ABSTRACT ZHOU, XIAOHU. Design and Control of Bi-Directional Grid-Interactive Converter for Plugin Hybrid Electric Vehicle Applications.

(Under the direction of Dr. Alex Q. Huang). The plug-in hybrid electric vehicle (PHEV) is a promising technology which provides a sustainable approach to transportation that is easily accessible to a large portion of the population that already relies on gasoline-fueled cars. Although the larger scale adoption of plug-in hybrid vehicles is still years away, politicians, electric utilities, and auto companies are eagerly awaiting the opportunities that will arise from reduced emissions, reduced gasoline consumption, new electric utility services, increased revenues, and new markets that will lead to the creation of new jobs. In addition, the electrification of the transportation system would lead to the creation of new avenues for researchers. In the case of power electronics researchers, plug-in hybrid electric vehicles would provide a new candidate for energy storage. Because energy storage is a component of so called smart grids, a topic of growing interest to the power engineering research community, PHEVs could be incorporated as a vital part of such a system. However, to enable this functionality, a power electronics interface between the vehicle and grid is required. The motivation of this dissertation is to design a grid-interactive smart charger to enable PHEV as distributed energy storage device which will play an important role in smart grid applications. For grid-connection applications of the proposed converter, adaptive virtual resistor control is proposed to achieve high power quality for plug-in hybrid electric vehicles integration with various grid conditions. High frequency resonance poses a challenge to controller design and moreover the various impedances lead to the variation of the resonant frequency which will make the control design more complicated. The proposed controller behaves as a controllable

resistor series with a filter capacitor but does not exist physically. It will be adjusted automatically based on grid conditions in order to eliminate high frequency resonance. For off-grid applications of the proposed converter, a new inductor current feedback controller based on active harmonic injection is proposed. An active harmonics injection loop is proposed to extract the harmonics from the load and add to the inductor current control loop. This method effectively improves the harmonics compensation capability for the inductor current feedback control and achieves a better output voltage with nonlinear loads. For a Solid State Transformer (SST) based smart grid with multiple plug-in hybrid electric vehicles, the instability issue is investigated. When the total demand power from the plug-in vehicles exceeds the capability of one SST, a new power management strategy is proposed in each vehicle to adjust its power demand in order to avoid voltage collapse of the SST. Gain scheduling technique is proposed to dispatch power to each vehicle based on batterys state of charge. A comprehensive case study is conducted to verify the proposed method. The proposed method can be used as a power electronics converter level control to improve the stability of a solid state transformer. For the DC/DC stage of the proposed converter a high order filter is proposed to be placed between the battery and the converter. The objective is to reduce the filter size which will further reduce the system cost and volume. Another major goal is to largely attenuate the current ripple of the charging current which will yield ripple free charging for a battery. Ripple free charging will eliminate the extra heat generated by the current ripple and will increase the battery life. The new controller is proposed to resolve the potential instability issue resulting from the high order filter. The control loop design and robustness analyses are conducted.

Design and Control of Bi-Directional Grid-Interactive Converter for Plug-in Hybrid Electric Vehicle Applications

by Xiaohu Zhou

A dissertation submitted to the Graduate Faculty of North Carolina State University in partial fulfillment of the requirements for the degree of Doctor of Philosophy Electrical Engineering

Raleigh, North Carolina 2011

APPROVED BY:

_______________________________ Dr. Alex Q. Huang Committee Chair ________________________________ Dr. Subhashish Bhattacharya

______________________________ Dr. Mo-Yuen Chow

________________________________ Dr. Srdjan Lukic

ii

DEDICATION To My Parents

Lili Huang and Zhigang Zhou

iii

BIOGRAPHY The author, Xiaohu Zhou, was born in Harbin, China. He received the B.S. and the M.S. degree from Harbin Institute of Technology, Harbin, China in 2004 and 2006, respectively, both in electrical engineering. Since fall of 2006, he started to pursue a Ph.D. degree at Semiconductor Power Electronics Center (SPEC) and later National Science Foundation funded Engineering Research Center: Future Renewable Electric Energy Delivery and Management Center (FREEDM), Department of Electrical and Computer Engineering, North Carolina State University, Raleigh.

iv

ACKNOWLEDGMENTS I would like to express my sincere appreciation to my advisor Dr. Alex Q. Huang for his guidance, encouragement and support. Dr. Huangs creative thinking, broad knowledge, insightful vision and warm character always inspires my work and study. To explore something new will always be rooted in my heart. Thank you for giving me this opportunity, I enjoy my study and work in FREEDM Systems Center very much. I am very grateful to my other committee members, Dr. Mo-Yuen Chow, Dr. Subhashish Bhattacharya and Dr. Srdjan Lukic for their valuable suggestion and helpful discussion during so many group and individual meetings. It is my great pleasure to work with you during these five years. I would like also to thank Dr. Gracious Ngaile for serving as the Graduate School Representative for my defense. I want to thank ERC program of the National Science Foundation and Advanced Transportation Energy Center for their financial support of my project and research. I would like to thank all the staff members at FREEDM Systems Center who provide an amazing environment for me to study and work. Special thanks go to Mr. Anousone Sibounheuang and Mrs. Colleen Reid for their help. I want to thank my student colleagues who have helped with many good discussions and gave me so many joyful times: Dr. Chong Han, Dr. Yan Gao, Dr. Bin Chen, Dr. Wenchao Song, Dr. Xiaojun Xu, Dr. Jinseok Park, Dr. Jeesung Jung, Dr. Yu Liu, Dr. Jun Wang, Dr. Jiwei Fan, Dr. Liyu Yang, Dr. Sungkeun Lim, Dr. Xin Zhou, Dr. Tiefu Zhao, Dr. Jun Li, Dr. Rong Guo, Dr. Xiaopeng Wang, Mr. Zhaoning Yang, Mr. Jifeng Qin, Mrs. Zhengping Xi, Mr. Sameer Mundkur, Mr. Zhigang Liang, Mr. Yu Du, Mr. Qian Chen, Mr. Gangyao Wang,

Mr. Xunwei Yu, Mr. Edward Van Brunt, Mr. Babak Parkhideh, Mr. Arvind Govindaraj, Mr. Sanzhong Bai, Mr. Zeljko Pantic, Mr. Xu She, Mr. Xingchen Yang, Mr. Yen-Mo Chen, Mr. Pochin Lin, my Project Partner Mr. Philip Funderburk, Mr. Zhuoning Liu, Miss. Zhan Shen, Miss. Mengqi Wang, Mr. Yalin Wang, Mr. Xing Huang, Mr. Li Jiang, Mr. Fei Wang, Mr. Kai Tan, Mr. Xiang Lu. Finally I want to give my heartfelt appreciation to my parents in China. You always encourage me to pursue my dreams and help me get through tough times. I am so grateful to you for your endless support, trust and love for all of these years.

vi

TABLE OF CONTENTS LIST OF TABLES ................................................................................................................... ix LIST OF FIGURES .................................................................................................................. x Chapter One Introduction ......................................................................................................... 1 1.1 1.2 Research Background: Plug-in Hybrid Electric Vehicles ....................................... 1 State of the Art of Technology ................................................................................ 7 1.2.1 1.2.2 1.2.3 1.2.4 1.3 1.4 Survey of SAE Standards for Battery Chargers ............................................ 7 Battery Charger Classifications .................................................................... 9 Bi-directional Charger Topology and Charging Station ............................. 12 Overview of Vehicle to Grid (V2G) Technology ....................................... 15

Research Motivation: Enable Integration of Distributed Energy Storage Devices Contributions and Dissertation Outline ................................................................. 19

(Plug-in Hybrid Electric Vehicles) with Smart Grid ...................................................... 16 Chapter Two Design a Grid-Interactive Converter for Plug-in Hybrid Electric Vehicles ......... .......................................................................................................................................... 23 2.1 2.2 2.3 Definition of Grid-Interactive Converter .............................................................. 23 Topology Selection of Proposed Grid-Interactive Converter................................ 25 Power Stage Design of Proposed Converter ......................................................... 29 2.3.1 2.3.2 2.4 2.5 Passive Components Design ....................................................................... 29 Efficiency Test ............................................................................................ 34

Control Structure of Proposed Converter .............................................................. 36 Summary of Chapter Two ..................................................................................... 42

Chapter Three High Frequency Resonance Mitigation for Plug-in Hybrid Electric Vehicles Integration with a Wide Range of Grids ................................................................................. 43 3.1 3.2 3.3 3.4 High Order Filter Formation and its Negative Impacts ......................................... 43 Review of Active Damping Methods .................................................................... 45 Large Scale Penetration of Plug-in Hybrid Electric Vehicles into Various Grids 48 Modeling and Design of Adaptive Virtual Resistor Controller ............................ 50

vii

3.5 3.6 3.7

Adaptive Virtual Resistor Control for a Wide Range of Grids ............................. 62 Verification of Proposed Adaptive Virtual Resistor Controller with Different Summary of Chapter Three ................................................................................... 87

Grids ............................................................................................................................... 77 Chapter Four New Inductor Current Control based on Active Harmonics Injection for Plug-in Hybrid Electric Vehicles Vehicle to Home Application ....................................................... 88 4.1 4.2 4.3 4.4 4.5 Review of Control Methods for Single Phase Inverter ......................................... 88 Theoretical Analysis of the Proposed Control Method ......................................... 92 Steady State Operation and Dynamic Response of the Proposed Controller ........ 99 Investigation of Inductor Current Transient Response with Different Controllers ... ............................................................................................................................. 107 Summary of Chapter Four ................................................................................... 119 Chapter Five Power Management Strategy for Multiple Plug-in Hybrid Electric Vehicles in FREEDM Smart Grid ........................................................................................................... 120 5.1 5.2 5.3 5.4 5.5 5.6 Architecture of PHEV Integration with Solid State Transformer based Smart Grid ............................................................................................................................. 120 The Issue of Multiple Plug-in Electric Vehicles Connected with Solid State Proposed Power Management Strategy to Avoid Instability of Solid State Gain Scheduling Technique to Dispatch Power based on State Charge of Vehicles ............................................................................................................................. 147 Load management of Solid State Transformer by Managing Power of PHEVs . 150 Summary of Chapter Five ................................................................................... 163 Transformer................................................................................................................... 121 Transformer................................................................................................................... 127

Chapter Six High-Order Filter for Compact Size and Ripple Free Charging ....................... 164 6.1 6.2 6.3 6.4 Design GoalCompact Filter Size and Ripple Free Charging .......................... 164 Filter Design and Comparison with Conventional Filter .................................... 165 Controller Design ................................................................................................ 168 Controller Robustness Analysis .......................................................................... 177

viii

6.5 6.6

Simulation and Experiment Results .................................................................... 189 Summary of Chapter Six ..................................................................................... 195

Chapter Seven Conclusion and Future Work........................................................................ 196 7.1 7.2 Conclusion ........................................................................................................... 196 Future Work ........................................................................................................ 199

REFERENCES ..................................................................................................................... 201

ix

LIST OF TABLES Table 1-1 Plug-in Hybrid Electric vehicle charging level ................................................ 8 Table 2-1 Component count for H-bridge converter and three-leg converter ................ 27 Table 2-2 Power stage components in experimental setup ............................................. 33 Table 3-1 System configuration ...................................................................................... 65 Table 4-1 Output voltage with different types of load .................................................. 104 Table 4-2 Performance comparison of the capacitor current feedback, the inductor current feedback and the proposed method .......................................................... 118 Table 6-1 Core volume and loss of L-type filters ......................................................... 167 Table 6-2 Core volume and loss of LCL-type filters .................................................... 167

LIST OF FIGURES Figure 1.1 U.S oil consumption by sectors ............................................................................... 2 Figure 1.2 Oil production and consumption in U.S .................................................................. 2 Figure 1.3 Green house gas reductions with the adoption of PHEVs ....................................... 4 Figure 1.4 Predicted shares of new car sales in U.S market ..................................................... 6 Figure 1.5 Major models of PHEV and PEV ............................................................................ 7 Figure 1.6 Conductive EV/PHEV charging station and J1772 connector .............................. 10 Figure 1.7 structure of inductive charging system .................................................................. 12 Figure 1.8 integrated charger topology ................................................................................... 12 Figure 1.9 PHEV and PEV as distributed energy storage device (DESD) in FREEDM smart grid .......................................................................................................................................... 18 Figure 1.10 PHEV and PEV in FREEDM Smart House ........................................................ 19 Figure 2.1 Infrastructure of PHEVs integration with FREEDM smart grid .......................... 24 Figure 2.2 Topology of the proposed bi-directional charger .................................................. 25 Figure 2.3 Three-leg converter phase output voltage and spectrum at 10 kHz ...................... 28 Figure 2.4 H-bridge converter phase output voltage and spectrum at 10 kHz ....................... 28 Figure 2.5 Correlation of voltage ripple, input inductor and dc capacitor .............................. 32 Figure 2.6 Correlation of current ripple, input inductor and dc bus voltage .......................... 33 Figure 2.7 3D modeling of the proposed converter ................................................................ 34 Figure 2.8 Lab prototype of the proposed converter............................................................... 35 Figure 2.9 Efficiency DC/AC Stage ....................................................................................... 36 Figure 2.10 Efficiency DC/DC Stage ..................................................................................... 36 Figure 2.11 Control Structure for Grid to Vehicle Function .................................................. 38 Figure 2.12 Control Structure for Vehicle to Grid Function .................................................. 38 Figure 2.13 Vehicle to grid with PR+HC controller ............................................................... 40 Figure 2.14 Controller performance comparison: PI, PR and PR+HC controller .................. 40 Figure 2.15 Grid to Vehicle with PR+HC controller .............................................................. 41 Figure 2.16 THD comparison: PI, PR and PR+HC controller................................................ 41

xi

Figure 2.17 Vehicle to grid: comparison between vehicles input current and IEEE 1547 standard ................................................................................................................................... 42 Figure 2.18 Grid to Vehicle: comparison between vehicles output current and IEEE 1547 standard ................................................................................................................................... 42 Figure 3.1 Formation of high-order filter by converter LC filter and grid impedance ........... 44 Figure 3.2 Passive damping methods to eliminate high-frequency harmonics ...................... 45 Figure 3.3 Frequency characteristics of different grid conditions .......................................... 49 Figure 3.4 Grid current with the control loop having different virtual resistor values ........... 50 Figure 3.5 Variable virtual resistor based adaptive damping method .................................... 52 Figure 3.6 (a) Virtual resistor controller in synchronous frame ............................................. 52 Figure 3.6 (b) Measured one control cycle operation times for controller in synchronous frame ....................................................................................................................................... 52 Figure 3.7 (a) Virtual resistor controller in stationary frame .................................................. 53 Figure 3.7 (b) Measured one control cycle operation times for controller in stationary frame ................................................................................................................................................. 53 Figure 3.8 Transfer function: converter output to grid current with/without virtual resistor . 59 Figure 3.9 Transfer function converter output to capacitor current with/without virtual resistor ..................................................................................................................................... 59 Figure 3.10 Transfer function grid voltage to grid current with/without virtual resistor ....... 60 Figure 3.11 Transfer function grid voltage to capacitor current with/without virtual resistor 60 Figure 3.12 Block diagram of control plant and proposed controller ..................................... 61 Figure 3.13 Control loop modeling ......................................................................................... 61 Figure 3.14 Control Parameter Characteristics: adaptive gain Kad ........................................ 65 Figure 3.15 Block diagram of controller with adaptive virtual resistor loop.......................... 66 Figure 3.16 Frequency detection function block .................................................................... 66 Figure 3.17 Root locus of control plant with various impedances (0.2mH to 2.5mH) without virtual resistor ......................................................................................................................... 66 Figure 3.18 Root locus of control plant with various impedances (0.2mH to 2.5mH) and fixed virtual resistor ......................................................................................................................... 67

xii

Figure 3.19 Root locus of control plant with various impedances (0.2mH to 2.5mH) with adaptive virtual resistor ........................................................................................................... 67 Figure 3.20 Root locus of control plant with 0.2mH adopts proper virtual resistor ............... 68 Figure 3.21 Root locus of control plant with 2.5mH adopts proper virtual resistor ............... 68 Figure 3.22 Grid impedance vs resonant frequencies ............................................................. 69 Figure 3.23 Resonant frequency vs proper adaptive gain Kad ............................................... 69 Figure 3.24 Relationship of adaptive gain Kad, control loop bandwidth and phase margin for stiff grid ................................................................................................................................... 70 Figure 3.25 Relationship of adaptive gain Kad, control loop bandwidth and phase margin for weak grid ................................................................................................................................. 70 Figure 3.26 Bode plot of control loop with 0.2mH impedance and with adaptive virtual resistor control ........................................................................................................................ 73 Figure 3.27 Bode plot of control loop with 2.5mH impedance and with adpative virtual resistor control ........................................................................................................................ 74 Figure 3.28 Bode plot of control loop with 1mH impedance and with adpative virtual resistor control ..................................................................................................................................... 74 Figure 3.29 Controller robustness analysis for stiff grid control loop with grid impedance 20% variation .......................................................................................................................... 75 Figure 3.30 Controller robustness analysis for weak grid control loop with grid impedance 20% variation .......................................................................................................................... 75 Figure 3.31 Resonant frequency detection to determine grid impedance ............................... 76 Figure 3.32 Resonant frequency detection to determine grid impedance during impedance transient ................................................................................................................................... 76 Figure 3.33 Converter-side current and grid-side current with the proposed controller enabled ................................................................................................................................................. 78 Figure 3.34 Converter-side current and grid-side current with the proposed controller enabled ................................................................................................................................................. 79 Figure 3.35 Converter-side current and grid-side current without the proposed controller ... 79 Figure 3.36 Spectrum of grid-side current without the proposed controller........................... 80 Figure 3.37 Converter-side current and grid-side current with the proposed controller ........ 80

xiii

Figure 3.38 Spectrum of grid-side current with the proposed controller enabled .................. 81 Figure 3.39 Spectrum comparison of grid-side current with IEEE 519 standard ................... 81 Figure 3.40 Converter-side current and grid-side current with the proposed controller enabled ................................................................................................................................................. 83 Figure 3.41 Converter-side current and grid-side current with the proposed controller enabled ................................................................................................................................................. 84 Figure 3.42 Converter-side current and grid-side current without the proposed controller ... 84 Figure 3.43 Spectrum of grid-side current without the proposed controller enabled ............. 85 Figure 3.44 Converter-side current and grid-side current with the proposed controller enabled ................................................................................................................................................. 85 Figure 3.45 Spectrum of grid-side current with the proposed controller enabled .................. 86 Figure 3.46 Spectrum comparison of grid-side current with IEEE 519 standard ................... 86 Figure 4.1 Capacitor current feedback control ....................................................................... 91 Figure 4.2 Inductor current feedback control ......................................................................... 91 Figure 4.3 Inductor and load current feedback control method .............................................. 92 Figure 4.4 Control block of the proposed method based on active harmonics injection ........ 93 Figure 4.5 Harmonics detection and extraction block ............................................................ 96 Figure 4.6 Active harmonics injection before the inner current loop ..................................... 96 Figure 4.7 Nonlinear load tests with the proposed control method ........................................ 97 Figure 4.8 Nonlinear load tests with inductor current feedback control................................. 97 Figure 4.9 Comparison of capacitor current spectrum: the proposed method and conventional controller ................................................................................................................................. 98 Figure 4.10 Comparison of output voltage spectrum: the proposed method, conventional controller and IEC62040-3 Standard ...................................................................................... 98 Figure 4.11 Simulation 1kW load test with the proposed method ........................................ 100 Figure 4.12 Experiment 1kW load test with the proposed method....................................... 100 Figure 4.13 Simulation no load test with the proposed method ........................................... 101 Figure 4.14 Experiment no load test with the proposed method .......................................... 101 Figure 4.15 Simulation RL test 1kW resistive load and 2.5mH inductor with the proposed method................................................................................................................................... 102

xiv

Figure 4.16 Experiment RL test 1kW resistor load with 2.5mH inductor test with the proposed method ................................................................................................................... 102 Figure 4.17 Simulation nonlinear loads with the proposed method ..................................... 103 Figure 4.18 Experiment Nonlinear load test with the proposed method .............................. 103 Figure 4.19 Experiment nonlinear load test with the proposed method ............................... 104 Figure 4.20 Simulation a 1kW load transient for dynamic response test of the proposed controller ............................................................................................................................... 105 Figure 4.21 Experiment a 1kW load transient for dynamic response test of the proposed controller ............................................................................................................................... 106 Figure 4.22 Simulation a 1kW load transient for dynamic response test of the proposed controller ............................................................................................................................... 106 Figure 4.23 Experiment a 1kW load transient for dynamic response test of the proposed controller ............................................................................................................................... 107 Figure 4.24 Inductor current to load current with the proposed control method .................. 113 Figure 4.25 Inductor current to load current with the capacitor current feedback control ... 113 Figure 4.26 dynamic response: output voltage, load current, capacitor current and inductor current ................................................................................................................................... 115 Figure 4.27 Inductor current overshoot during the load transient with the proposed control method at L=1mH ................................................................................................................ 116 Figure 4.28 Inductor current overshoot during the load transient with the proposed control method at L=0.5mH ............................................................................................................. 116 Figure 4.29 Inductor current overshoot during the load transient with the capacitor current control at L=1mH.................................................................................................................. 117 Figure 4.30 Inductor current overshoot during the load transient with the capacitor current control at L=0.5mH............................................................................................................... 117 Figure 5.1 FREEDM smart grid and Solid State Transformer based Intelligent Energy Management System ............................................................................................................. 120 Figure 5.2 Control loop model of inverter stage of solid state transformer .......................... 123 Figure 5.3 Bode plot of close loop of inner current loop ...................................................... 125 Figure 5.4 Bode plot of outer voltage loop open loop .......................................................... 125

xv

Figure 5.5 Bode plot of close loop of outer voltage loop ..................................................... 126 Figure 5.6 Bode plot of output impedance............................................................................ 126 Figure 5.7 Controller architecture of PHEV charger ............................................................ 130 Figure 5.8 Proposed power dispatch method based on frequency restoration ...................... 130 Figure 5.9 Implementation of power-frequency control in inverter stage of SST ................ 131 Figure 5.10 SST operation frequency ................................................................................... 132 Figure 5.11 Charging Power of two vehicles........................................................................ 133 Figure 5.12 Enlarged charging power of two vehicles ......................................................... 133 Figure 5.13 Voltage and current information of no.1 vehicle............................................... 134 Figure 5.14 Voltage and current information of no.2 vehicle............................................... 134 Figure 5.15 SST operation frequency ................................................................................... 136 Figure 5.16 Charging power of two vehicles ........................................................................ 136 Figure 5.17 Enlarged charging power of two vehicles ......................................................... 137 Figure 5.18 Voltage and current information of no.1 vehicle............................................... 137 Figure 5.19 Voltage and current information of no.2 vehicle............................................... 138 Figure 5.20 SST operation frequency ................................................................................... 139 Figure 5.21 Charging power of two vehicles ........................................................................ 140 Figure 5.22 Enlarged charging power of two vehicles ......................................................... 140 Figure 5.23 Voltage and current information of no.1 vehicle............................................... 141 Figure 5.24 Voltage and current information of no.2 vehicle............................................... 141 Figure 5.25 SST operation frequency ................................................................................... 143 Figure 5.26 Charging power of two vehicles ........................................................................ 143 Figure 5.27 Voltage and current information of no.1 vehicle............................................... 144 Figure 5.28 Voltage and current information of no.2 vehicle............................................... 144 Figure 5.29 SST operation frequency ................................................................................... 145 Figure 5.30 Charging power of two vehicles ........................................................................ 145 Figure 5.31 Voltage and current information of no.1 vehicle............................................... 146 Figure 5.32 Voltage and current information of no.2 vehicle............................................... 146 Figure 5.33 Relationship of dispatched power and integration gain Ki with one vehicle at urgent charging and another one with various conditions .................................................... 149

xvi

Figure 5.34 Relationship of dispatched power and integration gain Ki with one vehicle at normal charging and another one with various conditions ................................................... 149 Figure 5.35 Relationship of dispatched power and integration gain Ki with one vehicle at mild charging and another one with various conditions ....................................................... 150 Figure 5.36 System operation frequency .............................................................................. 155 Figure 5.37 Charging power of the vehicle .......................................................................... 155 Figure 5.38 Grid voltage, current and dc bus voltage of the vehicle .................................... 156 Figure 5.39 System operation frequency .............................................................................. 156 Figure 5.40 Charging power of the vehicle .......................................................................... 157 Figure 5.41 Injection current from renewable energy and possessive load current.............. 157 Figure 5.42 System operation frequency .............................................................................. 158 Figure 5.43 Charging power of the vehicle .......................................................................... 158 Figure 5.44 Injection current from renewable energy and possessive load current.............. 159 Figure 5.45 System operation frequency .............................................................................. 159 Figure 5.46 Charging and discharging power of the vehicle ................................................ 160 Figure 5.47 Grid current, voltage and dc bus voltage of the vehicle .................................... 160 Figure 5.48 Grid current, voltage and dc bus voltage of the vehicle (zoom-in) ................... 161 Figure 5.49 System operation frequency .............................................................................. 161 Figure 5.50 power of no.1 and no.2 vehicle ......................................................................... 162 Figure 5.51 grid voltage, current and dc bus voltage of no.1 vehicle ................................... 162 Figure 5.52 grid voltage, current and dc bus voltage of no.1 vehicle ................................... 163 Figure 6.1 volume comparison between LCL filter and L filter at 10A and 30A charging . 168 Figure 6.2 filter loss comparison between LCL filter and L filter at 10A and 30A charging ............................................................................................................................................... 168 Figure 6.3 System control loop model .................................................................................. 172 Figure 6.4 Bode plot of system open loop transfer function................................................. 172 Figure 6.5 Bode plot of control plant and notch filter .......................................................... 173 Figure 6.6 System bode plot with proposed notch filter ....................................................... 173 Figure 6.7 Root locus of system with proposed method ....................................................... 174 Figure 6.8 System with proposed low pass filter .................................................................. 175

xvii

Figure 6.9 Comparison of system with low pass filter and without low pass filter.............. 176 Figure 6.10 Root locus of system with proposed low pass filter .......................................... 176 Figure 6.11 System Bode plot with its filter capacitor variation from 0.5 to 1.5 of original value ...................................................................................................................................... 179 Figure 6.12 System Bode plot with its battery side inductance variation from 0.5 to 1.5 of original value ........................................................................................................................ 180 Figure 6.13 System with 100uF and 120uF capacitance with notch filter controller ........... 180 Figure 6.14 System stable with 100uF capacitance and notch filter controller .................... 181 Figure 6.15 System stable with 120uF capacitance and notch filter controller .................... 181 Figure 6.16 System unstable with 60uF capacitance and notch filter................................... 182 Figure 6.17 System unstable with 40uF capacitance and notch filter................................... 182 Figure 6.18 Bode plots of different notch filter transfer functions to improve controller robustness.............................................................................................................................. 183 Figure 6.19 System stable with 60uF capacitance and redesigned filter parameters............ 183 Figure 6.20 System still unstable with 40uF capacitance and redesigned filter ................... 184 Figure 6.21 System with 40uF capacitor with different notch filter parameters to make loop stable ..................................................................................................................................... 184 Figure 6.22 Notch filter and control plant with all capacitor values (0.5~1.5)..................... 185 Figure 6.23 Control robustness test: charging current with the filter capacitance change from 80uF to 120uF ....................................................................................................................... 185 Figure 6.24 Control robustness test: charging current with the filter capacitance change from 80uF to 100uF ....................................................................................................................... 186 Figure 6.25 Control robustness test: charging current with the filter capacitance change from 80uF to 60uF ......................................................................................................................... 186 Figure 6.26 Control robustness test: charging current with the filter capacitance change from 80uF to 40uF ......................................................................................................................... 187 Figure 6.27 Simulation waveforms of three currents without proposed control .................. 190 Figure 6.28 Simulation waveforms of three currents with proposed control ....................... 190 Figure 6.29 Zoom-in waveforms of three currents with proposed control ........................... 191

xviii

Figure 6.30 Experiment results of converter side current, charging current and capacitor current with proposed control ............................................................................................... 191 Figure 6.31 Experiment results of converter side current, charging current and capacitor current with proposed control (charging current AC coupled to show ripple) ..................... 192 Figure 6.32 Experiment results of converter side current, charging current and capacitor current with proposed control (zoom-in) .............................................................................. 192 Figure 6.33 Experiment results of current transient response: 1A (0.1C) to 10A (1C) step change ................................................................................................................................... 193 Figure 6.34 Experiment results of current transient response: 10A (1C) to 1A (0.1C) step change ................................................................................................................................... 193 Figure 6.35 Experiment results of pulse charging with 100Hz ............................................ 194 Figure 6.36 Experiment results of pulse charging with 200Hz ............................................ 194

Chapter One Introduction

1.1 Research Background: Plug-in Hybrid Electric Vehicles Currently, there are three significant issues challenging the conventional transportation method in the United States. The first issue is the nearly 100% dependence on the imported oil. The United States holds only 3% of global petroleum however it consumes one fourth of the worlds oil supply. According to the U.S Department of Energy, the consumption figure was 20.5 million barrels of oil per day in 2004, more than half of which came from imports [1]. Figure 1.1 concludes that about two-thirds of this oil is refined into gasoline and diesel fuel to power passenger vehicles and trucks in America. So we can see that most of this imported oil is consumed by the transportation system. As can been in figure 1.2 U.S domestic oil production has decreased 44% since the 1970s, the use of oil for transportation has increased 83% and this gap is still widening. Therefore, transportation in todays America largely depends on the imported oil, and this vulnerability of relying on an unstable part of the world continues to threaten national security. The second issue is that fuel price has increased to a critical point; oil price has increased 200% from 1998 to 2006 [3]. It is predicted by M. King Hubbert Center that world oil production will reach its peak within the next 5~15 years [4]. Even though the price of oil may fall temporarily, over the long term the price will continue to rise. The cost of transportation will also continue to increase. The last but not the least issue is the environmental concern. Transportation is currently the single largest source of carbon dioxide emissions in the U.S, contributing over 30% of total green house gasses emissions. The growth rate of emissions from the transportation sector has averaged 24% between 1993 and 2003, faster than the growth of any other sectors such as

electrical generation. In addition urban air pollutants brought by petroleum combustion such as carbon monoxide, nitrogen oxides, and volatile organic compounds adversely affect public health and air quality. To respond to these challenges a revolutionary, environment-friendly safe and sustainable transportation approach is required.

Figure 1.1 U.S oil consumption by sectors [2]

25 Domestic Production Domestic Consumption 20

Petroleum (mmb/day)

15

10

Source: U.S. Department of Energy, Energy Information Administration

0 1940

1950

1960

1970

1980

1990

2000

2010

Figure 1.2 Oil production and consumption in U.S [3]

One of most promising transportation solutions is a newly emerging concept: Plug-in Hybrid Electric Vehicle (PHEV). According to IEEE definition, a plug-in hybrid electric vehicle is: any hybrid electric vehicle which contains at least: (1) a battery storage system of 4 Kwh or more, used to power the motion of the vehicle; (2) a means of recharging that battery system from an external source of electricity; and (3) an ability to drive at least ten miles in all-electric mode, and consume no gasoline [5]. In the future with the advance of battery technology and electric motor design, the hybrid drive train may be replaced by a pure electric drive train. Plug-in Electric Vehicle (PEV) a concept similar to EV (electric vehicle) will also become a promising solution. Plug-in Hybrid Electric Vehicle introduces significant usage of electricity as transportation fuel. Like hybrid vehicles on the market today, these plug-in hybrids use battery power to supplement the power of its internal combustion engine. However conventional hybrids obtain all of their propulsion power from gasoline, PHEVs obtain most of their energy from an electric utility. Currently U.S energy price for the cost of gasoline is $3 per gallon and the national average cost of electricity is 8.5 cents per kilowatt-hour. So a PHEV runs on an equivalent of 75 cents per gallon. Given that 50% of vehicles on U.S roads are driven 25 miles a day or less, a plug-in hybrid vehicle can reduce petroleum consumption by about 60% [1]. There is another assessment from Pacific Northwest National Laboratory, it claims by changing the transportation fuel from gasoline to electricity the PHEV can reduce gasoline consumption by 85 billion gallons per year, potentially displacing 52% of U.S oil imports, saving $270 billion in gasoline consumption and reducing 27% of total U.S green house gas emissions [6, 7].

Regarding environmental benefits with a high penetration of PHEVs on the road, there is one reasonable question: Is that possible using plug-in vehicles with increasing power demand from power plant will reduce one source of pollution (burning fossil fuel) but increase another (fuel based power generation). EPRI examined this possibility in a very comprehensive way by assessing the environmental gain of using electric transportation in most America regions in this century. In collaboration with Natural Resources Defense Council (NRDC), the assessment focuses on the probable environmental impacts of bringing a large number of PHEVs to roads over the next half century. The results show that the cumulative green house gas emissions were reduced to 3.4 billion metric tons by 2050 [8]. The relationship of green house gas emissions reduction and the adoption of PHEV are shown in figure 1.3 [12]. Moreover, this report addresses the point that for most regions of the United States, the increased PHEV usage would result in modest but significant improvements in air quality through the reduction of various pollutants and the substantial reduction of ozone levels.

