Salimin, S 2014

LOW ORDER HARMONICS MITIGATION IN GRID CONNECTED,
PARALLEL PV INVERTERS
Suriana Binti Salimin
A Thesis Submitted for the Degree of
Doctor of Philosophy
School of Electrical, Electronics, and Computer Engineering
Newcastle University
October 2014
Abstract
Abstract
This research is concerned with the problem of network power quality when grid
connected systems are used to feed the network distribution grid. A parallel connection
of photovoltaic (PV) system is the main interest of study for this research. This type of
PV system uses power electronic components such as inverter and current controller that
produce harmonics which adversely affect the power quality of the distribution network.
Development and simulation of current controller using the proportional resonance (PR)
scheme is considered to overcome the harmonic problems in single and parallel PV
inverters. This scheme eliminates specific harmonic in the low order part. The control
parameter randomisation technique is added to the scheme to produce a more efficient
current controller system. Thus, the inverter harmonic performance (inverter output
current) is improved. This concept is extended to parallel inverter based systems, where
opportunities for harmonic cancellation improve the inverter harmonic performance
further. Experimental hardware setup using TMS320F2812 is also developed to verify
the promising simulation results.
Keywords: Harmonics, current controller, parallel, PV inverter, grid connected system.
i
Acknowledgements
Acknowledgements
First and foremost, I would love to express my sincere gratitude to my supervisors, Dr.
Matthew Armstrong and Dr. Bashar Zahawi. My main supervisor, Dr. Matthew
Armstrong is the most important person that helped, guided, supported, advised, and
determined to keep giving me valuable ideas and information throughout this project.
My utmost appreciation also goes to the staffs especially from the mechanical
workshop, electrical workshop, and electronic workshop for their guidance and
support. Thank you very much to Stephen Mitchell, Stephen Robson, Jeff, Darren, and
the others! Many thanks to all my colleagues in the Power Electronics, Drives and
Machines Research Group for their friendship and help.
Special thanks to my husband, Khairul, as well as my precious forever babies, Danish,

Daania dan Dalila for always be with me upon the completion of this study. Not to
forget, my family and friends who prayed for my success.
Thank you very much.
Without those whom I mentioned above, this PhD project will not be accomplished.
ii
Table of Contents
Table of Contents
Abstract...................................................................................................................... i
Acknowledgements......................................................................................................ii
Table of Contents........................................................................................................ iii
List of Figures........................................................................................................... viii
List of Tables...............................................................................................................xv
List of Abbreviations................................................................................................. xvi
Chapter 1: Introduction ............................................................................................... 1
1.1 Chapter Introduction ......................................................................................... 1
1.2 Overview on Distributed Generation (DG) ....................................................... 1
1.2.1 Electricity from Solar ................................................................................ 3
1.3 Background of grid connected PV Inverter ...................................................... 4
1.3.1 Parallel Inverter System ............................................................................ 9
1.4 Aims and Objectives ....................................................................................... 10
1.5 Thesis Outline ................................................................................................. 11
1.6 Chapter Summary............................................................................................ 11
Chapter 2: Literature Review .................................................................................... 13
2.1 Chapter Introduction ....................................................................................... 13
2.2 Reviews on Harmonic Injection ...................................................................... 13
2.2.1 Types of Harmonic .................................................................................. 15
2.3 Reviews of Previous Works ............................................................................ 17
2.3.1 Single Connected PV System ................................................................. 17
2.3.2 Parallel Connected PV System ............................................................... 21
iii
Table of Contents
2.3.3 General Review on Harmonic Performance Improvements ................... 23
2.4 Research Idea .................................................................................................. 24
2.5 Chapter Summary............................................................................................ 25
Chapter 3: Methodology ........................................................................................... 26
3.2 Identification of the system to be studied ....................................................... 26
3.2.1 PWM Technique ..................................................................................... 26
3.2.2 Digital Control System ............................................................................ 32
3.3 Current Controllers.......................................................................................... 33
3.3.1 Conventional PI Control Technique........................................................ 33
3.3.2 Proportional Resonance Controller Scheme (PR) ................................... 35
3.3.3 Modified Proportional Resonance (PR) Technique ................................ 38
3.4 Chapter Summary............................................................................................ 39
Chapter 4: Modelling of Grid Connected, PV Inverter System ................................ 41
4.1 Introduction ..................................................................................................... 41
4.2 Modelling of Grid Connected, Single PV Inverter System............................. 41
4.2.1 Modelling of Low Pass Filter with System Impedance .......................... 42
4.2.2 Modelling of PWM ................................................................................. 47
4.2.3 Modelling of Current Controllers............................................................ 49
4.2.4 Model of Grid Connected, Single PV Inverter System ........................... 54
4.3 Closed Loop Frequency Response .................................................................. 54
4.4 Discussion ....................................................................................................... 61
4.5 Chapter Summary............................................................................................ 61
iv
Table of Contents
Chapter 5: Simulations of Grid-Connected Inverter System with PI, PR and

Modified PR Control ................................................................................................... 63
5.2 Simulation of a single grid connected inverter ............................................... 63
5.2.1 Conventional PI control technique .......................................................... 67
5.2.2 Proportional resonance (PR) control technique ...................................... 70
5.2.3 Modified PR control technique ............................................................... 74
5.3 Simulation of two grid connected inverters in parallel ................................... 77
5.3.1 Conventional PI Current Controller ........................................................ 79
5.3.2 Proportional resonance (PR) control technique ...................................... 83
5.3.3 Modified PR Control Technique ............................................................. 87
5.4 Chapter Summary............................................................................................ 92
Chapter 6: Experimental Results of Stand Alone Inverter ....................................... 93
6.2 General Hardware Description........................................................................ 93
6.2.1 Inverter Board ......................................................................................... 93
6.2.2 TMS320F2812 Digital Signal Processor ................................................ 94
6.3 PI Experiments and Results ............................................................................ 95
6.3.1 Dead time Effect and Its Compensation.................................................. 96
6.3.2 Zero Crossing Detection (ZCD) ............................................................ 100
6.3.3 Results on Inverter 1 system ................................................................. 101
6.4 PR Experiments and Results ......................................................................... 105
6.4.1 Results on Inverter 1 System ................................................................ 107
6.5 Comparison between the PI and PR Controller System................................ 111
v
Table of Contents
6.6 Parameter Randomization Controller Technique .......................................... 114
6.6.1 Results on Randomized PI on Inverter 1 System .................................. 115
6.6.2 Results on Randomized PI on Inverter 2 System .................................. 118
6.6.3 Results on Randomized PR in Inverter 1 System ................................. 120
6.6.4 Results on Randomized PR in Inverter 2 System ................................. 122
6.7 Chapter Summary.......................................................................................... 124
Chapter 7: Experimental Results of Grid Connected Inverters .............................. 125
7.1 Introduction ................................................................................................... 125
7.2 Experimental results with conventional PI current control techniques ......... 125
7.2.1 Conventional PI Current Control of Individual Inverter Units ............. 126
7.2.2 Conventional PI Current Control of Two Inverters in Parallel ............. 128
7.2.3 Summary: Conventional PI Current Control ........................................ 130
7.2.4 Experimental Results with Randomised PI Current Control Techniques

...............................................................................................................130
7.2.5 Parameter Randomisation PI Current Control of Individual Inverter Units

...............................................................................................................131
7.2.6 Parameter Randomisation PI Current Control of Two Inverters in Parallel

...............................................................................................................133
7.2.7 Summary, Parameter Randomisation PI Current Controller................. 134
7.3 Experimental Results with Randomised PR Current Control Techniques.... 135
7.3.1 Fixed PR Current Control of Individual Inverter Units ........................ 135
7.3.2 Fixed PR Current Control of Two Inverters in Parallel ........................ 137
7.3.3 Summary: Fixed PR Current Controller ............................................... 138
7.3.4 Experimental Results with Randomised PR Current Control Techniques

...............................................................................................................139
7.3.5 Parameter Randomisation PR Current Control of Individual Inverter

Units ...............................................................................................................139
vi
Table of Contents
7.3.6 Parameter Randomisation PR Current Control of Two Inverters in

