and O. C. Onar, “ORNL Experience and

Challenges Facing Dynamic Wireless Power Charging of EV’s,” IEEE Circuits Syst.
Mag., vol. 15, no. 2, pp. 40–53, Secondquarter 2015.

[209] A. Shekhar, V. Prasanth, P. Bauer, and M. Bolech, “Economic Viability Study of

an On-Road Wireless Charging System with a Generic Driving Range Estimation
Method,” Energies, vol. 9, no. 2, p. 76, Jan. 2016.

[210] L. Bhaskar, A. Sahai, D. Sinha, G. Varshney, and T. Jain, “Intelligent traffic light
controller using inductive loops for vehicle detection,” in 2015 1st International
Conference on Next Generation Computing Technologies (NGCT), 2015, pp. 518–
522.

300
[211] “Polymer Li-Ion Cell: 3.7V 21Ah (1055275-2C, 77.7Wh, 42A rate) - UN38.3
Passed(DGR),” BatterySpace.com/AA Portable Power Corp. Tel: 510-525-2328.
[Online]. Available: http://www.batteryspace.com/polymer-li-ion-cell-3-7v-21ah-
1055275-2c-77-7wh-42a-rate---un38-3-passed-dgr.aspx. [Accessed: 29-Jul-2016].

[212] Y. Cao, R. C. Kroeze, and P. T. Krein, “Multi-Timescale Parametric Electrical

Battery Model for Use in Dynamic Electric Vehicle Simulations,” IEEE Trans.
Transp. Electrification, vol. PP, no. 99, pp. 1–1, 2016.

[213] O. US EPA, “Dynamometer Drive Schedules.” [Online]. Available:

https://www.epa.gov/vehicle-and-fuel-emissions-testing/dynamometer-drive-
schedules. [Accessed: 30-Jul-2016].

[214] M.-E. Choi, J.-S. Lee, and S.-W. Seo, “Real-Time Optimization for Power
Management Systems of a Battery/Supercapacitor Hybrid Energy Storage System in
Electric Vehicles,” IEEE Trans. Veh. Technol., vol. 63, no. 8, pp. 3600–3611, Oct.
2014.

[215] A. T. Al-Awami, E. Sortomme, G. M. A. Akhtar, and S. Faddel, “A Voltage-Based

Controller for an Electric-Vehicle Charger,” IEEE Trans. Veh. Technol., vol. 65, no.
6, pp. 4185–4196, Jun. 2016.

[216] “Electric Vehicle and Roadway.” [Online]. Available:

http://evr.usu.edu/news/press-releases/2014/jul17-test-track-approved. [Accessed:
09-Aug-2017].

[217] “Charged EVs | What’s up with wireless EV charging.” [Online]. Available:

https://chargedevs.com/features/whats-wireless-ev-charging/. [Accessed: 09-Aug-
2017].

301
 AHMED A S MOHAMED

01 January 1986 Born, Dakahliya, Egypt.

2002-2007 B.Sc. in Electrical Engineering, Zagazig University, Zagazig, Egypt.

2008-2011 M.S. in Electrical Engineering, Zagazig University, Zagazig, Egypt.

2014-2016 Research Assistant, Florida International University, Miami, Florida, USA.

Spring 2017 Doctoral Evidence Acquisition (DEA) Fellow, Florida International

University, Miami, Florida, USA.

Summer and Fall Dissertation Year Fellow (DYF), Florida International University, Miami,
2017 Florida, USA.

Fall 2017 The Outstanding PhD Graduate for College of Engineering and Computing

2008-2014 Research and Teaching Assistant, Zagazig University, Zagazig, Egypt.

PUBLICATION AND PRESENTATIONS

[P1] A. A. S. Mohamed, and Osama Mohammed, “Autonomous Two-layer Predictive Controller for
Bidirectional Inductive Power Transfer in EV Applications,” pending patent.

[P2] A. A. S. Mohamed, and Osama Mohammed, “Integrated Bidirectional Inductive/Conductive Electrical

Apparatus,” pending patent.

[J1] A. A. S. Mohamed, A. Berzoy, and O. Mohammed, “Design and Hardware Implementation of FL

MPPT Control for PV Systems Based on GA and Small-Signal Analysis,” IEEE Trans. Sustain. Energy,
vol. 8, no. 1, pp. 279-290, Jan. 2017.

[J2] A. A. S. Mohamed, A. Berzoy, and O. Mohammed, “Experimental Validation of Comprehensive

Steady-state Analytical Model of Bidirectional WPT System in EVs Applications,” IEEE Trans. Veh.
Technol., vol. 66, no. 7, pp. 5584-5594, July 2017.