Figure 1.3 Green house gas reductions with the adoption of PHEVs [12]

An economic assessment has also been conducted by EPRI. Assuming that within six major urban areas approximately about 50% of new-car sales are PHEVs. Substantial increases in household incomes are predicted from $188.7 million/year in the Birmingham region to $721.4 million/year in the Kansas region [8]. Based on the assumption of charging a vehicle at off-peak time, PHEVs provide a valuable potential of providing load leveling to utilities. By encouraging vehicle owners to recharge batteries at off-peak time, the grid could support a high level of PHEV penetration without the need for generating more power, and can improve power system efficiency by filling the generation valley at off-peak time. However there are still concerns that charging at off-peak time may result in another peak demand period if proper regulation is not adopted. This concern leverages the burgeoning field of smart grid technology. The NREL study indicates that no new power plants are required even with 50% PHEV market penetration [9], and according to Pacific Northwest National Lab report [6] given the average drive range of a car is 33 miles per day, the current U.S power grid capacity can fully supply approximately 70% of Americas passenger vehicles, that is, roughly 217 million cars. The predicted market penetration of PHEV from now to 2050 by EPRI is shown in figure 1.4 [10]. However whether the current utility generation capability can meet the growth of PHEVs in the future remains highly controversial. Finally, the auto industry and related research groups from national laboratories in the U.S show a strong interest in the PHEV. Test versions and modified versions of plug-in hybrids are already on the road. Some commercial versions will be available on the market by late 2011 or thereabouts. Worldwide, main-stream vehicle manufactures have joined in this new technology rapidly bringing many new products to the

market. In figure 1.5 the major models of PHEV and PEV in the current and near-future market are illustrated. The electric drive range is from 30miles to 300miles [11]. The booming market and new models show a bright tomorrow for PHEVs and PEVs.

Figure 1.4 Predicted shares of new car sales in U.S market [10]

From the above analysis we can see that PHEVs show great promise; they have the potential to curb emissions decrease gasoline usage and reduce the cost of transportation. Although large-scale adoption of plug-in vehicles is still a few years away, politicians, electric utilities, and auto companies are eagerly awaiting the opportunities that may arise from reduced emissions and gasoline consumption, new utility services and increased revenues, and new markets that will create new jobs. This is particularly exciting to electric utility companies, which can foresee substantial revenue growth through the electrification of transportation market. For consumers, plug-in vehicles will lower their operational costs when compared with traditional gasoline vehicles or todays gasoline-electric hybrids. The savings are potentially huge, as electricity cost per mile is calculated to be about one-quarter to one-third the cost of gasoline, depending on the region and price of gasoline. Thus PHEV

is a promising, efficient and sustainable solution to todays transportation challenge and a driving force to electrification of the transportation system.

Audi A1 E-tron

BMW ActiveE

BYD F3DM

Chevrolet Volt

Citroen Revolte

Fisker Karma

Ford Escape

Ford Focus

Honda Fit

Hyundai Blue-Will

Kia Ray

Mercedes S500 Vision

Nissan Leaf

Renault Fluence Z.E.

Suzuki Swift

Tesla Motors Roadster

Tesla Motors Model S

Toyota Prius

Toyota 2nd Gen RAV4

Volvo V60

Figure 1.5 Major models of PHEV and PEV

1.2 State of the Art of Technology 1.2.1 Survey of SAE Standards for Battery Chargers

SAE (North American Society of Automotive Engineer) is in charge of establishing all standards related with electric and hybrid electric vehicles in North America. According to SAE 2010s newest version of J1772 standard which is specially revised for the charging

infrastructure of plug-in hybrid electric vehicles, the charging infrastructure to PHEVs is classified into three levels, which is summarized in Table 1-1.
Table 1-1 Plug-in Hybrid Electric vehicle charging level [13]

Charge Method AC Level 1 AC Level 1 AC Level 2 DC Charging

Nominal Supply Voltage (Volts) 120V AC, 1-phase 120V AC, 1-phase 208 to240V AC, 1-phase Under development

Maximum Current (Amps-continuous) 12A 16A 80A

Branch Circuit Breaker rating (Amps) 15A 20A Per NEC 625

Definition of AC level I charging: a method of EV/PHEV charging that extends AC power from the utility to an on-board charger from the most common grounded electrical receptacle using an appropriate cord set. AC level I allows connection to existing electrical receptacles in compliance with the National Electrical Code-Article 625. Definition of AC level II charging: the primary method of EV/PHEV charging that extends AC power from the electric supply to an on-board charger. The electrical ratings are similar to large household appliances and can be utilized at home, workplace, and public charging facilities.The definition of DC level III charging: for PHEV application DC charging is still under development. It is cited in a previous version of J1772 that for EV level III charging is the conductive charging system architecture that provides a method for the provision of energy from an appropriate off-board charger to the EV in either private or public locations. The power available for DC Charging can vary from power levels similar to AC Level 1 and 2 to very high power levels that may be capable of replenishing more than of the capacity of the EV battery in as few as 10 minutes.

1.2.2

Battery Charger Classifications

There are several ways to classify battery chargers. Based on the type of connection there are conductive charging and inductive charging (contactless charging). Based on the power flow direction, there are uni-directional chargers and bi-directional chargers. Based on the utilization of the drive train, there are integrated chargers and stand alone chargers. Based on the location of battery charger, there are on-board chargers and off-board (stationary) chargers. Based on the number of power stages, there are two-stage chargers and single-stage chargers. Conductive charging is a direct coupling method which requires direct electrical contact between the charger and the batteries. It is achieved by connecting a charger to a power source with plug-in pads. Contrary to conductive charging, inductive charging is a contactless way of charging. It uses the electromagnetic field to transfer energy between the power source and the battery. Because there is a small gap between the primary coil and the secondary coil of the transformer, inductive charging can be considered one kind of shortdistance wireless energy transfer. It typically uses a primary coil to create an alternating electromagnetic field within a charging base, and a secondary coil in the moveable device picks up the power from the generated electromagnetic field and converts it back into electrical current, and finally this high frequency AC current will converted to DC to charge the battery. Inductive charging has the advantage of lower risk of electrical shock because the users are not exposed to conductors. High frequency inductive charging methods have been used in GM EV-1, Chevrolet S-10 EV and Toyota RAV4 EV. However, the disadvantage of this charging method is obvious: efficiency is the number one concern that the large air gap

10

between the primary side and secondary side reduces the coupling factor of the transformer and leads to lower efficiency. The reported efficiency of one industry product has achieved 86% [14] but still lower than conductive charging methods. In fact 2002 California Air Resources Board selected SAE J1772 conductive charging interface for electric vehicles in California. However high frequency inductive charging is still being investigated and improved [15-19], because it is preferred in some special application fields such as materials handling, clean factories for semiconductor manufacturing, liquid crystal displays, assembly plants and automatic movers in particularly harsh environment. An illustration of a conductive charging post and a 5-pin conductive connector in accordance to SAE J1772 is presented in figure 1.5 and the structure of inductive charging is shown in figure 1.6.

Figure 1.6 Conductive EV/PHEV charging station and J1772 connector

Compared to the uni-directional charger the bi-directional battery charger is preferred in future smart grid applications because it can achieve bi-directional power control, which is an essential element to enable the use of a renewable energy based smart grid. However, a bidirectional battery charger increases the cost of a whole vehicle and adds to control complexity as well. Power electronics equipment needs more active semiconductor switches, more gate drive circuits, and more powerful processors. The uni-directional battery chargers

11

available for PHEV in todays market are built by companies such as Delta Energy, Delphi, and Ford, etc. The Nissan Leaf has two charging ports [20], one AC charging port and one DC charging port. The AC charging port is used to connect level II electric vehicle supply equipment and a 6.6kW on-board charger. The DC charging port is connected to a DC offboard charging station in order to fast charge the battery. In order to reduce the cost of battery chargers for the whole vehicle system, the idea of utilizing the vehicle drive train inverter as the integrated charger instead of adding an additional charger is proposed [21-24]. When a vehicle is in an idle state, the inverter is not used. So its use as a battery charger wont affect the vehicle operation. Moreover the inverter naturally has bi-directional capability, so an integrated charger can be easily modified into a bi-directional charger. AC propulsions product AC-150 has an integrated 20 kW bidirectional grid power interface, which allows the power electronics and motor windings to be re-configured as a battery charger [21]. So far the concern to this integrated idea is whether the frequent utilization of the motor and inverter will affect the life time of the drive train. More test data regarding this impact are needed for the further analysis. However integrated charging is a creative and cost effective method for the massive production of plug-in hybrid vehicles. The integrated charger structure is shown in figure 1.7. The idea of utilizing the drive train as the battery charger is called integrated charger. This concept is implemented by AC Propulsion System and Oak Ridge National Lab and the charger system is drawn in figure 1.8.

12

Ls
Lp
Figure 1.7 structure of inductive charging system

Figure 1.8 integrated charger topology [25]

1.2.3

Bi-directional Charger Topology and Charging Station

A bi-directional battery charger can be either a two-stage solution or a single-stage solution. Two-stage indicates a bi-directional AC/DC stage and a bi-directional DC/DC stage. For a bi-directional AC/DC converter within level 1 and level 2 power rating and voltage level, an H-bridge based topology is usually adopted. The research area of converter

13

topology focuses on bi-directional DC/DC stage. In this stage, both non-isolated and isolated topology can be used. Half-bridge, full-bridge and push-pull are three building blocks which are used to construct the non-isolated topologies [26]. Half-bridge based buck boost topology is utilized [27]. Multi-phase half bridge based battery charger is proposed for higher power rating charging applications [28, 29]. The idea can also reduce the passive components. DCM modulation has been used to reduce the size of the magnetic components to increase power density [30]. With higher DC bus voltage, multilevel DC/DC converter is proposed to reduce the power losses and reduce the size of passive components [31]. Four-switch buck boost converter is proposed to handle a wide voltage range of battery [32]. For the non-isolated topology one point needs to be addressed. Due to safety requirement [29], all of the charging equipment must be isolated. So a charger with non-isolated DC/DC topology requires a line frequency transformer at the AC/DC stage. This low frequency transformer increases the weight and volume of the charger. A high frequency transformer is used in isolated topologies to achieve galvanic isolation and soft switching as well. The basic topologies include two sources at either primary side or secondary side: current source and voltage source [26]. Dual active bridge based topology is very popular [33]. Dual half bridge circuit is used in lower power rating applications [34]. For high power applications, a three phase dual active bridge is proposed [35, 36]. An inductor is used to form a current source in one side of the transformer and an active snubber is applied to reduce the voltage spike caused by this inductor [37, 38]. For isolated topologies, its advantages are higher power density and soft switching capable but the disadvantage is the loss of soft switching with different load conditions. Contrary to a two-stage structure, single stage topology will be a promising

14

solution which will reduce cost and increase system density. The controller design for singlestage topology will be also an interesting topic because the control structure is different compared to the well defined two-stage converter control structure. The charging station also called off-board charger is a concept for the fast charging approach. Until now the majority of commercial products called charging stations are actually level 2 charging equipment. The high power charging station for level 3 DC charging is a charging station for the Nissan Leaf built by Tokyo Electric Power Company (TEPCO). Besides the traditional method of using power from the utility to charge vehicles, ideas for using renewable energy resources at charging stations are proposed. Charging stations located in the public areas are designed for charging a large number of electric vehicles simultaneously. Moreover fast charging poses a high power demand to the utility. The current grid structure may not be suitable for such high power consumption units. By using the energy from renewable energy resources, the power systems burden is alleviated. The idea of solar energy based charging station is proposed to achieve green charging for vehicles [39-43] and the similar idea of using fuel cell to charge vehicles is proposed [44-48]. Municipal charging deck architecture [31] addresses the fast charging technique with the integration of available renewable resources and the utilization of ultra-capacitor to compensate for the power demand during the peak charging. In this dissertation based on SAE J1772 standard and the need for smart grid integration capability; a two-stage bidirectional stand alone conductive charger is proposed and its related control issues are investigated.

15

1.2.4

Overview of Vehicle to Grid (V2G) Technology

The average vehicle usage in U.S is only about one hour a day (in 2001, the average US driver drove 62.3 minutes/day) [49]. In other words, these cars are parked and idle most of the time. Suppose those cars were energy storage sources for the remaining 23 hours, and also suppose that the vehicle can discharge the batterys power back to the grid. The power system will have new service providers which are PHEVs. All of these techniques related to battery discharging and grid interaction belong to Vehicle to Grid (V2G) technology [49-58]. To understand vehicle to grid one can compare this technology to a solar power system. Vehicle to grid system is still in an early stage of research and development. The most important similarity between solar power and V2G is that V2G will probably connect to the grid in a highly distributed style at the same voltage level and with a similar power rating. The important difference is that V2G is bi-direction capable and may behave either as a source or a load. Solar power, for example, is only a source. Another important difference is that the primary goal of solar power is to generate power and supply to the load while the power generation function of V2G is only an additional function. With V2G capability, PHEVs can provide several services: spinning reserve, which contracts with the available PHEVs in order to provide power during unplanned outages of basic generators. Regulation service collects either real power or reactive power to help regulate systems voltage and frequency. Back-up service is one or more plug-in vehicles connected together to form as an autonomous grid during power outage in a certain area. Peak management is the service that a large number of plug-in vehicles are connected to help reduce system peak power demand. In addition, a large-scale adoption of PHEV V2G can be used to compensate for the

16

intermittent characteristics of renewable energy such as solar and wind by saving extra energy during peak times and releasing energy during valley times. Besides these functions PHEV should also have the capability to do islanding detection and have two-way communication with the grid. These are also the major functions of distributed generation. Moreover, the future vehicle should also have the low voltage ride through (LVRT) capability because at high penetration levels without ride through the anti-islanding tripping function will aggressively shut down many V2G vehicles. This will cause a momentary voltage sage, loss of loads and result in economic losses. The limitation for V2G technology is mainly due to the high cost of the plug-in hybrid electric vehicles. Two main costs are the components, notably the batteries and power electronics converter, and the labor for the conversion from gasoline power to electric power. Currently these high costs are predominantly due to low production volumes. However, unlike traditional power generation V2G can provide fast regulation service from clean energy. This clean power capability in addition to oil-free personal transportation tool and the future potential to support intermittent renewable energy sources will provide the environmental, system reliability, and energy security benefits. Thus, there is a case for finding policy support to initialize the fleets that are capable of V2G technology. Policy mechanisms might include tax credits granted for the purchase of plug-in vehicles. 1.3 Research Motivation: Enable Integration of Distributed Energy Storage Devices (Plug-in Hybrid Electric Vehicles) with Smart Grid It is believed that in the future the power grid will undergo revolutionary change; the new system will become more distributed with integration of large scale of renewable energy

17

sources and energy storage devices [59-61]. The widespread utilization of distributed energy at residential and industrial levels is a major paradigm shift for the electric power industry, moving away from todays centralized power generation paradigm toward a distributed generation based new grid [62]. With advanced communication methods and intelligent grid control the distributed grid will be upgraded to a smart grid. The distributed energy storage device (DESD) is an indispensible element in the formation of this future smart grid. It will complement distributed renewable energy resources (DRER), enable various types of grid regulation and supply backup power in islanding operations. The infrastructure of large scale interconnection of PHEVs with a power grid and other renewable resources is shown in figure 1.9. The interconnection of PHEVs with a home to form a smart home is shown in figure 1.10. It is critical to note that large scale penetration of PHEVs into the power grid cannot work without help from information technology. Advanced metering infrastructure (AMI), revenue-grade meters, various communications methods such as WIFI, cellular, Ethernet, Power line carrier and Zigbee are very important for the power management of these distributed storage devices. The smart grid impacts many of the operational and enterprise information systems, including supervisory control and data acquisition (SCADA), feeder and substation automation, customer service systems, planning, engineering and field operations, grid operations, scheduling, and power marketing. It is expected that there will be a significant number of plug-in vehicles and solar generation integrated into the distributed grid around 2012~2014 [64]. This will result in system overloads, voltage distortion, increased harmonics, increased line losses and unbalanced phase. To mitigate these issues

18

and to maintain system stability, coordinated voltage and reactive power control, automatic switches and extensive monitoring will be needed. Moreover a combination of distributed and centralized intelligent control, congestion management strategies, and market based dynamic pricing strategy will also be needed. The integration of DESD (PHEV and PEV) with the power grid is an absolute requirement in the future power system. The interface between the power grid and PHEVs power electronics technology will play a key role in this integration. Therefore the motivation of this dissertation is to design a power electronics interface to treat the PHEV as a distributed energy storage device (DESD), and integrate DESD with a future smart grid, and enhance the performance of this interaction, as well as manage the power among multiple DESDs (vehicles).

Figure 1.9 PHEV and PEV as distributed energy storage device (DESD) in FREEDM smart grid [63]

19

Figure 1.10 PHEV and PEV in FREEDM Smart House [65]

1.4 Contributions and Dissertation Outline There are several issues and challenges related to PHEVs interaction with the grid. The challenges can be divided into two groups: the power electronics level for individual vehicle and the power management level for multiple vehicles. At the power electronics level grid connection, a high performance grid connection controller is required. High quality current during either charging or discharging is essential to grid connection. However variable grid impedances compromise the control performance and complicate the control loop design. High frequency resonance appears on the grid charging/discharging current. Moreover the grid impedance is unknown and will compromise the controller with fixed compensation parameters. A new controller is needed to ensure high quality current in any kind of grid for the future large penetration of PHEVs. During off-grid operation, a new controller is needed to improve the quality of output voltage when connected with non-linear loads. In power

20

management level when multiple vehicles are connected together supplied under a power electronics transformer (solid-state transformer), the power dispatch control is important to the stability of the solid state transformer. If the total power demand of all the vehicles exceeds the capacity of the transformer it will cause transformer voltage collapse. High speed communication and system level intelligent control is the traditional method to address this issue. However due to the communication delay, congestion and the rapid speed of power electronics, the collapse may still happen. So a power electronics control scheme without communication is proposed to automatically adjust power demand of each vehicle in order to avoid voltage collapse. On the battery side, the charging current with low current ripple is highly preferred because low current ripple will reduce the heat and lengthen the lifetime of a battery. Normally, to reduce current ripple either passive components should be increased or switching frequency need to be increased. However, this will either increase the system size or reduce system efficiency. A method with compact size and low current ripple is desired. In this dissertation, the research efforts are directed to deal with these issues and new solutions are proposed to meet these challenges. The dissertation is organized as below: In Chapter I, the research background is introduced. State of the art of technology is reviewed. The research motivation is given. The research contributions and dissertation outline are presented. In Chapter II a grid-interactive smart charger for plug-in hybrid electric vehicles in smart grid applications is proposed. The proposed converter has three major functions: grid to vehicle, vehicle to grid and vehicle to home. The system infrastructure of PHEVs with grid is

21

proposed. The converter power stage is designed. The control structure for three functions is designed. In Chapter III a new adaptive virtual resistor controller is proposed to achieve high performance of power quality to assist large scale penetration of plug-in hybrid electric vehicles into various power grids. The modeling of the proposed controller is derived and analyzed. The control loop design for different grid conditions is proposed. The proposed method acts as a controllable resistor at various grid impedances. The control loop robustness is examined with control parameters mismatched at different grid impedances. The simulation and experiment results verify the proposed controller. In Chapter IV a new inductor current feedback control based on active harmonics injection concept is proposed for vehicle to home application of PHEVs. The active injection loop is designed and plugged into the loop to improve the harmonics compensation capability for nonlinear loads. The inductor current overshoot during the load transient is investigated for both inductor current feedback control and capacitor current feedback control. The inductor current based control can limit the current overshoot with an even smaller inductor value while the capacitor current based control cannot limit the current overshoot. The capacitor current feedback control has the potential to cause core saturation with a smaller inductor. So the proposed control method can be used to further reduce the passive components and optimize the volume and weight of the converter. In Chapter V a new power management strategy is proposed to solve the voltage instability issue of the Solid State Transformer (SST) which supplies multiple PHEVs. When multiple PHEVs are plugged into one SST based smart grid and the total demand power

22

exceeds the capability of SST, a new power dispatch method is proposed in each PHEV to adjust its power demand in order to avoid voltage collapse of SST. Gain scheduling technique is proposed to dispatch power to each vehicle based on batterys state of charge. The battery with low state of charge will get more power. A comprehensive case study is conducted to verify the proposed method. The proposed method can be used as the power electronics converter level control to improve the stability of solid state transformer. In Chapter VI: a high order filter is proposed for use in DC/DC stage of the battery charger. The objective is to reduce the filter size which will further reduce the system cost and volume. Another major goal is to effectively attenuate the current ripple of the charging current which will yield an almost ripple free charging for battery. Ripple free charging will eliminate the extra heat caused by the current ripple and increase battery life. The filter based controller is proposed to deal with the potential instability issue brought by the high order filter. The control loop design and robustness analyses are conducted and presented. The simulation and experiment results verify the proposed controller. In chapter VII, the final conclusions are drawn and future research topics are discussed.

23

Chapter Two Design a Grid-Interactive Converter for Plug-in Hybrid Electric Vehicles
2.1 Definition of Grid-Interactive Converter The term smart converter in this discussion refers to the converter with bi-directional power flow capability and the related functions. With bi-directional power flow capability the proposed grid-interactive converter for plug-in hybrid electric vehicles in household applications is presented [66]. The infrastructure of a PHEV integrated with an American House is shown in figure 2.1. In [67, 68] the circuitry configuration for a traditional American house is drawn in details, the AC mains and all connections are taken out of the house in order to show the wiring connection of PHEV with the house. In the United States electrical distribution scheme, one house receives input power from a split-phase distribution transformer that converts 13.2kV (line to line voltage) to a split-phase 240V/120V. The center-tapped transformer supplies 120V to normal home loads and 240V to heavy duty appliances such as electric oven and dryer. For a house installed with renewable energy capabilities, the generated electricity can be sold back to the grid so a bi-directional smart meter is used to calculate the net power consumption of this house and this meter is also known as net metering[69-71]. Regarding V2G function, both real power and reactive power control can be implemented. However, the grid code for PHEVs is not well established and reactive power control at residential level is still controversial. So only the real power control is designed for the V2G function, other functions related to reactive power such as power factor correction, reactive

24

power supplying and active power filter can be implemented but not included. Regarding the real power control, the grid code IEEE 1547-2008 [72] for distributed generator (DG) requires the disconnection of the DG from the house during the grid fault in order to provide for the safety of the maintenance personnel. However by adding the extra electric breaker and wire connection the DG can be still utilized to supply critical home loads during faults such as an uninterrupted power supply (UPS). This function is very important considering the power outage cases recently experienced due to natural disasters such as snow storms in the Northeast and hurricanes or floods in the Southern region. Therefore this capability of behaving as a UPS is also added in the proposed grid-interactive converter called Vehicle to Home (V2H). In total, the major functions for the proposed converter are grid to vehicle, vehicle to grid and vehicle to home.

Figure 2.1 Infrastructure of PHEVs integration with FREEDM smart grid

25

2.2 Topology Selection of Proposed Grid-Interactive Converter The topology of the proposed bi-directional battery charger is shown in figure 2.2. This bidirectional charger has a two-stage topology: stage 1 is a grid-side converter; stage 2 is a battery-side converter. A split-phase three-leg converter [73-77] is used as the grid-side converter in order to fit the household circuitry configuration. Compared with a traditional split-capacitor H-bridge, the center point of a three-leg converter is tapped to the middle point of the third leg rather than the middle point of the dc capacitors. The two half-bridge branches of the three-leg converter have the same uni-polar sinusoidal pulse width modulation method as an H-bridge converter, and the third half-bridge is controlled to keep the two 120V output voltage balance. Compared with a split-capacitor H-bridge converter, the three-leg converter has the following advantages: 1) no DC capacitor voltage balance issue; 2) comparatively smaller output filter size; 3) smaller DC bus current ripple; 4) higher utilization of DC bus voltage. The topology for the battery-side converter is a bi-directional buck-boost converter.

Figure 2.2 Topology of the proposed bi-directional charger

At grid to vehicle function, the converter transfers the power from the grid to charge the battery. The grid-side converter uses different half-bridges to converter AC power based on different input voltages. As shown in figure 2.2 if the input voltage is 240V the half-bridge

26

LA and LB will operate and if the input voltage is 120V AC the half-bridge LA or LB and LN will operate. In the proposed charger the power rating is set to 120V/5kW and 240V/10kW. Although the current rating exceeds the rating of home circuitry branches, higher power can help implement fast charging algorithms. The battery pack is composed of 90 lithium-ion battery cells and its terminal voltage is from 180V to 360V. At V2G and V2H function, the power inside the battery is inversely fed back to either the grid or the loads. At V2G mode, the grid-side converter operates in current-mode control which regulates the grid current to be a low-harmonics sinusoidal current. While at V2H the converter operates in voltage-mode control which regulates the output voltage to be sinusoidal with any type of load. The battery-side converter in both modes regulates the DC bus voltage by operation in the boost mode. The power feeding-back to grid is determined by the state of charge of the battery monitored by the battery management system (BMS) through CAN bus and the power demand from power system. Unlike the photovoltaic or fuel cell system which are utilized as much as possible, it is not desirable to use V2G systems continuously for a long time in order to preserve the health of the vehicles battery pack. When the power system needs power from the vehicles in a certain area the operators should choose vehicles with healthy batteries. Thus at V2G the duration of the service had better not be too long like the time scale for frequency regulation and spinning reserve.

27

Table 2-1 Component count for H-bridge converter and three-leg converter

IGBT Module(dual) Gate Drivers(dual) 2 3 DC link capacitor 2 1 (550V electrolytic capacitor) DC link voltage 2 1 sensor AC current sensor 2 2 Controller 1 1 A component count table for these two topologies is given in table 2-1. The current sensor and voltage sensor number for these two topologies are almost the same. The H-bridge converter needs two voltage sensors to monitor the DC link voltage to avoid unbalanced voltage. The phase output voltages and spectrums of the three-leg converter and the H-bridge converter at 10 kHz switching frequency are shown in figure2 and figure 3. The magnitude of the three-leg converters 120V phase output voltages dominant harmonics are reduced from 74.9% to 49% compared to that of the H-Bridge due to the modulation of the neutral branch. The neutral branch has the same switching frequency as the H-bridge branches and is synchronized with the H-bridge branches. This is because of the additional modulation of the neutral branch which supplies a zero-voltage level. The neutral branch will have the same switching frequency as the H-bridge branches. Note that normally uni-polar PWM generates harmonics at its multiple switching frequency however the neutral leg modulation generates harmonics at the switching frequency.

H bridge converter 2

Three-leg converter 3

28

500

400

300

200

100

-100

-200

-300

-400

-500 0.05

0.052

0.054

0.056

0.058 Time

0.06

0.062

0.064

0.066

Figure 2.3 Three-leg converter phase output voltage and spectrum at 10 kHz
250 200

150

100

50

-50

-100

-150

-200

-250 0.05

0.052

0.054

0.056

0.058 Time

0.06

0.062

0.064

0.066

Figure 2.4 H-bridge converter phase output voltage and spectrum at 10 kHz

Control effort for the two topologies is almost the same because the number of the voltage and current sensors, the PWM ports and the control algorithms for these two converters are similar. However, the control scheme differs with regards to the dc link capacitors voltage balance for unbalanced loads. For the three-leg converter there is no need to balance the capacitors, but in the case of the H-bridge converter it is very hard to balance the capacitors voltages unless there is an external circuit to charge the capacitors individually. In similar applications such as Photovoltaic and fuel cell power conditioning systems, a high frequency multi-winding transformer [78-82] in the dc/dc stage can be used to supply power to the different capacitors to keep the voltage balanced.

29

2.3 Power Stage Design of Proposed Converter 2.3.1 Passive Components Design

The first step is to design dc bus capacitor. When designing the dc bus capacitor for a single phase rectifier, the major disadvantage is the second-order harmonic on the dc bus, which needs a fairly large bus capacitor to smooth the dc voltage. Considering this capacitor an energy buffer between input AC power and output dc power, the capacitor value can be calculated and chosen based on its stored energy. Assuming the converter has unity power factor, the input power is:

Pin = u g ig =

UI UI cos 2t 2 2

(1)

Where the current and voltage are:

u g = U sin t

(2)

ig = I sin t The energy stored in the input inductor is: E= 1 2 L ( I sin t ) 2

(3)

(4)

Instantaneous power stored in the input inductor is:

1 2 (5) PL = E t = ( L ( I sin t ) ) t = LI 2 sin t cos t 2 The energy first passes through the input inductor and then the H-bridge finally charges the dc capacitor. Without considering devices power loss, the energy stored in the dc capacitor is the difference between the input energy and the energy stored in inductor: PC = Pin PL = UI UI cos 2t LI 2 sin t cos t 2 2 (6)

30

The dc component in (6) is supplied to the DC output, while the left second-order components would charge and discharge the capacitor which leads to the DC bus voltage ripple. By manipulating (6) to (7) and integrating the instantaneous power for a half cycle, the ripple energy is derived in (8):

LI 2 UI UI cos 2t + LI 2 sin t cos t = cos 2t + sin 2t 2 2 2

U 2 I 2 2 L2 I 4 U = + sin(2t + arctan ) LI 4 4
U 2 I 2 2 L2 I 4 + U I LI 4 4 sin 2tdt = + 4 4
2 2 2 2 4

(7)

EC =

(8)

During one switching cycle the energy difference which equals the energy stored in the capacitor is given in (9):
U 2 I 2 2 L2 I 4 + 1 4 4 EC = c[(Vdc + Vdc ) 2 (Vdc Vdc ) 2 ] = 2c Vdc Vdc = 2

(9)

From the ripple energy stored in capacitor we can derive the correlation between dc capacitor, dc bus voltage ripple and input inductor, given by equation (10) and graphically presented in figure 2.5.
U 2 I 2 2 L2 I 4 + 4 4 C= 2 Vdc Vdc

(10)

Set the dc bus voltage ripple cannot exceed more than 5% of the nominal dc bus voltage; the dc capacitor value is selected as 2mF.

31

The second step is to choose the filter components. The filter inductor for the input/output filter is designed based on the current ripple on that inductor. At any given time, the ripple current can be calculated based on worst case ripple current.

I pk =

V t (VDC U sin t ) D = L 2L fs

(11)

Where the duty cycle D is determined in (12): D= U sin t VDC M sin t = = M sin t VDC VDC (12)

Then the peak current can be represented by the function of dc bus voltage and modulation index M in (13):
I pk = VDC (1 M sin t ) M sin t 2Lf fs

(13)

Set the current ripple to a proper percentage of the rated current, the inductor value is determined in (14):
Lf = VDC (1 M sin t ) M sin t 2 f s I pk

(14)

Here, UDC is the bus voltage with voltage ripple, Usint is the instantaneous value of AC input voltage at the positive cycle, and fs is the switching frequency, U is the peak magnitude of AC input voltage, and M is the PWM modulation index. Based on equation (14) the correlation between the dc bus voltage, input inductor and current ripple is described the by 3-D drawing in figure 2.6. The inductor value is chosen to be 0.75mH and the ripple current is 6.6A which is around 10% of the peak output current (58.9A). To calculate the filter capacitor, the LC filter is to damp the harmonics of the output voltage. Equation (15) shows it can achieve better performance with higher LC value. However, the output capacitor value could not be too large otherwise too much power will be stored in the capacitor. It is normally said that less than 10% of the rated power could be stored in the capacitor. The

32

filter capacitor value is calculated and chosen as 50uF. The power stage components and parameters are listed in table 2-2.
1 1 Cf = 2 f res L f
2

(15)

Also the corner frequency of the LC filter, the capacitor value should not exceed 5% the system base capacitance otherwise it will affect the power factor and absorb too much reactive power on the capacitor

5% Pn (16) n Vn2 Here, n is the fundamental frequency, Vn is the output voltage and f res is the filter corner C frequency.
correlation of voltage ripple, input inductor and dc capacitor
-3

x 10 6 5 dc capacitor(F) 4 3 2 1 0 3

2 x 10
-3

250 200 1 150 100 0 50 0 voltage ripple(Volt)

input inductor(H)

Figure 2.5 Correlation of voltage ripple, input inductor and dc capacitor

33

correlation of input inductor, current ripple and DC bus voltage

440

DC bus voltage (V)

420

400

380

360 15 10 5 current ripple (A) 0 1 0 input inductor (H) 3 2 x 10

-3

5 4

Figure 2.6 Correlation of current ripple, input inductor and dc bus voltage Table 2-2 Power stage components in experimental setup

Power Rating

10kW (240V network) 5kW (120V network)

Power Devices IGBT Gate Driver DC Bus Capacitor AC Filter Inductor AC Filter Capacitor DC Filter Inductor DC Filter Capacitor

CM150DY-12NF BG2B 2mF 1mH/40mohm 50uF 0.5mH/20mohm 220uF

Powerex Powerex Electronics Concept Magnetics Kool Mu GE-Regal Capacitor Magnetics Kool Mu Epcos Corporation

34

2.3.2 Efficiency Test The prototype is designed based on the structure of a commercial inverter. The 3-D modeling is done by COSMOS SolidWorks and shown in figure 2.7. The hardware prototype is constructed in laboratory and shown in figure 2.8.