Parallel. .............................................................................................................. 140
7.3.7 Summary: Parameter Randomisation PR Current Controller ............... 142
7.4 Chapter Summary.......................................................................................... 143
Chapter 8: Conclusions ........................................................................................... 144
8.1 Conclusions ................................................................................................... 144
Appendix A Multisim and Ultiboard Designing Tool ............................................... 147
Appendix B Screen Shot of the PCBs ....................................................................... 149
Appendix C Schematic Diagram of Single Inverter System ..................................... 151
Appendix D PWM Generation .................................................................................. 152
Appendix E Dead-time .............................................................................................. 155
Appendix F Code Composer Studio (CCS) .............................................................. 156
Appendix G Testing an Open Loop System .............................................................. 158
Appendix H Current Sensor ....................................................................................... 160
References .................................................................................................................164
vii
List of Figures
List of Figures
Figure 1-1: Average annual growth rates of renewable energy capacity, end 2004 to
2009[9]. ......................................................................................................................... 3
Figure 1-2: Basic concept of grid connected inverter. .................................................. 4
Figure 1-3: Centralised technology PV arrangement. ................................................... 6
Figure 1-4: Single-string technology PV arrangement. ................................................ 6
Figure 1-5: Multi-string technology PV arrangement. .................................................. 6
Figure 1-6: AC cell PV arrangement [24]. .................................................................... 7
Figure 1-7: Parallel arrangement of grid connected system.......................................... 9
Figure 2-1: The current loop of PV inverter with PR and HC. ................................... 18
Figure 2-2 Harmonic trend line of parallel inverter system using conventional control
and parameter randomization control technique [33]. ................................................ 22
Figure 3-1: Bipolar PWM switching. .......................................................................... 27
Figure 3-2: Waveforms of (a) 𝑉𝑐𝑜𝑛𝑡𝑟𝑜𝑙 > 𝑉𝑡𝑟𝑖 which turns on Gate 1; ................. 28
Figure 3-3: Unipolar PWM unit. ................................................................................. 29
Figure 3-4: Unipolar PWM switching. ....................................................................... 30
Figure 3-5: Waveforms at (a) Out1 and Out2 (b) Out3 and Out4. .............................. 31
Figure 3-6: Inverter output voltage. ............................................................................ 32
Figure 3-7: Digital control system. ............................................................................. 33
Figure 3-8: A PI current controller block diagram. .................................................... 34
Figure 3-9: A proportional resonance (PR) current controller block diagram. ........... 37
Figure 3-10: A modified PR current controller block diagram. .................................. 39
Figure 4-1: Basic arrangement of grid connected inverter system. ............................ 41
Figure 4-2: Equivalent circuit for low pass filter and the grid. ................................... 42
Figure 4-3: Block diagram of the low pass filter and grid impedance branch. ........... 44
Figure 4-4: Single update PWM with triangular carrier waveform. ........................... 48
Figure 4-5: Open loop Bode plot for PI current controller scheme shows flat response
at 50 Hz indicating no harmonic rejection at fundamental frequency. ....................... 50
Figure 4-6: Open loop Bode plot of PR current controller scheme shows high
magnitude at 50 Hz indicating good harmonic rejection at fundamental frequency. . 51
viii
List of Figures
Figure 4-7: Open loop Bode plot of PR current controller scheme when (a)
𝜔𝑐=1: wider magnitude response at 50 Hz shows less sensitive controller, and (b)
𝜔𝑐=1, 5, 20, 50: sensitivity of the controller is reduced with increasing value of 𝜔𝑐.
..................................................................................................................................... 52
Figure 4-8: Open loop Bode plot of PR+HC current controller scheme with 𝝎𝒄=10.53
Figure 4-9: Model of grid connected inverter system with PI current controller. ...... 54
Figure 4-10: Closed loop response of the system with PI current controller. (a) Bode
plot shows the attenuation of the low order harmonics, and controller resonant point
at approximately 7 kHz, and (b) Root locus confirms the stability of the closed loop
controller for the applied gains. .................................................................................. 58
Figure 4-11: Closed loop response of the system with PR current controller. (a) Bode
plot shows improved attenuation of low order harmonics, and lower impact of
resonance (now at 10 kHz), and (b) Root locus confirms the stability of the closed
loop controller for the applied gains. .......................................................................... 59
Figure 4-12: Close loop response of the system with PR+HC current controller.
(a) Bode plot shows attenuation of low order harmonics, and further improvement of
the resonant effect at 10 kHz, and (b) Root locus confirms the stability of the closed
loop controller for the applied gains. .......................................................................... 60
Figure 5-1: Simulink model of grid-connected inverter system. ................................ 64
Figure 5-2: Grid model which contains a background THD of 2.45% [44]. .............. 65
Figure 5-3: Conventional PI current controller digital Simulink model. .................... 68
Figure 5-4: Screen shot of (a) Reference current of PI controller with 20 A peak
value, (b) Inverter output current of PI controller which shows a distorted waveform,
and (c) Large magnitude error between reference current and inverter output current
of PI controller. ........................................................................................................... 69
Figure 5-5: Low order harmonic spectrum of inverter output current with PI control
(𝐾𝑃 = 0.03 , 𝐾𝐼 = 0.03) ............................................................................................ 70
Figure 5-6: Simulink model of current controller using PR technique. ...................... 71
Figure 5-7: Screen shot of (a) Reference current of PR controller with 20 A peak
value, (b) Inverter output current of PR controller which shows smoother waveform
compared to PI controller, and (c) Reduced error between reference current and
inverter output current of PR controller. ..................................................................... 72
ix
List of Figures
Figure 5-8: Low order harmonic spectrum of inverter output current with PR current
controller. .................................................................................................................... 73
Figure 5-9: Comparison in the harmonic profile of PI and PR current controllers
shows better reduction of harmonics magnitude with PR current controller. ............. 74
Figure 5-10: Simulink model of current controller using the modified PR control
technique. .................................................................................................................... 75
Figure 5-11: Screen shot of (a) Random signal generation of the modified PR
controller, and (b) The randomized proportional gain, 𝑅𝐾𝑃 signal which varies
within a set limit. ......................................................................................................... 76
Figure 5-12: Harmonic profile of inverter output current with modified PR current
controller shows better outcome compared to the conventional PR current controller.
..................................................................................................................................... 77
Figure 5-13: Parallel inverter system model. .............................................................. 78
Figure 5-14: Current error signal of (a) Inverter 1, and (b) Inverter 2, with PI current
controller which shows big magnitude errors. ............................................................ 79
Figure 5-15: Waveforms of (a) Inverter 1 output current, (b) Inverter 2 output current,
and (c) Grid current with PI controller. ....................................................................... 80
Figure 5-16: Harmonic profiles of (a) Inverter 1 output current, (b) Inverter 2 output
current, and (c) Grid current. All shows significant harmonic emissions between the
3rd and 19th harmonic orders........................................................................................ 82
Figure 5-17: Harmonic profiles of inverter 1, inverter 2, and grid current with PI
current controller which shows similar harmonic spectrum exists except for the 15th,
17th and 19th order. ...................................................................................................... 82
Figure 5-18: The error signal between reference and the actual current of ................ 83
and (c) Grid current with PR current controller. ......................................................... 84
Figure 5-20: Harmonic profiles with the respective per unit magnitude of ................ 86
Figure 5-21: Harmonic profiles of Inverter 1, Inverter 2, and Grid current with PR
controller in one graph shows harmonics addition and cancellation occurred. .......... 86
Figure 5-22: The error signal between reference and the actual current of ................ 87
and (c) Grid current with modified PR current controller........................................... 88
Figure 5-24: Harmonic profiles with the respective per unit magnitude of (a) Inverter
1 output current, (b) Inverter 2 output current, and (c) Grid current. ........................ 90
x
List of Figures
Figure 5-25: Harmonic profiles of inverter 1, inverter 2, and grid current with
modified PR controller shows harmonic cancellation and addition leading to a further
THD reduction compared to the PR current controller technique. ............................. 90
Figure 5-26: Current harmonic profile using PI, PR and modified PR controllers.
THD of grid current with modified PR controller shows the lowest value. ............... 91
Figure 6-1: Schematic diagram of power inverter board. ........................................... 94
Figure 6-2: Block diagram of the eZdspTM F2812 [81]. ............................................. 95
Figure 6-3: Screen shot of current output waveform. ................................................. 96
Figure 6-4: FFT of output current. .............................................................................. 96
Figure 6-5: Screen shot of the output current waveform with (a) dead time 0.00,
(b) dead time 0.01, (c) dead time 0.02, and (d) dead time 0.03. ................................. 99
Figure 6-6: Output current FFT for different dead time compensation unit. .............. 99
Figure 6-7: Output waveforms of ZCD and inverter system in phase. ..................... 101
Figure 6-8: Harmonic data of inverter 1 output current. ........................................... 101
Figure 6-9: Screen shot of (a) Inverter 1 output current, and (b) Inverter 1 current
error. .......................................................................................................................... 102
Figure 6-10: Harmonic data of inverter 2 output current using PI control. .............. 104
Figure 6-11: Harmonic spectrums of Inv 1 and Inv 2 with PI current control.......... 104
Figure 6-12: Harmonic data of inverter 1 output current using PR control. ............. 107
Figure 6-13: Screen shot of (a) output current waveform with reference 3 A peak ;
and (b) current error waveform which shows approximately 0.2 A peak. ................ 108
Figure 6-14: Harmonic data of inverter 2 output current using PR control .............. 109
Figure 6-15: Harmonic spectrums of Inv 1 and Inv 2 with PR current control. ....... 110
Figure 6-16: Compare output current waveform of two controller systems (a) Output
current from PI controller shows distortion at the top, and (b) output current from PR
controller shows smooth curve at the top. ................................................................. 111
Figure 6-17: Compare error of two controller systems (a) current error with PI
controller system is approximately 0.5 A peak; and (b) current error with PR
controller system is reduced to approximately 0.2 A peak. ...................................... 112
Figure 6-18: Compare FFT of two controller systems. ............................................. 113
Figure 6-19: Random signal generation (a) generation of random signal more
scattered around before filtering process ; and (b) random signal becomes smoother
like a waveform after filtering process...................................................................... 114
xi
List of Figures
Figure 6-20: Output current harmonic spectrum of inverter 1 system using a