[J3] A. A. S. Mohamed, A. A. Marim and O. A. Mohammed, "Magnetic Design Considerations of

Bidirectional Inductive Wireless Power Transfer System for EV Applications," in IEEE Transactions on
Magnetics, vol. 53, no. 6, pp. 1-5, June 2017.

[J4] A. A. S. Mohamed, C. R. Lashway and O. Mohammed, "Modeling and Feasibility Analysis of Quasi-
Dynamic WPT System for EV Applications," in IEEE Transactions on Transportation Electrification,
vol. 3, no. 2, pp. 343-353, June 2017.

[J5] A. A. S. Mohamed, A. Berzoy, F. G. N. de Almeida, and O. Mohammed, “Modeling and Assessment

Analysis of Various compensation topologies in Bidirectional IWPT system for EV Applications,” IEEE
Transaction on Industry Applications , vol. 53, no. 5, pp. 4973-4984, Sept.-Oct. 2017.

302
[J6] A. A. S. Mohamed and O. Mohammed, “Physics-Based Co-simulation Platform with Analytical and
Experimental Verification for Bidirectional IPT System in EV Applications,” IEEE Trans. Veh. Technol.,
vol. pp, no. 99, pp. 1-1, 2017.

[J7] A. A. S. Mohamed, Siguang An, and O. Mohammed, “Coil Design Multi-Objective Optimization of
Power Pad in IPT System for Electric Vehicle Applications,” IEEE Transaction on Magnetics, vol. pp,
no. 99, pp. 1-1, 2017.

[J8] A. Berzoy, A. A. S. Mohamed and O. Mohammed, "Complex-Vector Model of Interturn Failure in

Induction Machines for Fault Detection and Identification," in IEEE Transactions on Industry
Applications, vol. 53, no. 3, pp. 2667-2678, May-June 2017.

[J9] Alberto Berzoy, A. A. S. Mohamed, and Osama Mohammed, “Optimizing power converter PCB
design for lower EMI,” COMPEL - Int. J. Comput. Math. Electr. Electron. Eng., vol. 34, no. 5, pp. 1364–
1380, Sep. 2015.

[J10] Hany F. Habib, A. A. S. Mohamed, Mohamad El Hariri, Osama A. Mohammed, “Utilizing

Supercapacitors for Resiliency Enhancements and Adaptive Microgrid Protection Against
Communication Failures”, Electric Power Systems Research, vol. 145, pp. 223-233, Apr. 2017.

[J11] A. Berzoy, A. A. S. Mohamed and O. Mohammed, "Impact of Inter-Turn Short-Circuit Location on

Induction Machines Parameters Through FE Computations," in IEEE Transactions on Magnetics, vol. 53,
no. 6, pp. 1-4, June 2017.

[J12] Samy Fadel, A. A. S. Mohamed, and Osama A. Mohammed, “Fuzzy Logic-Based Autonomous
Controller for Electric Vehicles Charging in Residential Distribution Systems,” Electric Power Systems
Research, vol. 148, pp. 48-58, July 2017.

[J13] A. Berzoy, A. A. S. Mohamed, and O. Mohammed, “Design considerations and predictive direct
current control of Active Regenerative Rectifiers for harmonic and current ripple reduction,” under review
in IEEE Transaction on Power Electronics.

[C1] A. A. S. Mohamed, A. Berzoy, F. G. N. de Almeida, and O. Mohammed, “Steady-state performance

assessment of different compensation topologies in two-way IWPT system for EV ancillary services,” in
2016 IEEE Industry Applications Society Annual Meeting, 2016, pp. 1–8.

[C2] A. A. S. Mohamed, A. Berzoy, and O. A. Mohammed, “Physics-Based Co-Simulation Platform for

EMC Analysis of Two-Way Inductive WPT System in EV Applications,” in 2016 IEEE Vehicle Power
and Propulsion Conference (VPPC), 2016, pp. 1–6.

[C3] A. A. S. Mohamed, D. Allen, T. Youssef, and O. Mohammed, “Optimal design of high frequency H-
bridge inverter for wireless power transfer systems in EV applications,” in 2016 IEEE 16th International
Conference on Environment and Electrical Engineering (EEEIC), 2016, pp. 1–6.

[C4] A. A. S. Mohamed, A. Berzoy, and O. Mohammed, “Power Flow Modeling of Wireless Power
Transfer for EVs Charging and Discharging in V2G Applications,” in 2015 IEEE Vehicle Power and
Propulsion Conference (VPPC), 2015, pp. 1–6.

[C5] A. A. S. Mohamed, T. Youssef and O. Mohammed, "Vehicle side predictive power-flow control of
bidirectional WPT system for EV ancillary services," 2017 IEEE Applied Power Electronics Conference
and Exposition (APEC), Tampa, FL, USA, 2017, pp. 3211-3217.

303