Figure 2.7 3D modeling of the proposed converter

35

Figure 2.8 Lab prototype of the proposed converter

After the construction of the prototype, an efficiency test is performed to test the converter reliability and measure efficiency. For the DC/AC stage the DC bus voltage is 400V, switching frequency is 10 kHz and modulation index is 0.85. The DC/DC stage is tested in the same way. Note that the efficiency data coincide with the value calculated by a software package provided by Mitsubishi [83]. The only difference is that the passive components loss is not included in the software. The measured efficiency for the DC/AC stage and DC/DC stage is shown in figure 2.9 and figure 2.10 respectively.

36

AC/DC stage efficiency 96.50% 96.00% 95.50% 95.00% 94.50% 94.00% 93.50% 93.00% 1000W 2000W 5000W 6500W 8000W 10000W

Figure 2.9 Efficiency DC/AC Stage

DC/DC stage efficiency

98.50% 98.00% 97.50% 97.00% 96.50% 96.00% 95.50% 95.00% 94.50% 1000W 2000W 5000W 6500W 8000W 10000W

Figure 2.10 Efficiency DC/DC Stage

2.4 Control Structure of Proposed Converter The control structures for both Grid to Vehicle and Vehicle to Grid belong to grid-tied current mode controller [84-86]. Typically double-control-loop structure is used: inner current loop and outer voltage loop. The inner current loop achieves good current tracking, low current harmonics and fast transient response. The outer voltage loop regulates the DC bus voltage. The control structure for grid to vehicle is shown in Figure 2.11; it is composed of a double-loop structure for the AC/DC stage and a single loop for DC/DC stage. The DC/DC stage has a single loop structure for grid to vehicle. Battery management system

37

monitors the battery pack and acquires a serial of parameters such as state of charge (SOC), state of health (SOH), voltage and temperature. The controller will decide how to charge this battery based on its condition. Different charging algorithms are implemented in the controllers which are constant-current constant-voltage charging algorithm, pulse charging algorithm and Reflex charging algorithm [87, 88]. These charging algorithms dont pose a challenge to power electronics controller because the battery is a very stiff plant with slow response. Regarding pulse charging algorithm, it is believed to be the best charging algorithm [89] because it uses battery AC impedance theory and tries to charge the battery at the lowest impedance. Variable AC impedance theory has already been verified in the field of Electrochemistry [90-94]. So to design a good charging algorithm it is essential to work closely with batteries electrochemical characteristics. Figure 2.12 shows the multi-loop structure for vehicle to grid function. Compared to distributed generation controller, the outer voltage loop is regulated at the DC/AC stage and only the inner current loop is needed. The current reference is generated by the power system. When the power system needs vehicles to do the service such as frequency regulation and load leveling, an operator will arrange a number of plug-in vehicles and send the power requirement. The reason to use DC/AC to support the dc bus voltage is that during the mode transfer between G2V and V2G, this control structure will lead to a smooth transition.

38

ig

idc

Vgrid

* idc

* Vdc

* ig

idc

Vdc

Vgrid

ig

Figure 2.11 Control Structure for Grid to Vehicle Function

DC bus DC stage Battery GFI Breaker AC stage

idc

Vgrid
AC Grid

Battery Management System power system demand

CAN feedback current

* idc +

dc

PI

PWM Boost operation

* Vdc+

* ig

Vdc

PI

+ -

CR

SPWM

Vgrid

PLL

ig

Figure 2.12 Control Structure for Vehicle to Grid Function

Regarding the current regulator, proportional plus Resonant (PR) controller [95-98] has been proven that it has better performance than a proportional integral (PI) controller in stationary frame. More theoretical analysis shows that the proportional plus resonant controller in stationary frame is equivalent to proportional integral controller in rotating frame. So in stationary frame a PR controller is more effective at achieving zero steady-state error and improves the reference tracking capability. The controller equation is written in equation (19): Gc ( s ) = K p + K i 2c s s + 2c s + 02
2

(17)

39

Here Kp determines the dynamic response of the controller, Ki adjusts the gain of the setting frequency, with a higher gain the error is reduced, c is the cutoff frequency which is much smaller than 0, and 0 is the resonant frequency which is set to 376.8 rad/s. Based on PR controller, a series set of resonant blocks are utilized in particular to eliminate several selected low order odd harmonics [99]. Similar to PR controller, a series of resonant controller cascaded together are tuned to the desired low order odd frequencies to further reduce current harmonics. The PR controller with selective harmonics elimination blocks are shown in (20): Gh ( s ) = K p + K i 2c s 2c s + K ih 2 2 s + 2c s + 0 h =3,5,7 s + 2c s + (h0 ) 2
2

(18)

Here Kih determines the gains of the low order odd harmonic, h is the odd harmonics. In this section the V2G function and G2V function are tested with three different types of controller in stationary frame. First the V2G function is tested with PI, PR and PR+HC controller with 1.2kW. The result with PR+HC controller is shown in figure 2.13. With stationary frame PI the control error, especially the phase error, still exists. With stationary PR controller the phase error is largely reduced because of the resonant gain at the fundamental frequency, however the low-order harmonics such as 3rd, 5th and 7th harmonics still exist. With PR+HC controller both the phase error and the low-order harmonics are reduced. The detailed comparison for the individual harmonic among the three controllers is shown in figure 2.14. With the spectrum of the output current shown in the figure the PR+HC has the best performance.

40

Figure 2.13 Vehicle to grid with PR+HC controller (Purple curve: grid voltage; Red curve: output current)
PR+HC 4.50% 4.00% 3.50% 3.00% 2.50% 2.00% 1.50% 1.00% 0.50% 0.00% 3rd 5th 7th 9th 11th 13th 15th 17th 19th PR PI

Figure 2.14 Controller performance comparison: PI, PR and PR+HC controller

After V2G test, the G2V experiment is conducted. The current controller is the same as that used in V2G function. A steady-state operation for 1kW charging is shown in figure 2.15. In figure 2.16 the total harmonics for three controllers are used in G2V function with different input current. Finally the current performance for both V2G function and G2V function is listed and compared with the standard for distributed generation IEEE 1547-2008.

41

It can be seen from the figure 2.17 and figure 2.18 at 1.2kW the total harmonics and the individual harmonic meet the standard.

Figure 2.15 Grid to Vehicle with PR+HC controller (Blue curve: dc bus voltage; Red curve: grid voltage; Purple curve: input current)
PR+HC 9.00% 8.00% 7.00% 6.00% 5.00% 4.00% 3.00% 2.00% 1.00% 0.00% 2.5A 5A 7.5A 10A 15A PR PI

Figure 2.16 THD comparison: PI, PR and PR+HC controller

42

IEEE1547 5% 4% 4% 3% 3% 2% 2% 1% 1% 0%

output current

Figure 2.17 Vehicle to grid: comparison between vehicles input current and IEEE 1547 standard
IEEE 1547 5% 4% 4% 3% 3% 2% 2% 1% 1% 0%
3r d 5t h 7t h 9t h 11 th 13 th 15 th 17 th 19 th 21 th 23 th 25 th 27 th 29 th 31 th 33 th 35 th 37 th 39 th

Figure 2.18 Grid to Vehicle: comparison between vehicles output current and IEEE 1547 standard

2.5 Summary of Chapter Two In this chapter, a bi-directional grid-interactive converter with multiple functions for PHEV is proposed. It achieves three major functions: grid to vehicle (G2V), vehicle to grid (V2G) and vehicle to home (V2H). The system infrastructure, operational principles are illustrated. The hardware design and control structure for different functions are presented.

3r d 5t h 7t h 9t h 11 th 13 th 15 th 17 th 19 th 21 th 23 th 25 th 27 th 29 th 31 th 33 th 35 th 37 th 39 th

input current

43

Chapter Three High Frequency Resonance Mitigation for Plug-in Hybrid Electric Vehicles Integration with a Wide Range of Grids

3.1 High Order Filter Formation and its Negative Impacts The proposed converter has an LC output filter to provide high quality current and voltage, but with the connection of grid impedance the converter is connected with an LCL filter. The inductor-capacitor-inductor (LCL) filter is widely utilized in the grid integration application such as renewable energy interconnection, high performance regenerative rectifier, etc. [100-103]. It is placed between a voltage source converter and the grid. Normally, using L-type filter to fulfill existing grid codes such as IEEE 519-1992 [104], IEC61000-3-4 [105] and IEEE 1547 [106], a large inductor value should be used. However, the large inductor reduces the dynamic performance of the converter, also increases the system cost and volume. The LCL filter is a third-order low-pass filter which effectively attenuates the current ripples, with a smaller converter side inductor the current through the grid is almost ripple free. Furthermore in high power application where the switching frequency is limited due to the power losses, using LCL filter can help improve the output current quality with lower switching frequency. To summarize the advantages of LCL-type filter over L-type filter, two major aspects need to be addressed. First is higher attenuation of the harmonics. The attenuation rate of LCL filter is 60dB per decade compared to 20dB per decade of L-type filter. The second point is lower inductance compared to L-type filter and better dynamic response. In the proposed converter, grid impedances will vary with different grid conditions, i.e., the leakage inductance of isolated transformers and the inductance from

44

a long charging cable. The figure 3.1 also shows the converter is connected with the grid through variable inductances and its LC filter.

if
Lf
Cf

ig

Lg

Figure 3.1 Formation of high-order filter by converter LC filter and grid impedance

Although an LCL filter has the advantages of low inductor value and current ripple, its drawbacks are notable. Because of its high order resonance characteristics once excited by high frequency harmonics, the filter will lead to resonant oscillation. At the natural resonant frequency of the filter, a filter capacitor can be considered very low impedance almost to the point of short circuit so it will draw harmonic currents around resonant frequency. It will lead to high voltage oscillation on the capacitor voltage and the grid current. This oscillation will distort the grid current, increase power losses, and trigger the converter protection and may even lead to the system instability. The resonant frequency can come from both the converter PWM voltage output and the grid voltage harmonics though the grid side harmonics will be much lower than the converter output. To solve this low impedance issue at the resonant frequency, passive damping methods are proposed. The passive damping method normally utilizes a power resistor series with the filter capacitor to increase the impedance of this branch at the resonant frequency. But it greatly increases system efficiency, for example with 120/1.5kW converter 10ohm damping resistor the power loss accounts to 3.41%. The modification method shown in figure 3.2 has been proposed to replace the only resistor with

45

an inductor paralleled with a capacitor and then series with a resistor. In this way the power loss on the resistor is largely reduced. But this will increase the system cost and volume and more importantly the tuning frequency cannot be changed. Because of the simplicity and high reliability, the passive solutions are adopted in industry. However, additional power losses and the cost of inductor and resistor are the major drawbacks of this method. Another notable drawback is that usually the passive damping method is effective in a certain range of resonant frequency but cannot be effective in a large range of resonant frequency. But the grid impedance cannot remain constant so the resonant frequency varies. A passive filter with fixed frequency cannot effectively attenuate the resonant. Thus an active means of attenuation with loss free and adaptive tuning capability is a preferred and promising method.

Lf
Vconv
Cf

Lg

Lf
Vconv
Cf

Lg

Figure 3.2 Passive damping methods to eliminate high-frequency harmonics

3.2 Review of Active Damping Methods Researchers have proposed lots of good active damping methods to address the unstable problem. The filter based controller is proposed to extract and eliminate the resonant components in the control loop. A genetic algorithm is proposed to be used as the tool to choose the right parameters for the filter based controller [107]. Further development based on the band-pass filter, virtual flux is used to eliminate addition current sensors [108]. A

46

digital infinite impulse type of filter is proposed to improve stability [109]. Different types of filter based controllers are studied and compared [110]. The control parameters e.g., gain and time constant are tuned to improve the phase margin at the resonant frequency in order to make control loop stable [111]. The effect of the sampling frequency to the stability of the control loop is analyzed. Compared to the resonant frequency, a much faster sampling frequency is needed [111, 112]. A hybrid controller which has a virtual harmonic damper and a three-step posicast compensator is proposed to damp the resonance resulting from the LC input filter [113]. In order to maintain control loop stable, it is beneficial for the controller to know as many control variables as possible. Beside the single grid current control loop, the converter side current and the filter capacitor current or voltage can be used [114-123]. An LCL filter had been altered to an LCCL filter so two control variable, the grid current and part of filter capacitor current can be measured by one current sensor [124]. By sensing the grid current and the filter capacitor voltage, an additional admittance feed-forward path is added for fuel cell applications [125]. A direct power control (DPC) is modified to integrate active damping control with sensing the capacitor filter voltage [126]. With multiple control variables sensed and used in the controller, PI state space controller is proposed [112, 127130]. The state variable estimator is combined with state space controller [131-132]. Predictive current control has been applied with state space controller [133, 176-179]. A robust controller based on H-infinity theory is proposed to tune the control parameters for different grid impedances [180]. Passivity theory has been applied to examine the convergence of the controllers state trajectories [181]. Discrete sliding mode current controller is proposed for active damping [182-184]. Compared to PI based controllers, these

47

controllers based on modern theories have not yet been well accepted in practical applications. The LCL filter can be also utilized in microgrid applications which switch between grid-connection operation and stand-alone operation [185, 186]. Virtual resistor based damping method [118, 187-190] is very similar to the passive damping concept. The controller senses either the filter capacitor current or voltage to emulate a resistor in series with the filter capacitor to resolve the resonant issue. However, the virtual resistor based controller has the drawback that its virtual resistor value highly relies on the value of LCL filter. If the values of an LCL filter change, the performance of virtual resistor controller is compromised. Moreover, in order to connect the PHEVs with a wide range of grids the controller needs to be stable with different grid conditions. It means that the virtual resistor controller should not only have variable resistor values to enhance the controller performance at different grid conditions, but also the adaptive tuning capability to adjust the virtual resistor value based on the grid conditions automatically. Some of the literature regarding the virtual resistor controller has discussed about one set of controller at the different resonant frequency mainly caused by filter capacitor. However, this type of controller has seldom been adopted and examined in practical grid conditions for PHEV applications. The stability of the virtual resistor controller at different grid conditions needs to be examined. The virtual resistor controller with adaptive tuning capability behaves like a controllable resistor. This method is desired for this type of application which is for a wide range of gird impedance. In this chapter, an adaptive virtual resistor controller to achieve high power quality for plug-in hybrid electric vehicles is proposed. Since current literature lacks the stability analysis of

48

controller of PHEVs under different grid conditions, this paper analyzes and designs the controller to work with a large set of grids stably. 3.3 Large Scale Penetration of Plug-in Hybrid Electric Vehicles into Various Grids With increased number of commercial plug-in vehicles available on the market, there will be a large number of PHEVs connected to the power grid in the future. One remarkable issue of the large scale penetration of PHEVs to the grid is the power balance issue. An intelligent power allocation and management method leveraged with cyber technology is proposed to alleviate this grid collapse [134]. The issue caused by the power electronics converter needs to be addressed. When connected with the grid, the filter of the bi-directional charger combining with the grid impedance forms a high order filter. This high order filter will generate high frequency oscillation. When the PHEVs are largely adopted into the grid the vehicles will be connected with a diversity of grids. For example in a remote area (rural area) or in an isolated location far from the distribution transformer, the grid is highly inductive and is also referred to weak grid [135-137]. The grid configuration becomes even more complicated when the home appliances are considered. Because there is an equivalent capacitance correcting power factor [138]. In figure 3.3 the Bode plot of control plant with different grids is shown. The control plant means the transfer function of the control input to the grid current. The grid impedance in this figure is 0.2mH, 0.5mH and 2.5mH respectively. As we can see that with the variations of grid impedance, the control plant also changes. Therefore, a single virtual resistor value which is specially designed for one grid impedance cannot compensate for all the grid conditions. In figure 3.4, the grid current with control loop having different virtual resistor values is shown. If the control loop designed for the grid

49

impedance 2.5mH is used at a 0.2mH grid, the control loop is not stable. The grid current changes from stable operation to oscillation. The high frequency resonance appears on the grid current. Thus, the control loop with a fixed virtual resistor value cannot be applied to a wide range of grids.

Bode Diagram

50

L=0.2mH L=0.5mH L=2.5mH

Magnitude (dB)

-50

-100 -90 -135

Phase (deg)

-180 -225 -270 1 10

10

2 Frequency (Hz)

10

10

Figure 3.3 Frequency characteristics of different grid conditions

50

15 grid current

10

-5

-10

-15

0.24

0.26

0.28
Time

0.3

0.32

0.34

0.36

Figure 3.4 Grid current with the control loop having different virtual resistor values

3.4 Modeling and Design of Adaptive Virtual Resistor Controller Inside a high order filter the capacitor path is the weakest path which always allows high frequency current to flow through. To eliminate the weakest patch various passive damping methods [139] place resistors, inductors and capacitors either series or parallel with the capacitor in order to absorb the high frequency components. This idea will enlighten the active controller design to come out with a variable virtual resistor to be placed in series with the capacitor. The capacitor current is sensed and multiplied with an adaptive gain which represents a variable resistor. The multiplication of capacitor current and the gain is added to the input of the PWM generator to eliminate the resonant frequency components before they go into the PWM generator to make a converter output as resonance excitation source. The virtual resistor only exists in the control loop not physically connected in the converter

51

system. The virtual resistor value can be changed based on different grid impedances. In this paper an adaptive virtual resistor method is proposed. Figure 3.5 the idea of proposed adaptive virtual resistor method is drawn. As shown in this figure the proposed controller emulates a variable resistor to be connected with the capacitor. The LCL filter is in between with two voltage sources: converter PWM output voltage and grid voltage. The virtual resistor based controller for the single phase application can be divided into two types: Stationary Frame Control and Synchronous Frame Control. Figure 3.6 (a) describes the architecture of virtual resistor based controller in synchronous frame. 3.6(b) shows the measured one control cycle operation time for proposed controller in synchronous frame. One control cycle takes around 34us. Figure 3.7 (a) describes the architecture of virtual resistor based controller in stationary frame. 3.7 (b) shows the measured one control cycle operation time for proposed controller in stationary frame. It takes around 12us. So to compare the controller in these two control frames the stationary frame has faster calculation and shorter time duration. So it occupies fewer of the digital processors resources. Synchronous frame takes more time because it contains a single phase d-q transformation. Regarding the performance the stationary frame with PI controller cannot compete with synchronous frame because the control objective is at 60Hz not dc component. But proportional resonant (PR) largely improves the gain at 60Hz and the steady-state error for both the magnitude and phase can be eliminated theoretically. PR controller in [97] has been proved to be equivalent to PI in synchronous frame. To save the resource of digital process virtual resistor controller in stationary frame with PR and HC (harmonic cancellation) control is used.

52

Figure 3.5 Variable virtual resistor based adaptive damping method

Figure 3.6 (a) Virtual resistor controller in synchronous frame

Figure 3.6 (b) Measured one control cycle operation times for controller in synchronous frame

53

Vgrid Vdc Vdc-ref +

PLL

sin
PI

ig-ref

+ -

PR+HC +

SPWM

gating signals

ig icf
Ga

Stationary frame

Figure 3.7 (a) Virtual resistor controller in stationary frame

Figure 3.7 (b) Measured one control cycle operation times for controller in stationary frame

A. Modeling of Virtual Resistor Controller For single phase stationary frame application the transfer functions can be derived directly such variables are in dc format [191]. As shown in figure 3.5 that the grid current ig is the control objective. It has two voltage sources affecting it based on superposition. The transfer function for grid current to converter output voltage and grid voltage should be derived. The capacitor current ic is also important to the virtual resistor based method because it emulates a resistor in series with the capacitor. So the relationship of capacitor current to converter output voltage and grid voltage is derived. The transfer function will be used in the next step to help control loop design. Use the circuit shown in figure 3.1 and ignore the inductor

54

winding resistance. The equations show the relation between the grid voltage, grid current and the converter output. di f (t ) dt dig (t ) dt
= uconv (t ) ucf (t )

Lf

(1)

Lg

= ucf (t ) u g (t )

(2)

cf

ducf (t ) dt

= ic (t )

(3)

i f (t ) = ic (t ) + ig (t )

(4)

Convert equations (1) ~ (4) from time domain to frequency domain to get equations (5) ~ (8): s L f i f ( s ) = uconv ( s ) ucf ( s ) s Lg ig ( s ) = ucf ( s ) u g ( s ) s c f ucf ( s ) = ic ( s ) i f ( s ) = ic ( s ) + ig ( s) (5) (6) (7) (8)

To consider the transfer function of grid current to converter output, the grid voltage is considered constant without any disturbance. So in small signal model the grid voltage is zero. So put (6) into (7) and (8) the grid current ig can be used to substitute capacitor voltage and converter side current. sLg ig ( s ) = ucf ( s ) (9)

55

s 2 c f Lg ig ( s ) = ic ( s )

(10)

With the capacitor current and grid current derived, the converter side current can be represented by grid current also.
( s 2 c f Lg + 1)ig ( s ) = i f ( s )

(11)

Substitute equation (11) to (5) to get the transfer function of converter output to grid current:
uconv ( s ) = sL f ( s 2 Lg c f + 1)ig ( s ) + sLg ig ( s ) = ( s 3 Lg L f c f + s ( Lg + L f ))ig ( s )

(12)

ig ( s ) uconv ( s )

1 Lg L f c f s + ( Lg + L f ) s
3

(13)

After derivation of converter output to grid current, the converter output to capacitor current is easily derived:
uconv ( s ) = sL f ( s 2 Lg c f + 1)ig ( s ) + sLg ig ( s ) = ( s 3 Lg L f c f + s ( Lg + L f )) s 2 c f Lg

ic ( s )

(14)

Lg c f s ic ( s ) = uconv ( s ) Lg L f c f s 2 + Lg + L f

(15)

In the next step, the virtual resistor loop is plugged into the controller. The capacitor current is sensed and multiplied with adaptive gain Kad, so the equation (5) will be rewritten as: s L f i f ( s ) = uconv ( s) kad ic ( s) ucf ( s ) (16)

56

So to substitute (11) to (16), the transfer function with virtual resistor is included in (17): uconv ( s ) = sL f ( s 2 Lg c f + 1)ig ( s ) + sLg ig ( s ) + s 2 kad Lg c f ig ( s ) = ( s 3 Lg L f c f + s 2 kad Lg c f + s ( Lg + L f ))ig ( s ) The transfer of the converter output to the grid current with virtual resistor: ig ( s ) uconv ( s ) = 1 Lg L f c f s + Lg c f kad s 2 + ( Lg + L f ) s
3

(17)

(18)

Similar to equations (16), (17) and (18), the transfer function of converter output voltage and capacitor current with virtual resistor Kad is derived in (19) and (20).
uconv ( s ) = sL f ( s Lg c f + 1)ig ( s ) + sLg ig ( s ) + kad ic ( s ) =
2

( s 3 Lg L f c f + s 2 c f Lg kad + s ( Lg + L f )) s 2 c f Lg

ic ( s )(19)

Lg c f s ic ( s ) = 2 uconv ( s ) Lg L f c f s + Lg c f K ad s + Lg + Lc

(20)

The next step is to analyze the impact from grid voltage to both grid current and capacitor current. In addition to being excited by the converter output voltage, the excitation from the grid voltage should also be taken into account. But it is obvious that grid voltages impact is not significant because to meet the grid code the high frequency components of grid voltage are much lower than the converter output voltage. Consider the converter output voltage to be constant the relationship of grid voltage and grid current and capacitor current can be derived. s L f i f ( s ) = ucf ( s ) s Lg ig ( s ) = ucf ( s ) u g ( s ) (21) (22)

57

s c f ucf ( s ) = ic ( s ) i f ( s ) = ic ( s ) + ig ( s )
u g ( s ) = 1 ig ( s ) sLg ig ( s ) Lf c f s2 + 1

(23) (24)

(25)

The transfer function of grid voltage to grid current is:

ig ( s ) ug ( s) = Lf c f s2 + 1 Lg L f c f s 3 + ( Lg + L f ) s

(26)

After derivation of the grid voltage to grid current, the grid voltage to capacitor current is easily derived:
u g ( s ) = sLg ( 1 1 1)ic ( s ) ic ( s ) 2 Lf c f s sc f

(27)

The transfer of the grid voltage to the capacitor current is:

Lf c f s ic ( s) = u g ( s ) Lg L f c f s 2 + Lg + L f

(28)

The virtual resistor loop is plugged into the controller. This proposed method cannot only take effect with converter output voltage but also the grid voltage. The capacitor current is sensed and multiplied with adaptive gain Kad, so the equation (5) will be rewritten: s L f i f ( s ) = kad ic ( s ) ucf ( s ) Simplify the equation (29) to (30): s L f i f ( s ) = sc f kad ucf ( s ) ucf ( s ) (30) (29)

58

The relation of grid current and capacitor voltage is: ig ( s ) = ( sc f kad + 1)ucf ( s ) s Lf sc f ucf ( s )

(31)

Substitute (31) to (22), the transfer function of grid voltage to grid current with virtual resistor:
ig ( s ) ug ( s) = L f c f s 2 + c f K ad s + 1 Lg L f c f s 3 + Lg c f K ad s 2 + ( Lg + L f ) s

(32)

Use capacitor current to represent grid current: ig ( s ) = ( sc f kad + 1)ic ( s ) s2 Lf c f ic ( s )

(33)

Substitute (33) to (32), the transfer function of grid voltage to capacitor current with virtual resistor: Lf c f s ic ( s) = 2 u g ( s ) Lg L f c f s + Lg c f K ad s + Lg + L f

(34)

The comparison of the transfer function with virtual resistor and without virtual resistor is drawn in figures 3.8, 3.9, 3.10 and 3.11 respectively. As we can see from figures 3.8 and 3.10 that with virtual resistor control, the high frequency resonant from both converter output voltage and grid voltage is eliminated. Figures 3.9 and 3.11 also prove that with virtual resistor the capacitor current doesnt have the resonant frequency. So from the circuit perspective the capacitor path has increased its impedance to this specific resonant frequency and no longer has low impedance.

59

150

Bode Diagram

100

Magnitude (dB) Phase (deg)

50

-50

-100

-150 -90

-135

-180

-225

-270 10
2

10

10 Frequency (Hz)

10

Figure 3.8 Transfer function: converter output to grid current with/without virtual resistor

150

Bode Diagram

100

Magnitude (dB) Phase (deg)

50

-50

-100 90

45

-45

-90 10
2

10

10 Frequency (Hz)

10

Figure 3.9 Transfer function converter output to capacitor current with/without virtual resistor

60

150 100 50 Magnitude (dB) 0 -50 -100 -150 -200 90

Bode Diagram

45 Phase (deg)

-45

-90 10
1

10

10 Frequency (Hz)

10

Figure 3.10 Transfer function grid voltage to grid current with/without virtual resistor

150

Bode Diagram

100 Magnitude (dB) Phase (deg)

50

-50 90

45

-45

-90 10
2

10

10 Frequency (Hz)

10

Figure 3.11 Transfer function grid voltage to capacitor current with/without virtual resistor

61

B. Analysis of Control Loop of Virtual Resistor Controller Virtual resistor loop only affects the inner current loop so the control loop is designed in details especially for the current control loop. From the analysis above we know that LCL filter is connected with two voltage sources, the converter output Uconv and grid Ug. Each of them will have influence on both the grid current and capacitor current. The inner current loop is designed to regulate grid current ig and capacitor current ic is used to damp the high frequency resonance of the grid current loop. Based on the relationship among the converter output, gird voltage, grid current and capacitor current, the diagram of control plant with proposed controller is drawn in figure 3.12. To further simplify the block diagram the control loop model is achieved in figure 3.13.

ug

ig
ic

uconv

Figure 3.12 Block diagram of control plant and proposed controller

Figure 3.13 Control loop modeling

Gcomp ( s ) = k p +

ki1 s ki 2 s ki 3 s + 2 + 2 2 2 s + 20s + s + 20s + (3 ) s + 20s + (5 ) 2

2

(35)

62

Gd ( s ) =

1 1 + Ts s

(36)

Ga ( s ) = kad

(37)

The current controller is stationary frame PR plus HC control. Kp is the proportional gain and Ki1, Ki2, Ki3 is the integration gain and is the fundamental frequency. Harmonics cancellation is used to eliminate low frequency odd order harmonics. According to Liserre [86] that for single phase application the 3rd and 5th harmonics are more important so HC controller is designed to reduce 3rd and 5th harmonics. The transport delay and computational delay are also included. Ts is defined as one switching period. Kad is the adaptive gain to represent virtual resistor. After the modeling of the control loop in the next segment the control loop design will be presented. The adaptive virtual resistor controller design will be narrated in detail. 3.5 Adaptive Virtual Resistor Control for a Wide Range of Grids The major issue of large scale penetration of PHEVs into the grid has been described. Since the issue is that various sets of grid impedances compromise the performance of the damping control, the damping controller must have the capability of tuning its parameters automatically. The grid impedance extraction method is proposed by using the resonant frequency of LCL filter to calculate the grid impedance [140, 141]. In order to address a wide range of grid impedances an effective way of damping is to ensure that the active damping controller adaptive capability. The adaptive gain Kad performs as a controllable resistance in series with the capacitor. To demonstrate why a variable Kad is important to the control loop performance, the control plant with adaptive gain is drawn in figure 3.14. We can see from

63

this figure that the phase margin changes with the variation of Kad. The Kad acts like a damping factor in the control loop, with different versions of Kad the Bode plot can be lightly damped or heavily damped. However the phase margin and the gain margin are highly dependent on this factor. If Kad is too high, the phase margin will be too low and the loop may not be stable. If the phase margin is too high, such as 100 degree the loop response will be very slow. If Kad is too low, the gain margin will be too low and the loop is close to its stability boundary and any interference may cause it unstable Next segment explains how to design an adaptive gain Kad based on the variation of grid impedances. In this paper, grid impedances change with different grid conditions. Lower impedance results in a stiff grid and higher impedance yields a weak grid. Based on system power rating and converter parameters, the grid impedance is chosen to be between 0.2mH (0.8% pu) and 2.5mH (9.8% pu). The system parameters and grid impedances are presented in Table 3-1. The controller structure is shown in figure 3.15. In this figure the entire control loop includes the outer dc bus regulation loop and the inner current synchronization loop. Based on the sensed grid current, the resonant frequency is extracted and detected by a frequency detection block. This function is edge triggered at every positive edge of the detected frequency signal so that the time duration between the first positive edge and the second positive edge is obtained. By inverting this time duration, we can get the frequency of the measured signal. Zero crossing function can be also added to prevent the mistrigger and increase the accuracy of the frequency detection. In figure 6.31 0.2mH grid impedance is measured based on its resonant frequency. In figure 6.32 the transient of grid impedance is tested.

64

In figure3.17, the pole-zero map for the dominant poles of the controller without virtual resistor compensation is drawn. The lowest impedance is 0.2mH, and the highest impedance is 2.5mH. The impedance increases incrementally with 0.1mH each time. In figure 3.18, the dominant poles of the control loop with different grid impedance are used to show the advantages of using adaptive virtual resistor over using fixed resistor value. We can see that the controllers dominant poles do not locate at the boundary or outside the unity circle, this means the controller is stable with one fixed virtual resistor value, but even so, the control loop performance is not good for some impedance because of the positions of the dominant poles. Some dominant poles has lower phase margin especially for the lower impedance cases. While for some higher impedance cases, the control loop can guarantee higher phase margin but have lower gain margin concurrently. Use phase margin, bandwidth and gain margin as the specification to design virtual resistor Kad. Virtual resistor at each impedance point should locate the dominant poles at the proper locations. By saying proper locations, it means that the control loop have the higher value of phase margin, gain margin and control loop bandwidth. The pole-zero map for the dominant poles of the controller with adaptive Kad is drawn in figure 3.19. In this figure, the dominant poles are located at the proper positions to ensure the performance of the control loop. Figure 3.20 and figure 3.21 show the cases of the stiffest and the weakest grid conditions, the control plant can achieve a good phase margin with different Kad values. The relationship of grid impedance and resonant frequency is shown in figure 3.22 and the relationship of grid impedance with proper resistor value Kad is shown in figure 3.23.