randomized proportional gain of PI control technique. ............................................ 116
randomized integral gain of PI control technique. .................................................... 116
Figure 6-22: Harmonic spectrum of inverter 1 system using a conventional,
randomized proportional, and randomized integral gain of PI control technique..... 117
randomized proportional, and randomized integral gain of PI control technique..... 119
randomized proportional gain of PR control technique. ........................................... 120
randomized resonant gain of PR control technique. ................................................. 121
randomized proportional and randomized resonant gain of PR control technique. .. 121
randomized proportional gain of PR control technique. ........................................... 122
randomized resonant gain of PR control technique. ................................................. 123
Figure 6-31: spectrum of inverter 2 system using a conventional, randomized
proportional, and randomized resonant gain of PR control technique. ..................... 123
Figure 7-1: Experimental H-Bridge inverter. Single inverter grid connected system
................................................................................................................................... 126
Figure 7-2: Low order harmonic spectra of (a) Inverter 1 output current with
conventional PI control: predominant harmonics appear between the 3rd and 15th
harmonic and (b) Inverter 2 output current with conventional PI control: predominant
harmonics appear between the 3rd and 15th harmonic. ............................................ 127
Figure 7-3: Experimental, parallel connected inverter system comprising of two
inverter units.............................................................................................................. 128
Figure 7-4: Low order harmonic spectra of parallel inverters with PI current control.
(a) Results of inverter 1 output current; (b) Results of Inverter 2 output current; and
xii
List of Figures
(c) Results of the parallel system output current. All shows strong correlation between
them with predominant harmonics appear between the 3rd and 15th harmonic....... 129
randomised PI control; and (b) Inverter 2 output current with randomised PI control;
where difference trend lines are observed between the two inverters. ..................... 132
Figure 7-6: Harmonic spectra of parallel inverters with randomised PI current control.
(c) Results of the parallel system output current. Different trend lines are observed.
................................................................................................................................... 133
conventional PR control; and (b) Inverter 2 output current with conventional PI
control. Results show some degree of correlation in harmonic profile for each of the
two inverters. ............................................................................................................. 136
Figure 7-8: Low order harmonic spectra of parallel inverters with PR current control.
(c) Results of the parallel system output current. Small cancellation and addition of
harmonics are observed at the system output current. .............................................. 137
randomised PR control; and (b) Inverter 2 output current with randomised PR control.
Different trend line patterns indicate the reduction in the degree of correlation in the
harmonic profile. ....................................................................................................... 139
Figure 7-10: Low order harmonic spectra of parallel inverters with randomised PR
current control. (a) Results of inverter 1 output current; (b) Results of Inverter 2
output current; and (c) Results of the parallel system output current. Harmonics
magnitude reduction are observed at the parallel system output current between the
7th and 19th order. ...................................................................................................... 141
Figure A 1: Screen shot of Multisim designing tool ................................................. 147
Figure A 2: Screen shot of Ultiboard designing tool ................................................ 148
Figure B 1: Screen shot of the PCBs ......................................................................... 150
Figure C 1: Circuit diagram of the power inverter board ......................................... 151
Figure D 1: Diagram to determine GPTIMER_Freq ................................................ 152
Figure D 2: Output pin PWM1 with different duty cycles........................................ 154
Figure E 1: Signals of PWM1 and PWM2 with approximately 1.7µs dead-time..... 155
Figure F 1: CCS Software ......................................................................................... 156
xiii
List of Figures
Figure F 2: Output of PWM1 pin after the triangular waveform is compared to the
modulating signal ( ma = 0.95 ) ................................................................................ 157
Figure G 1: Waveform of output filter without load ................................................. 158
Figure G-2: Current probe ......................................................................................... 159
Figure G 3: Waveform of output filter after a resistive load is connected................159
xiv
List of Tables
List of Tables
Table 1-1 Population and Energy Consumption Over Time. ........................................ 2

Table 1-2: Advantages and disadvantages of PV array configuration [22, 26, 27]. ..... 8
Table 2-1: IEEE standard on harmonic limitation [45]. .............................................. 15
Table 3-1 System parameters. ..................................................................................... 32
Table 4-1: List of parameters of grid connected model. ............................................. 57
Table 5-1: System parameters. .................................................................................... 65
Table 5-2: Grid model harmonic data [44]. ................................................................ 66
Table 7-1: THD figures for independently operated, grid connected, inverters ....... 128
Table 7-2: THD Performance of Two Inverter Parallel System with conventional PI
control ....................................................................................................................... 130
Table 7-3: THD Performance of Two Inverter Parallel System with conventional
control ....................................................................................................................... 132
Table 7-4: THD Performance of Two Inverter Parallel System with random
parameter control ...................................................................................................... 134
Table 7-6: THD Performance of Two Inverter Parallel System with conventional PR
control. ...................................................................................................................... 138
Table 7-8: THD Performance of Two Inverter Parallel System with randomised PR
control. ...................................................................................................................... 142
xv
List of Abbreviations and Symbols
Abbreviations
DG Distributed Generation
PV Photovoltaic
DC Direct Current
AC Alternating Current
PWM Pulse Width Modulation
L Inductor
LC Inductor and Capacitor
LCL Inductor, Capacitor and Inductor
MPPT Maximum Power Point Tracking
RMS Root Mean Square
IEEE Institute of Electrical and Electronic Engineers (USA)
PI Proportional and Integral
PR Proportional and Resonant
HC Harmonic Compensator
DSP Digital Signal Processing
PR+HC Proportional Resonant with Harmonic Compensator
THD Total Harmonic Distortion
LV Low Voltage
UPS Uninterruptible Power Supply
WAC Weighted Average Current
PLL Phase Locked Loop
CPWM Centered Pulse Width Modulation
SHMPWM Selective Harmonic Mitigation Pulse Width Modulation
SHEPWM Selective Harmonic Elimination Pulse Width Modulation
xvi
OSPWM Optimal Sinusoidal Pulse Width Modulation

SMC Sliding Mode Control
FFT Fast Fourier Transform
THIPWM Third Harmonic Injection Pulse Width Modulation
ACSC Average Current Sharing Center
NSPWM Non-sinusoidal Pulse Width Modulation
NSHVC Non-sinusoidal Hysterisis Voltage Control
APF Active Power Filter
DIRS Direct Injection Random Signal
IGBT Insulated Gate Bipolar Transistor
ADC Analogue to Digital Converter
DAC Digital to Analogue Converter
ZOH Zero Order Hold
PID Proportional Integral Derivative
P Proportional
I Integral
ESR Equivalent Series Resistance
xvii
Symbols
𝑓 Frequency
kWp KiloWatt power
𝐺𝑃𝑅 Proportional-Resonant Current Controller
𝐺𝐻𝐶 Harmonic Compensator Current Controller
𝐾𝑃 Proportional gain
𝐾𝐼 Integral gain
𝑉𝑃𝑉 Output voltage of Photovoltaic
𝑅𝑃 Random gain component
𝑉𝑑𝑐 DC output voltage
𝑉𝑐𝑜𝑛𝑡𝑟𝑜𝑙 /𝑉𝑠𝑖𝑛 Voltage of control signal
𝑉𝑡𝑟𝑖 Voltage of triangular waveform
𝐺𝑃𝐼 Proportional-Integral Current Controller
𝜔𝑜 Fundamental frequency (rad/s)
𝑇 Sampling time (s)