65

Table 3-1 System configuration

Power rating Voltage Filter inductor and capacitor Stiff grid condition Weak grid condition

1.5kW 120V 1mH, 50uF 0.2mH (0.8% pu) 2.5mH (9.8% pu)

Bode Diagram

0
Magnitude (dB)

Kad=7.5 Kad=2.0 Kad=15.0 Kad=0

-50

-100

-90 -135
Phase (deg)

-180 -225 -270 1 10

10

10

10

10

Frequency (Hz)

Figure 3.14 Control Parameter Characteristics: adaptive gain Kad

66

Figure 3.15 Block diagram of controller with adaptive virtual resistor loop

Figure 3.16 Frequency detection function block

0.30/T 0.8 0.35/T

Pole-Zero Map 0.25/T 0.20/T 0.1 0.15/T 0.2 0.3

0.6

0.40/T

0.4 0.5 0.6 0.7

0.10/T

0.4 0.45/T 0.2 Imaginary Axis

0.8 0.9

0.05/T

0.50/T 0.50/T

-0.2 0.45/T -0.4 0.40/T 0.10/T 0.05/T

-0.6

-0.8

0.35/T 0.30/T 0.20/T 0.25/T -1 -0.5 0 Real Axis

0.15/T

-1 -1.5

0.5

1.5

Figure 3.17 Root locus of control plant with various impedances (0.2mH to 2.5mH) without virtual resistor

67

0.30/T 0.8 0.35/T

Pole-Zero Map 0.25/T 0.20/T 0.1 0.15/T 0.2 0.3

0.6

0.40/T

0.4 0.5 0.6 0.7

0.10/T

0.4 0.45/T 0.2 Imaginary Axis

0.8 0.9

0.05/T

0.50/T 0.50/T

-0.2 0.45/T -0.4 0.40/T 0.10/T 0.05/T

-0.6

-0.8

0.35/T 0.30/T 0.20/T 0.25/T -1 -0.5 0 Real Axis

0.15/T

-1 -1.5

0.5

1.5

Figure 3.18 Root locus of control plant with various impedances (0.2mH to 2.5mH) and fixed virtual resistor

0.30/T 0.8 0.35/T

Pole-Zero Map 0.25/T 0.20/T 0.1 0.15/T 0.2 0.3

0.6

0.40/T

0.4 0.5 0.6 0.7

0.10/T

0.4 0.45/T 0.2 Imaginary Axis

0.8 0.9

0.05/T

0.50/T 0.50/T

-0.2 0.45/T -0.4 0.40/T 0.10/T 0.05/T

-0.6

-0.8

0.35/T 0.30/T 0.20/T 0.25/T -1 -0.5 0 Real Axis

0.15/T

-1 -1.5

0.5

1.5

Figure 3.19 Root locus of control plant with various impedances (0.2mH to 2.5mH) with adaptive virtual resistor

68

0.30/T 0.8 0.35/T

Pole-Zero Map 0.25/T 0.20/T 0.1 0.15/T 0.2 0.3

0.6

0.40/T

0.4 0.5 0.6 0.7

0.10/T

0.4 0.45/T 0.2 Imaginary Axis

0.8 0.9

0.05/T

0.50/T 0.50/T

-0.2 0.45/T -0.4 0.40/T 0.10/T 0.05/T

-0.6

-0.8

0.35/T 0.30/T 0.20/T 0.25/T -1 -0.5 0 Real Axis

0.15/T

-1 -1.5

0.5

1.5

Figure 3.20 Root locus of control plant with 0.2mH adopts proper virtual resistor

0.30/T 0.8 0.35/T

Pole-Zero Map 0.25/T 0.20/T 0.1 0.15/T 0.2 0.3

0.6

0.40/T

0.4 0.5 0.6 0.7

0.10/T

0.4 0.45/T 0.2 Imaginary Axis

0.8 0.9

0.05/T

0.50/T 0.50/T

-0.2 0.45/T -0.4 0.40/T 0.10/T 0.05/T

-0.6

-0.8

0.35/T 0.30/T 0.20/T 0.25/T -1 -0.5 0 Real Axis

0.15/T

-1 -1.5

0.5

1.5

Figure 3.21 Root locus of control plant with 2.5mH adopts proper virtual resistor

69

grid impedance vs resonant frequency 1800

1700

1600

1500 resonant frequency (Hz)

1400

1300

1200

1100

1000

900

800

0.5

1 1.5 grid impedance (mL)

2.5 x 10
-3

Figure 3.22 Grid impedance vs resonant frequencies

18

16

14 Resistor value Kad

12

10

4 800

900

1000

1100

1200 1300 1400 Resonant frequency

1500

1600

1700

1800

Figure 3.23 Resonant frequency vs proper adaptive gain Kad

70

800 700 600 500 400 300 200 100 0 63.5 62.5 61.6 60.6 59.8 58.9 58.1 57.3 56.5 55.5 100-200 200-300 300-400 400-500 500-600 Bandwidth P.M 700-800

0-100

600-700

Figure 3.24 Relationship of adaptive gain Kad, control loop bandwidth and phase margin for stiff grid
500 450 400 350 300 250 200 150 100 50 0 52.7 51 49.3 47.5 45.7 Bandwidth 44 42.3 40 38.9 37.2 P.M 300-350

0-50 350-400

50-100 400-450

100-150 450-500

150-200

200-250

250-300

Figure 3.25 Relationship of adaptive gain Kad, control loop bandwidth and phase margin for weak grid

Regarding the selection virtual resistor values, there are tradeoffs between the control loop bandwidth, phase margin and gain margin. If the virtual resistor is set too high, the phase

71

margin and bandwidth will increase, but the gain margin will decrease. So the value of virtual resistor is chosen at different impedance to get proper bandwidth, phase margin and gain margin with enough design margin for real system. These virtual resistor values become a group of parameters which can guarantee the control loop stability and also improve the control performance. Each of impedance from 0.2mH to 2.5mH corresponds to one designated virtual resistor. The mathematical equation between the resonant frequency and virtual resistor value (Ga) can be obtained by curve fitting and written in equation (38):
kad = Ga = f ( x = freq) = 1.383 1013 x5 0.91109 x 4 + 2.381106 x3 3.088 103 x 2 +1.997 x 508.6
(38)

The next segment is to analyze and design the control loop for different grid conditions. Here three grid impedances are used as example. The extreme case 0.2mH and 2.5mH are used and 1mH are used as random impedances in the middle of the impedance range. Figure 3.24 shows the relationship of Kad with control loop bandwidth and phase margin at 0.2mH. As Kad increases the gain margin increases, the bandwidth increases and phase margin decreases. So the first case is to design the control loop for 0.2mH grid impedance. Figure 3.26 shows that the adaptive gain loop is added with the PR+HC current controller. The control loop bandwidth is 700Hz less than 1/10 of the switching frequency. The phase margin is 60 degree and gain margin is 5.37 dB. The magnitude at fundamental frequency and 3rd and 5th order harmonics are well regulated. Figure 3.27 shows an adaptive gain loop is added to the controller designed for 2.5mH. The control loop bandwidth is 441Hz, the phase margin is 45.5 degree the stable margin for control system and gain margin is 5.21dB. The reason the performance is not as good as 0.2mH is that with larger impedance the

72

resonant frequency moves to lower frequency so the bandwidth is limited to a lower frequency. In figure 3.28, the Bode plot is drawn for the system with 1mH impedance. It is very important to test the robustness of the controller to the variation of the grid impedance. The proposed method uses the resonant frequency to determine the virtual resistor gain. But practically, the grid impedance will not keep constant and always has variations based on the load condition. If the variations of the grid impedance are still limited by the virtual resistor set by the initial impedance, the adaptive tuning loop may not change the virtual resistor value. In addition, if the damping gain is not precisely chosen based on the derived equation the control parameter will not correspond to the grid impedance very well. In other words, the virtual resistor gain Kad may not be the exact value for the grid impedance. So the robustness of the proposed controller needs to be examined. In the proposed system the grid impedances vary from 0.2mH to 2.5mH shown in Table I. Assume +20% ~-20% variations of grid impedance and the control loop parameters are still the parameters designed for 0.2mH and 2.5mH. The control loop performance for different grids is investigated. For a stiff grid with 0.2mH impedance the grid impedance will change from 0.16mH to 0.24mH. For a weak grid with 2.5mH impedance the grid impedance will change from 2mH to 3mH. In figure 3.29 the stiff grid control parameters designed for 0.2mH have been applied to the control loop with 0.16mH and 0.24mH. Both control systems are still stable. The control loop parameters for 0.16mH are phase margin 62.8 degree, bandwidth 718Hz and gain margin 5.20dB. For 0.24mH the control parameters are phase margin 58.4 degree, bandwidth 683Hz and gain margin 5.32dB. In figure 3.30 the weak grid control parameters designed for 2.5mH have been applied to the control loops with 2mH and 3mH.

73

Both control systems are still stable. The control loop parameters for 2mH are phase margin 38.3 degree, bandwidth 534Hz and gain margin 3.93dB. For 3mH the control parameters are phase margin 49.2 degree, bandwidth 377Hz and gain margin 6.32dB. Based on the above controller design and robustness analyses we can predict that proposed adaptive resistor controller can make the control loop stable even with the parameters mismatch or the variation of the grid impedance. But the control loop performance is compromised, especially at the weak condition. In the next section, the simulation and experimentation based on all the grid conditions mentioned here will be conducted to verify the proposed control method.

100

Bode Diagram

50

Magnitude (dB)

System: G Gain Margin (dB): 5.37 At frequency (Hz): 1.57e+003 Closed Loop Stable? Yes

-50

-100

-45 -90 -135 Phase (deg) -180 -225 -270 -315 -360 10
1

System: G Phase Margin (deg): 60.4 Delay Margin (sec): 0.000239 At frequency (Hz): 701 Closed Loop Stable? Yes

10

10

10

10

Frequency (Hz)

Figure 3.26 Bode plot of control loop with 0.2mH impedance and with adaptive virtual resistor control

74

100

Bode Diagram

50

System: G Gain Margin (dB): 5.21 At frequency (Hz): 776 Closed Loop Stable? Yes

Magnitude (dB)

-50

-100

-150

-200 -45 -90 -135 Phase (deg) -180 -225 -270 -315 -360 10
1

System: G Phase Margin (deg): 45.5 Delay Margin (sec): 0.000287 At frequency (Hz): 441 Closed Loop Stable? Yes

10

10

10

10

Frequency (Hz)

Figure 3.27 Bode plot of control loop with 2.5mH impedance and with adaptive virtual resistor control

100

Bode Diagram

50

Magnitude (dB)

System: G Gain Margin (dB): 5.13 At frequency (Hz): 929 Closed Loop Stable? Yes

-50

-100

-150 -45 -90 -135 Phase (deg) -180 -225 -270 -315 -360 10
1

System: G Phase Margin (deg): 51 Delay Margin (sec): 0.000284 At frequency (Hz): 500 Closed Loop Stable? Yes

10

10

10

10

Frequency (Hz)

Figure 3.28 Bode plot of control loop with 1mH impedance and with adaptive virtual resistor control

75

100

Bode Diagram

50

L=0.2mH L=0.16mH L=0.24mH

Magnitude (dB) Phase (deg)

-50

-100

-150

-200 0 -45 -90 -135 -180 -225 -270 -315 -360 10

1

10

10

10

10

Frequency (Hz)

Figure 3.29 Controller robustness analysis for stiff grid control loop with grid impedance 20% variation

100 50 0 -50 -100 -150 -200 0 -45 -90

Phase (deg)

L=2.5mH L=2.0mH L=3.0mH

Magnitude (dB)

-135 -180 -225 -270 -315 -360 1 10

2 3 4

10

10

10

Frequency (Hz) 105

Figure 3.30 Controller robustness analysis for weak grid control loop with grid impedance 20% variation

76

Figure 3.31 Resonant frequency detection to determine grid impedance

Figure 3.32 Resonant frequency detection to determine grid impedance during impedance transient

77

3.6 Verification of Proposed Adaptive Virtual Resistor Controller with Different Grids In this section, two grid conditions are examined which aims at investigating the effectiveness of proposed controller. In case I the PHEV is connected with a stiff bus-0.2mH; and in case II the PHEV is connected with a weak bus-2.5mH. The grid voltage is the actual voltage from the feeder with THD around 3.8%~4%. The objective of using the actual voltage is to test the ability of the proposed controller to reject disturbance from the grid side. In case I, the grid is a stiff bus so the impedance, sets at 0.2mH, is very low. Note the grid impedance can be even lower as in the case of an infinity bus but there will be no more inductance and no resonance at all. The transformers leakage inductance is around 40uH which does not affect the total impedance significantly. The simulation results with the proposed adaptive resistor plugged into the control loop are shown in figure 3.33. In this figure the converter-side current and grid-side current are shown. It can be seen that with the damping function enabled in the control loop the high frequency resonance is mitigated effectively. The experimental results in figure 3.34 shows with proposed controller enabled in the loop the resonance on the grid-side current and converter-side current is mitigated. In figure 3.35 the grid-side current and converter-side current without proposed method control are shown. The spectrum of the grid-side current is analyzed and plotted in figure 3.36 and converterside current is not plotted because the grid-side current is the control objective. It can be seen that the resonant frequency is around 25th~27th harmonics which matches with the grid impedance value perfectly. The resonant frequency is calculated:

78

f res =

1103 + 0.24 103 = = 1618.67 Hz = 26.9 60 Hz 1103 0.24 103 50 106 L f Lg c f

L f + Lg

(39)

In figure 3.37 the grid-side current and converter-side current with proposed control is shown. In figure 3.38 the spectrum of the grid-side current is analyzed and it is obvious that high frequency components (25th~27th harmonics) are eliminated. Finally in figure 3.39 the grid current with proposed control method is compared with IEEE 519 harmonics standard. The results show that both total harmonics and individual harmonic of grid current with the proposed method meet the requirement of grid code.

15 converter current 10 5 0 -5 -10 -15 0.1 0.11 0.12 0.13 0.14 0.15 0.16 0.17 0.18 0.19 0.2

15 10 5 0 -5 -10 -15 0.1 0.11 0.12 0.13 0.14 0.15 0.16 0.17 0.18 0.19 0.2 grid current

Figure 3.33 Converter-side current and grid-side current with the proposed controller enabled

79

Figure 3.34 Converter-side current and grid-side current with the proposed controller enabled (curve1: converter-side current; curve2: grid-side current; curve3: command to enable the proposed control)

Figure 3.35 Converter-side current and grid-side current without the proposed controller (curve1: converter-side current; curve2: grid-side current)

80

Current Spectrum 10% 9% 8% 7% 6% 5% 4% 3% 2% 1% 0% 1st 3rd 5th 7th 9th 11th 13th 15th 17th 19th 21th 23th 25th 27th 29th

Figure 3.36 Spectrum of grid-side current without the proposed controller

Figure 3.37 Converter-side current and grid-side current with the proposed controller (curve1: converter-side current; curve2: grid-side current)

81

Current spectrum 10% 9% 8% 7% 6% 5% 4% 3% 2% 1% 0% 1st 3rd 5th 7th 9th 11th 13th 15th 17th 19th 21th 23th 25th 27th 29th THD

Figure 3.38 Spectrum of grid-side current with the proposed controller enabled
IEEE 519 5% 4% 4% 3% 3% 2% 2% 1% 1% 0% proposed controller

3rd

5th

7th

9th

11th

13th

15th

17th

19th

21th

23th

25th

27th

29th

31th

33th

35th

37th

Figure 3.39 Spectrum comparison of grid-side current with IEEE 519 standard

39th

82

In case II, the proposed controller is investigated with a weak bus. Here the weak bus means very high value grid impedance which will make the resonance frequency very low. In this paper the grid inductance is set at 2.5mH. Since the resonant frequency is in the low frequency range and the crossover frequency must be set before the resonant peak, the system bandwidth will be limited. The simulation results are shown in figure 3.40 with the converter-side current and grid-side current. The experimental results in figure 3.41 show that with the proposed control enabled in the loop, the resonance on the grid-side current and converter-side current is eliminated. In figure 3.42 the grid-side current and converter-side current without the proposed control are shown. The spectrum of the grid-side current is analyzed and plotted in figure 3.43. The resonant frequency is around the 14th harmonics which matches with the current grid impedance value. L f + Lg L f Lg c f 1103 + 2.54 103 = 840.69 Hz = 14 60 Hz 1103 2.54 103 50 106

f res =

(40)

In figure 3.44 the grid-side current and converter-side current with proposed control are shown. In figure 3.45 the spectrum of the grid-side current is analyzed and the high frequency components are eliminated after proposed control is enabled. Finally in figure 3.46 the grid current with proposed control method is compared with IEEE 519 harmonics standard. The results show that total harmonics and individual harmonic of grid current with the proposed method meet the requirement of grid code.

83

40 Converter-side current 20 0 -20 -40 0.16

0.18

0.2

0.22

0.24

0.26

0.28

0.3

30 20 10 0 -10 -20 -30 0.16 0.18 0.2 0.22

Time

Grid current

0.24

0.26

0.28

0.3

Figure 3.40 Converter-side current and grid-side current with the proposed controller enabled

84

Figure 3.41 Converter-side current and grid-side current with the proposed controller enabled (curve1: converter-side current; curve2: grid-side current; curve3: command to enable the proposed control)

Figure 3.42 Converter-side current and grid-side current without the proposed controller (Red curve: converterside current; Purple curve: grid-side current)

85

THD 10% 9% 8% 7% 6% 5% 4% 3% 2% 1% 0%
1st 3rd 5th 7th 9th 11th 13th 15th 17th 19th 21th 23th 25th 27th 29th

THD

Figure 3.43 Spectrum of grid-side current without the proposed controller enabled

Figure 3.44 Converter-side current and grid-side current with the proposed controller enabled (Red curve: converter-side current; Purple curve: grid-side current)

86

THD 10% 9% 8% 7% 6% 5% 4% 3% 2% 1% 0%
1st 3rd 5th 7th 9th 11th 13th 15th 17th 19th 21th 23th 25th 27th 29th

THD

Figure 3.45 Spectrum of grid-side current with the proposed controller enabled

IEEE 519 5% 4% 4% 3% 3% 2% 2% 1% 1% 0% 3rd 5th 7th 9th 11th 13th 15th 17th 19th

proposed controller

21th

23th

25th

27th

29th

31th

33th

35th

37th

Figure 3.46 Spectrum comparison of grid-side current with IEEE 519 standard

39th

87

3.7 Summary of Chapter Three In this chapter the challenge of integration of PHEVs with a large set of grid impedances is described first. An adaptive virtual resistor control is proposed to solve the high frequency resonance issue. The modeling of the control loop is derived. The controller design and tuning guideline is presented. This virtual resistor loop can be plugged into the controller and tuned automatically with the grid impedance. It can allow the PHEV controller to work with a wide range of grid impedances. Symbolic case studies have been conducted to verify the proposed controller. The stiff grid, weak grid and a grid in the medium have been examined. The proposed control method eliminates the high frequency resonance effectively. This proposed method greatly enhances the performance for large adoption of PHEVs into the grid.

88

Chapter Four New Inductor Current Control based on Active Harmonics Injection for Plug-in Hybrid Electric Vehicles Vehicle to Home Application
4.1 Review of Control Methods for Single Phase Inverter An off-grid application (Vehicle to Home function) performs as a line-interactive UPS [142]. The converter is shunt into the system to supply backup power during grid faults. So control methods for a single phase inverter can be used. The first step is to analyze the existing control methods for single phase inverter application. Based on the control frame, the control method can be divided into two groups: stationary frame which uses the instantaneous value and synchronous frame which uses DQ transformation. For the DQ transformation, a faked imaginary axis can be used to form the DQ together with a real axis [143]. Another way of doing this is to command the output voltage and capacitor current to be DQ axis because these two values are naturally perpendicular to each other. The new transformation in a three-dimension method for multiple H-bridge legs is able to achieve good performance [144]. Without considering the control frames, many control theories and methods have been applied to control single phase or three phase inverter. A robust controller based on adaptive theory and dissipativity theory is proposed in [145-147]. Dead-beat adaptive hysteresis current control is used as the current loop for the inverter [148] and an improved deadbeat current controller is proposed in [149]. The control delay caused by digital processors is minimized to achieve high performance for the inverter [150]. The repetitive controller is proposed to compensate for the harmonics of the output voltage [151-153]. Multiloop structure with the digital predictive control is

89

proposed for inverter control [154]. A synchronous frame controller with the individual compensation blocks is proposed to eliminate low-order harmonics of the output voltage [155], and a similar method is proposed for the active filter application [156]. Beyond the complex control theories, one good way to classify the control methods is based on their feedback signals [157]. As we know, a perfectly sinusoidal output voltage is the ultimate goal for an inverter. However, normally a double-loop control structure is adopted for the inverter control. An outer voltage loop is used to regulate output voltage while the inner current loop is intended to improve the dynamic speed for the load transient. To sense the control signal, for the outer loop the output voltage should be measured and monitored with precision. Regarding the inner current loop, several choices can be made, which bring us up to four methods. Based on the sensed current signal, the control methods are inductor current feedback control, capacitor current feedback control, load current feedback control and combined inductor current and load current feedback control. The concept back-EMF decoupling is also applied to these control methods [158]. This decoupling method can be considered a type of feedforward control by adding the output voltage signal to the control signal together to generate PWM signals. The advantage of this concept is that output voltage is constant even during the load transient so the control parameters dont need to be adjusted drastically. This also allows the control to be easily tuned. This loop can be used to decouple the impact of the DC bus on the controller. The last feedforward part plays a less important role in the loop. It is added to the voltage command in order to compensate for the voltage drop across the inductor especially when the load is heavy and voltage drop across the inductor is not small. But the total value of this part only account for 1/1000 of the total

90

voltage command. Thus when a resonant type controller is used in the voltage loop and the resonant gain is set to a higher value this feedforward part is unnecessary. Among four control methods, the capacitor current feedback method is verified to be the best control method with good steady-state and dynamic performance [159]. The capacitor current feedback control is first proposed by Ryan [158]. This method uses an inexpensive current transformer to sense the current through the filter capacitor and utilizes this signal as the control feedback signal. A filter should be applied to filter out the high frequency harmonics on the capacitor current otherwise these harmonics will be transmitted into the current control and propagated thus polluting the entire loop. This methods structure is output voltage loop and capacitor current loop. One way to understand this method is without considering high frequency harmonics, that the capacitor current is always a sinusoidal current. This fundamental current does not change much when connected with any type of load and if this current is well controlled, the output voltage which is the integration of the capacitor current will be automatically regulated to a sinusoidal voltage. However, this control method still has drawbacks in that this method cannot sense the inductor current to protect the converter. An enhanced Luenberger observer is proposed in place of sensing the capacitor current [157]. Inductor current feedback control senses the inductor current as the inner loop control variable. Although the control performance is not as good as capacitor current based control, this method provides superior protection. Whether there is a large current overshoot on the inductor or a load short-circuit fault the control method can guarantee the converter stage is safe. The load current control directly senses the load current as the control variable, but it

91

can not protect power devices if a fault occurs on the converter side e.g., resonant oscillation or wrong control logic. When inductor current feedback is used disturbance input decoupling also known as disturbance feedforward control, can be implemented. This method incorporates with load current sensing to reject the load disturbance effectively. This method senses both inductor current and load current as the feedback variables; its control performance is promising. So this method can be considered an alternative to capacitor current feedback, because capacitor current is the sum of the inductor current and load current. With more current sensors the protection is also achieved but the cost of the total system increases. This may compromise this method. To eliminate this extra sensor, a load current prediction algorithm can be used [157]. Three basic control structures without disturbance decoupling are shown in figures 4.1, 4.2 and 4.3 respectively.

Vout
Vout

Vout
Gc ( s )
GPI ( s )
1 sL

if

ic

ic

Vout

1 sc f

Vout

rL

iLoad

Figure 4.1 Capacitor current feedback control

Vout
Vout

Vout
Gc ( s )
GPI ( s )
1 sL

if

ic

if

Vout

1 sc f

Vout

rL

iLoad

Figure 4.2 Inductor current feedback control

92

Vout
Vout

Vout
Gc ( s )
GPI ( s )
1 sL

if

ic

if

1 sc f

Vout

rL
Vout
sL

iLoad

Figure 4.3 Inductor and load current feedback control method

In this chapter a new control method based on active harmonics injection and inductor current feedback is proposed. The reason for choosing inductor current feedback is because inductor current is the essential variable for protection. With the active harmonics injection, control performance is improved especially for a nonlinear type of load. Moreover inductor current feedback and capacitor current feedback are compared and the advantages of inductor current are analyzed and verified. Inductor current feedback can limit the current overshoot at the load transients which may cause the inductor core saturation. So with the inductor current feedback control higher switching frequency and smaller passive components can be used to provide an optimized design for the inverter stage. 4.2 Theoretical Analysis of the Proposed Control Method Inductor current feedback control senses the inductor current as the inner control variable. The outer voltage loop supplies a sinusoidal voltage reference to the inner loop; however, when a nonlinear load such as a diode bridge rectifier is connected with the inverter the inverter needs to provide high percentage of odd-order harmonics to the load. This odd-order harmonics challenges the control loop design because the inner loop reference should have the sinusoidal component for the output voltage and odd-order harmonics to cancel out the harmonics brought by the inductor current. Theoretically, the output voltage loop can be

93

designed with a very high bandwidth which can cover higher order harmonics e.g. 15th, 17th or even higher. However, in actual control loop design, the bandwidth is very limited, especially for the outer voltage loop. The voltage loop is only above the fundamental frequency. The inner current loop has a much faster bandwidth. So, according to this reason, it is usually very easy to get good output voltage waveforms in simulation but the waveforms are distorted in experiment. Since the outer loop is always very limited only dealing with the fundamental frequency, the harmonics coming with the inductor current will pollute the whole loop and output voltage as well. One effective way to reduce the harmonics on the capacitor is to sense the capacitor current as the feedback signal which is the capacitor current feedback control as discussed previously. Based on inductor current feedback control, an extra harmonics injection loop is proposed as a plug-in to the original loop.

Vout
Vout

Vout
Gv ( s )

Gi ( s )

1 sL

if

ic

if
Vout

1 sc f

Vout

Vh

rL
GPI ( s )

iLoad

Vh*
Figure 4.4 Control block of the proposed method based on active harmonics injection

First the inductor current equals the sum of the load current and the capacitor current, and the capacitor current contains the fundamental component and harmonics components. iL = iLoad + ic = iLoad + icf + ich (1)

94

The inverter output voltage equals the sum of the output voltage and the voltage drop on the inductor. v pwm = iL ls + vout = (iLoad + icf + ich ) ls + (icf + ich ) 1 1 = iLoad ls + (icf + ich ) (ls + ) cs cs (2)

With simple mathematic manipulation it can be seen that if the inverter output voltage can generate harmonics voltage of the capacitor the output voltage will be sinusoidal. So a harmonics extraction block is used to get the output voltage harmonics because these harmonics are directly related to the harmonics current on the capacitor. v pwm ich (ls + 1 cs 1 1 ) = iLoad ls + icf (ls + ) cs cs (3)

vh = ich

(4)

After the harmonics voltage is detected and extracted, these harmonics will go to the loop which controls the amount of harmonics injection. Since the ultimate goal is to eliminate the harmonics current on the capacitor so the harmonics voltage reference is set to zero. The injection loop parameters are adjusted to make the loop output equal the additional item on the left side of the equation (3). Based on the above analysis an active harmonics injection loop can be added in Figure 4.4. This loop first extracts the harmonics from the output voltage and forces the extracted harmonics to equal zero with the reference voltage setting at zero. The generated harmonics are injected to be combined with the inner current loop output as the modulation signal This harmonics output will cancel the load harmonics in such a way

95

that the capacitor current will have few or even no harmonics components. With no distortion on the capacitor current the output voltage will be perfectly sinusoidal. With this control method, one important function is to detect and extract the harmonics. Similar to the process used in an active filter [160, 161] the voltage signal can be transformed into the synchronous frame and transmitted through a high pass filter. This is a suitable solution for three-phase applications. However for single phase applications another method is used. An Enhanced PLL [162] is used to sense the output voltage and generate a sinusoidal voltage reference which follows the output voltage instantaneously. A peak detection function is used to detect and follow the peak value of the output voltage and multiplies it by with the sinusoidal reference to form a modified output voltage reference. The voltage difference between the instantaneous output voltage and the modified voltage reference is the extracted harmonics. The reason to use peak diction is to make sure there are no fundamental components in the injection loop to affect the control accuracy and Kp is used to compensate the error from both the sensing and the conditioning process. The harmonics extraction block is shown in figure 4.5. A similar idea can be designed and implemented in figure 4.6. Both of the methods inject the harmonics to the control loop, and the harmonics are controlled by the output voltage. The only difference is the point where the harmonics are injected. The good thing is that this loop doesnt have bandwidth limitation like the outer voltage loop. So it can be designed be fast enough to cover high order harmonics. The experiment is conducted with the conventional inductor current control and the proposed control method. The nonlinear load is a single phase diode-bridge rectifier with the filter inductor and dc load at the dc bus. The output voltage is 120V, and the load peak

96

current is 14A. The results with the conventional control and the proposed control method are shown in figure 4.7 and 4.8 respectively. The output voltage THD with the conventional control method is 6.11% while with the proposed control the output voltage THD has been improved to 4.68%. Figure 4.9 shows the capacitor current spectrum comparison between two control methods. The performance of the proposed controller is further examined by comparing the harmonics with IEC standards for UPS: IEC 62040-3 [192, 193]. In figure 4.11, the harmonics of the output voltage regulated with conventional control method are compared with IEC 62040-3.

Vout

Vh

kp
Figure 4.5 Harmonics detection and extraction block

Vout
Vout

Vout
Gv ( s )

Gi ( s )

1 sL

if

ic

if
Vout

1 sc f

Vout

rL

iLoad

Vh

GPI ( s )

Figure 4.6 Active harmonics injection before the inner current loop

97

Voltage error

Output

Load current

Figure 4.7 Nonlinear load tests with the proposed control method (blue curve: output voltage 200V/dim; Purple curve: voltage error between the reference and actual voltage; Green curve: load current 10A/dim)

Voltage error

Output

Load current

Figure 4.8 Nonlinear load tests with inductor current feedback control (blue curve: output voltage 200V/dim; Purple curve: voltage error between the reference and actual voltage; Green curve: load current 10A/dim)

98

with injection
60.00%

without injection

50.00%

40.00%

30.00%

20.00%

10.00%

0.00% THD 3rd 5th 7th 9th 11th 13th 15th 17th 19th 21th 23th 25th 27th 29th

Figure 4.9 Comparison of capacitor current spectrum: the proposed method and conventional controller
proposed controller 14.00% conventional controller IEC 62040-3

12.00%

10.00% 8.00%

6.00%

4.00%

2.00% 0.00% 3rd 5th 7th 9th 11th 13th 15th 17th 19th 21th 23th 25th 27th 29th

Figure 4.10 Comparison of output voltage spectrum: the proposed method, conventional controller and IEC62040-3 Standard

99

4.3 Steady State Operation and Dynamic Response of the Proposed Controller According to the analysis in the previous section, the controller performance should be superior to the inductor current feedback control method especially for nonlinear load. Four types of loads have been used to examine the proposed control method. They are a 1kW resistive load, no load, 1kW resistor and 2.5mH RL and nonlinear load. First the simulation results are shown in figure 4.11, figure 4.13, figure 4.15 and figure 4.17 respectively. After the simulations are conducted, the experiment is conducted based on the same loads mentioned above. The experimental results are shown in figure 4.12, figure 4.14, figure 4.16 and figure 4.18 respectively. In figure 4.19 the actual output voltage and the output voltage reference are shown. In this figure the purple curve is the voltage error between the actual output voltage and the voltage reference. It can be seen that the voltage error only has the high frequency components and the fundamental frequency component is hardly found. This will prove the accuracy of the proposed control method. The output voltage THD for different types of loads is summarized in table 4-1.

100

200 0 -200 0.3 10 0 -10 0.3 20 0 -20 0.3 5 0 -5 0.3 0.305 0.31 0.315 0.32 0.325 0.33 0.335 0.34 0.345 0.35 0.305 0.31 0.315 0.32 0.325 0.33 0.335 0.34 0.345 0.35 0.305 0.31 0.315 0.32 0.325 0.33 0.335 0.34 0.345 0.35 0.305 0.31 0.315 0.32 0.325 0.33 0.335 0.34 0.345 0.35

Figure 4.11 Simulation 1kW load test with the proposed method (curve 1: output voltage; curve 2: load current; curve 3: inductor current; curve 4: capacitor current)

Figure 4.12 Experiment 1kW load test with the proposed method (Green curve: load current 20A/dim; Blue curve: output voltage 200V/dim; Red curve: inductor current 20A/dim; Purple curve: capacitor current 10A/dim)

101

200 0 -200 0.15 10 0 -10 0.15 5 0 -5 0.15 5 0 -5 0.15

0.155

0.16

0.165

0.17

0.175

0.18

0.185

0.19

0.195

0.2

0.155

0.16

0.165

0.17

0.175

0.18

0.185

0.19

0.195

0.2

0.155

0.16

0.165

0.17

0.175

0.18

0.185

0.19

0.195

0.2

0.155

0.16

0.165

0.17

0.175

0.18

0.185

0.19

0.195

0.2

Figure 4.13 Simulation no load test with the proposed method (curve 1: output voltage; curve 2: load current; curve 3: inductor current; curve 4: capacitor current)

Figure 4.14 Experiment no load test with the proposed method (Green curve: load current 20A/dim; Blue curve: output voltage 200V/dim; Red curve: inductor current 20A/dim; Purple curve: capacitor current 10A/dim)

102

200 0 -200 0.15 10 0 -10 0.15 20 0 -20 0.15 5 0 -5 0.15 0.155 0.16 0.165 0.17 0.175 0.18 0.185 0.19 0.195 0.2

0.155

0.16

0.165

0.17

0.175

0.18

0.185

0.19

0.195

0.2

0.155

0.16

0.165

0.17

0.175

0.18

0.185

0.19

0.195

0.2

0.155

0.16

0.165

0.17

0.175

0.18

0.185

0.19

0.195

0.2

Figure 4.15 Simulation RL test 1kW resistive load and 2.5mH inductor with the proposed method (curve 1: output voltage; curve 2: load current; curve 3: inductor current; curve 4: capacitor current)

Figure 4.16 Experiment RL test 1kW resistor load with 2.5mH inductor test with the proposed method (Green curve: load current 20A/dim; Blue curve: output voltage 200V/dim; Red curve: inductor current 20A/dim; Purple curve: capacitor current 10A/dim)

103

200 0 -200 0.15 20 0 -20 0.15 20 0 -20 0.15 5 0 -5 0.15

0.155

0.16

0.165

0.17

0.175

0.18

0.185

0.19

0.195

0.2

0.155

0.16

0.165

0.17

0.175

0.18

0.185

0.19

0.195

0.2

0.155

0.16

0.165

0.17

0.175

0.18

0.185

0.19

0.195

0.2

0.155

0.16

0.165

0.17

0.175

0.18

0.185

0.19

0.195

0.2

Figure 4.17 Simulation nonlinear loads with the proposed method (curve 1: output voltage; curve 2: load current; curve 3: inductor current; curve 4: capacitor current)

Figure 4.18 Experiment Nonlinear load test with the proposed method (Red curve: DC bus voltage 200V/dim; Purple curve: load current 10A/dim; Blue curve: output voltage 200V/dim; Green curve: capacitor current 10A/dim)

104

Figure 4.19 Experiment nonlinear load test with the proposed method (Purple curve: output voltage error; Blue curve: output voltage 200V/dim; Red curve: output voltage reference; Purple curve: load current 10A/dim) Table 4-1 Output voltage with different types of load

Load Condition No Load 1kW resistive load RL load Nonlinear load

Output Current THD [%] 3.59% 71.7%

Output Voltage THD [%] 2.90% 2.68% 3.01% 4.68%

After the steady-state test the dynamic response of the proposed controller is examined with a 1kW load transient. It should be mentioned that the proposed method belongs to the inductor current based control group. The proposed method focuses on improving the

105

harmonics compensation capability for the inductor current feedback control but it is not designed to improve the system dynamic response. However with the simulation results and experimental results it can be seen that the response of the proposed controller is still good. A 1kW load transient test is first simulated in figure 4.20. Then the experimental results are shown in figure 4.21, the response time for the output voltage after the load transient is approximately 3~4 cycles. More results show the voltage error during the load transient in figure 4.22 in simulation and figure 4.23 in experiment.

200 0 -200 20 0 -20 20 0 -20 10 0 -10 0.22 0.23 0.24 0.25 0.26 0.27 0.28

0.22

0.23

0.24

0.25

0.26

0.27

0.28

0.22

0.23

0.24

0.25

0.26

0.27

0.28

0.22

0.23

0.24

0.25

0.26

0.27

0.28

Figure 4.20 Simulation a 1kW load transient for dynamic response test of the proposed controller (curve 1: output voltage; curve 2: load current; curve 3: inductor current; curve 4: capacitor current)

106

Figure 4.21 Experiment a 1kW load transient for dynamic response test of the proposed controller (Green curve: load current 10A/dim; Blue curve: output voltage 200V/dim; Red curve: inductor current 20A/dim; Purple curve: capacitor current 10A/dim)
200 0 -200 0.2 20 0 -20 0.2 20 0 -20 0.2 0.2 0 -0.2 0.2

0.21

0.22

0.23

0.24

0.25

0.26

0.27

0.28

0.29

0.3

0.21

0.22

0.23

0.24

0.25

0.26

0.27

0.28

0.29

0.3

0.21

0.22

0.23

0.24

0.25

0.26

0.27

0.28

0.29

0.3

0.21

0.22

0.23

0.24

0.25

0.26

0.27

0.28

0.29

0.3

Figure 4.22 Simulation a 1kW load transient for dynamic response test of the proposed controller

107

Figure 4.23 Experiment a 1kW load transient for dynamic response test of the proposed controller (Green curve: load current 20A/dim; Red curve: inductor current 20A/dim; Blue curve: output voltage 200V/dim; Purple curve: output voltage error)

4.4 Investigation of Inductor Current Transient Response with Different Controllers The proposed method is similar to the inductor current controller during transient because its third loop compensates only the harmonics. So in this section the comparison focuses on two major control methods: inductor current and capacitor current. Although it has been already proven that the capacitor current feedback control has better steady-state and dynamic state performance [159], an issue that needs to be addressed is that the inductor current feedback control has the capability to limit the inductor current overshoot during the load transient which has not yet been discussed. Although it is not addressed, the capability to limit the inductor current overshoot during the load transient is very important, because current overshoot may lead to the saturation of the magnetic cores. This issue becomes even

108

more critical with the claim that in a low power rating application (the power rating in the range of tens of kW) the magnetic cores are required to operate close to the saturation point on the B-H curve to reduce the cost and the size [163, 164]. To begin with, assume the stationary frame is used and the load transient happens at the moment when the output voltage reaches the peak value because the capacitor voltage is 90 degree lag of the capacitor current. When the current goes to zero the voltage reaches the peak value so at this time the load transient provides the highest current overshoot in of all the transients. In the analysis the load is set as a resistive load for simplification. The detailed analysis of capacitor current feedback control starts from the inner current loop. Assuming the sampling frequency equals the switching frequency the load transient can be detected by the digital processor immediately. After the transient goes into the interrupt of the processor the inner loop feedback signal changes first, because during the load change it is the capacitor that supplies the current. As seen in Figure 4.26 at the moment of transient the capacitor current has a significant drop in the opposite direction due to discharge. The inner loop reference is subtracted from the feedback signal to obtain an error and this error is sent to the inner current loop. As the feedback signal decreases, the inner loop output increases and the inverter output voltage increases. The next important event is that the output voltage drops due to the discharge of the capacitor. So the outer loop generates an increased reference signal for the inner loop this will even further increase the inner loop error which will lead to a higher converter output. As we know the voltage on the inductor is the difference between the converter output and the output voltage. With the converter output increasing and the output voltage decreasing if the inductor value is further reduced the

109

current overshoot on the inductor will be further increased, because the nature of this control does not provide any limitation on the inductor current. Note that the inner current loop with different inductor value can be varied. A slow inner loop may be used to mitigate the current overshoot however in this way the control will suffer from a slow response. The next step is to analyze the transient for the inductor current feedback control. During the transient, the inductor current cannot change immediately so the changing variable in the control loop is only the output voltage drop due to the capacitor discharge. However when the inner loop reference increases due to the outer voltage loop the inner current feedback signal also increases because the inductor current increases. So these two increasing values will compete to get an error which will go through the inner loop to generate the converter output. Obviously the error will be increased rather than being decreased. Otherwise the converter output wont increase and the control loop becomes a positive feedback. But compared to the capacitor current control the trend of this increasing is still low because of an increasing feedback signal in the inner loop. In both control methods the outer voltage loop provides an increasing reference for the inner current loop. The capacitor current control makes the inner loop error increase, while the inductor current control counteracts this increase to make the inner loop error rise at a slower rate because of the increasing inductor current. With a smaller inductor the inductor current overshoot intends to rise higher however this will in turn make the current feedback increase higher than the inner loop error becomes lower, which in turn limits the rise of the inductor current. So with a smaller inductor the inductor current control can limit the overshoot current. It also can be seen from another angle that the dynamic response of the capacitor current feedback has already been proven to

110

be faster than the inductor current feedback because the inner loop error of the capacitor current feedback is much greater than the inductor current feedback otherwise it cannot respond faster than the inductor current feedback control. This can also be considered the justification for the above analysis. Similar to the concept dynamic stiffness which is defined and derived by Ryan [157], the relationship between the inductor current and the load current are derived using small signal model. The same relationship has been used in the DVR application for dynamic response analysis [165]. This relationship is used to evaluate the inductor current at the point of the load current transient. The inductor current to the load current with proposed method is derived in equation (5), while the relation for the capacitor current control is derived in (6).
1 1 G pr ( s )G pi1 ( s ) + GH ( s )G pi 2 ( s ) cs cs = 1 1 iLoad G pr ( s )G pi1 ( s ) + GH ( s )G pi 2 ( s ) + G pi1 ( s ) + ls cs cs 4 3 2 As + Bs + Cs + Ds1 + E = 6 Fs + Gs 5 + As 4 + Bs 3 + C s 2 + Ds1 + E iL
A = k pr k p1 + k p 2 ; B = k pr k p1c + k pr k p1 f h + k p1kr + k pr ki1 + k p 2c + ki 2 ; C = k pr k p1 2 + k pr k p1c f h + k p1kr f h + k pr ki1c + k pr ki1 f h + kr ki1 + k p 2 2 + ki 2c ; f f D = k pr k p1 f h 2 + k pr ki1 2 + k pr ki1c f h + kr ki1 f h + ki 2 2 ; f f f E = k pr ki1 f h 2 f

(5)

111

A = k pr k p1 + k p 2 + k p1c c + k p1 f h c + ki1c + lc 2 + lcc f h ; f B = k pr k p1c + k pr k p1 f h + k p1kr + k pr ki1 + k p 2c + ki 2 + k p1 2 c + k p1c f h c + ki1c c f + ki1 f h c + lcf h 2 ; f C = k pr k p1 2 + k pr k p1c f h + k p1kr f h + k pr ki1c + k pr ki1 f h + kr ki1 + k p 2 2 + ki 2c f f + k p1 f h 2 c + ki1 2 c + ki1 f hc c; f f D = k pr k p1 f h 2 + k pr ki1 2 + k pr ki1c f h + kr ki1 f h + ki 2 2 + ki1 f h 2 c; f f f f E = k pr ki1 f h 2 f F = lc; G = lcc + lcf h + k p1c

Here k pr and kr is the proportional and integration gain of the resonant controller for the outer voltage loop, c and f is the cut-off frequency and the fundamental frequency of the resonant controller respectively, k p1 and ki1 is the proportional and integration gain of the PI controller for the inner current loop, k p 2 and ki 2 is the proportional and integration gain of the PI controller for the harmonics injection loop f h is the cut-off frequency of the high-pass filter. From this equation we can see that F and G are so small that they can be ignored because of the small value of inductor and capacitor. So the order for the denominator and the numerator of this equation can be considered the same. 1 G pr ( s )G pi1 ( s ) + GL ( s )G pi1 ( s ) cs

1 G pr ( s )G pi1 ( s ) + GL ( s )G pi1 ( s ) + ls cs A1s 4 + B1s 3 + C1s 2 + D1s1 = F1s 6 + G1s 5 + A1s 4 + B1s 3 + C1s 2 + D1s1 + E1 iLoad

iL

(6)

112

A1 = k k 1 + k 1 f L c; pr p p B1 = k k 1c + k k 1 f L + k 1kr + k ki1 + k 1 f Lc c + ki1 f L c; pr p pr p p pr p C1 = k k 1 2 + k k 1c f L + k 1kr f L + k ki1c + krki1 + k 1 f L 2 c + ki1 f L 2 c; pr p f pr p p pr p f f D1 = k k 1 f L 2 + k ki1 f L 2 + k ki1 f Lc + krki1 f L + ki1 f L 2 c; f pr p f pr f pr A1 = k k 1 + k 1 f L c + lc 2 + lcc f L ; pr p p f B1 = k k 1c + k k 1 f L + k 1kr + k ki1 + k 1 f Lc c + ki1 f L c; pr p pr p p pr p C1 = k k 1 2 + k k 1c f L + k 1kr f L + k ki1c + k ki1 f L 2 + krki1 + k 1 f L 2 c + ki1 f Lc c; pr p f pr p p pr pr f p f D1 = k k 1 f L 2 + k ki1 f L 2 + k ki1 f Lc + krki1 f L + ki1 f L 2 c; f pr f pr f pr p F1 = lc; G1 = lcc + lcf L ; Here k and kr is the proportional and integration gain of the resonant controller for the pr outer voltage loop, c and f is the cut-off frequency and the fundamental frequency of the resonant controller respectively, k 1 and ki1 is the proportional and integration gain of the PI p controller for the inner current loop, f L is the cut-off frequency of the low-pass filter. As we can see from the Bode plot drawn in Figure 4.24 the lower the inductor value used the lower the inductor current overshoots will be. Inductor current overshoot is not directly related to the inductor value. This is important because the trend is to have a smaller inductor so that the current overshoot is not too high to cause core saturation. With the inductor current as the feedback signal, the inductor current is well regulated. Note that this derivation is not related with switching frequency. Even if the inductor is reduced and switching frequency is kept unchanged the current overshoot can still be limited. In this way more current ripples are generated which will cause more heating on the cables and loads. To avoid the extra loss higher switching frequency can be applied.

113

Figure 4.24 Inductor current to load current with the proposed control method

Figure 4.25 Inductor current to load current with the capacitor current feedback control

114

Contrary to the Bode plot drawn in Figure 4.25 with the capacitor current the lower the inductor value is used the higher the inductor current overshoot will be. This will impose a challenge to the capacitor current based control because this method cannot limit the current overshoot during the load transient. When advanced semiconductor devices which can handle a higher switching frequency are applied to the inverter design, a lower inductor will be used to optimize the system volume; however, with capacitor current control the current overshoot will be increased with a smaller inductor and may cause core saturation. This is the drawback of the capacitor current based controllers but the advantage of the inductor current based controllers. A 1kW load transient has been simulated and tested with the proposed controller using three inductor values: 1mH, 0.5mH and 0.25mH. The illustrated simulation for the inductor current overshoot during the load transient response is shown in figure 4.26. A 1mH and a 0.5mH inductor are tested. The inductor current overshoot with 1mH and 0.5mH are shown in Figure 4.27 and Figure 4.28 respectively. As we can see, the current overshoot of both cases are almost the same which is at about 15A. Using the capacitor current feedback control method the same load transient test has been conducted. A 1mH and a 0.5mH inductor are tested in the experiment. In Figure 4.29 and Figure 4.30 the same transient test is conducted with the capacitor current feedback control. As we can see that at L=1mH the inductor current overshoot is 17A however at L=0.5mH the inductor current overshoot has been increased to 20A, which verifies the above analysis. Finally, based on the analysis, the performance of the inductor current feedback control, capacitor current feedback control and the proposed method is compared and summarized in table 4-2.

115

200 0 -200 20 0 -20 10 0 -10 0.23 20 0 -20 0.232 0.234 0.236 0.238 0.24 0.242 0.244 0.246

0.23

0.232

0.234

0.236

0.238

0.24

0.242

0.244

0.246

0.23

0.232

0.234

0.236

0.238

0.24

0.242

0.244

0.246

0.23

0.232

0.234

0.236

0.238

0.24

0.242

0.244

0.246

Figure 4.26 dynamic response: output voltage, load current, capacitor current and inductor current (curve 1: output voltage; curve 2: load current; curve 3: capacitor current; curve 4: inductor current)

116

Figure 4.27 Inductor current overshoot during the load transient with the proposed control method at L=1mH

Figure 4.28 Inductor current overshoot during the load transient with the proposed control method at L=0.5mH

117

Figure 4.29 Inductor current overshoot during the load transient with the capacitor current control at L=1mH

Figure 4.30 Inductor current overshoot during the load transient with the capacitor current control at L=0.5mH

118

Table 4-2 Performance comparison of the capacitor current feedback, the inductor current feedback and the proposed method

Capacitor current feedback Steady-state operation (linear load) Harmonics compensation (nonlinear load) Dynamic response Current sensor requirement Current overshoot limitation Further reduction of Passive components Over-current protection May cause core saturation Needs an extra current sensor Better 1 CT 1 LEM No Good Good

Inductor current feedback Good

Proposed control method Good

Poor

Good

Good 1 LEM

Good 1 LEM

Yes

Yes

Yes

Yes

Integrated with controller

Integrated with controller

119

4.5 Summary of Chapter Four In this chapter a new inductor current feedback controller based on active harmonics injection is proposed for the stand-alone application of plug-in hybrid electric vehicle application which can be also called vehicle to home application. This new controller can improve the harmonics compensation capability for the controller which senses the inductor current. The inductor current overshoot for both the capacitor current control and the proposed controller is investigated. It is proven that the inductor current control can limit the inductor current overshoot with an even smaller inductor. While using the capacitor current control the current overshoot will be higher with a smaller inductor which may increase the probability of core saturation. Therefore the proposed method will be used to optimize passive components for the power stage design of the converter and further reduce the total power stage volume and weight.

120

Chapter Five Power Management Strategy for Multiple Plug-in Hybrid Electric Vehicles in FREEDM Smart Grid

5.1 Architecture of PHEV Integration with Solid State Transformer based Smart Grid The Solid State Transformer (SST) is considered the key unit for power processing and conversion in the distributed renewable energy internetFREEDM system as shown in figure 5.1. Within IEM (intelligent energy management) the role of SST is to interface with and enable the active management of distributed energy resources, energy storage devices and different types of loads at either household level or industry level. The basic idea of an SST is to use a power electronics converter to replace the conventional bulky and nonintelligent transformer. In addition the SST has the capability to deal with utility issues such as voltage sag, power factor correction, etc [63]. At the household level one SST converts a 12 kV distribution level voltage to a 120V/240V split-phase voltage for a residence.

Figure 5.1 FREEDM smart grid and Solid State Transformer based Intelligent Energy Management System [63]

As the key element for distributed energy storage device (DESD) at level I and II charging, PHEV will interface with the inverter stage of the SST. In this system a solid state

121

transformer supplies power to the chargers of PHEVs. The charger has two power stage (AC/DC and DC/DC) and its model and control algorithm has been developed in the previous chapters. So the scenario of this system is that the source of this system is the inverter stage of SST and the loads are parallel operated PHEVs. Considering only PHEV charging functions, power management strategy is needed when the total possessive power requirement exceeds the limit of the inverter of the SST. The proposed method will be analyzed and system modeling will be presented in the next sections. 5.2 The Issue of Multiple Plug-in Electric Vehicles Connected with Solid State Transformer A smart transformer (solid state transformer) is used as intelligent energy management for the FREEDM smart grid. Its primary goal is to enable power processing and management. A distribution level solid state transformer is designed to supply power to 1~2 US families based on power requirements and consumption. A SST (solid state transformer) is able to supply power to multiple plug-in hybrid electric vehicles. However there is one problem for the power management. If the power demand of plug-in vehicles is higher than the SST and there is no effective communication and power allocation method, the SST voltage will collapse. This will cause problems for other loads supplied by the same SST. This issue can be resolved with two-way communication between the vehicle and the SST. In addition a power allocation algorithm should be applied to ensure that the total demand power will not exceed the safe operating limit of the SST. From the power electronics control point of view, if a better control method is proposed without the need for communication, it will be a very helpful and promising method. Because communication methods may not work or sometimes experience delays. A power electronics converter operates rapidly (in kHz range), so a fast

122

control method is desired to deal with power management at the power converter level rather than the system level. In other words, a smart grid such as FREEDM grid needs a hierarchy controller which includes very high level control including a system plan and real time allocation algorithm and also power electronics level control to guarantee the power management. In this chapter, a new power management method for multiple PHEVs/PEVs operating in FREEDM smart grid is proposed. It will achieve automatic power allocation when the possessive power demand of the vehicles is higher than the power limitation of SST. This method can be used as a converter level control strategy to deal with system instability in the worst case scenario. The worst case means the system level control and communication is disabled. In summary, in this chapter the research focuses on resolving the following issues: 1) Voltage collapse when the possessive load (demand of charging PHEV) exceeds the capability of SST. Power electronics level control strategy to avoid system collapse; 2) Power management relies on communication; its performance suffers from the delay and congestion resulting from the communication method; 3) PHEV needs the guidance to adjust the power in a distributed manner, not based on centralized command; 4) How to enable load management with the help of PHEV. The interface between the SST and the PHEV chargers is the inverter stage of the SST. The control loop model is shown in figure 5.2. The inverter controller includes an outer voltage loop and an inner current loop. First the inner current loop is analyzed and modeled.

123

Figure 5.2 Control loop model of inverter stage of solid state transformer

In the inner current loop the Gcomp_c is the inner PI controller and L is the filer inductor, r is the winding resistance of the inductor. So the closed loop transfer function for the inner current is written as:
k 1 1 ( k p 1 + i1 ) k p1s + ki1 Ls + r = s Ls + r = Ginner ( s ) = k 1 1 Ls 2 + ( k p1 + r ) s + ki1 1 + G pi ( s ) 1 + ( k p 1 + i1 ) Ls + r s Ls + r G pi ( s )

(1)

The Bode plot is drawn in figure 5.3 and as we can see that it can be consider a first order system with a corner frequency around 2kHz. This is also the bandwidth of the inner current loop. The next step is drawn the Bode plot for both of inner and outer loop. In order for the compensator of the outer voltage loop to increase the DC gain at 60Hz, a proportional resonant controller is used as Gcomp_v. So the system open loop transfer function is written as:
Gouter ( s ) = Gcomp _ v ( s ) Ginner ( s ) Gd ( s ) 1 sC f

(2)

In this equation, the Gd is the system delay and Cf is output filter capacitor. The entire system Bode plot is shown in figure 5.4. The bandwidth is 844Hz around 1/10 of the switching frequency and phase margin is 50 degree. The dc gain at 60Hz is boosted to 49dB

124

which reduces the stead-state error. With the open loop transfer function derived the close loop transfer function is plotted in figure 5.5. The output impedance is important for the inverter because the lower the output impedance has less of an effect on load current impacts on the output voltage. The output impedance is derived based on the passive components.
di f dt = uconv i f r uout

(3)

cf

duout = i f iLoad dt

(4)

From control loop another equation can be derived:

uconv = (uref uout ) Gcomp _ v ( s ) Ginner ( s )

(5)

Substitute the equation (3) and (4) into equation (5) and the relationship of load current, output voltage and voltage reference can be derived:
uout = Gcomp _ v ( s ) Ginner ( s ) Gcomp _ v ( s ) Ginner ( s ) + Ls + c f s + 1
2

uref

Ls + r iLoad Gcomp _ v ( s ) Ginner ( s ) + Ls 2 + c f s + 1

(6)

As we can see from equation (6) that if the voltage loop has high magnitude and the output impedance is very low the output voltage will equal the reference voltage. The output impedance can be written as:
Z out = Ls + r Gcomp _ v ( s ) Ginner ( s ) + Ls 2 + c f s + 1

(7)

125

5 0 -5
Magnitude (dB)

Bode Diagram

-10 -15 -20 -25 -30 -35 0

-30
Phase (deg)

-60

-90 0 10

10

10

10
Frequency (Hz)

10

10

Figure 5.3 Bode plot of close loop of inner current loop

60 40 20
Magnitude (dB)

System: Gvoltage Frequency (Hz): 60 Magnitude (dB): 49

Bode Diagram

System: Gvoltage Frequency (Hz): 844 Magnitude (dB): -0.000867

0 -20 -40 -60 -80 -100 -120 0 -45 -90

System: Gvoltage Frequency (Hz): 844 Phase (deg): -131

Phase (deg)

-135 -180 -225 -270 0 10

10

10

10
Frequency (Hz)

10

10

Figure 5.4 Bode plot of outer voltage loop open loop

126

10 0 -10 Magnitude (dB) -20 -30 -40 -50 -60 0

Bode Diagram

-45 Phase (deg)

-90

-135

-180 10
0

10

10

10 Frequency (Hz)

10

10

Figure 5.5 Bode plot of close loop of outer voltage loop

20

Bode Diagram

0
Magnitude (dB) Phase (deg)

-20

-40

-60

-80 180 135 90 45 0 -45 -90 -135 -180 0 10

1 2 3 4 5

10

10

10
Frequency (Hz)

10

10

Figure 5.6 Bode plot of output impedance

127

The Bode plot of the output impedance of the inverter is shown in figure 5.6. The negative peak at 60Hz is caused by the PR controller. The impedance is very low which means a good decoupling from the load effect. However even the control loop shows the good dynamic and voltage regulation performance there is no way the controller can make the system stable when the load power is higher than its thermal capability. 5.3 Proposed Power Management Strategy to Avoid Instability of Solid State Transformer Although the converters of PHEV are operating in parallel, the proposed method cannot rely on droop control [166-169], because droop control can be only used in voltage source parallel operations. However, during grid connected operation the converters are controlled as current sources. It is also difficult to change the power demand from the inverter side because as a voltage source, its output power is only decided by its loading information. So the question comes to the point that to lower the power demand and avoid power collapse the only possible solution is to reduce the charging rate of the chargers. Then another two questions arise: first, how to inform all the chargers to reduce the power demand without communication. Second is how to reduce the power demand if the chargers are informed. To answer the first question, the frequency will be used as the medium for communication between the inverter and the chargers, because all of the chargers are equipped with phase lock loop (PLL) to measure the frequency and synchronize with grid. Similar idea of utilizing the system frequency as the signal to communicate with others is proposed in [170] Even though there is only one inverter to supply the loads; frequency droop can be applied to the inverter control scheme. In this way the both inverter and chargers know the power information even without communication. In this method only frequency droop is discussed

128

because the chargers all have power factor correction capability and only consume real power. So the inverter output frequency will droop with the increase of the output power, the frequency will change in a reasonable range (60.5Hz~59.5Hz) [72]. All of the equipment in addition to the charger can operate within this frequency. Exceeding this range will trip the protection of the inverter. To answer the second question frequency restoration method is proposed. When the frequency of the inverter falls below a certain point it will trigger another control loop inside all of the chargers. Note that control loop is built inside the charger rather than in the inverter because the voltage source cannot change its output power while the current source can. The control loop will have frequency restoration capability which means it will control the inverters output frequency back to a preset safe point. To move the frequency back to the preset point the inverter power output will also reduce to the safe level. So the power collapse crisis is resolved. The frequency restoration method is different from method of Divan and Iravani [171, 172]. In their methods the frequency is used to push the frequency back to the normal value after the frequency droop of several parallel inverters. In this method frequency restoration is implemented in a current controlled source. The frequency target is not the grid frequency but the preset save frequency. The proposed frequency restoration controller is implemented in DC/DC stage of the charger because the AC/DC stage of the charger cannot control the power demand. Normally the AC/DC stage of the charger regulates the DC bus voltage and corrects the power factor. DC/DC stage of the charger determines the charging rate which is the power demand. In normal operation the PHEV user will choose a charging rate and charge the vehicle. When

129

multiple vehicles are connected and charged simultaneously, the total power demand equals the power output of the inverter output. If the power demand exceeds the capability of this inverter and triggers the frequency threshold, the proposed controller will be enabled and reduce the power demand in order to move the frequency up to the predetermined value. All the chargers connected with this inverter will be controlled by the proposed controller at the DC/DC stage. The frequency will move gradually back to the predetermined value and operate in stable state. In other words when multiple vehicles are connected with an inverter and their power demand is higher than the inverters capability the power demand of each vehicle will be reduced gradually to avoid the power collapse. The control architecture for AC/DC and DC/DC stage of the charger is drawn in figure 5.7. The frequency restoration method used in DC/DC stage is shown in figure 5.8. The frequency-power relationship is implemented in the controller of the inverter in figure 5.9. In this operation mode when some vehicles are unplugged from the feeder (inverter) and the total power demand drops, the frequency will increase. The control scheme is designed to let the frequency difference become negative and after the integrator of the proposed controller works for sometime the frequency restoration controller output will drop to zero. This means that the charging rate can return to the user defined value so at this time the control mode goes back to normal operation. The inverter output frequency will not be kept at the preset value but will be determined by its actual output power. Once again, if the loads are increased higher than the limit, the control mode will switch to frequency restoration mode to regulate the output power. Therefore the mode transfer can be automatically achieved.

130

The test system is built with a detailed switching model in Matlab Simulink. The Solid State Transformer is represented by its low-voltage inverter part. Its power rating is set to 10kW and output voltage is 240V. Two PHEVs are AC level II charger with rated power 6kW and are connected together to an SST. A comprehensive case study is conducted, which includes normal operation, total power demand higher than SST power rating, and mode transfer operation.

Figure 5.7 Controller architecture of PHEV charger

Figure 5.8 Proposed power dispatch method based on frequency restoration

= o k Pout

(8)

In this equation, o is the initial frequency 60.5Hz and slope rate parameter K is defined that the power increases from 0 to 10kW with the frequency decreasing from 60.5Hz to 59.5Hz. The K is selected 0.00008 Hz/W.

131

Figure 5.9 Implementation of power-frequency control in inverter stage of SST

Case Study I is the normal operation of charging PHEVs. One PHEV is charging at very beginning with 4kW and another one is plugged into the grid at 0.2s randomly and charging with 4kW. The total power equals to the power capability of SST so each vehicle is charged with its required power. The frequency is measured by a single phase PLL. Although the PLL has short time duration of dynamic response and during that duration the frequency exceeds the frequency limit, a blanking function is used to delay the detection function and prevent the mis-trigger of the power dispatch operation. In figure 5.10 the system operation frequency is shown and this frequency exceeds the limit for around 0.2s and because of the blanking function this frequency didn't trigger the proposed power dispatch method. The frequency after the transient finally stabilized at 59.5Hz which means 10kW output for the SST. In figure 5.11 the charging power for both vehicles is shown and the enlarged waveform is shown in figure 5.12. The charging power for each vehicle is 6kW and for normal operation charging power can be any value as long as the total power doesnt exceed the 10kW power limit. The grid voltage, current and dc bus voltage of no.1 and no.2 vehicle is shown in figure 5.13 and figure 5.14 respectively. As we can see the grid current is in

132

phase with the grid voltage and dc bus is well maintained at 400V. So the control objective for both AC/DC and DC/DC stage is achieved.

60.2 system operation frequency 60.1

60

59.9

59.8

59.7

59.6

59.5

59.4

0.2

0.4

0.6
Time

0.8

1.2

Figure 5.10 SST operation frequency

133

Charging power of No.1 Vehicle 6000 5000 4000 3000 2000 1000 0 0 0.2 0.4 0.6 0.8 1 1.2

Charging power of No.2 Vehicle 6000 5000 4000 3000 2000 1000 0 0 0.2 0.4 0.6
Time

0.8

1.2

Figure 5.11 Charging Power of two vehicles

Charging power of No.1 Vehicle 5000 4995 4990 4985 4980 0.8

0.85

0.9

0.95

1.05

1.1

1.15

1.2

Charging power of No.2 Vehicle 5000 4995 4990 4985 4980 0.8

0.85

0.9

0.95

1
Time

1.05

1.1

1.15

1.2

Figure 5.12 Enlarged charging power of two vehicles

134

Grid current of No.1 Vehicle 50

-50

0.2

0.4

0.6

0.8

1.2

Grid voltage of No.1 Vehicle 500

-500

0.2

0.4

0.6

0.8

1.2

DC bus voltage of No.1 Vehicle 400 300 200

0.2

0.4

0.6
Time

0.8

1.2

Figure 5.13 Voltage and current information of no.1 vehicle

Grid current of No.2 vehicle 50

-50

0.2

0.4

0.6

0.8

1.2

Grid voltage of No.2 vehicle 500

-500

0.2

0.4

0.6

0.8

1.2

DC bus voltage of No.2 Vehicle 400 300 200

0.2

0.4

0.6
Time

0.8

1.2

Figure 5.14 Voltage and current information of no.2 vehicle

135

Case II is for power dispatch control. Initially, the first vehicle is charging at very beginning with full power rating 6kW and the second one is plugged into the grid charging also with full power rating 6kW at 0.2s. The total power rating exceeds the preset power limit. The proposed power dispatch control is enabled. The dispatched power to each vehicle is controlled to be equal and each vehicle is charging with 5kW. In figure 5.15 the system operation frequency is shown. This frequency exceeds the limit after the blanking function so the proposed power dispatch controller is enabled. The power for each vehicle is designed to be equally shared with 5kW. The frequency is stabilized at 59.5Hz which represents the total power for SST 10kW. The equal sharing power for no.1 and no.2 vehicle is shown in figure 5.16 and an enlarged waveform is shown in figure 5.17. The grid voltage, grid current and dc bus voltage for no.1 and no.2 vehicle are shown in figure 5.18 and 5.19 respectively. As we can see that the control performance is good. The dc bus regulation is very stable and the variation is very low with the change of the power. So in this case, the plug-in vehicles have the equal power sharing. Assuming that the vehicles are in a similar condition which means a similar state of charge (SOC), state of health (SOH) and the charging priority, the proposed method works well. For vehicles with different conditions the dispatched power can be different.

136

60 System operation frequency 59.9

59.8

59.7

59.6

59.5

59.4

59.3

59.2

0.5
Time

1.5

Figure 5.15 SST operation frequency

Charging power of No.1 vehicle 8000 6000 4000 2000 0

0.5

1.5

Charging power of No.2 vehicle 8000 6000 4000 2000 0

0.5
Time

1.5

Figure 5.16 Charging power of two vehicles

137

Charging power of No.1 vehicle 4996 4994 4992 4990 4988 4986 1.1 1.15 1.2 1.25 1.3 1.35 1.4 1.45 1.5

Charging power of No.2 vehicle 5016 5014 5012 5010 5008 5006 1.1 1.15 1.2 1.25 1.3
Time

1.35

1.4

1.45

1.5

Figure 5.17 Enlarged charging power of two vehicles

Grid current of No.1 vehicle 50

-50

0.2

0.4

0.6

0.8

1.2

Grid voltage of No.1 vehicle 500

-500

0.2

0.4

0.6

0.8

1.2

DC bus voltage of No.1 vehicle 400 300 200

0.2

0.4

0.6
Time

0.8

1.2

Figure 5.18 Voltage and current information of no.1 vehicle

138

Grid current of No.2 vehicle 50

-50

0.2

0.4

0.6

0.8

1.2

Grid voltage of No.2 vehicle 500

-500

0.2

0.4

0.6

0.8

1.2

DC bus voltage of No.2 vehicle 400 300 200

0.2

0.4

0.6
Time

0.8

1.2

Figure 5.19 Voltage and current information of no.2 vehicle

Case III is for power dispatch control. The first vehicle is charging at very beginning with 6kW and the second one is plugged into the grid charging with 6kW at 0.2s. The total power rating exceeds the preset power limit. The proposed power dispatch control is enabled. Since the stage of charge (SOC) of two vehicle battery is different, the controller is designed to dispatch more power to the vehicle with lower SOC. The first vehicle is dispatched 6kW and the second one is dispatched 4kW. In figure 5.20 the system operation frequency is shown. In this case the frequency also triggers the proposed controller and is stabilized at 59.5Hz. Based on different control parameters the dispatched power for no.1 and no.2 vehicles is shown in figure 5. 21 and enlarged waveform is shown in figure 5.22. The dispatched power is controlled at 4kW and 6kW very accurately. The grid voltage, grid current and dc bus voltage for no.1 and no.2 vehicles is shown in figures 5.23 and 5.24 respectively. As we can

139

see that the control performance is good. The dc bus regulation is very stable and the variation is very low with the change of the power. So the vehicles can have unequal power sharing based on the condition of SOC of batteries.