Hz Hertz (frequency)
𝐺𝑃𝑅𝑚𝑜𝑑𝑖𝑓𝑖𝑒𝑑 Modified Proportional-Resonant Current Controller
𝑅𝐾𝑃 Randomized proportional gain
𝐶𝑓 Filter capacitance value
𝐿𝑓 Filter inductance value
𝐿𝑔 Grid inductance value
𝑅𝐿𝑓 Resistance value of the inductor of the filter
𝑅𝐶𝑓 Resistance value of the capacitor of the filter
𝑅𝑔 Resistance value of the grid
𝑉𝑃𝐶𝐶 Voltage at the point of common coupling
𝑉𝑥 Voltage at inverter output
xviii
𝐻𝐿𝑓 Transfer function of the filter
𝐼𝑥 Output inverter current
𝐼𝑔 Grid current
𝐼𝐶𝑓 Filter capacitance current
𝐼𝑜 Output current from the filter
𝐻𝐿𝑔 Transfer function of the grid impedance
𝑐𝑝𝑘 Peak of the carrier signal
𝑚𝑠 (𝑡) Modulating signal
𝑐(𝑡) Carrier signal
𝑑 Duty cycle
𝜔𝑐 Cut-off frequency
𝐾𝑐3 Resonant gain at three times of fundamental frequency
𝐾𝑐5 Resonant gain at five times of fundamental frequency
𝐾𝑐7 Resonant gain at seven times of fundamental frequency
𝑉𝐺 Background grid voltage
𝐼𝑟𝑒𝑓 Reference current
xix
Introduction
Chapter 1: Introduction
1.1 Chapter Introduction
The first chapter of this thesis provides an overview on distributed generation (DG).
This is the technical area in which this work aims to make a valid contribution to
knowledge. The main advantages of DG technology are described and the recent annual
growth statistics for application of renewable energy systems is presented.
This thesis will specifically concentrate on small scale grid connected photovoltaic
(PV) systems. Therefore, the second half of the chapter provides a brief discussion on
Photovoltaic (PV) systems, highlighting the technological challenges which exist
within these systems and the opportunities for further research. The final section of this
chapter summarises the aims and objectives of this research project.
1.2 Overview on Distributed Generation (DG)
Throughout the whole of the modern era, each century has seen an increase in energy
consumption, which is intrinsically linked to the continual increase in worldwide
human population. This can be seen in Table 1-1 as proposed by R. DeGunther [1].
With the well recognised depletion of traditional fossil fuels such as coal, oil and gas,
distributed generation technologies or embedded generation is rapidly becoming a
significant and important matter worldwide. Typically, these generation systems are
connected at strategic places within the power distribution network and much closer to
the end users than conventional power stations. Based on P. Chiradeja and Y. Ma, et al
[2, 3], the key advantages of this type of electricity generation are as follows:
i- It can reduce transmission losses as the plant is installed near its application.
ii- It has satisfying power supply demand where the energy can always be
utilized whenever needed by users. For instance in the event of line outage
or scheduled interruption.
1
Introduction
Table 1-1 Population and Energy Consumption Over Time.
Consumption
Date Population (billions)
(kWh/day)
5000 bc 0.1 9.4
0 ad 0.3 10.1
1850 1.3 12.0
1980 4.4 51.0
2000 6.0 230
2050 9.0 1000?
Distributed generation uses renewable energy sources that include solar, wind, wave,
hydro power and biomass. They convert these natural sources to usable electricity.
Often, they are of small scale, scattered and close to the loads and their control devices
[4, 5]. Renewable sources play an important role in order to reduce the environmental
pollution and thus decrease the worries of global warming and climate changes [6]. As
known by many, traditional energy sources such as petroleum, natural gas and coal are
major contributors’ to the pollution of the environment. It has been stated in literature
that:
“All fossil fuels produce carbon dioxide, a greenhouse gas when burned. When
there is insufficient air, all fossil fuels produce highly toxic carbon monoxide” [7].
2
Introduction
1.2.1 Electricity from Solar

Generally, there is an upward trend in the utilisation of renewable energy sources for
electricity generation. However, as shown in Figure 1-1, grid connected photovoltaic
(PV) systems have experienced a particularly rapid growth (60%) in demand between
the end of year 2004 and 2009 with a 53% increase in capacity for 2009 alone. Being
so popular, photovoltaic generation systems are now readily integrated into distribution
systems worldwide [8].
Figure 1-1: Average annual growth rates of renewable energy capacity, end 2004 to
2009[9].
It is estimated that the future lifespan of the Sun is approximately 4 to 5 billion years.
For this reason, it is generally assumed that the Sun represents an energy that is to all
intents and purposes, inexhaustible [10]. Furthermore, it is one of the cleanest energy
sources. With regards to grid connected systems, PV technology is generally
acknowledged to be one of the more reliable and low maintenance solutions, compared
to alternative systems such as wind turbine generators. When operating, there are
typically no moving parts which need to be replaced when broken. Instead, the PV
panels generally only require scheduled inspection and cleaning. Furthermore, PV
systems are very flexible and can be installed anywhere significant levels of sun can
reach such as rooftops of industrial, commercial and residential buildings. The
electricity can also be sold to the utility grid if extra power is generated.
3
Introduction
Below are the lists of the advantages according to Foster R. et al [10]:
a) High reliability.
b) Low operating costs.
c) Local fuel (not imported fossil fuel).
d) Long, useful life.
e) Clean energy.
f) No on-site operator required.
g) Low operation and maintenance costs.
More explanation on PV systems is provided in the next section.
1.3 Background of grid connected PV Inverter
Renewable energy generating systems can be readily used to supply standalone load
systems. However, there is an increasing trend to also connect PV systems to the utility
grid. In such systems, the grid receives the electrical energy generated by the PV
inverter system [11]. The basic concept of a PV grid connected inverter system is
illustrated in Figure 1-2.
Figure 1-2: Basic concept of grid connected inverter.
4
Introduction
Based on Figure 1-2, the basic concept of grid connected inverter consists of:
a) PV array – A PV array is assembled from several PV modules made of multiple

solar cells to capture the sunlight and converts it to a DC voltage [12, 13].
b) DC/DC boost converter – The input of the converter varies due to the DC
voltage variation from the PV array whilst the output from the converter is
controlled to follow a constant desired voltage value [14]. The purpose is to
increase the low input voltage to a standard DC bus voltage value; i.e.
400 V(DC) [15]. The performance depends on its design as well as its operating
conditions. Regardless of the weather and temperature, maximum output power
can be achieved through the implementation of a good maximum power point
tracking (MPPT) algorithm [16].
c) DC/AC inverter – An inverter converts the DC voltage to an AC voltage,
typically using Pulse Width Modulation (PWM) switching techniques. In
addition, the inverter must be controlled to ensure that the voltage is
synchronised to the grid supply voltage. This ensures unity power factor
operation and optimum real power transfer to the grid.
d) Filter – A low pass filter is used to block and prevent high frequency
components from entering the grid. These high frequency components come
from the PWM switching frequency of the DC/AC inverter, which is typically
in the kilohertz range. The types of filter used can be of inductive (L) type,
inductive and capacitive (LC) type and inductive, capacitive, and inductive
(LCL) type. Hanju Cha and Trung-Kien Vu [17] detail the advantages and
disadvantages of each type of filter.
e) Controller – This inverter current controller minimises the error between the
inverter output current and controller reference current. The intention is to keep
the signal sinusoidal, with very low distortion, and to maintain a power factor
close to unity [18-20].
Typically, PV arrays can be configured into four different types of topology;

centralized technology, single-string technology, a multi-string technology, and an AC-
cell or AC module technology [21]. In the paper by Xiaojin Wu et al [22], the first three
arrangements are described. A centralized technology is a conventional arrangement
where several PV modules are connected in series to make a string of PV modules.
5
Introduction
Then, strings of PV modules are connected in parallel to form the PV array and feed
the inverter. This is illustrated in the Figure 1-3.
Figure 1-3: Centralised technology PV arrangement.
Figure 1-4: Single-string technology PV arrangement.
Figure 1-5: Multi-string technology PV arrangement.
6
Introduction
The second configuration is the single-string technology shown in Figure 1-4. It

consists of several PV modules connected together in series to form the PV array. The
third configuration is the multi-string technology shown in Figure 1-5, which consists
of a PV module with its own DC/DC converter, connected in parallel, and fed to a
single DC/AC inverter. All three configurations have a maximum power point tracking
(MPPT) controller which is used to draw the maximum power from the array [23].
Figure 1-6: AC cell PV arrangement [24].
The last configuration is also called micro-inverter technology [25] (Figure 1-6). It is a
system which combines a PV module and an AC module inverter to feed the grid. All
of these arrangements and configurations have their own advantages and disadvantages
comprising of complexity, efficiency, reliability, and more. The key points are
summarised in Table 1-2.
7
Introduction
Table 1-2: Advantages and disadvantages of PV array configuration [22, 26, 27].
Advantages Disadvantages
Centralized 1. PV array arrangement 1. The MPPT controller

Technology generates higher voltage and considers the PV modules
current suitable for large scale arrangement as one whole
systems (>1 MW). thus makes the output power
smaller.
2. Simple and low cost as only
one MPPT controller is used.
Single-string 1. PV array arrangement 1. One shaded PV module

Technology generates higher voltage and low can cause large power
current which is easy to dissipation on that module as
assemble based on power the photocurrent in that
demands. particular module may drop.
Multi-string 1. Each PV module has its own 1. Complicated and costly