60 System operation frequency 59.9

59.8

59.7

59.6

59.5

59.4

59.3

59.2

0.5
Time

1.5

Figure 5.20 SST operation frequency

140

Charging power of No.1 vehicle 8000 6000 4000 2000 0

0.5

1.5

Charging power of No.2 vehicle 8000 6000 4000 2000 0

0.5
Time

1.5

Figure 5.21 Charging power of two vehicles

Charging power of No.1 vehicle 6010 6005 6000 5995 5990 1.4

1.41

1.42

1.43

1.44

1.45

1.46

1.47

1.48

1.49

1.5

Charging power of No.2 vehicle 4030 4025 4020 4015 4010 1.4

1.41

1.42

1.43

1.44

1.45
Time

1.46

1.47

1.48

1.49

1.5

Figure 5.22 Enlarged charging power of two vehicles

141

Grid current of No.1 vehicle 50

-50

0.5 Grid voltage of No.1 vehicle

1.5

500

-500

0.5

1.5

DC bus voltage of No.1 vehicle 400 300 200 100 0 0.5

Time

1.5

Figure 5.23 Voltage and current information of no.1 vehicle

Grid current of No.2 vehicle 50

-50

0.5 Grid voltage of No.2 vehicle

1.5

500

-500

0.5

1.5

DC bus voltage of No.2 vehicle 400 300 200 100 0 0.5

Time

1.5

Figure 5.24 Voltage and current information of no.2 vehicle

142

Case IV is for power dispatch control. Mode transfer is also examined. The first vehicle is charging with 6kW initially and the second one is plugged into the grid charging with 6kW at 0.2s. The total power rating exceeds the preset power limit. The proposed power dispatch control is enabled. After a while the first vehicle is unplugged by its user emulating a very normal and random PHEV charging scenario. At that time the power demand is lower than the SST power rating so the normal operation can be reactivated. The power dispatch control at no.2 vehicle will move the charging power back to its setting value and here the power is 6kW. The system frequency will be determined by the power it supplies to the loads according to the frequency and power curve in fig. 5.4. In figure 5.25 the frequency first is controlled at 59.5Hz initially and then moved to 59.9Hz after the disconnection of no.1 vehicle. This proves that the proposed power dispatch method can be automatically switched back to normal charging method. In figure 5.26 the charging power for no.1 vehicle is 6kW at first and 5kW due to power dispatch and then 0kW, while for no.2 vehicle first is 6kW and then 5kW due to power dispatch and finally return to the normal charging rate of 6kW. All the grid voltage, grid current and dc bus voltage of no.1 and no.2 vehicle are shown in figure 5.27 and 5.28 respectively. After the disconnection of one vehicle, the total power drops to a point lower than 12.5kW another vehicle can return to its normal charging rate.

143

60.1 system operation frequency 60 59.9 59.8 59.7 59.6 59.5 59.4 59.3 59.2

0.5
Time

1.5

Figure 5.25 SST operation frequency

Charging power of No.1 vehicle 7000 6000 5000 4000 3000 2000 1000 0 0 0.5 1 1.5

Charging powre of No.2 vehicle 8000 6000 4000 2000 0

0.5
Time

1.5

Figure 5.26 Charging power of two vehicles

144

Grid current of No.1 vehicle 50

-50

0.5 Grid voltage of No.1 vehicle

1.5

500

-500

0.5

1.5

DC bus voltage of No.1 vehicle 400 300 200 100 0 0.5

Time

1.5

Figure 5.27 Voltage and current information of no.1 vehicle

Grid current of No.2 vehicle 50

-50

0.5 Grid voltage of No.2 vehicle

1.5

500

-500

0.5

1.5

DC bus voltage of No.2 vehicle 400 300 200 100 0 0.5

Time

1.5

Figure 5.28 Voltage and current information of no.2 vehicle

145

60.1 system operation frequency 60

59.9

59.8

59.7

59.6

59.5

59.4

59.3

0.5
Time

1.5

Figure 5.29 SST operation frequency

Charging power of No.1 vehicle 6000 5000 4000 3000 2000 1000 0 0 0.5 1 1.5

Charging power of No.2 vehicle 7000 6000 5000 4000 3000 2000 1000 0 0 0.5
Time

1.5

Figure 5.30 Charging power of two vehicles

146

Grid current of No.1 vehicle 50

-50

0.5 Grid voltage of No.1 vehicle

1.5

500

-500

0.5

1.5

DC bus voltage of No.1 vehicle 400 300 200 100 0 0.5

Time

1.5

Figure 5.31 Voltage and current information of no.1 vehicle

Grid current of No.2 vehicle 50

-50

0.5 Grid voltage of No.2 vehicle

1.5

500

-500

0.5

1.5

DC bus voltage of No.2 vehicle 400 300 200 100 0 0.5

Time

1.5

Figure 5.32 Voltage and current information of no.2 vehicle

147

Case V is for power dispatch control. The first vehicle is charging at very beginning with 8kW and the second one is plugged into the grid charging with 6kW at 0.2s. The total power rating exceeds the preset power limit. The proposed power dispatch control is enabled. After a while the second vehicle is unplugged by its user attempting to emulate a very normal and random PHEV charging scenario. At that time the power demand is lower than the SST power rating so normal operation can be switched back. The power dispatch control at no.1 vehicle will move the charging power back to its setting value and, at this point the power is 8kW. The system frequency will be determined by the power it supplies according to the frequency and power curve in fig. 5.4 and for this example it is 59.86Hz. In figure 5.29 the frequency first is controlled at 59.7Hz and then moves to 59.86Hz after unplugging no.2 vehicle. 59.86Hz means the charging power of no.1 vehicle goes back to 8kW again. It proves that the proposed power dispatch method can be automatically switched back to normal charging method. In figure 5.30 the charging power for no.1 vehicle is 8kW and then goes to 6kW because of power dispatch control and again back to 8kW, while for no.2 vehicle first is 6kW and then 4kW because of power dispatch control and finally 0kW. All of the grid voltage, grid current and dc bus voltage of no.1 and no.2 vehicle is shown in figure 5.31 and 5.32 respectively. After the unplugging of one vehicle, the total power drops to lower than 12.5kW another vehicle can come back to its normal charging rate. 5.4 Gain Scheduling Technique to Dispatch Power based on State Charge of Vehicles Based on different state of charge of batteries the controller can be designed to charge the battery with different charging rates during the proposed power dispatch operation. The integration gain of the frequency restoration loop will affect the power dispatched to each

148

vehicle. With different vehicle states of charging tuning integration gain can make the vehicle with less SOC dispatch more power. Since the proposed power dispatch method doesnt need any communication between vehicles the gain scheduling for the integration Ki should be decided by each vehicle without knowing others information. Based on the battery state of charge the integration gain can be simplified into three groups: urgent charge, regular charge and mild charge. Urgent charge means the user needs to charge the vehicle as soon as possible even the power limited by the feeders capability. Regular charge means the user is not in a hurry and wont need as much as power as possible. Mild charge means the vehicle can wait and needs a small amount of power. In urgent charge integration gain Ki is defined as 200, and in regular charge integration gain Ki is 800 and in mild charge integration gain Ki is 1500. Then different integration gain values will be applied in order to find out the power dispatched to each vehicle in six cases. The relationship of integration gain and dispatched power will be used as guidance for each vehicle to share power. One vehicle is set at three conditions, while the other one is controlled to let its state of change vary from 30% to 70% and let its integration gain change in a wide range (100 to 4000). So the first condition is one vehicle in urgent condition which assumes its state of charge is only 30% and its integration gain is 200. The second condition is vehicle in normal condition with its state of charge 50% and its integration gain 1000. The third condition is one vehicle in urgent condition with its state of charge 70% and its integration gain 1500. In each case how to choose the proper integration gain to dispatch the designated power will be shown. The first condition is presented in figure 5.33, the second condition is presented in figure 5.34 and the third condition is shown in figure 5.35.

149

7000 SOC=30% SOC=40% SOC=50% SOC=60% SOC=70%

6500

6000

5500

5000

4500

4000

3500

3000

2500

2000

500

1000

1500

2000

2500

3000

3500

4000

Figure 5.33 Relationship of dispatched power and integration gain Ki with one vehicle at urgent charging and another one with various conditions
8000 SOC=30% SOC=40% SOC=50% SOC=60% SOC=70%

7000

6000

5000

4000

3000

2000

500

1000

1500

2000

2500

3000

3500

4000

Figure 5.34 Relationship of dispatched power and integration gain Ki with one vehicle at normal charging and another one with various conditions

150

8000 SOC=30% SOC=40% SOC=50% SOC=60% SOC=70%

7500

7000

6500

6000

5500

5000

4500

4000

3500

3000

500

1000

1500

2000

2500

3000

3500

4000

Figure 5.35 Relationship of dispatched power and integration gain Ki with one vehicle at mild charging and another one with various conditions

5.5 Load management of Solid State Transformer by Managing Power of PHEVs In the previous sections, a new power management strategy is proposed for multiple PHEVs connected with a solid state transformer. This method is designed to regulate the power of controllable loads e.g., PHEV. This method should be also examined with noncontrollable loads or loads whose power cannot be changed. So in this section, the proposed method is applied to a system, which has an SST, non-controllable loads and a PHEV. The proposed method will manage the power of the PHEV in order to assist the SST to do the load management. Five cases are studied. Case I is the SST has both a non-controllable load and the PHEV connected. The possessive power of the non-controllable load and the PHEV is higher than the power capability of SST. The PHEV will reduce its charging power in order to maintain the system operation frequency at 59.5Hz. In this way the power

151

requirement of the non-controllable load is fulfilled and the SST is fully utilized. This case can be extended to the case, which has several PHEVs connected with SST. The operation principle and control strategy is the same. Case II and III are described that the SST has both a non-controllable load and the PHEV connected. Concurrently, the SST gets the power from a renewable resource e.g. solar or fuel cell generator. So in this case, the SST is not only the power generation unit. With the power coming from external resources, the power requirement of the non-controllable load should be met at first. Then the proposed method will control the PHEV to get the rest of the power. If the rest of the power doesnt meet the need of the PHEV, the frequency will be maintained at 59.5Hz. If the rest of the power meets the need of the PHEV, the frequency will be determined based on the power-frequency curve (figure 5.5). Case IV is the SST has both a non-controllable load and the PHEV connected. In addition, no external generator is connected into the system. However, the non-controllable load is even higher than the power capability of the SST. In this case, the PHEV should discharge rather than charge. It will discharge the power back to the load in order to maintain the system frequency 59.5Hz. The assumption of this case is that the PHEV is willing to send power back to the grid (vehicle to grid) and the vehicle has enough power inside its battery pack. In the actual case, the PHEVs battery may not be in good condition and may have low state of charge; it may not send enough power back to the grid. But this is not the concern of the proposed method. This case should be investigated together with battery management system and system level intelligent control.

152

Case V is the SST has both a non-controllable load and two PHEVs connected. Contrary to Case IV, the PHEVs in case V will share the discharging power. The discharging power can be regulated by control the integration gain Ki in proposed frequency restoration loop. The way of designing gain Ki is similar to the gain scheduling technique which is discussed in previous section. With different integration gain, the vehicle can choose the amount of power discharged to the grid. The total discharging power is determined by the system frequency. In order to fully utilize the SST, the frequency should be maintained at 59.5Hz. The following results demonstrate the three cases. In figure 5.36, the system operation frequency is shown. The non-controllable load is 6kW at first and then the vehicle is plugged into the system. The frequency drops lower than 59.5Hz and the proposed method is enabled. Because the non-controllable load cannot adjust its own power, the vehicle will adjust its power to maintain the system frequency 59.5Hz. At this point, the vehicles power is 4kW, which is drawn in figure 5.37. Figure 5.38 shows the system voltage, input current and dc bus voltage of the vehicle. With the proposed method, the vehicles input current and dc bus voltage are regulated very well. In figure 5.39, the operation system frequency is shown for case II. Since possessive power of SST and renewable energy is higher than the power of load and vehicle, the vehicle is fully charged. SST is not working at its full rating because it is determined by the possessive load power subtracting the power of the renewable energy. If the renewable energys power is higher than the load power, SST will send power back to the utility rather than output any power to the load. In figure 5.40, the charging power of the vehicle is shown. The charging power first tries to reach the rated power 6kW. But it trips the proposed method

153

and the power drops to 4kW to regulate the system frequency. After the power injection from renewable energy, the charging power can go back to rated power again. In figure 5.41, the possessive load current and current injected by renewable energy are shown. In figure 5.42, the operation frequency is shown for case III. In this case, the possessive power of SST and renewable energy is lower than the power of load and vehicle, the vehicle cannot be fully charged. In figure 5.43, the charging power of the vehicle is shown. The non-controllable load is 8kW and the charging power is 2kW without external power. After 2kW power injection of the renewable energy, the charging power is regulated at 4kW to maintain frequency 59.5Hz. SST is fully loaded in this case. In figure 5.44, the possessive load current and current injected by renewable energy are shown. The case IV is verified with the vehicles power discharged back to the grid. At first, SST has no loads. Then the vehicle is plugged into the system and charged at 6kW. Then a 12kW non-controllable load is plugged into the system. The vehicle will discharge the power back to SST in order to keep it stable. The operation frequency is shown in figure 5.45, it is 60.5Hz at first and then 59.9Hz and finally regulated at 59.5Hz. The charging power is shown in figure 5.46. It is 6kW at first and then changes to -2kW. The vehicles input current, grid voltage and dc bus voltage are shown in figure 5.47 and the zoom-in waveform is shown 5.48. The dc bus is regulated very well. The current is synchronized with the grid voltage. At first it is in phase with the voltage and then out-of phase with the voltage. The transition is short and smooth. From the above cases, the proposed method can successfully enable the load management for solid state transformer.

154

Finally the case V is verified with two vehicles power discharged back to the grid. At first, two vehicles are charged at SST and the each vehicle is charged equally with 5kW. Then a 15kW non-controllable load is plugged into the system, the vehicles have to change the mode from charging to discharging. The mode transfer from charging to discharging is determined by the current reference which is generated by the proposed frequency restoration loop. When the frequency drops to a frequency point much lower than 59.5Hz, the current reference will become negative which means the vehicle needs to discharging rather than charge now. The charging loop needs to be disabled and the discharging loop is enabled. In the actual operation, this case can rarely happen because the non-controllable load exceeds the SSTs capability. However, from another aspect, the proposed method can extend the operation range of SST with discharging PHEVs. In figure 5.49, the system operation frequency is shown. In figure 5.50, the power for two vehicles is drawn. No.1 vehicle first charges 5kW and then discharges 2kW. No.2 vehicle first charges 5kW and then discharge 3kW. In figure 5.51 and 5.52, the grid voltage, current and dc bus voltage for no.1 and no.2 vehicle is shown respectively. With discharging vehicles, SST can supply load which is higher than its power capability without any voltage collapse. One thing needs to be addressed that in this case the vehicles must have the capability to discharge and also the health of the batteries are not affected. The decision-making method even game theory can be used to combine the battery information to determine how much power should be discharged to the grid for the vehicle.

155

60 system operation frequency 59.9

59.8

59.7

59.6

59.5

59.4

59.3

59.2

0.5
Time

1.5

Figure 5.36 System operation frequency

7000

6000

5000

4000

3000

2000

1000

0.5
Time

1.5

Figure 5.37 Charging power of the vehicle

156

50

-50 500

0.5

1.5

-500

0.5

1.5

400 300 200

0.5
Time

1.5

Figure 5.38 Grid voltage, current and dc bus voltage of the vehicle
60 59.9 59.8 59.7 59.6 59.5 59.4 59.3 59.2 59.1

0.2

0.4

0.6

0.8
Time

1.2

1.4

1.6

Figure 5.39 System operation frequency

157

7000

6000

5000

4000

3000

2000

1000

0.2

0.4

0.6

0.8
Time

1.2

1.4

1.6

Figure 5.40 Charging power of the vehicle

Gird voltage 500

-500

0.2

0.4

0.6

0.8

1.2

1.4

1.6

Current injection from renewable energy 50

-50

0.2

0.4

0.6

0.8

1.2

1.4

1.6

Possesive load current 100

-100

0.2

0.4

0.6

0.8
Time

1.2

1.4

1.6

Figure 5.41 Injection current from renewable energy and possessive load current

158

60.2 system operation frequency 60

59.8

59.6

59.4

59.2

59

58.8

0.2

0.4

0.6

0.8

1
Time

1.2

1.4

1.6

1.8

Figure 5.42 System operation frequency

6000

5000

4000

3000

2000

1000

0.2

0.4

0.6

0.8

1
Time

1.2

1.4

1.6

1.8

Figure 5.43 Charging power of the vehicle

159

Grid voltage 500

-500

0.2

0.4

0.6

0.8

1.2

1.4

1.6

1.8

Current injection from renewable energy 20

-20

0.2

0.4

0.6

0.8

1.2

1.4

1.6

1.8

Possesive load current 100

-100

0.2

0.4

0.6

0.8

1
Time

1.2

1.4

1.6

1.8

Figure 5.44 Injection current from renewable energy and possessive load current

60.6 system operation frequency 60.4 60.2 60 59.8 59.6 59.4 59.2 59 58.8

0.5

1
Time

1.5

2.5

Figure 5.45 System operation frequency

160

7000 powre from vehicle 6000 5000 4000 3000 2000 1000 0 -1000 -2000 -3000

0.5

1
Time

1.5

2.5

Figure 5.46 Charging and discharging power of the vehicle

Grid current of vehicle 50

-50

0.5

1 Grid voltage of vehicle

1.5

500

-500

0.5

1 DC bus voltage of vehicle

1.5

400 300 200

0.5

1
Time

1.5

Figure 5.47 Grid current, voltage and dc bus voltage of the vehicle

161

Current from vehicle 40 20 0 -20 -40 0.3 500 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8

System AC bus voltage

-500 0.3 440 420 400 380 0.3

0.35

0.4

0.45

0.5

0.55

0.6

0.65

0.7

0.75

0.8

DC bus voltage of vehicle

0.35

0.4

0.45

0.5

0.55
Time

0.6

0.65

0.7

0.75

0.8

Figure 5.48 Grid current, voltage and dc bus voltage of the vehicle (zoom-in)
60.2 system operation frequency 60 59.8 59.6 59.4 59.2 59 58.8 58.6 58.4

0.5

1
Time

1.5

2.5

Figure 5.49 System operation frequency

162

power of no.1 vehicle 8000 6000 4000 2000 0 -2000 -4000 0 0.5 1 1.5 2 2.5

power of no.2 vehicle 8000 6000 4000 2000 0 -2000 -4000 0 0.5 1
Time

1.5

2.5

Figure 5.50 power of no.1 and no.2 vehicle

grid current of no.1 vehicle 50

-50 0.8 500

0.85

0.9

0.95

1.05

1.1

1.15

1.2

1.25

1.3

grid voltage of no.1 vehicle

-500 0.8

0.85

0.9

0.95

1.05

1.1

1.15

1.2

1.25

1.3

dc bus voltage of no.1 vehicle 420 400 380 0.8 0.85 0.9 0.95 1 1.05
Time

1.1

1.15

1.2

1.25

1.3

Figure 5.51 grid voltage, current and dc bus voltage of no.1 vehicle

163

grid voltage of no.2 vehicle 50

-50 0.8 500

0.85

0.9

0.95

1.05

1.1

1.15

1.2

1.25

1.3

grid current of no.2 vehicle

-500 0.8

0.85

0.9

0.95

1.05

1.1

1.15

1.2

1.25

1.3

dc bus voltage of no.2 vehicle 420 400 380 0.8

0.85

0.9

0.95

1.05
Time

1.1

1.15

1.2

1.25

1.3

Figure 5.52 grid voltage, current and dc bus voltage of no.1 vehicle

5.6 Summary of Chapter Five In this chapter, a new power management strategy is proposed to solve the power collapse issue of multiple PHEVs operating in a Solid State Transformer (SST) based smart grid. When multiple PHEVs are plugged into one SST based smart grid and the total demand power is higher than the power capability of SST, a frequency restoration based controller is adopted in each PHEV to reduce its power demand to move the frequency back to a stable point. Gain scheduling technique is proposed to dispatch power of each vehicle based on batterys state of charge. The battery with a low state of charge will be dispatched with more power. The proposed power dispatch method is faster than a two-way communication oriented system level control for the smart grid. So it can be used as a power electronics converter level control to improve the stability of a solid state transformer.

164

Chapter Six High-Order Filter for Compact Size and Ripple Free Charging
6.1 Design GoalCompact Filter Size and Ripple Free Charging High order filter can not only be used in AC application to reduce the filter size but can also be used in battery charging application. In using this type of filter, two primary goals are achieved. First is the reduction of the filter size compared to conventional LC or L filter and yield a compact battery charger. Second is the guarantee that the charging current has almost zero switching ripples. The ripple free charging can avoid the negative effects brought by the high frequency current ripples. Although there is no convincing evidence on the impact of ripple current on batteries, it is widely believed that ripple current may harm the health of batteries because it leads to temperature increases due to the additional internal heating and power losses. Moreover, ripple currents may speed up positive grid corrosion and cause premature failures. Therefore, most battery manufacturers recommend that the ripple current should not exceed 5% of the battery Ahr capacity. Even though ripple currents may have little impact on conventional charging, their effect will be quite notable for the higher current charging or even fast charging [172-175]. Another important concern is the cooling system for the battery bank. In automotive application, it is not practical to design an extra cooling system for battery bank. So it is not easy to take out the heat of the battery. Ripple free charging can reduce the heat of the battery, which alleviates the burden of the cooling system. In this chapter ripple free current charging is achieved by applying an LCL filter and this filter has a much lower value and size than conventional filter with ripple currents. So a compact size and low volume charger can be achieved.

165

6.2 Filter Design and Comparison with Conventional Filter The spectrum of the PWM output voltage of the dc/dc converter is written in (1) u (t ) = D Vdc + 2Vdc sin(h D) cos(hswt ) h h =1

(1)

Set 48V battery with dc bus 100V to design the converter side inductor, the charging current is defined as 10A. The ripple current which is the half of the peak to peak current is 6A, so the converter side inductor can be chosen as a smaller value. iL _ PK = (Vdc Vbatt ) D 2 L1 f s (2)

D=

Vbatt Vdc
(Vdc Vbatt ) (V V ) V (100 48) 48 D = dc batt batt = = 0.208mH 3 2iL _ PK f s 2iL _ PK f s Vdc 2 6 10 10 100

(3)

L1 =

(4)

The converter side current has ripple current 6A and then the ripple current flows through a second order filter network composed by the capacitor C and battery side inductor L2. The ripple current will greatly attenuate after this LC network. To design this L2 and C, the transfer functions of converter side current to converter output and battery side current to converter output are used. Without considering the resistance of both inductors the transfer function are derived: i1 uconv = 1 + L2Cs 2 L1 L2Cs 3 + ( L1 + L2 ) s

(5)

166

i2 uconv

1 L1 L2Cs + ( L1 + L2 ) s
3

(6)

From these two transfer functions, the attenuate ratio between the converter side current and the battery side current. In order to achieve ripple current lower than 0.5A, this ratio is set to 1/12. i2 1 1 = = 2 i1 1 + L2Cs 12 (7)

So the multiplication of L2 and C is 11 and the value can vary but by reducing the L2 value the capacitor value can be higher. In this paper the inductor L2 is set 0.05mH and the capacitor is calculated to be 55uF. Finally the actual value of the LCL filter is: L1=0.2mH, L2=0.05mH, C=80uF. With this filter setup the actual current ripple will be further attenuated to around 0.4A. The peak to peak current ripple to the charging current ratio is less than 1%. While with the same ripple current 0.5A, the conventional L filter is calculate as:
L1 = (Vdc Vbatt ) (V V ) V (100 48) 48 D = dc batt batt = = 2.496mH 3 2iL _ PK f s 2iL _ PK f s Vdc 2 0.5 10 10 100

(8)

The conventional method must use 2.5mH inductor to get the same current ripple. And this inductor is 10 times the total inductance of L1 an L2. The inductor size for converter side inductor is actually larger than theoretical size because of large current ripple on the inductor. The author used Kool Mu core from Magnetics to design these inductors, with different charging current the comparison of LCL filter and L-type filter is shown in table 6.1 and 6.2. First case the charging current equals 10A, all the inductors are designed with E cores. Second case the charging current is increased to 30A. The comparison charts are shown in

167

figure 6.1 and 6.2. We can see that the volume reduction of using LCL filter is around 3~4 times and the loss reduction is around 2~3 times.
Table 6-1 Core volume and loss of L-type filters L-type filter 10A Core volume (mm3) Core loss (W) 19.5 L=2.5mH E-core 237000 L-type filter 30A Core volume (mm3) Core loss (W) 181.2 L=2.5mH E-core 865000

Table 6-2 Core volume and loss of LCL-type filters LCL-type filter 10A Core volume (mm3) Core loss (W) LCL-type filter 30A Core volume (mm3) Core loss (W) 41.6 13.8 6.70 L1=0.2mH E-core 86500 0.86 L2=0.05mH E-core 86500 L1=0.2mH E-core 28600 L2=0.05mH E-core 5340

168

volume of proposed filter at 10A charging

volume of L filter at 10A charging

volume of proposed filter at 30A charging

volume of L filter at 30A charging

250000 200000 150000 100000 50000 0

900000 800000 700000 600000 500000 400000 300000 200000 100000

Figure 6.1 volume comparison between LCL filter and L filter at 10A and 30A charging
loss of proposed filter at 10A charging loss of L filter at 10A charging
loss of proposed filter a 30A charging loss of L filter at 30A charging

20 18 16 14 12 10 8 6 4 2 0 1

200 180 160 140 120 100 80 60 40 20 0 1

Figure 6.2 Filter loss comparison between LCL filter and L filter at 10A and 30A charging

6.3 Controller Design A constant current charging algorithm is designed and implemented in digital controller. Its design and modeling procedure is quite similar to the design of a grid-connected converter with LCL filter. To avoid high frequency oscillation on the filter current, conventional controller needs to be modified. Since high frequency oscillation is caused by the converter PWM output and the resonant frequency of LCL filter, a filter which is designed to extract and eliminate resonant frequency is plugged into the control loop. The merits of this filter are that it functions without extra current or voltage sensor, and it is easy to implement. For the filter type, either notch filter or low pass filter can be used. The control loop model is shown in figure 6.3. The plant is derived as below:

169

di1 L1 dt = uconv uc di2 = uc ubatt L2 dt duc c dt = i1 i2

(9)

In equation (9), L1 is the converter side inductor, L2 is the battery side inductor and C is the filter capacitor. Uconv is the converter output voltage, Uc is the voltage on the filter capacitor, Ubatt is the battery voltage. And i1 is the converter side current and i2 is the battery side current and also the charging current. Transfer the equation (9) from time domain to frequency domain to get equation (10):
L1 i1s = uconv uc L2 i2 s = uc ubatt c u s = i i 1 2 c

(10)

To get the small signal model, the battery voltage is assumed to be constant. Equation (9) can be further derived to:
uconv = L1 i1s + uc uc = L2 i2 s i = i + c u s c 2 1

(11)

From equation (11) substitute i1 and Uc with i2 and then derive the transfer function of charging current to converter output: G plant ( s ) = i2 uconv = 1 L1 L2Cs + ( L1 + L2 ) s
3

(12)

The compensator used here is PI controller in equation (13):

170

GPI ( s ) = k p +

ki s

(13)

The sampling and propagation delay is modeled in equation (14) Gd ( s ) = 1 1+ T s (14)

In equation (14), the delay T is the switching cycle. So the open loop transfer function of the system is written in equation (15) and the bode plot is drawn in figure 6.4 G ( s ) = G pi ( s ) Gd ( s ) G plant ( s ) (15)

As we can see in figure 6.4 that Bode plot of the system with only PI compensation has a negative gain margin and is not stable. The high frequency gain in the system will yield high frequency resonant current on the charging current. In order to solve this issue, filter based method is proposed. The method can extract the resonant frequency and eliminate this frequency. Two types of filters are proposed. The first filter is a second order notch filter and its ideal transfer function is: G filter =
2 s 2 + res s 2 s 2 + res + res Q1

(16)

In the real system, this ideal transfer function will generate a very high negative magnitude at the setting frequency. So the transfer function is modified to:

171

G filter _ notch =

d c res s 2 s2 + + res d

s2 +

res s

2 + res

(17)

The second filter used is a second order low pass filter and its transfer function is: G filter _ low =
2 res

s +
2

res s
Q2

(18)
2 res

After plugging the notch filter transfer into the control loop the whole system transfer function: G ( s ) = G pi ( s ) Gd ( s ) G plant ( s ) G filter _ notch ( s ) (19)

The control plant and notch filter are drawn in figure 6.5. The notch filter resonant frequency matches the plants resonant frequency exactly. From figure 6.6 we can see that after applying notch filter in the control loop, the phase margin is 55 deg close to 60 deg the ideal phase margin for control system. The loop bandwidth is 700Hz a little less than 1/10 of the switching frequency. The gain margin is 10.4 and the high frequency resonant peak is eliminated. The dc gain of the system is almost infinity because of a pole placed at zero frequency which forms a first order system. With this controller the system will have high dc regulation accuracy and less steady-state error. The controller will have good dynamic response and a short settling time because it is directly related to a high phase margin. In figure 6.7 the root locus of the system also proves that the double-pole, which causes the system instability, is compensated with two zeros.

172

* i2

i2

Figure 6.3 System control loop model

40

Bode Diagram

20
Magnitude (dB) Phase (deg)

-20

-40

-60 -90

-135

-180

-225

-270 3 10

10

10

Frequency (rad/sec)

Figure 6.4 Bode plot of system open loop transfer function

173

Bode Diagram

50

0
Magnitude (dB)

-50

-100

-150 90 45 0
Phase (deg)

-45 -90 -135 -180 -225 -270 2 10 10

3 Frequency (Hz)

10

10

Figure 6.5 Bode plot of control plant and notch filter

150

Bode Diagram

100

Magnitude (dB)

50

0 System: Gout2c_damp Gain Margin (dB): 10.4 At frequency (Hz): 1.85e+003 Closed Loop Stable? Yes

-50

-100

-150 -90 -135 -180 Phase (deg) -225 -270 -315 -360 10
-2

System: Gout2c_damp Phase Margin (deg): 55.5 Delay Margin (sec): 0.000218 At frequency (Hz): 708 Closed Loop Stable? Yes

10

-1

10

10

10

10

10

10

Frequency (Hz)

Figure 6.6 System bode plot with proposed notch filter

174

1 0.30/T 0.8 0.35/T

Pole-Zero Map 0.25/T 0.20/T 0.1 0.2 0.15/T

0.6

0.3 0.40/T 0.4 0.5 0.6 0.7 0.45/T 0.8 0.9 0.50/T 0.50/T 0.05/T 0.10/T

0.4

0.2 Imaginary Axis

-0.2 0.45/T -0.4 0.40/T 0.10/T 0.05/T

-0.6

-0.8

0.35/T 0.30/T 0.20/T 0.25/T -0.8 -0.6 -0.4 -0.2 0 Real Axis 0.2 0.4

0.15/T

-1 -1

0.6

0.8

Figure 6.7 Root locus of system with proposed method

The low pass filter is the second proposed method. It aims at reducing all the magnitude after its corner frequency. With its corner frequency tuned at the right value the high frequency peak is largely reduced. The control system with low pass filter is written in equation (20) G ( s ) = G pi ( s ) Gd ( s ) G plant ( s ) G filter _ low ( s ) (20)

In figure 6.8 the Bode plot of the system is drawn. From this figure we can see that the low pass filter based method has the serious drawback of low control bandwidth. The control bandwidth is only 300Hz much slower than the notch filter based method. In addition with the low pass filter the level shifts to all frequencies after only after its corner frequency but does not compensate for the resonant frequency peak. So there is possibility of gain peaking at the resonant frequency. As explained in figure 6.9, once amplified by improper gain or

175

system parameters the resonant gain will move higher than zero dB line which will lead to negative gain margin and an unstable system. In figure 6.10 the root locus plot also proves that the resonant frequency is not compensated for because the double-pole is very close to the boundary of the unit circle and oscillation is easily generated.

150 100 50 0 Magnitude (dB) -50 -100 -150 -200 -250 -300 -350 -90 -135 -180 -225 Phase (deg) -270 -315 -360 -405 -450 -495 -540 10
-2 -1 0 1

Bode Diagram

System: Gout2c_damp Gain Margin (dB): 16.9 At frequency (Hz): 1.13e+003 Closed Loop Stable? Yes

System: Gout2c_damp Phase Margin (deg): 52.5 Delay Margin (sec): 0.000481 At frequency (Hz): 303 Closed Loop Stable? Yes

10

10

10

10

10

10

10

10

Frequency (Hz)

Figure 6.8 System with proposed low pass filter

176

150 100 50 0 Magnitude (dB) -50 -100 -150 -200 -250 -300 -350 -90 -135 -180 -225 Phase (deg) -270 -315 -360 -405 -450 -495 -540 10
-2 -1 0 1

Bode Diagram System: Gout2c_undamp Gain Margin (dB): -12.5 At frequency (Hz): 2.79e+003 Closed Loop Stable? No

10

10

10

10

10

10

10

10

Frequency (Hz)

Figure 6.9 Comparison of system with low pass filter and without low pass filter

1
0.30/T

Pole-Zero Map 0.25/T 0.20/T 0.1 0.2 0.15/T

0.8

0.35/T

0.6

0.3 0.40/T 0.4 0.5 0.6 0.7 0.45/T 0.8 0.9 0.50/T 0.05/T 0.10/T

0.4

0.2
Imaginary Axis

0 0.50/T -0.2
0.45/T 0.05/T

-0.4
0.40/T 0.10/T

-0.6

-0.8

0.35/T 0.30/T 0.20/T 0.25/T

0.15/T

-1 -1

-0.8

-0.6

-0.4

-0.2

0
Real Axis

0.2

0.4

0.6

0.8

Figure 6.10 Root locus of system with proposed low pass filter

177

Based on the analysis above the notch filter based method can achieve better performance than the low filter based method. The notch filter based method is also more robust than the low pass filter based method. In the next section the robustness analysis and dynamic response for notch filter based controller will be examined. 6.4 Controller Robustness Analysis The controller with the notch filter is proven to have better performance than the low pass filter. But usually the notch filter is limited since it can only be designed to be effective at a narrow frequency band. So if a system parameter such as inductance or capacitance has a sudden change, the controller performance of this method is questionable. How to extend the frequency range and whether the extended frequency range affects controller performance will also be discussed. In this section, the controller robustness is analyzed and examined with the variation of capacitance or inductance. In figure 6.11 the control system model with capacitance variation from 0.5 to 1.5 of its original value. This is a very challenging test. In figure 6.12 the control model with inductance variation from 0.8 to 1.2 of its original value is tested. The variation of inductance should emulate the variation of battery internal impedance. In figure 6.13 the control loop with the notch filter set at the original value is connected with both 120uF and 100uF capacitor. The individual Bode plot for 100uF and 120uF is shown in figure 6.14 and 6.15 respectively. When the capacitance becomes larger the resonant frequency moves to a lower frequency. As we can see that prior to the notch filter resonant frequency there is a resonant peak and that peak will move across the zero dB line because the notch filter gain at that frequency is not enough to compensate for that

178

frequency. But before this resonant frequency the phase has already passed -180 degree gradually because of the slow phase drop of the notch filter. So at this resonant frequency, even though the resonant peak crosses over the zero dB line, the phase has already dropped significantly lower than -180. The gain margin of this system is 8.55dB with 100uF and 6.34dB with 120uF. With multiple zero dB crossings, the first crossing is used to measure the bandwidth and phase margin. In figure 6.16 and 6.17 the capacitance is reduced to 60uF and 40uF respectively. With the reduction of the capacitance the resonant frequency has been moved to a higher frequency range. As we can see from the Bode plots, after the compensation from the notch filter the phase has a sudden jump above the -180 degree because of the effect of resonant frequency. At the -180 degree crossing point the magnitude of the system is higher than 0dB so the negative gain margin leads to an unstable system. The notch filter has two parameters c and d in equation (17). They can be used to control the magnitude and frequency range of the notch filter. By tuning c and d the frequency damping range is widened. A notch filter with different c and d parameters is shown in figure 6.18. With a redesigned notch filter, the capacitances 60uF and 40uF are examined again. In figure 6.19 system with 60uF capacitance is stable because with a wider frequency damping range the magnitude is damped to be lower than 0dB even if the phase jumps back to -180 degree line. So the gain margin is 4.34dB and system has returned to a stable state. The bandwidth and phase of the system is determined by the first zero crossing point. But in figure 6.20 the system with 40uF is still not stable because the frequency has moved to a higher level where the notch filter is ineffective for damping its magnitude. So the gain margin becomes negative again and system is not stable. One way to make it stable is to retune the parameters

179

to make the frequency damping range wider. However, as can be seen from figure 6.18, the wider the damping range the lower the boosting phase. The phase will be reduced with a wider damping range and in figure 6.19 we can see that although the system is stable the phase margin has already dropped to 36 degree which may lead to longer settling time and higher overshoot to the control loop.

40 20 0 Magnitude (dB) -20 -40 -60 -80 -100 -120 -90

Bode Diagram C=80uF C=40uF C=120uF

-135 Phase (deg)

-180

-225

-270 10
2

10

10 Frequency (Hz)

10

Figure 6.11 System Bode plot with its filter capacitor variation from 0.5 to 1.5 of original value

180

40 20 0 Magnitude (dB) -20 -40 -60 -80 -100 -120 -90

Bode Diagram L=50uH L=25uH L=75uH

-135 Phase (deg)

-180

-225

-270 10
2

10

10 Frequency (Hz)

10

Figure 6.12 System Bode plot with its battery side inductance variation from 0.5 to 1.5 of original value

150

Bode Diagram

100

Magnitude (dB)

50

0 System: Gout2c_damp Gain Margin (dB): 6.34 At frequency (Hz): 1.82e+003 Closed Loop Stable? Yes

-50

-100

-150 -90 -135 -180 Phase (deg) -225 -270 -315 -360 -405 10
-2

System: Gout2c_damp Phase Margin (deg): 54.9 Delay Margin (sec): 0.000212 At frequency (Hz): 719 Closed Loop Stable? Yes

10

-1

10

10

10

10

10

10

Frequency (Hz)

Figure 6.13 System with 100uF and 120uF capacitance with notch filter controller

181

150

Bode Diagram

100

Magnitude (dB)

50

0 System: Gout2c_damp Gain Margin (dB): 8.55 At frequency (Hz): 1.83e+003 Closed Loop Stable? Yes

-50

-100

-150 -90 System: Gout2c_damp Phase Margin (deg): 54.9 Delay Margin (sec): 0.000212 At frequency (Hz): 719 Closed Loop Stable? Yes

-180 Phase (deg)

-270

-360

-450 10
-2

10

-1

10

10

10

10

10

10

Frequency (Hz)

Figure 6.14 System stable with 100uF capacitance and notch filter controller

150

Bode Diagram

100

Magnitude (dB)

50

0 System: Gout2c_damp Gain Margin (dB): 6.34 At frequency (Hz): 1.82e+003 Closed Loop Stable? Yes

-50

-100

-150 -90 -135 -180 Phase (deg) -225 -270 -315 -360 -405 10
-2

System: Gout2c_damp Phase Margin (deg): 54.2 Delay Margin (sec): 0.000205 At frequency (Hz): 732 Closed Loop Stable? Yes

10

-1

10

10

10

10

10

10

Frequency (Hz)

Figure 6.15 System stable with 120uF capacitance and notch filter controller

182

150

Bode Diagram

100 System: Gout2c_damp Gain Margin (dB): -6.59 At frequency (Hz): 3.26e+003 Closed Loop Stable? No

Magnitude (dB)

50

-50

-100

-150 -45 -90 -135 Phase (deg) -180 -225 -270 -315 -360 10
-2

System: Gout2c_damp Phase Margin (deg): 56.1 Delay Margin (sec): 0.000224 At frequency (Hz): 697 Closed Loop Stable? No

10

-1

10

10

10

10

10

10

Frequency (Hz)

Figure 6.16 System unstable with 60uF capacitance and notch filter

150

Bode Diagram

100

Magnitude (dB)

50

System: Gout2c_damp Gain Margin (dB): -13.6 At frequency (Hz): 3.98e+003 Closed Loop Stable? No

-50

-100

-150 -45 -90 -135 Phase (deg) -180 -225 -270 -315 -360 10
-2

System: Gout2c_damp Phase Margin (deg): 56.6 Delay Margin (sec): 0.000229 At frequency (Hz): 687 Closed Loop Stable? No

10

-1

10

10

10

10

10

10

Frequency (Hz)

Figure 6.17 System unstable with 40uF capacitance and notch filter

183

Bode Diagram

-10

Magnitude (dB) Phase (deg)

-20

-30

-40

-50

-60 90

45

-45

-90 10
1

10

10

10 Frequency (Hz)

10

10

Figure 6.18 Bode plots of different notch filter transfer functions to improve controller robustness

150 100 50 Magnitude (dB) 0 -50 -100 -150 -200 -250 -45 -90 -135 Phase (deg) -180 -225 -270 -315 -360 10
-2

Bode Diagram System: Gout2c_damp Gain Margin (dB): 4.47 At frequency (Hz): 3.26e+003 Closed Loop Stable? Yes

System: Gout2c_damp Phase Margin (deg): 36.2 Delay Margin (sec): 0.00019 At frequency (Hz): 531 Closed Loop Stable? Yes

10

-1

10

10

10

10

10

10

10

Frequency (Hz)

Figure 6.19 System stable with 60uF capacitance and redesigned filter parameters

184

150 100 50 Magnitude (dB) 0 -50 -100 -150 -200 -45 -90 -135 Phase (deg) -180 -225 -270 -315 -360 10
-2

Bode Diagram

System: Gout2c_damp Gain Margin (dB): -2.95 At frequency (Hz): 3.99e+003 Closed Loop Stable? No

System: Gout2c_damp Phase Margin (deg): 36.5 Delay Margin (sec): 0.000192 At frequency (Hz): 527 Closed Loop Stable? No

10

-1

10

10

10

10

10

10

10

Frequency (Hz)

Figure 6.20 System still unstable with 40uF capacitance and redesigned filter

150 100 50 Magnitude (dB) 0 -50 -100 -150 -200 -250 -45 -90 -135 Phase (deg) -180 -225 -270 -315 -360 10
-2

Bode Diagram

System: Gout2c_damp2 Gain Margin (dB): -13.6 At frequency (Hz): 3.98e+003 Closed Loop Stable? No

System: Gout2c_damp1 Phase Margin (deg): 19.1 Delay Margin (sec): 0.000174 At frequency (Hz): 304 Closed Loop Stable? Yes

10

-1

10

10

10

10

10

10

10

Frequency (Hz)

Figure 6.21 System with 40uF capacitor with different notch filter parameters to make loop stable

185

60 40 20 Magnitude (dB) 0 -20 -40 -60 -80 -100 -120 90 45 0 Phase (deg) -45 -90 -135 -180 -225 -270 10
1

Bode Diagram c=80u c=100u c=120u c=60u c=40u filter

10

10

10

10

Frequency (Hz)

Figure 6.22 Notch filter and control plant with all capacitor values (0.5~1.5)
16 14 12 10 8 6 4 0.08 0.09 0.1 0.11 0.12 0.13 0.14 0.15 0.16

1.5

0.5

0 0.08

0.09

0.1

0.11

0.12
Time

0.13

0.14

0.15

0.16

Figure 6.23 Control robustness test: charging current with the filter capacitance change from 80uF to 120uF

186

16 14 12 10 8 6 4 0.08 0.09 0.1 0.11 0.12 0.13 0.14 0.15 0.16

1.5

0.5

0 0.08

0.09

0.1

0.11

0.12
Time

0.13

0.14

0.15

0.16

Figure 6.24 Control robustness test: charging current with the filter capacitance change from 80uF to 100uF
16 14 12 10 8 6 4 0.08 0.09 0.1 0.11 0.12 0.13 0.14 0.15 0.16

1.5

0.5

0 0.08

0.09

0.1

0.11

0.12
Time

0.13

0.14

0.15

0.16

Figure 6.25 Control robustness test: charging current with the filter capacitance change from 80uF to 60uF

187

16 14 12 10 8 6 4 0.08 0.09 0.1 0.11 0.12 0.13 0.14 0.15 0.16

1.5

0.5

0 0.08

0.09

0.1

0.11

0.12
Time

0.13

0.14

0.15

0.16

Figure 6.26 Control robustness test: charging current with the filter capacitance change from 80uF to 40uF

As we can see, that the controller works well with low resonant frequency but has some drawbacks when the resonant frequency shifts to a higher frequency. By tuning the notch filters parameters the performance at the high frequency can be improved. However, with this procedure the phase margin is sacrificed. Tuning the notch filter can make the effective damping range wider but concurrently, the control system loses phase margin because of the notch filter. The controller is robust to the capacitance variation from 0.75 to 1.5 of its original value. At 0.75 C, the phase margin is reduced to only 36 degree. In figure 6.21 an unstable system with 40uF and a stable system with 40uF as well are compared. By further extending the effective frequency range of the notch filter, the system with 40uF capacitor can be made stable. The notch filter is tuned to widen the damped frequency range around the resonant frequency. However, concurrently, the control loop phase margin drops

188

dramatically in the red curve and the phase margin is only 19 degree. This phase margin is not reasonable in a practical design. So here is a tradeoff between the extension of the effective damped frequency range and control loop performance. To make the control loop robust to a wide range of parameters variation the control loop has too suffer from the phase margin reduction. In figure 6.22 the notch filter transfer function and the transfer function of control plant with all of the inductance (25uH to 75uH) is shown. From the control loop design point of view, the variation of capacitance is the same as the variation of inductance because both cases result in the movement of the resonant frequency. The inductance variation mainly comes from the inductor change because the internal inductance of the battery cannot be changed drastically. From figure 6.21 and 6.22 we can see that the frequency variation resulting from the capacitance variation is wider than that of the inductance variation. So the capacitance variation can represent both the inductance and capacitance variation. The inductance variation analysis is the same as the capacitance variation so it is not included. The simulation is conducted to prove the above analysis. In figure 6.23 the capacitance changes from 80uF to 100uF the charging current almost doesnt change. This matches the Bode plot in that the controller for both 80uF and 100uF has the same DC gain, phase margin and bandwidth. In figure 6.24 the capacitance changes from 80uF to 120uF the charging current is well regulated but some low current ripples can be observed. With the resonant frequency moving toward an even lower frequency oscillation may be aroused. In figure 6.25 with the redesigned notch filter the control system is stable. The charging current is regulated at 10A the current ripples are higher because the filter capacitance becomes smaller and absorbs less ripple currents. In figure 6.26 with the

189

capacitance change from 80uF to 40uF the control system is no longer stable and the charging current begins to oscillate. 6.5 Simulation and Experiment Results The simulation is conducted in Matlab Simulink with a 48V lead-acid battery pack. The circuit parameters are converter side inductor 0.2mH, the battery side inductor 0.05mH and the filter capacitor 80uF. The charging current is 10A, dc bus voltage is 100V and the switching frequency is 10kHz. The simulation results first show the control without the proposed filter in figure 6.27. As we can see the high frequency oscillation appears on all currents. The blue curve is the converter side current, the purple curve is the capacitor current and the red curve is the charging current. In figure 6.28 the proposed controller is plugged into the control loop, all currents are stable. The zoom-in waveforms of three currents are shown in figure 6.29. Experimental result for proposed control loop is shown in figure 6.30 and the measured ripple current is shown in figure 6.31. Zoom-in waveforms of all three currents are shown in figure 6.32. So from the experiment results, we can see that converter side current has the peak-to-peak current ripple 12A, the capacitor current is 6A and the charging current has peak-to-peak current ripple 0.6 less than 1% of the charging current. So the target of ripple free charging is achieved.

190

converter-side inductor current 30 20 10 0 0.1 0.101 0.102 0.103 0.104 0.105 0.106 0.107

capacitor current 20 0 -20 0.1 40 20 0 -20 0.1 0.101 0.102 0.103 0.104
Time

0.101

0.102

0.103

0.104

0.105

0.106

0.107

battery-side inductor current

0.105

0.106

0.107

Figure 6.27 Simulation waveforms of three currents without proposed control

Converter side inductor current 20

10

0 0.1 10

0.101

0.102

0.103

0.104

0.105

0.106

0.107

0.108

0.109

0.11

Capacitor current

-10 0.1 11

0.101

0.102

0.103

0.104

0.105

0.106

0.107

0.108

0.109

0.11

Battery side inductor current

10

9 0.1

0.101

0.102

0.103

0.104

0.105
Time

0.106

0.107

0.108

0.109

0.11

Figure 6.28 Simulation waveforms of three currents with proposed control

191

Converter side inductor current 20

10

0 0.1 10

0.1001

0.1002

0.1003 Capacitor current

0.1004

0.1005

0.1006

-10 0.1 10.5 10 9.5 0.1

0.1001

0.1002

0.1003

0.1004

0.1005

0.1006

Battery side inductor current

0.1001

0.1002

0.1003
Time

0.1004

0.1005

0.1006

Figure 6.29 Zoom-in waveforms of three currents with proposed control

Figure 6.30 Experiment results of converter side current, charging current and capacitor current with proposed control

192

converter side current

peak to peak 12A

filter capacitor current

charging current

peak to peak 0.6A

Figure 6.31 Experiment results of converter side current, charging current and capacitor current with proposed control (charging current AC coupled to show ripple)

converter side current

filter capacitor current charging current

Figure 6.32 Experiment results of converter side current, charging current and capacitor current with proposed control (zoom-in)

193

Figure 6.33 Experiment results of current transient response: 1A (0.1C) to 10A (1C) step change

Figure 6.34 Experiment results of current transient response: 10A (1C) to 1A (0.1C) step change

194

Figure 6.35 Experiment results of pulse charging with 100Hz

Figure 6.36 Experiment results of pulse charging with 200Hz

195

In figure 6.33 and 6.34 the control loop dynamic response is tested. In figure 6.33 the charging current is changed from a low charging rate 1A (0.1C) to a high charging rate 10A (1C). The control objective is a battery bank which has a quite slow response but the control response speed is quite good. The settling time is less than 1ms and the current overshoot is less than 10%. In figure 6.34 the charging current is changed from a high charging rate 10A to a low charging rate 2A. In figure 6.35 and 6.36, the pulse charging algorithm is implemented in the control loop. Since the tested battery is a lead-acid battery bank the pulse frequency only needs several hundred hertz. So both 100Hz and 200Hz current pulse charge are tested. 6.6 Summary of Chapter Six In this chapter, a high-order filter is proposed for dc/dc stage of interactive converter to facilitate the charging function. The major achievement of this high-order filter is to reduce the filter size and system volume and moreover significantly reduce the current ripples of the charging current which yield ripple free charging. Ripple free charging can reduce the heat generated by the ripple current and internal resistance of the battery so it can lengthen the batterys lifetime. To solve the resonance issue of the high-order filter, both the notch filter and low-pass filter based controllers are proposed and compared. Comprehensive robustness analysis is conducted on the notch filter based controller. Simulation and experiment results verify the performance of the proposed filter and new controller.

196

Chapter Seven Conclusion and Future Work

7.1 Conclusion In this dissertation a grid-interactive smart charger is proposed for plug-in hybrid electric vehicle in the smart grid applications. The major contributions focus on improving the performance of PHEV for both grid-connection application and off-grid application, and power management strategies of multiple PHEVs in smart grid application. In chapter II a bi-directional grid-interactive charger for plug-in hybrid electric vehicles in household environment is proposed. The infrastructure of a PHEV integrated with the power grid at an American home power circuitry is presented. The proposed converter has three major functions: grid to vehicle (G2V), vehicle to grid (V2G) and vehicle to home (V2H). The detailed converter power stage design for a 10kW lab prototype is reported including passive components design, 3-D modeling of the power stage and the efficiency test of the power stage. The control of three major functions is designed and experimental results verify the performance of the controller. In chapter III an adaptive virtual resistor controller to achieve high power quality for large scale penetration of plug-in hybrid electric vehicles into various power grids is proposed. The trend of high penetration of plug-in hybrid electric vehicles into the power grid of the future seems to be imperative and is the best way to utilize PHEVs as energy storage devices. However when largely connected to the grid the grid impedance varies tremendously and the grid impedance with LC filter of the converter will form a high order filter which will lead to high frequency resonance. This resonance will result in a serious power quality issue which

197

limits the grid-connection for PHEVs. Both the active damping method and passive damping methods are effective ways to resolve this issue. The active damping method has more advantages over the passive one in that it has no power consumption and more control flexibility. Virtual resistor based series active damping method is proposed to solve the resonance issue. A detailed design and control loop analysis is proposed. An auto-tuning capability is proposed to be plugged into the controller to achieve automatic adjustment of the damping parameter for various sets of grid impedances. The auto-tuning method uses the filter capacitor current to extract the resonant current. Then the resonant frequency is detected and is used as the tuning criterion. The proposed control method and auto-tuning method are verified by the simulation and the experimental results. In chapter IV a new control method is proposed as a stand-alone application for plug-in hybrid electric vehicles. The inverter control methods are thoroughly reviewed and classified into three groups. The inductor current feedback control and the capacitor current feedback control are two most widely used control methods. A new method is proposed based on inductor current feedback control. An active harmonics loop is proposed to be a third loop for a double-loop control structure. This loop detects and extracts the harmonics from the output voltage and uses this signal as the feedforward control for the whole loop. The proposed method greatly enhances the harmonics compensation capability for the inductor current feedback control. An experiment for steady-state and the transient is conducted to examine the controller performance. The proposed control method has a better output voltage especially with nonlinear loads. The capacitor current feedback control and inductor current feedback control are compared with a different perspective. The capacitor current feedback

198

control cannot limit the inductor current overshoot during the load transient while the inductor current feedback control can limit the inductor current overshoot even with a smaller inductor. So we conclude that the inductor current based controller can be used to optimize passive components and reduce the total volume and weight of the system. By using the capacitor current based controller the current overshoot will become even higher with a smaller inductor. This trend of increased inductor current overshoot will increase the possibility of magnetic core saturation especially in the low power applications where the cores are designed to operate close to saturation point. The simulation and experimental results verify the proposed idea and analysis. In Chapter V a new power management strategy is proposed to resolve the power collapse issue of multiple PHEVs operating in the Solid State Transformer (SST) based smart grid. When multiple PHEVs are plugged into one SST based smart grid and the total demand for power is higher than the power capability of an SST, a new controller is adopted in each PHEV to reduce its power demand in order to avoid power collapse of the SST. A gain scheduling technique is proposed to dispatch power to each vehicle based on batterys state of charge. The battery with a low state of charge will get more power. This method doesnt require a communication method so it will reduce the dependence of the power electronics converter on communication. It can be used as the converter level control strategy to deal with voltage instability in the worst case scenario. The worst case means both communication and system level control are disabled. The proposed power dispatch method is faster than the two-way communication oriented system level control for the smart grid. So

199

it can be used as the power electronics converter level control to improve the stability of the solid state transformer. In Chapter VI: a high order filter is proposed to be use in DC/DC stage of bi-directional battery charger. The objective is to reduce the filter size which will further reduce the system cost and volume. Another major goal is to attenuate current ripple of the charging current which will yield nearly ripple free charging for batteries. Ripple free charging will reduce the extra heat caused by the current ripple and increase the battery life. The filter based controller is proposed to deal with the potential instability issue brought by a high order filter. The filter based method has the advantages of easily implementation and no need of extra current or voltage sensors. Both low pass filter based controller and notch filter based controller are analyzed and compared. The notch filter based controller has better performance. The control loop design and robustness analyses are conducted and presented. The simulation and experiment results verify the proposed controller. 7.2 Future Work Future work I: a system level study of large scale of PHEV penetration into the power grid. This work will focus on the harmonics interaction among a large number of plug-in vehicles. Even though the individual vehicle meets the grid codes such as IEC61000 and IEEE 519, there may still be some harmonics issues. The issues are caused by series and parallel resonance of grid impedance, parasitic circuits of power electronics. The work should first deliver a switching model of converter which meets the grid code and then large numbers of this model will be integrated together with different grid impedances and grid voltage. The power quality issue with a high penetration of PHEVs will be investigated. Then

200

system level solutions to this harmonics issue such as active filter and converter level solution e.g., random harmonics cancellation should be studied and compared to find a better approach. Future work II: modern control theory such as H-Infinity theory and Sliding Mode theory can be applied to grid connection control. Adaptive tuning based on different impedances can be achieved. Better performance and less usage of sensing circuitry compared to PI voltage oriented control (VOC) or PI state space control can be expected. Future work III: define a frequency based standard for the connection of PHEVs with solid state transformer. Based on this frequency, the voltage collapse protection and load management will be achieved. Besides using the frequency, the voltage can be also used to explore if the reactive power is needed by the loads. The reactive power sharing can be supplied by both the solid state transformer and PHEVs.

201

REFERENCES
[1] [2] [3] L. Sanna, Driving the solutions the plug-in hybrid vehicle, EPRI Journal, fall, 2005. U.S. Energy Information Administration, International Energy Outlook 2010: http://www.eig.gov. T. Markel, Plug-in hybrid electric vehicle current status, long-term prospects and key challenges, Presented at Clean Cities Congress and Expo, May 2006. [4] R. Udall, S. Andrews, M. King Hubbert center for petroleum supply studies, Hubbert Center Newsletter vol. 1, 1999. [5] [6] IEEE USA, Plug-in electric hybrid vehicles, 15 June 2007. M. Kintner-Meyer, K. Schneider, R. Pratt, Impacts assessment of plug-in hybrid vehicles on electric utilities and regional U.S power grids Part I: technical analysis, Pacific Northwest National Laboratory, November, 2007. [7] M. Scott, M. Kintner-Meyer, D.B. Elliott, W.M. Warwick, Impacts assessment of plug-in hybrid vehicles on electric utilities and regional U.S power grids Part II: economic assessment, Pacific Northwest National Laboratory, November, 2007. [8] [9] M. Duvall, Plug-in hybrids on the horizon building a business case, EPRI Journal, spring, 2008. K. Parks, P. Denholdm T. Markel, Costs and emissions associated with plug-in hybrid electric vehicle charging in the Xcel Energy Colorado service territory, Technical Report NREL/TP-640-41410 May 2007. [10] EPRI, Environmental assessment of plug-in hybrid electric vehicles, volume 1: Nationwide Greenhouse Gas Emissions, Final Report, July 2007. [11] [12] Plug-in America Website: http://www.pluginamerica.com. T. Markel, Plug-in hybrid electric vehicle current status, long-term prospects and key challenges, Presented at Clean Cities Congress and Expo, May 2006. [13] North American Society of Automotive Engineer (SAE), SAE electric vehicle and plug in hybrid electric vehicle conductive charge coupler, J1772 Standard, 2010.

202

[14] [15]

Advanced technology vehicles, WM7200 Inductive Charger Owners Manual. Chwei-sen Wang, O.H. Stielau, G.A. Covic, Design considerations for a contactless electric vehicle battery charger, IEEE Trans. Ind. Electron., vol. 52, no.5, pp. 1308-1314, 2005.

[16]

N.H. Kutkut, D.M. Divan, D.W. Novotny, R.H. Marion, Design consideration and topology selection for a 120-kW IGBT converter for EV fast charging, IEEE Trans. Power Electron., vol. 13, no.1, pp. 169-178, 1998.

[17]

M.G. Egan, D.L. OSullivan, J.G. Hayes, M.J. Willers, C.P. Henze, Power-factor-corrected single-stage inductive charger for electric vehicles batteries, IEEE Trans. Ind. Electron., vol. 54, no.2, pp. 12171216, 2007.

[18]

J.T. Boyes, G.A.J. Elliott, G.A. Covic, An appropriate magnetic coupling co-efficiency for the design and comparison of ICPT pickups, IEEE Trans. Power Electron., vol. 22, no.1, pp. 333-135, 2007.

[19]

J.G. Hayes, M.G. Egan, J.M.D. Murphy, S.E. Schulz, J.T. Hall, Wide-load-range resonant converter supplying the SAE J-1773 electric vehicle inductive charging interface, IEEE Trans. Ind. Appl., vol. 35, no.4, pp. 884-895, 1999.

[20] [21] [22]

Nissan Leaf Website: http://www.nissanusa.com/leaf-electric-car. AC Propulsion tzeroTM Technology Website: http://www.acpropulsion.com/products-tzero.html. G. Pellegrino, E. Armando, P. Guglielmi, An integral battery charger with power factor correction for electric scooter, IEEE Trans. Power Electron., vol. 25, no.3, pp. 751-759, 2010.

[23]

Young-Joo Lee, A. Khaligh, Ali Emadi, Advanced integrated bidirectional AC/DC and DC/DC converter for plug-in hybrid electric vehicles, IEEE Trans. Veh. Technol., vol. 58, no.8, pp. 3970-3980, 2009.

[24]

Seung-Ki Sul, Sang-Joon Lee, An integral battery charger for four-wheel drive electric vehicle IEEE Ind. Appl. vol. 31, no.5, pp. 1096-1099, 1995.

[25]

Gui-Jia Su, Utilizing the traction drive power electronics system to provide plug-in capability for HEVs, Advanced Power Electronics and Electric Machinery Program Annual Report, 2007.

203

[26]

Jih-Sheng Lai, D. J. Nelson, Energy management power converters in hybrid electric and fuel cell vehicles, IEEE Proceedings, vol. 95, no. 4, pp.766-777, 2007

[27]

Xiaohu Zhou, S. Lukic, S. Bhattacharya, A. Huang, Design and control of grid-connected converter in bi-directional battery charger for plug-in hybrid electric vehicle application, in Proc IEEE VPPC09, pp. 1716-1721, 2009.

[28]

D. Aggeler, F. Canales, H. Zelaya-De La Parra, A. Coccia, N. Butcher, O. Apeldoorn, Ultra-fast DCcharge infrastructure for EV-mobility and future smart grids, in Proc IEEE ISGT Europe 2010, pp. 1-8, 2010.

[29]

D. P. Urciuoli, C. W. Tipton, Development of a 90 kW Bi-directional DC-DC Converter for Power Dense Applications, in Proc IEEE APEC06 pp. 1375-1378, 2006.

[30]

M. Pepper, K. Mansfield, J. Elmes, K. Rustom, etc Bi-directional DCM DC to DC Converter for Hybrid Electric Vehicles, in Proc IEEE PESC08 pp. 3088-3092, 2008.

[31]

Yu Du, Xiaohu Zhou, Sanzhong Bai, S. Lukic, A. Huang, Review of non-isolated bi-directional DC-DC converters for plug-in hybrid electric vehicle charge station application at municipal parking decks, Proc IEEE APEC10 pp. 1145-1151, 2010

[32]

S. Waffler, J. W. Kolar, A Novel Low-Loss Modulation Strategy for High-Power Bidirectional Buck+Boost Converter, IEEE Trans. Power Electron., vol. 24, no.6, pp. 1589-1599, 2009.

[33]

Mustansir H. Kheraluwala, High-Power High-Frequency DC-To-DC Converters, Dissertation, University of Wisconsin-Madison, 1991.

[34]

Fang Z. Peng, Hui Li, Gui-Jia Su an Jack S. Lawler, A new ZVS bidirectional DC-DC converter for fuel cell and battery application, IEEE Trans. Power Electron., vol. 19, no.1, pp. 54-65, 2004.

[35]

M N. Kheraluwala, R. W Gascoigne, D. M. Divan, E. D. Baumann, Performance characterization of a high-power dual active bridge dc-to-dc conveter, IEEE Trans. Ind. Appl., vol. 28, no.6, pp. 1294-1301, 1992.

204

[36]

R. W. A.A De Doncker, D. M. Divan, M. H. Kheraluwala, A three-phase soft-switched high-powerdensity dc/dc converter for high-power applications, IEEE Trans. Ind. Appl., vol. 27, no.1, pp. 63-73, 1991.

[37]

Lizhi Zhu, A novel soft-commutating isolated boost full-bridge zvs-pwm dc-dc converter for bidirectional high power applications, IEEE Trans. Power Electron., vol. 21, no.2, pp. 422-429, 2006.

[38]

Lizhi Zhu, Kunrong Wang, Fred C. Lee, Jin-Sheng Lai, New start-up schemes for isolated full-bridge boost converters, IEEE Trans. Power Electron., vol. 18, no.4, pp. 946-951, 2003.

[39]

F. Boico, B. Lehman, Solar battery chargers for NiMH batteries, IEEE Trans. Power Electron., vol. 22, no.5, pp. 1600-1669, 2007.

[40]

S.J. Chiang, Hsin-Jang Shieh, Ming-Chieh Chen, Modeling and control of PV charger system with SEPIC converter, IEEE Trans. Ind. Electron., vol. 56, no.11, pp. 4344-4353, 2009.

[41]

R. Gules, De Pellegrin Pacheco, H.L. Hey, J. Imhoff, A maximum power point tracking system with parallel connection for PV stand-alone applications, IEEE Trans. Ind. Electron., vol. 55, no.7, pp. 26742683, 2008.

[42]

M.J.V. Vazquez, J.M.A. Marquez, F.S. Manzano, A methodology for optimizing stand-alone PVsystem size using parallel-connected DC/DC converters, IEEE Trans. Ind. Electron., vol. 55, no.7, pp. 2664-2673, 2008.

[43]

Ying-Chun Chuang, Yu-Lung Ke, Hung-Shiang Chuang, Hung-Kun Chen, Implementation and analysis of an improved series-loaded resonant DC-DC converter operating above resonance for battery chargers, IEEE Trans. Ind. Appl., vol. 45, no.3, pp. 1052-1059, 2009.