Technology MPPT controller thus maximum
power is drawn. 2. Shading problem is
reduced as only one module
in series.
Micro inverter 1. The output from this 1. Has an additional up-front

technology technology can be directly cost.
connected to AC system.
Regardless of the PV array arrangement, these inverter systems do have their

drawbacks that generally arise from the switching process and non-linearity in the
components of the power electronic converter system. Briefly, the purpose of these
electronic components is to integrate the PV system into the supply grid and ensure
that it complies with a host of power quality recommendations and standards [28]. The
inverter for instance has to fulfil several important functions before the current may be
fed into the supply grid [29]. Firstly, it must shape the current into a sinusoidal
waveform where a controller is used and secondly, it must invert the DC current into
an AC current with a correct switching. In doing so, it is very difficult to avoid levels
of distortion in the conversion and control process. The result is undesirable current
and voltage harmonics appearing in the inverter output which can potentially contribute
to a lower power quality grid network with increased risk of disruption.
8
Introduction
Power quality is defined as the voltage and current provided by the grid that can be
successfully utilized by the user without interference or interruption [30]. For good
power quality operation, it is important that the current harmonics produced at the
output of the inverter do not exceed the national Point of Common Coupling standards
[31]. The impact of these harmonics include variation in root mean square (RMS)
voltage, disturbance of electronic components and stress on insulation materials [32].
Other impacts are discussed in Chapter 2. Therefore, it is vital to reduce the harmonic
level to be under the limit as recommended in IEEE 929-2000 [31].
1.3.1 Parallel Inverter System
Figure 1-7: Parallel arrangement of grid connected system.
Parallel inverter systems are commonly used to increase the power rating of the PV
system and maximise the return of power to the network. In this system, several
individually controlled inverter units are connected to the grid via a common point of
coupling. Each inverter output voltage should be synchronised to the grid voltage and
inject unity power factor current into the network. The synchronization method of
individual inverter units, together with the interaction between closely coupled systems
can potentially increase the production of low order harmonics [33]. This is also agreed
by Cespedes M. And J. Sun [34] where it is stated that unintended interactions may
occur because each of the inverters has its own dynamic system. He also grouped the
unintended interaction into the low frequency region; not more than the fundamental
frequency, medium frequency region; between the fundamental frequency and the
9
Introduction
crossover frequency, and the high frequency region; more than the crossover
frequency. Each of them comes from different sources with its own ways of mitigation.
The main point is that the unintended interaction between the inverters may affect the
power quality of the network. Figure 1-7 shows an example arrangement of several PV
inverter systems connected in parallel with the supply grid. It can be seen that each PV
system has its own inverter and controller system.
1.4 Aims and Objectives
This chapter has highlighted the types of PV inverter system which are commonly used
in practice. The need for good power quality operation has been described and the
causes and impacts of harmonic generation are stated. Even today, with modern devices
and control methods, the issue of power quality remains a problem which all
manufacturers must address. For this reason, this research aims to propose a method to
eliminate the low order harmonics production in a parallel PV inverter system
connected to the supply grid. As a result, the work primarily concentrates on the output
side (AC) of the system which consists of the DC/AC inverter, the current controller,
the low pass (LC) filter and the supply grid.
With this in mind, the thesis has the following objectives:
a) To demonstrate the low-order harmonic performance of a single-phase grid

connected inverter using the classical proportional integral (PI) control
methods.
b) To show that the low order harmonic performance of a grid-connected inverter

can be improved by replacing the PI controller with a proportional resonance
(PR) current controller scheme.
c) To explore that the addition of harmonic compensators (HC) and controller

parameter randomisation techniques can further improve the harmonic
performance of a parallel connected inverter systems.
Studies on proportional resonant controllers with harmonic compensators

(PR+HC) have been carried out previously. A full literature review on this topic
10
Introduction
is presented in Chapter 2. However, to the best of the authors’ knowledge, no

studies have attempted to model the PR+HC controller and make better analysis
specifically to grid connected system.
d) To validate the above points in simulation and experimentally.
1.5 Thesis Outline
This thesis consists of 8 chapters. This Chapter has discussed the background of PV
inverter system and has set out the aims and objectives of the research project. Chapter
2 presents a literature review of relevant previous work in the field of PV inverter
systems. Chapter 3 focuses on the different control methods to be applied in this
project; the conventional PI controller scheme, the PR+HC controller scheme and the
modified PR+HC controller scheme. Chapter 4 is written to model the grid connected
inverters. Chapter 5 demonstrates the simulation results when using the three controller
schemes. Chapters 6 and Chapter 7 demonstrate the experimental results when using
the three controller schemes with a stand-alone system and a grin connected system.
Finally, Chapter 8 presents the conclusion of this research project.
1.6 Chapter Summary
This chapter has described the background to distributed generation and the reasons for
its continuing popularity compared to traditional forms of fossil fuel based generation.
The concept of single and parallel PV inverter systems has been discussed.
Furthermore, relevant to this thesis, the basic causes of harmonic generation have been
set out, and the need to meet the appropriate grid connection standards has been set out.
At the end of this chapter, the aims and objectives of the research project are identified.
In this research project, the main contributions to knowledge include:
a) Modelling the inverter system with a PR+HC current controller. This controller
includes additional harmonic compensation terms, e.g. for the 3rd, 5th, and 7th
harmonic orders, which can reduce low order harmonics injection into the grid. In
Chapter 4, frequency domain results demonstrate the increased impedance to these
components, and the resulting power quality benefits.
11
Introduction
b) For a stand-alone inverter, improved harmonic performance can be achieved by

randomizing, within limits, the integral gain of a PI controller. Hence, the technique
can also be considered for grid connected applications. Previous research has typically
concentrated on modifying the proportional controller only.
c) In this thesis, a PR controller employing parameter randomisation techniques is

employed. The author believes that this is the first time that such a technique has been
applied. In this thesis, simulation and practical results demonstrate that a modified PR
controller technique improves the harmonic performance of individual and parallel
connected inverter systems. In Chapter 7, the harmonic profiles of the parallel inverters
system show no correlation between the inverters. This leads to a better total harmonic
value (THD) value. Overall a 19% improvement, compared to a conventional PI current
controller, is achieved.
This research has resulted in the following publication:
- S. Salimin, M. Armstrong and B. Zahawi, "Randomized Integral Gain of Pi Current

Controller for A Single Pv Inverter System," Engineering, Vol. 5 No. 1B, 2013, pp.
221-225.
12
Literature Review
Chapter 2: Literature Review
The chapter begins with an overview of harmonic injection. A brief explanation on

what harmonics are; as well as their potential impacts, is discussed. Following this, four
types of harmonics are identified. The work from a significant number of papers is
discussed in the area of harmonic injection and harmonic mitigation techniques for both
single and multiple/parallel grid-connected inverter systems. On a broader level,
additional papers concentrating on harmonic generation from engineering fields other
than grid-connected inverters are then reviewed. Finally, from the literature review, the
direction and objectives of this research project are compiled and explained.
2.2 Reviews on Harmonic Injection
The most significant discussion in any grid-connected system is the current or voltage
harmonic injection into the grid. In general, harmonics are considered to be undesirable
voltage or current frequency components that can arise in an electrical system or power
network. They normally appear at frequencies which are an integer multiples of the
fundamental frequency of the system [35]. In grid connected inverter systems, the
fundamental frequency is normally 50 Hz or 60 Hz, depending on the country of
concern. The mathematical definition of a single harmonics is:
𝑓𝑛 = 𝑛𝑓1 ( where 𝑓1 is the fundamental frequency and n = 1,2,3,… )
When this harmonic is added to the magnitude of fundamental frequency, a distorted

current or voltage waveform is produced. In grid-connected systems, harmonic
emissions not only depends on the characteristics of the grid network, but also on the
location of PV installation within the network and the behaviour of the inverter system
itself [36, 37]. For instance, installing the PV inverter system closer to the secondary
side of the pole transformer, the effects of the harmonics are less than installing the PV
closer to loads [38].
13
Literature Review
In PV grid connected system, concerns have been raised by several relevant bodies
about the injection of harmonics into the network. Impacts of harmonic currents
according to [33, 39-41] include:
a) Communication interference.
b) Heating problems that lead to over-current, insulation breakdown, cable
corrosion –due to ‘skin effects’ of copper.
c) Solid state device malfunctions.
d) Voltage distortion that will reduce reliability electrical and electronic systems.
Papaioannou et al [42] describe the impact of the connection of PV inverter systems to

the low voltage (LV) distribution network. The investigation took place in Greece
involving a 20 kWp PV system connection under different weather conditions. Results
from the harmonic behaviour examination show the significant presence of 3rd, 5th, and
7th order harmonics current. Furthermore, Kotsopoulos et al., 2005 [43] reported that
when many inverters are operating in parallel, resonance caused by the interaction
between the inverter output filter capacitance with the grid is also a big issue. A case
in point is failure to the power factor correction capacitors. A report by Caamano et al
[32] mentioned a wide range of harmonic impacts to the network components and AC
side of the inverter system. These include variation in RMS voltage and flicker,
disturbances of electronic component, overloading of passive filters, stress on
insulation materials, and many more.
H. Soo et al [44] stated a different view about low-order harmonics; one of four types
of harmonics injection. The paper demonstrated that the low order harmonic profile of
the inverter output current can be affected by the operating conditions of the grid
voltage, variations in the grid impedance, and the connection of other non-linear loads
to the network. IEEE 929-2000 [31] sets limitation on the level of harmonic injection
into the grid which is permissible. Table 2-1 shows the recommended individual current
harmonic limitations for general distribution systems. The current THD must also be
less than 5% of the fundamental frequency at the rated output power of the inverter. In
addition, an individual voltage distortion limit of 3% and overall voltage THD of less
than 5% is defined by IEEE 519-1992 [45].
14
Literature Review
Table 2-1: IEEE standard on harmonic limitation [45].
Harmonic Order (Odd) Limit
3rd – 9th < 4%
11th – 15th < 2%
17th – 21st < 1.5%
23rd – 33rd < 0.6%
Above 33rd < 0.3%
These recommendations apply to single inverter installations only. The author is

unaware of any recommendations relating to parallel connected systems. Indeed, this
may be difficult to enforce, as it is often the case that individual inverter units are
connected to the grid separately, and co-incidentally form a parallel operating system
with existing installations. For example, consider the case where a domestic customer
has a PV installation at home. The engineer installing the system will typically only
concern themselves with the immediate installation, not the neighbouring devices.
The various types of harmonics are often being categorised as PWM or switching
harmonics [33], low-order harmonics [33], and DC current injection [46]. A. Testa et
al [47] has also demonstrated that inter-harmonics and sub-harmonics can be included
as an additional source of harmonic pollution which may need to be considered. A brief
explanation of these types of harmonic is described in the following section.
2.2.1 Types of Harmonic
a) Low-order Harmonics
Low-order harmonics are components at frequencies close to the fundamental

frequency. Their origin is often from the deficiencies in the inverter output current
controller. Besides, extrinsic sources such as connection to a distorted grid can also
produce low-order harmonics. As stated by Abeyasekera et al [48], in order to remove
low-order harmonics, it requires a bulky and costly output filter. Therefore, it is
common to ensure that a robust current control loop is applied. Several PWM inverter
15
Literature Review
switching techniques and control methods have been proposed to specifically reduce
and eliminate low-order harmonics. They are broadly discussed in sub-section 2.3.
b) PWM/Switching Harmonics
Switching harmonics are the high order harmonics generally in kHz range. They are
generated because of the side effect of high frequency inverter switching and can be
filtered relatively easily by using a low pass filter at the inverter output. In a study by
H.Zhang et al [49], equations to calculate switching harmonic components are derived
for the purpose of minimizing the dc link capacitor. S. Ma et al [50] has proposed two
techniques as a solution for switching harmonics problem. They are the zero voltage
switching half cycle and voltage phase compensation and their effectiveness is
confirmed.
c) Inter-harmonics and Sub-harmonics
Inter-harmonics and sub-harmonics on the other hand are components at variable

frequencies that are not in integer multiples of fundamental frequency. The detection
and measurement of the frequencies are difficult to trace and there is still unanswered
questions on how to accurately measure and eliminate them from systems. Voltage or
current distortion from inter-harmonics may arise from effects such as sub-synchronous
oscillations, voltage fluctuation and disturbance in fluorescent lamps, arcing loads,
variable speed electric drives, static converters and ripple controllers [47, 51]. Chun Li
et al [52] have carried out experiment work and field tests on inter-harmonics where
the result illustrated the difficulties in the analysis. Z. Leonowizc [53] has also shown
a method called a root-music method that can be used to estimate the signal parameters
of sub-harmonics and inter-harmonics; however, the computation is complex.
d) DC Current Injection
The last type of harmonic is DC current injection. Power electronic converters are one
of the major sources of DC current being injected into the grid. DC components can
shift the operating point of distribution transformers [54] and this is highly undesirable.
For this reason, network operators often set a limit of close to zero for the DC
component. Often, a mains frequency transformer is used to couple the inverter to the
network, and prevent any DC current components from entering the grid. [55].
16
Literature Review
Avoiding the use of an isolation transformer in PV systems is now being the trend
where the cost and weight can be reduced [56].
2.3 Reviews of Previous Works

To date, a considerable amount of literature has been published to improve the
harmonic performance of grid connected inverters, all with their associated merits and
disadvantages. It is always possible to make the improvements by improving the power
electronic converter hardware; inverter topology, PWM switching schemes, filter
arrangements, and so on. Alternatively, it is also possible to enhance the performance
and robustness of the inverter current controller.
The following discussion is divided into three sections. In the first section, several
papers on harmonic performance improvements of a single grid-connected PV inverter
system are described. This is then followed by a review of the harmonic performance
of parallel grid-connected inverter systems. The final section is a discussion of
harmonic performance in other application areas, such as uninterruptible power supply
(UPS) and AC motor drives.
2.3.1 Single Connected PV System

In 1973, a generalized technique of harmonic elimination in the half-bridge and full-
bridge output waveforms was proposed by Patel and Hoft [57]. In this work, output
waveforms for both the half-bridge and full-bridge are ‘chopped’ M times per half
cycle. An M equation is then obtained through several steps of derivation which can
best be solved by computational techniques. Whilst complex, solutions for eliminating
the 5th, 7th, 11th, 13th, and 17th can be obtained.
Then, in the study by Teodorescu et al [58], a new control structure was proposed to
mitigate high harmonic distortion problems which arise from the imperfect
compensation action of a grid voltage feed forward PI controller. This new structure
uses a P+Resonant (PR) controller to control the fundamental current and several
generalized integrators in a harmonic compensator (HC) for THD level reduction. The
harmonics of interest were the 3rd, 5th, and 7th components.
17
Literature Review
Figure 2-1: The current loop of PV inverter with PR and HC.
Figure 2-1 illustrates the block diagram of the proposed control strategy where GPR is
the PR current controller and GHC is the harmonic compensator. PR controller was
designed first without the compensator until the correct tuning of the proportional gain,
KP and resonant gain, KR is determined. This was then followed with the addition of
HC. From the paper [59], Observation from the Bode graph for PR + HC shows a peak
gain exists at the harmonics frequencies of interest where this gain cannot be found in
the PR graph. Interestingly, controller dynamic remains unchanged and this is being
the key point to compensate the selective harmonic components. Tests to compare the
spectrum for PI, PR and the proposed structure, PR + HC, were done. Results show
much improvement in the interested harmonics order and the THD level is also
decreased [59].
In another study by Su et al [59], a proportional resonant (PR) controller scheme is also

used to achieve a high quality sinusoidal output current waveform in a single phase PV
inverter. By simulation, it is shown that the THD obtained for the output current is
2.72%, which is clearly acceptable [60].
A different research study on the proportional resonant (PR) control system has also
been carried out by Guoqiao et al [60, 61]. In his papers, a new current feedback method
for the PR controller is proposed. In the first paper, instead of taking the inverter output
current or the grid current as the feedback, he suggests to use the weighted average
value (WAC) of the inverter output current and the grid current as the current feedback
method. In the second paper, he recommends splitting the capacitor of the LCL filter
into two parts (C1 and C2) and to take the current flowing between the two capacitors
as the feedback current for the PR controller. From the experimental results for both
18
Literature Review
proposed methods [61, 62], two conclusions are drawn. Firstly, the control system is
reduced from a third order system to a first order system; this can be proved by deriving
the transfer function solution for the system. Secondly, less phase error and lower
current THD are achieved when compared to using the inverter output current
feedback.
Castilla M. et al [62] proposed a different way of using the resonant harmonic

compensator. Instead of connecting the resonant harmonic compensator in parallel with
the tracking regulator, the paper presented the compensator to be in series with the
tracking regulator. Experimental results show that this location prevents the need of
PLL algorithm. This is because this technique alone can provide accurate
synchronization with grid voltage. In addition, the grid current distortion is also slightly
lower when compared to the conventional harmonic compensator connection.
A novel harmonic elimination PWM technique is introduced by Villarreal-Ortiz et al

[63] where the location of the commutation angles is derived from a closed-form
equation based on a centroid algorithm. This technique is named Centroid PWM
(CPWM) and is suitable to use for online control applications. In the study, Villarreal-
Ortiz et al achieve lower THD levels when using the proposed technique compared to
PWM techniques carried out in previous studies.
Leopoldo et al [64] proposed a new method to generate the switching pattern of three-
level PWM inverters. This is called selective harmonic mitigation PWM (SHMPWM).
Initially, the paper revised the selective harmonic elimination PWM (SHEPWM)
technique from previous works where the need for tuned filters is essential to minimise
uncontrolled non-eliminated harmonics (thus increasing the volume, weight and cost
of the system). Using the proposed strategy, an advance on previous methods is
achieved by considering the harmonics, THD, and local grid standards as a global
problem. The switching pattern is reformulated as a nonlinear optimization problem
and solved using a general purpose optimization heuristic algorithm. The method is
claimed to be very versatile and can be tuned to meet any grid code in the calculation
process. The study considered grid codes EN 50160 and CIGRE WG 36-05 and 15
switching angles. Results highlighted low harmonic levels when using the SHMPWM
technique, hence avoiding the use of the costly tuned filters in the system. Experiment
19
Literature Review
in order to compare the technique with previous ones that are optimal sinusoidal PWM
(OSPWM) and SHEPWM methods was carried out [65]. At first, result for SHEPWM
was studied and showed zero harmonic content up to the 43rd order but the 45th and 49th
order had important content, 8.1% and 16.9% respectively. Result was slightly different
with OSPWM method where the switching harmonic was inside the restrictive band.
Both previous methods did not comply with the harmonic grid codes limitation.
Conversely, grid codes stated earlier was fulfilled with the proposed SHMPWM
technique.
Following that, in another study by Qunhai Huo et al [65], innovation of the

conventional PI current control mode was made due to poor control effects in PI current
loop control where the exponential sliding mode control (SMC) is proposed to be used
in the control loop. In the paper, the chosen sliding model and existence of SMC as
well as SMC stability analysis using Lyapunov theorem was derived. Simulation was
done and output current waveforms were compared between the proposed method and
conventional PI controller. Results showed a high degree of ripple and sawtooth-like
waveform in the output current of conventional PI controller. In contrast, a smooth sin
waveform can be seen in the result showing greatly reduced ripple phenomenon. In
addition, the FFT analysis showed an improvement in THD to 0.91% rather than
24.75%. To conclude, the system stability is theoretically confirmed their report of the
poor control effects when using the conventional one. However, this study only
provides strong theoretical support in designing hardware but has not been able to
demonstrate and implement it.
The most recent research was presented by Jevraj and Nasrudin [18]. They claimed that
there is a limitation in harmonic reduction when using a typical single-phase three level
inverter. This is because the harmonic components of output voltage are determined by
the carrier frequency and the switching functions. Therefore, in order to tackle this
limitation, a five level inverter topology is introduced for PV systems where instead of
one reference signal, the new inverter topology uses two where they took turn to be
compared with the carrier signal at one time. The idea behind this topology is to
generate five level of output voltage; +VPV, +1/2VPV, 0, -1/2VPV, -VPV; where VPV is the
voltage across the DC-DC boost converter. Furthermore, the proposed topology uses
an auxiliary circuit between the DC-DC boost converter and the inverter. Results are
20
Literature Review
compared with the results of three-level inverter system. The lower THD measurement
for the proposed inverter proved that harmonic content can be reduced as the number
of output levels increases. However, the measured PV system efficiency is low because
of the auxiliary circuit.
2.3.2 Parallel Connected PV System