[44]

Zhenhua Jiang, R.A. Dougal, Control strategies for active power sharing in a Fuel-Cell-powered battery-charging station, IEEE Trans. Ind. Appl., vol. 40, no.3, pp. 917-924, 2004.

[45]

F. Koyanagi, Y. Uriu, R. Yokoyama, Possibility of fuel cell fast charger and its arrangement problem for the infrastructure of electric vehicles, in Proc IEEE PPT01 2001.

205

[46]

Zhenhua Jiang, R. Leonard, R. Dougal, H. Figueroa, A. Monti, Processor-in-the-loop simulation, realtime hardware-in-the-loop testing, and hardware validation of a digitally-controlled, fuel cell power battery-charging station, in Proc IEEE PESC04, pp. 2251-2258, 2004.

[47]

T. Jacobs, J. Beukes, Suitability of fuel cell technology for electricity utility standby power applications, in Proc IEEE INTELEC06 pp. 1-7, 2006.

[48]

W.A. Adams, J.D. Blair, K.R. Bullock, C.L. Gardner, Laishui Li, Cost/benefit analyses of a new battery pack management technique for telecommunication applications: future directions with fuel cell/battery systems, in Proc IEEE INTELEC04, pp. 73-82, 2005.

[49]

J. Tomic, W. Kempton, Using fleets of electric-drive vehicles for grid support, Journal of Power Source, 168 pp. 459-468, 2007.

[50]

W. Kempton, J. Tomic, Vehicle-to-grid power fundamentals: calculating capacity and net revenue, Journal of Power Source, 144 pp. 268-279, 2005.

[51]

W. Kempton, J. Tomic, Vehicle-to-grid power fundamentals: from stabilizing the grid to supporting large-scale renewable energy, Journal of Power Source, 144 pp. 280-294, 2004.

[52]

S. Han, K. Sezaki, Development of an optimal vehicle-to-grid aggregator for frequency regulation, IEEE Trans. Smart Grid, 2010.

[53]

I. Cvetkovic, T. Thacker, Dong Dong, G. Francis, V. Podosinov, D. Boroyevich, F. Wang, R. Burgos, G. Skutt, J. Lesko, Future home uninterruptible renewable energy system with vehicle-to-grid technology, in Proc. IEEE ECCE09, pp. 2675-2681, 2009.

[54]

B. Kramer, S. Chakraborty, B. Kroposki, A review of plug-in vehicles and vehicle-to-grid capability, in Proc IEEE IECON08, pp. 2278-2283, 2008.

[55]

M.E. Ropp, Similarities between vehicle-to-grid interfaces and photovoltaic systems, in Proc IEEE VPPC09, pp. 1331-1335, 2009.

[56]

A. Rubaai, V. Udo, An adaptive control scheme for load-frequency control of multi-area power system part I and part II, Electric Power System research., 24, pp. 183-197, 1992.

206

[57]

W. Kempton, S. Letendre, Electric vehicles as a new source of power for electric utilities, Transportation Research Part D: Transport and Environment, 2(3), pp. 157-175, 1997.

[58]

A. Bubaai, V. Udo, Self-tuning load frequency control: multilevel adaptive approach, in IEEE Proc. Generation, Transmission and Distribution, 141(4), 1994.

[59]

S.M. Lukic, Jian Cao, R.C. Bansal, F. Rodriguez, A. Emadi, Energy storage systems for automotive applications, IEEE Trans. Ind. Electron., vol. 55, no.6, pp. 2258-2267, 2008.

[60]

R. Lawrence, S. Middlekauff, The new guy on the block: applying distributed generation tools in power design systems, IEEE Mag. Ind. Appl., 2005.

[61]

J.M. Guerrero, F. Blaabjerg, T. Zhelev, K. Hemmes, E. Monmasson, S. Jemei, M.P. Comech, R. Granadino, J.I. Frau, Distributed generation: toward a new energy paradigm, IEEE Mag. Ind. Electron. Vol. 4, no.1, pp. 44-52, 2010.

[62]

Alex Q. Huang, Gerald Heydt, Steiner Dale, Jim Zheng, and Mariesa Crow, Energy Internet-Future Renewable Electric Energy Delivery and Management (FREEDM) Systems, Power Electronics Society NEWSLETTER Fourth Quarter, 2008 pp. 8-9.

[63]

Alex. Q. Huang, Mariesa L. Crow, Gerald Thomas Heydt, Jim P. Zheng and Steiner J. Dale, The future renewable electric energy delivery and management (FREEDM) system: the energy internet, IEEE Proceedings, pp.133-148, 2010.

[64] [65] [66]

Ali Ipakchi, F. Albuyeh, Grid of the future, IEEE Mag. on Power & Energy, pp. 52-62, Sep, 2009. FREEDM Smart House: http://www.freedm.ncsu.edu/. Xiaohu Zhou, Gangyao Wang, S. Lukic, S. Bhattacharya, A. Huang, Multi-function bi-directional battery charger for plug-in hybrid electric vehicle applications, in Proc. IEEE ECCE09, 2009

[67] [68]

Peter Gevorkian, Alternative Energy Systems in Building Design, McGraw-Hill 2010 Solar Energy International: Photovoltaics: Design and Installation Manual: renewable energy education for a sustainable future New Society Publisher 2004.

207

[69]

Net metering: hearing before the Subcommittee on Energy of the Committee on Energy and Natural Resources, United States Senate, One Hundred Eleventh Congress, first session. United States. Congress. Senate. Committee on Energy and Natural Resources. Subcommittee on Energy. May 7, 2009

[70]

S. Horowitz, C. Christensen, R. Anderson, Searching for the optimal mix of solar and efficiency in zero net energy buildings Electronic resource

[71]

M. Seneviratne, A practical approach to water conservation for commercial and industrial facilities, Elsevire/Butterworth-Heinemann, 2007

[72]

IEEE Application Guide for IEEE std 1547, IEEE Standard for Interconnecting Distributed Resources with Electric Power Systems December 2008

[73]

Jin Wang, Peng, F.Z, J. Anderson, A. Joseph, and R. Buffenbarger, Low cost fuel cell converter system for residential power generation, IEEE Trans. Power Electron., vol. 19, no.5, pp. 1315-1322, 2004.

[74] [75]

Fuel Cell Handbook, 6th ed., Nat. Energy Technol. Lab, U.S. Dept. Energy, Washington, DC, 2002. Yi Huang, Miaosen Shen, Peng, F.Z, Jin Wang, Z-source inverter for residential photovoltaic systems, IEEE Trans. Power Electron., vol. 21, no.6, pp. 1776-1782, 2006.

[76]

X. Yu, M. R. Starke, L. M. Tolbert and B. Ozpineci, Fuel cell power conditioning for electric power applications: a summary, IET Electric Power Applications, vol. 49, pp. 643-656, Sep. 2007.

[77]

Xiaohu Zhou, S. Lukic, A. Huang, A comparison of converter topologies for vehicle to grid applications: three-leg converter versus H-bridge converter in Proc. IEEE IECON09, 2009.

[78]

S. B. Kjaer, J. K. Pedersen, F. Blaabjerg, A review of single-phase grid-connected inverters for photovoltaic modules, IEEE Trans. Ind. Appl., vol. 41, no.5, pp. 1292-1306, 2005.

[79]

F. Blaabjerg, Zhe Chen, S. B. Kjaer, Power electronics as efficient interface in dispersed power generation systems, IEEE Trans. Power Electron., vol. 19, no.5, pp. 1184-1194, 2004.

[80]

T. Shimizu, K. Wada, N. Nakamura, Flyback-Type single-phase utility interactive inverter with power pulsation decoupling on the dc input for an AC photovoltaic module system, IEEE Trans. Power Electron., vol. 21, no.5, pp. 1264-1272, 2006.

208

[81]

T. Kerekes, M. Liserre, R. Teodorescu, C. Klumpner, M. Sumner, Evaluation of three-phase transformerless photovoltaic inverter topologies, IEEE Trans. Power Electron., vol. 24, no.9, pp. 22022211, 2009.

[82]

Yaosuo Xue, Liuchen Chang, S.B. Kjaer, J. Bordonau, T. Shimizu, Topologies of single-phase inverters for small distributed power generators: an overview, IEEE Trans. Power Electron., vol. 19, no.5, pp. 1305-1314, 2004.

[83] [84]

Melcosim_ver_08_10 Power loss simulation software: Mitsubishi F. Blaabjerg, R. Teodorescu, M. Liserre, A.V. Timbus, Overview of control and grid synchronization for distributed power generation systems, IEEE Trans. Ind. Electron., vol. 53, no.5, pp. 1398-1409, 2006.

[85]

A. Timbus, M. Liserre, R. Teodorescu, P. Rodriguez, F. Blaabjerg, Evaluation of current cotnroller for distributed power generation systems, IEEE Trans. Power Electron., vol. 24, no.3, pp. 654-664, 2009.

[86]

M. Liserre, R. Teodorescu, F. Blaabjerg, Multiple harmonics control for three-phase grid converter systems with the use of PI-RES current controller in a rotating frame, IEEE Trans. Power Electron., vol. 21, no.3, pp. 836-841, 2006.

[87]

Liang-Rui Chen, Neng-Yi Chu, Chau-Shing Wang, Ruey-Hsun Liang, Desing of a reflex-based bidirectional converter with the energy recovery function, IEEE Trans. Ind. Electron., vol. 55, no.8, pp. 3022-3029, 2008.

[88]

L. R. Chen, C. S Wang, W. R. Yang, N. Y Chu, A novel bidirectional converter with reflex charging function, in Proc. IEEE ICIT08 2008.

[89]

R.C. Cope, Y. Podrazhansky, The art of battery charging, The Fourteenth Annual Battery Conference on Applications and Advance, pp. 223-235, 1999.

[90]

Min Chen, G.A. Rincon-Mora, Accurate electrical battery model capable of predicting runtime and I-V performance, IEEE Trans. Energy Convers., vol. 21, no.2, pp. 504-511, 2006.

209

[91]

Kazuo Onda, Masato Nakayama, Kenichi Fukuda, Kenji Wakahara, Takuto Araki, Cell impedance measurement by Laplace transformation of charge or discharge current-voltage, Journal of the Electrochemical Society, 153(6), pp.1012-1018, 2006.

[92]

P.L. Moss, G. Au, E.J. Plichta, J.P. Zheng, An electrical circuit for modeling the dynamic response of Li-Ion Polymer Batteries, Journal of the Electrochemical Society, 155(12), pp.986-994, 2008.

[93]

Mark.W. Verbrugge, Ping Liu, Electrochemical characterization of high-power lithium ion batteries using triangular voltage and current excitation sources, Journal of Power Sources, 174 pp. 2-8, 2007.

[94]

Yoshiyasu Saito, Thermal behaviors of lithium-ion batteries during high-rate pulse cycling, Journal of Power Source, 146, pp. 770-774, 2005.

[95]

R. Teodorescu, F. Blaabjerg, M. Liserre, P.C. Loh, Proportional-resonant controllers and filters for gridconnected voltage-source converters, Electric Power Applications, IEE Proceedings, vol. 153, no.5, pp. 750-762, 2006.

[96]

A.V. Timbus, M. Ciobotaru, R. Teodorescu, F. Blaabjerg, Adaptive resonant controllers for gridconnected conveters in distributed power generation systems, in Proc. IEEE, APEC06 2006.

[97]

D.N. Zmood, D.G. Holmes, Stationary frame current regulation of PWM inverters with zero steadystate error, IEEE Trans. Power Electron, vol. 18, no.3, pp. 814-822, May, 2003.

[98]

X. Yuan, W. Merk, H .Stemmler, J. Allmeling, Stationary-frame generalized integrators for current control of active power filters with zero steady-state error for current harmonics of concern under unbalanced and distorted operating conditions, IEEE Trans. Ind. Appl., vol. 38, no. 2, pp. 523532, Mar./Apr. 2002.

[99]

C. Lascu, L. Asiminoaei, I. Boldea, and F. Blaabjerg, High performance current controller for selective harmonic compensation in active power filters, IEEE Trans. Power Electron., vol. 22, no. 5, pp. 1826 1835, Sep. 2007.

[100] M. Liserre, F. Blaabjerg, S. Hansen, Design and control of an LCL-filter based three-phase active rectifier, IEEE Trans. Ind. Appl., vol. 41, no.5, pp. 1281-1291, 2005.

210

[101] K. Jalili, S. Bernet, Design of LCL filters of active-front-end two-level voltage-source converters, IEEE Trans. Ind. Electron., vol. 56, no.5, pp. 1674-1689, 2009. [102] Michael Lindgren, Modeling and control of voltage source converters connected to the grid, Ph.D Dissertation, Chalmers University of Technology, Goteborg, Sweden, November 1998. [103] M. Liserre, A. DellAquila, F. Blaabjerg, An overview of three-phase voltage source active rectifier interfacing the utility, 2003 IEEE Bologna Power Tech Conference Proceedings, 2003. [104] IEEE Standard 519-1992, IEEE recommended practices and requirements for harmonic control in electric power systems, Institute of Electrical and Electronics Engineers, Inc. 1993. [105] IEEE Application Guide for IEEE std 1547, IEEE Standard for Interconnecting Distributed Resources with Electric Power Systems December 2008. [106] IEC 61000-3-4: Part 3-4: Limits for harmonic current emissions (equipment input current greater than 16A per phase), International Electrotechnical Commission, first edition, 1998-10. [107] M. Liserre, A. DellAquila, F. Blaabjerg, Genetic algorithm-based design of the active damping for an LCL-filter three-phase active rectifier, IEEE Trans. Power Electron., vol. 19, no.1, pp. 76-86, 2004. [108] M. Malinowski, S. Bernet, A simple voltage sensorless active damping scheme for three-phase PWM converter with an LCL filter, IEEE Trans. Ind. Electron., vol. 55, no.4, pp. 1876-1880, 2008. [109] C. Dick, S. Richter, M. Rosekeit, J. Rolink and R. De Doncker, Active Damping of LCL Resonance with Minimum Sensor Effort by means of a Digital Infinite Impulse Response Filter, in Proc. IEEE EPE07, Sept. 2007, pp. 1-8. [110] J. Dannehl, M. Liserre, F. W. Fuchs, Filter-based Active Damping of Voltage Source Converters with LCL Filter, IEEE Trans. Ind. Electron., vol. 58, no.8, pp. 3623-3633, 2011. [111] V. Blasko, V. Kaura, A Novel Control to Actively Damp Resonance in Input LC Filter of a Three-Phase Voltage Source Converter, IEEE Trans. Ind. Appl., vol. 33, no.2, pp. 542-550, 1997. [112] I. J. Gabe, V. F. Montagner, H. Pinheiro, Design and Implementation of a Robust Current Controller for VSI Connected to the Grid Through an LCL Filter, IEEE Trans. Power Electron., vol. 24, no.6, pp. 1444-1452, 2009.

211

[113] Yun Wei Li, Bin Wu, N.R. Zargari, J.C. Wiseman, D. Xu, Damping of PWM Current-Source Rectifier Using a Hybrid Combination Approach, IEEE Trans. Power Electron., vol. 22, no.4, pp. 1383-1393, 2007. [114] P. Loh, D. Holmes, Analysis of Multiloop Control Strategies for LC/CL/LCL-Filtered Voltage-Source and Current-Source Inverters, IEEE Trans. Ind. Appl., vol. 41, no.2, pp. 644-654, 2005. [115] E. Twining, D. Holmes, Grid Current Regulation of a Three-Phase Voltage Source Inverter with an LCL Input Filter, IEEE Trans. Power Electron., vol. 18, no.3, pp. 888-895, 2003. [116] J. Dannehl, C. Wessels, F. W. Fuchs, Limitations of Voltage-Oriented PI Current Control of GridConnected PWM Rectifier with LCL Filters, IEEE Trans. Ind. Electron., vol. 56, no.2, pp. 380-388, 2009. [117] M. H. Bierhoff, F. W. Fuchs, Active Damping for Three-Phase PWM Rectifiers with High-Order LineSide Filters, IEEE Trans. Ind. Electron., vol. 56, no.2, pp. 371-379, 2009. [118] J. Dannehl, F.W. Fuchs, S. Hansen, P.B. Thogersen, Investigation of Active Damping Approaches for PI-based Current Control of Grid-Connected PWM Converters with LCL Filters, IEEE Trans. Ind. Appl., vol. 46, no.4, pp. 1509-1517, 2010. [119] J. Wiseman, Bin Wu, Active Damping Control of a High-Power PWM Current-Source Rectifier for Line-Current THD Reduction, IEEE Trans. Ind. Electron., vol. 52, no.3, pp. 758-764, 2005. [120] F. Magueed, J. Svensson, Control of VSC Connected to the Grid through LCL-Filter to Achieve Balanced Currents, in Proc IEEE IAS05, 2005, pp. 572-578. [121] Zhiling Qiu, Jie Kong, Guozhu Chen, A Novel Control Approach for LCL-based Shunt Active Power Filter with High Dynamic and Steady-State Performance, in Proc IEEE PESC08, 2008, pp. 3306-3310. [122] Hea-Gwang Jeong, Kyo-Beum Lee, Sewan Choi, Woojin Choi, Performance Improvement of LCLFilter-based Grid-Connected Inverters Using PQR Power Transformation, IEEE Trans. Power Electron., vol. 25, no.5, pp. 1320-1330, 2010.

212

[123] Fie Liu, Yan Zhou, Shanxu Duan, Jinjun Yin, Bangyin Liu, Fangrui Liu, Parameter Design of a TwoCurrent-Loop Controller Used for a Grid-Connected Inverter System with LCL Filter, IEEE Trans. Ind. Electron., vol. 56, no.11, pp. 4483-4491, 2009. [124] Guoqiao Shen, Dehong Xu, Luping Cao, Xuancai Zhu, An Improved Control Strategy for GridConnected Voltage Source Inverters with an LCL Filter, IEEE Trans. Power Electron., vol. 23, no.4, pp. 1899-1906, 2008. [125] Sung-Yeul Park, Chien-Liang Chen, J.S. Lai, Seung Ryul Moon, Admittance Compensation in Current Loop Control for a Grid-Tie LCL Fuel Cell Inverter, IEEE Trans. Power Electron., vol. 23, no.6, pp. 1716-1723, 2008. [126] L. A. Serpa, S. Ponnaluri, P. M. Barbosa, J. W. Kolar, A Modified Direct Power Control Strategy Allowing the Connection of Three-Phase Inverters to the Grid Through LCL Filters, IEEE Trans. Ind. Appl., vol. 43, no.5, pp. 1388-1400, 2007. [127] J. Dannehl, F. W. Fuchs, P. B. Thogersen, PI State Space Current Control of Grid-Connected PWM Converters with LCL Filters, IEEE Trans. Power Electron., vol. 25, no.9, pp. 2320-2330, 2010. [128] Eric Wu, P. W. Lehn, Digital Current Control of a Voltage Source Converter with Active Damping of LCL Resonance, IEEE Trans. Power Electron., vol. 21, no.5, pp. 1364-1373, 2006. [129] C. Ramos, A. Martins, A. Carvalho, Complex State-Space Current Controller for Grid-Connected Converters with an LCL Filter, in Proc IEEE IECON09, pp. 296-301, 2009. [130] P. T. Cheng, J. M. Chen, C. L. Ni, Design of a State-Feedback Controller for Series Voltage-Sag Compensator, IEEE Trans. Ind. Appl., vol. 45, no.1, pp. 260-267, 2009. [131] B. Bolsens, K. De Brabandere, J. Van Den Keybus, J. Driesen, R. Belmans, Model-based Generation of Low Distortion Currents in Grid-Coupled PWM-Inverters Using an LCL Output Filter, IEEE Trans. Power Electron., vol. 21, no.4, pp. 1032-1040, 2006. [132] I. J. Gabe, H. Pinheiro, Multirate State Estimator Applied to the Current Control of PWM-VSI Connected to the Grid, in Proc IEEE IECON08, pp. 2189-2194, 2008.

213

[133] H. Miranda, R. Teodorescu, P. Rodriguez, L. Helle, Model Predictive Current Control for High-Power Grid-Connected Converters with Output LCL Filter, in Proc IEEE IECON09, pp. 633-638, 2009. [134] P. Kulshrestha, Lei Wang, Mo-Yuen Chow, S. Lukic, Intelligent energy management system simulator for PHEVs at municipal parking deck in a smart grid environment, in Proc IEEE PES09, 2009. [135] J.O.G. Tande, K. Uhlen, Wind turbines in weak grids-constraints and solutions, in Proc. CIRED 2001, vol. 4, pp. 18-21, 2001. [136] H. Bindner, P. Lundsager, Integration of wind power in the power system, in Proc. IECON02, pp. 3309-3316, 2002. [137] M.P. Palsson, T. Toftevaag, K. Uhlen, J.O.G. Tande, Large-scale wind power integration and voltage stability limits in regional networks, in Proc. PES Summer Meeting, pp. 762-769, 2002. [138] J. H. R. Enslin, P. J. M. Heskes, Harmonic interaction between a large number of distributed power inverters and the distribution network, IEEE Trans. Power Electron., vol. 19, no.6, pp. 1586-1593, 2004. [139] C.T. Wang, Z. Ye, G. Sinha, X. Yuan, Output filter design for a grid-interconnected three-phase inverter, in Proc IEEE PESC03, 2003. [140] M. Liserre, F. Blaabjerg, R. Teodorescu, Grid impedance estimation via excitation of LCL-filter resonance, IEEE Trans. Ind. Appl., vol. 43, no.5, pp. 1401-1406, 2007. [141] L. Asiminoaei, R. Teodorescu, F. Blaabjerg, U. Borup, A digital controlled PV-inverter with grid impedance estimation for ENS detection, IEEE Trans. Power Electron., vol. 20, no.6, pp. 1480-1490, 2005. [142] Ali Emadi, Uninterruptible power supplies and active filter, CRC Press Boca Raton, Fla, 2005. [143] R. Zhang, M. Cardinal, A grid simulator with control of single-phase power converters in D-Q rotating frame, in IEEE Proc PESC02, vol. 3, pp. 1431-1436, 2002. [144] M.J. Ryan, R.D. Lorenz, R. De Doncker, Modeling of multileg sine-wave inverters: a geometric approach, IEEE Trans. Ind. Electron., vol. 46, no.6, pp. 1183-1191, 1999.

214

[145] P. Mattavelli, G. Escobar, A. M. Stankovic, Dissipativity-based adaptive and robust control of UPS, IEEE Trans. Ind. Electron., vol. 48, no.2, pp. 334-343, 2001. [146] G. E. Valderrama, A. M. Stankovic, P. Mattavelli, Dissipativity-based adaptive and robust control of UPS in unbalanced operation, IEEE Trans. Power Electron., vol. 18, no.4, pp. 1056-1062, 2003. [147] G. Escobar, P. Mattavelli, A. M. Stankovic, A. A Valdez, J. Leyva-Ramos, An adaptive control for UPS to compensate unbalance and harmonic distortion using a combined capacitor/load current sensing, IEEE Trans. Ind. Electron., vol. 54, no.2, pp. 839-847, 2007. [148] S. Buso, S. Fasolo, L. Malesani, P. Mattavelli, A dead-beat adaptive hysteresis current control, IEEE Trans. Ind. Appl., vol. 36, no.4, pp. 1174-1180, 2000 [149] P. Mattavelli, An improved deadbeat control for UPS using disturbance observers, IEEE Trans. Ind. Electron., vol. 52, no.1, pp. 206-212, 2005. [150] P. Mattavelli, F. Polo, F. Dal Lago, S. Saggini, Analysis of control-delay reduction for the improvement of UPS voltage-loop bandwidth, IEEE Trans. Ind. Electron., vol. 55, no.8, pp. 2903-2911, 2008. [151] P. Mattavelli, F.P. Marafao, Repetitive-based control for selective harmonic compensation in active power filters, IEEE Trans. Ind. Electron., vol. 51, no.5, pp. 1018-1024, 2004 [152] G. Escobar, A.A. Valdez, J. Leyva-Ramos, P. Mattavelli, A repetitive-based controller for UPS using a combined capacitor/load current sensing, in IEEE Proc. PESC05, pp. 955-961, 2005. [153] G. Escobar, P.R. Martinez, Leyva-Ramos, P. Mattavelli, A negative feedback repetitive control scheme for harmonic compensation, IEEE Trans. Ind. Electron., vol. 53, no.4, pp. 1383-1386, 2006 [154] S. Buso, F. Fasolo, P. Mattavelli, Uninterruptible power supply mulitloop control employing digital predictive voltage and current regulators, IEEE Trans. Ind. Appl., vol. 37, no.6, pp. 1846-1854, 2001. [155] P. Mattavelli, Synchronous-frame harmonic control for high-performance AC power supplies, IEEE Trans. Ind. Appl., vol. 37, no.3, pp. 864-872, 2001. [156] P. Mattavelli, A closed-loop selective harmonic compensation for active filters, IEEE Trans. Ind. Appl., vol. 37, no.1, pp. 81-89, 2001.

215

[157] M.J. Ryan, W.E. Brumsickle, and R.D. Lorenz, Control topology options for single-phase UPS inverters, IEEE Trans. Ind. Appl., vol. 33, pp. 493501, Mar./Apr. 1997. [158] M.J. Ryan, R.D. Lorenz, A high performance sine wave inverter controller with capacitior current feedback and back-EMF decoupling, in Proc IEEE PESC95, pp. 507-513, 1995. [159] Poh Chiang Loh, M. J. Newman, D. N. Zmood and D. G. Holmes, A comparative analysis of multiloop voltage regulation strategies for single and three-phase UPS systems, IEEE Trans. Power Electron., vol. 18, no.5, pp. 1176-1185, Sep 2003. [160] M. Aredes, E. Watanabe, H. Akagi, Instantaneous power theory and applications to power conditioning, Wiley-IEEE press 2007. [161] S. Bhattacharya, T.M. Frank, D.M. Divan, B. Banerjee, Active filter system implementation, IEEE Magazine on Ind. Appl., vol. 4, no.5, pp. 47-63, 1998 [162] M. Karimi-Ghartemani, M. R. Iravani, A nonlinear adaptive filter for online signal analysis in power systems: applications, IEEE Trans. Power Del., vol. 17, no.2, pp. 617-622, 2002. [163] H. Koizumi, T. Mizuno, T. Kaito, Y. Noda, N. Goshima, M. Kawasaki, K. Nagasaka, and K. Kurokawa, A novel microcontroller for grid-connected photovoltaic systems, IEEE Trans. Ind. Electron., vol. 53, no. 6, pp. 18891897, Dec. 2006. [164] R. A. Mastromauro, M. Liserre, A. DellAquila, Study of the effects of inductor nonlinear behavior on the performance of current controllers for single-phase PV grid converters, IEEE Trans. Ind. Electron., vol. 55, no. 5, pp. 2043-2052. [165] Yun Wei Li, Poh Chiang Loh, F. Blaabjerg, D.M. Vilathgamuwa, Investigation and improvement of transient response of DVR at medium voltage level, IEEE Trans. Ind. Appl., vol. 43, no.5, pp. 13091319, 2007. [166] Josep M. Guerrero, Juan C. Vasquez, Jose Matas, Luis Garcia de Vicuna and Miguel Castilla, Hierarchical control of droop-controlled ac and dc microgrids- A general approach toward standardization, IEEE Trans. Ind. Electron., vol. 58, no. 1, pp. 158172, Jan. 2011.

216

[167] Yasser Abdel-Rady Ibrahim Mohamed, Ehab F. El-Saadany Adaptive decentralized droop controller to preserve power sharing stability of paralleled inverters in distributed generation microgrids, IEEE Trans. Power Electron., vol. 23, no. 6, pp. 28062816, Nov. 2008. [168] J. M. Guerrero, L. G de Vicuna, J. Matas, M. Castilla and J. Miret, A wireless controller to enhance dynamic performance of parallel inverters in distributed generation systems, IEEE Trans. Power Electron., vol. 19, no. 5, pp. 12051213, Sept. 2004. [169] M. C. Chandorkar, D. M. Divan Control of parallel connected inverters in standalone AC supply system, IEEE Trans. Ind. Appl., vol. 29, no. 1, pp. 136-143, Jan./Feb. 1993. [170] Emanuel Serban, Helmine Serban A control strategy for a distributed power generation microgrid application with voltage and current-controlled source converter, IEEE Trans. Power Electron., vol. 25, no. 12, pp. 29812992, Nov. 2010. [171] M. C. Chandorkar, D. M. Divan, B. Banerjee Control of disturbed UPS sytsems, in Proc IEEE PESC94, pp. 197-204, 1995. [172] F. Katiraei, M. R. Iravani Power management strategies for amicrogrid with multiple distributed generation unites, IEEE Trans. Power Syst., vol. 21, no.4, pp. 1821-1831, 2006. [173] A. I. Harrison, Batteries and AC phenomena in UPS systems: the battery point of view, in IEEE 1989 Telecommunications Energy Conference, 1989. [174] Effects of AC ripple current on VRLA battery life. Emerson Network Power. [175] Nasser Kutkut, Output AC ripple effects, Technical Note, Power Designers. [176] S. Mariethoz, M. Morari, Explicit Model-Predictive Control of a PWM Inverter with an LCL Filter, IEEE Trans. Ind. Electron., vol. 56, no.2, pp. 389-399, 2009. [177] Jose M. Espi, Jaime Castello, Rafael Garcia-Gil, Gabriel Garcere, Emilio Figueres, An Adaptive Robust Predictive Current Control for Three-Phase Grid-Connected Inverters, IEEE Trans. Ind. Electron., vol. 58, no.8, pp. 3537-3546, 2011.

217

[178] Qingrong Zeng, Liuchen Chang, An Advanced SVPWM-based Predictive Current Controller for ThreePhase Inverters in Distributed Generation Systems, IEEE Trans. Ind. Electron., vol. 55, no.3, pp. 12351246, 2008. [179] K. H. Ahmed, A. M. Massoud, S. J. Finney, B. W. Williams, A Modified Stationary Reference Framebased Predictive Current Control with Zero Steady-State Error for LCL Coupled Inverter-based Distributed Generation Systems, IEEE Trans. Ind. Electron., vol. 58, no.4, pp. 1359-1370, 2011. [180] Shuitao Yang, Qin Lei, Fang Z. Peng, Zhaoming Qian, A Robust Control Scheme for Grid-Connected Voltage Source Inverters, IEEE Trans. Ind. Electron., vol. 58, no.1, pp. 202-212, 2011. [181] L. Harnefors, L. Zhang, M. Bongiorno, Frequency-Domain Passivity-based Current Controller Design, Power Electronics, IET, vol.1, no. 4, pp. 455-465, 2008. [182] J. Dannehl, F. W. Fuchs, Discrete Sliding Mode Current Control of Three-Phase Grid-Connected PWM Converter, in Proc IEEE EPE09, 2009, pp. 1-10. [183] J. Dannehl, F. W. Fuchs, Flatness-based Control of an Induction Machine Fed via Voltage Source Inverter-Concept, Control Design and Performance Analysis, in Proc IEEE IECON06, 2006, pp. 51255130. [184] F. Fuchs, J. Dannehl, F. W. Fuchs, Discrete Sliding Mode Current Control of Grid-Connected ThreePhase PWM Converters with LCL Filter, in Proc IEEE ISIE10, 2010, pp. 779-785. [185] Yasser Abdel-Rady Ibrahim Mohamed, Ehab F. El-Saadany, A Control Scheme for PWM VoltageSource Distributed-Generation Inverters for Fast Load-Voltage Regulation and Effective Mitigation of Unbalanced Voltage Disturbances, IEEE Trans. Ind. Electron., vol. 55, no.5, pp. 2072-2084, 2008. [186] J. Vasquez, J. Guerrero, A. Luna, R. Rodriguez, R. Teodorescu, Adaptive Droop Control Applied to Voltage-Source Inverters Operating in Grid-Connected and Islanded Modes, IEEE Trans. Ind. Electron., vol. 56, no.10, pp. 4088-4096, 2009. [187] P.A. Dahono, A Control Method for DC-DC Converter That Has an LCL Output Filter based on New Virtual Capacitor and Resistor Concepts, in Proc IEEE PESC04, pp. 36-42, 2004.

218

[188] C. Wessels, J. Dannehl, F.W. Fuchs, Active Damping of LCL-Filter Resonance based on Virtual Resistor for PWM Rectifier-Stability Analysis with Different Filter Parameters, in Proc IEEE PESC08, pp.3532-3358, 2008. [189] A. Rahimi, A. Emadi, Active Damping in DC/DC Power Electronic Converters: A Novel Method to Overcome the Problems of Constant Power Loads, IEEE Trans. Ind. Electron., vol. 56, no.5, pp. 14281439, 2009. [190] W. Gullvik, L. Norum, R. Nilsen, Active Damping of Resonance Oscillation in LCL-Filters based on Virtual Flux and Virtual Resistor, in Proc. IEEE EPE07, 2007. [191] M. P. Kazmierkowski, R. Krishnan, F. Blaabjerg, Control in Power Electronics. Selected Problems. New York: Academic, 2002. [192] F. Botteron, H. Pinheiro, A Three-Phase UPS that Complies with the Standard IEC 62040-3, IEEE Trans. Ind. Electron., vol. 54, no.4, pp. 2120-2136, 2007. [193] Uninterruptible Power Systems (UPS)-Part 3: Method of Specifying the Performance and Test Requirements, First Edition 1999-03, International Standard IEC 62040-3.