PV systems which are connected in parallel are commonly found in many networks.
Such an arrangement is often used to increase the power delivered to the network. In
this topology, each PV array has its own inverter and controller system, but is
connected in parallel at a common point of coupling to the supply network. However,
research studies into low order harmonics injection in parallel inverter systems, and
methods to eliminate them, are not so popular compared to single connected inverter
systems.
In a study by Younis et al [66, 67], a new configuration of parallel inverter system is

presented. The design is based on a parallel three inverter system, where a resistor is
used in series at the inverter output to minimise circulating current flow in the inverters.
The paper also explains how the use of a third harmonic injection PWM (THIPWM)
can be employed for further THD reduction. Results compare single inverter operation
to parallel inverter operation, and show an improvement in current and voltage THD
performance when the inverters are connected in parallel.
Another study by Armstrong et al [33] presents an inverter current controller with

simple parameter randomization as an alternative to inter-inverter communications, for
the purpose of introducing harmonic cancellation between individual inverter units. In
the study, three independent inverters with conventional current control technique are
tested and shown to exhibit very similar harmonic performance due to a strong
correlation with the grid synchronisation process and the dynamics of the grid itself;
harmonics between the 3rd and 15th are particularly predominant. When the three
inverters are connected together in parallel with a common point of coupling to the
network, a similar trend in harmonic performance is observed in each inverter unit.
21
Literature Review
Figure 2-2 Harmonic trend line of parallel inverter system using conventional control
and parameter randomization control technique [33].
In order to prevent these harmonics from becoming stronger when more and more
inverters were connected in parallel, a modification of the current controller was made.
The new technique is called the current controller parameter randomization technique
where a random gain component, Rp, was added to the original proportional gain, Kp.
With this technique, results showed an uncorrelated trend lines for both individual
inverter and parallel inverter system. From Figure 2-2, although the harmonics between
the 3rd and 11th remain the dominant one, magnitude of these harmonics was reduced
on average and additional reduction was also achieved in the higher harmonics order.
This confirmed a lower THD level. However, a limitation of the method proposed can
be observed from the result shown in Figure 2-2 where a greater value can be seen in
the 3rd and 5th harmonic orders.
Recently, A.M. Roslan et al [68] introduced an adaptive gain scheduling in the

controller of parallel inverters. Instead of using the conventional PI current controller,
it uses a PR current controller in the study because of its many advantages compared
to conventional PI. The study focused on the instantaneous average current sharing
between two inverters when the condition of line impedance varied. Two gain
schedulers are used in two different location of each inverter. The model of the parallel
system using the proposed technique is explained clearly in the paper. The impact of
this technique can be seen in the inverter output current of each inverter. Final results
show a better performance regarding the current and power sharing of parallel inverter
22
Literature Review
systems under variable line impedance but a study of low order harmonics performance
has not been carried out.
M. Jafari et al [69] has also presented a method regarding the average current sharing.
The proposed method uses a reference current which is taken from the average current
sharing centre (ACSC) and is claimed to be the most robust scheme in average current
sharing literature group. This is because it provides robust stability and performance
under filter and load parameter variations. The study begins with analysis of a single
inverter system both with linear and non-linear load and followed with the multi-
inverter systems analysis. However, evidence from hardware experimental results is
not stated for verification of simulation findings.
2.3.3 General Review on Harmonic Performance Improvements
In addition to reviewing the performance of inverters focusing in grid connected PV

system, several papers proposing ideas on improving the inverter performance in
different areas have also been noticed. This includes improvement in the controller for
islanding inverters, UPS as well as AC motor drives and vector-controlled drive.
During the islanding mode, instead of using the non- sinusoidal PWM (NSPWM),
Khodsuz M. and A. Sheikholeslami [70] has presented a different technique known as
non-sinusoidal hysteresis voltage control (NSHVC). The paper briefly explained how
the nonlinear load can cause problems especially when the supply grid is off. Several
advantages and features of the proposed technique have also been listed. Simulation of
NSHVC and NSPWM are performed and compared. When the new controller
technique is used, a significant reduction in THD and the harmonic components are
noticed from the results shown.
In a recent study, B. Geethalakshmi et al [71] has proposed a shunt active power filter
(APF) in order to compensate for the current harmonic components produced by
nonlinear loads such as current and voltage source inverter. The APF is realized by
using a 5 level H-bridge inverter. From the simulation results, based on several APF
performance verification including unbalanced supply voltage and dynamic load
variation, the proposed technique has proved to reduce the magnitude of significant
harmonics as well as the line current THD.
In her study on harmonics reduction, Najwa Mahamad et al [72] divided the harmonics
generated into three types; the positive-sequence harmonics which are of the 7th, 13th,
23
Literature Review
and 19th components, the negative-sequence harmonics which are of the 5th, 11th, and
17th components that can totalled up the THD current of the system, and lastly the zero-
sequence harmonics which are of the triplen harmonics such as the 3rd, 9th, and 15th
harmonic components and can cause overheating on the neutral wire of the three phase
power distribution system. In order to reduce the harmonics, the paper proposed an
employment of a transformer and LC filter. From the experimental result, using the
transformer alone can only reduce a small amount of harmonics current magnitude.
However, after implementing the combination of the transformer and the LC filter, the
harmonic in line current and neutral current is reduced to the standard of IEEE.
In 2008, A. Zabihinejad and J.S. Moghani [73] has proposed a new method called
Direct Injection of Random Signal (DIRS) in order to reduce the inverter output current
and voltage THD of an induction motor drive. In their study, a random signal is applied
in the control scheme and has caused a variance in the output current and voltage THD.
Results show that varying the random signal by 30% has best improved the output
voltage harmonic spectrum characteristics. In addition, the harmonic components are
also well distributed.
2.4 Research Idea
In this research, a “hybrid controller” is proposed where the features from promising
techniques or schemes above will be combined to produce a new control scheme
capable of minimising the low order harmonics in the parallel grid connected inverter
system. The positive features of selective harmonic compensation scheme; which can
compensate the low-order harmonics of particularly the 3rd, 5th, and 7th, and the random
signal injection; which can reduced the harmonic magnitude of the 9th to 17th orders are
used together in the current controller system for this research project in order to
produce a computationally efficient controller. So far, no research studies have been
found to apply these two methods together and use it in any inverter applications.
Therefore, the combining methods are proposed here to reduce the harmonic magnitude
of the low order harmonics spectrum between the 3rd and 19th of the grid connected
system.
24
Literature Review
2.5 Chapter Summary
From this chapter, an overview of the harmonics injection has been explained. The
definition of harmonics as well as its impacts is discussed. Besides, the four types of
harmonics that are the low order harmonics, switching harmonics, dc current injection
and inter harmonics are also mentioned in this chapter. This chapter has also discussed
a large amount of papers on previous works about the harmonics injection and the
solving methods with first; on a single grid connected inverter system and second; on
a parallel grid connected inverter system. Furthermore, a few papers about harmonics
injection from other area that grid connected inverter are also reviewed. At the end of
this chapter, the idea of the research project is clarified.
25
Methodology
Chapter 3: Methodology

At the beginning of this chapter, two forms of PWM technique are discussed. Following
that, the advantages of a digital system is explained. Three types of controller systems;
the conventional PI current controller, the PR current controller, and the modified PR
current controller, are also discussed in detail towards the end of this chapter.
3.2 Identification of the system to be studied

In order to start simulating the project models, these two main parts or systems must
be identify and understood first; the PWM technique and the inverter current control
method. PWM is used as a switching technique to drive the gates of the inverter.
Whereas, the current control technique is used to respond and compensate for any error
in the inverter current. Synchronization to the grid voltage is usually desirable and will
also be implemented by the current controller; this is vital to have a unity power factor
injection. Details are discussed next.
3.2.1 PWM Technique

There are two forms of switching techniques that can be used to drive the gate signals
for inverters; the bipolar PWM switching technique and the unipolar PWM switching
technique. Here, both of them is briefly discussed.
26
Methodology
1. Bipolar PWM Switching Technique
Figure 3-1: Bipolar PWM switching.
This technique is normally used with a half-bridge inverter where it compares a sine
wave that acts as the control wave with a carrier wave which is normally in a triangular
form. It is shown in Figure 3-1 above. The process is quite simple. Whenever the
voltage of the sine wave is bigger than the carrier wave, switch 1 will turn on and the
𝑉𝑑𝑐⁄
output will then be + 2 . But when the voltage of the sine wave is smaller than the
𝑉
carrier wave, switch 2 will turn on and the output will become − 𝑑𝑐⁄2. This process
is illustrates in Figure 3-2.
27
Methodology
(a)
(b)
(c)
Figure 3-2: Waveforms of (a) 𝑽𝒄𝒐𝒏𝒕𝒓𝒐𝒍 > 𝑽𝒕𝒓𝒊 which turns on Gate 1;
(b) 𝑽𝒄𝒐𝒏𝒕𝒓𝒐𝒍 < 𝑽𝒕𝒓𝒊 which turns on Gate 2 ; and (c) Output voltage.
28
Methodology
2. Unipolar PWM Switching Technique
Figure 3-3: Unipolar PWM unit.
In this project, a unipolar PWM technique is used as the switching technique for the H-
bridge inverter. This is because this switching technique offers a higher efficiency and
higher power output than bipolar switching technique [74]. Moreover, less electro-
magnetic interference (EMI) can be achieved when using unipolar switching [75].
T.Abeyasekera [76] also mentioned two advantages of a unipolar PWM technique.
First, it has less dv/dt stress on the load as the voltage goes from 0 to +𝑉𝑑𝑐 or from 0 to
−𝑉𝑑𝑐 . Second, it has significant reduction in DC link voltage ripple thus reducing the
size of the capacitors. The model of an inverter with a unipolar switching technique is
as Figure 3-3. This PWM unit is then connected to an H-bridge inverter (see Figure
3-4). In the early stage where basic simulation is run, IGBTs with internal diodes were
used as the switching devices. For later simulations (Chapter 5) and experimental
hardware (Chapter 6 and Chapter 7) purpose, MOSFETs are used because they are
suitable for the conditions of the research; use less than 200 V DC supply, less than 1
kW system, and using 20 kHz switching frequency. 20 kHz is a typical switching
frequency for a grid connected inverter system. It is chosen because it represents a good
balance between conduction and switching losses in typical MOSFET switched
applications. Furthermore, it is beyond the audible range of humans, which is highly
desirable. Therefore, throughout this thesis, unless otherwise stated the chosen PWM
29
Methodology
switching frequency for all simulation and experimental work is 20 kHz. This system
works as described in the next page.
Figure 3-4: Unipolar PWM switching.
Based on Figure 3-3 and Figure 3-4, if
𝑉𝑠𝑖𝑛 > 𝑉𝑡𝑟𝑖 ; Out 1 will turn on Switch 1.
𝑉𝑠𝑖𝑛 < 𝑉𝑡𝑟𝑖 ; Out 2 will turn on Switch 2.
−𝑉𝑠𝑖𝑛 > 𝑉𝑡𝑟𝑖 ; Out 3 will turn on Switch 3.
−𝑉𝑠𝑖𝑛 > 𝑉𝑡𝑟𝑖 ; Out 4 will turn on Switch 4.
30
Methodology
(a)
(a)
(b)
(b)
Figure 3-5: Waveforms at (a) Out1 and Out2 (b) Out3 and Out4.
(b)
31
Methodology
Figure 3-5 are the examples of the waveforms that are being the gate signal for IGBT1,
IGBT2, IGBT3, and IGBT4. When the combination of turning on and off above is
compared, output results should be in a form of a modified sine wave with 𝑉𝑝𝑒𝑎𝑘 must
equal to 𝑉𝑑𝑐 . Therefore, to ensure the PWM technique is working in the correct way,
all the parameters involved are set as in Table 3-1 and the model is run for 0.2 s with
maximum step size in the configuration parameters being changed to 1.0x10-6. This is
to confirm the sampling rate is enough without missing any important information.
Output result agrees with the theory and is shown in Figure 3-6.
Table 3-1 System parameters.
Sine Wave Frequency = 50 Hz

Ma = 0.8
Carrier Wave Frequency = 20 kHz

Output values = [-1 1 -1]
Inverter Source 400 𝑉𝑑𝑐
Figure 3-6: Inverter output voltage.
3.2.2 Digital Control System

The crucial part in the project is concerned with the control process. Better regulation
of the current control system is important so that a unity power factor current is injected
into the system. The idea of this project is to use a digital system rather than the
analogue system. This is because digital system has the advantage of flexibility,
accuracy, and easiness to monitor the parameters. Compared to the simple analogue
controller system, the digital system uses a microprocessor as a main element with
other small systems before and after it. These small systems are; sample and hold which
32
Methodology
holds analogue signal at a constant value, analogue to digital converter (ADC) which
converts analogue signals to digital number and digital to analogue converter (DAC)
which converts the digital number back to an analogue signal [77]. The block diagram
of digital control system in Figure 3-7 illustrates the process.
i/p
Sampling & ZOH ADC Microprocessor DAC ZOH o/p
Figure 3-7: Digital control system.
The process starts as soon after a signal is identified. This signal is then sampled at a
certain constant intervals periodically and held throughout the same intervals by a zero-
order hold ( ZOH ) device to form a staircase-like sample. An analogue to digital
converter is then used to convert the sample to discrete values in a finite time as
demanded by the microprocessor. The major job of microprocessor is to control the
incoming values so that it appears as nearly as demanded by the user. When the job is
done, it has to be converted back to analogue signal via a digital to analogue converter.
Again, a zero-order hold ( ZOH ) is a necessity.
3.3 Current Controllers

As mentioned in the previous chapters, this research project will demonstrate the
efficiency of three different current controllers; the conventional PI control technique,
the proportional resonance (PR) control technique and the modified proportional
resonance control technique. All three of them will be discussed as in the following
sub-sections.
3.3.1 Conventional PI Control Technique

In this project, rather than the proportional, integral and derivative (PID) system, only
the proportional (P) and integral (I) terms are used in the current controller system. This
type of controller is the most common controller applied as the inverter current
feedback process. The term P will give an output that is proportional to the system
error, which is the difference between the system output value and the desired value. It
has a gain, 𝐾𝑃 , which will multiply the error and response to it. The purpose is to reduce
the rise time of the system. However, using this proportional term alone will result in
having a system stationary error. In order to eliminate this error and complete the P
33
Methodology
based control, the integral part is used. The output of the integral part is the
multiplication of a gain, 𝐾𝐼 and the summing of the previous errors to the current system
error. This is a continuous process which will stop if the system signal or the system
output value matches the desired value demanded by the user. The analogue PI transfer
function is
𝐾𝐼
𝐺𝑃𝐼 (𝑠) = 𝐾𝑃 +
𝑠
(Eq. 3-1)
In the z-domain, the transfer function becomes
𝐾𝐼
𝐺𝑃𝐼 (𝑧) = 𝐾𝑃 +
1 − 𝑧−1
(Eq. 3-2)
Figure 3-8: A PI current controller block diagram.
Figure 3-8 illustrates the PI current controller block diagram. The system actual current
is compared to the reference current and will be used in the controlling process. The
output signal after the process has been taken that is the Out1 is then used for the PWM
inverter switching.
34
Methodology
3.3.2 Proportional Resonance Controller Scheme (PR)

This second technique is called the proportional resonance control technique. In this
technique, several harmonic current regulators work in parallel. By using this
technique, the weakness of the PI current control technique regarding the steady state
error can be dealt with [78]. This means that the measured output current will be equally
same to the reference current or the demand current. Based on previous paper [58], the
analogue transfer function of the PR current controller is defined as:
𝑠
𝐺𝑃𝑅 (𝑠) = 𝐾𝑃 + 𝐾𝑅
𝑠2 + 𝜔𝑜2
(Eq. 3-3)
By using the Tustin transformation, the analogue equation above is changed to the
𝟐 𝟏−𝒛−𝟏
z domain known as the discrete function. This is done by substituting 𝒔 with .
𝑻 𝟏+𝒛−𝟏
Based on that, Eq. 3.3 is then transformed to:
2 1 − 𝑧 −1
𝑇 1 + 𝑧 −1
𝐺𝑃𝑅 (𝑧) = 𝐾𝑃 + 𝐾𝑅
2 1 − 𝑧 −1 2
(𝑇 ) + 𝜔𝑜2
1 + 𝑧 −1
2
2 1 − 𝑧 −1 2 1 − 𝑧 −1
𝐾𝑃 [( ) + 𝜔𝑜2 ] + 𝐾𝑅 ( )
𝑇1+𝑧 −1 𝑇 1 + 𝑧 −1
𝐺𝑃𝑅 (𝑧) = 2
2 1 − 𝑧 −1
(𝑇 ) + 𝜔𝑜2
1 + 𝑧 −1
4 1 − 2𝑧 −1 + 𝑧 −2 2 2𝐾𝑅 1 − 𝑧 −1
𝐾𝑃 [( ) ( ) + 𝜔 ] +
𝑇 2 1 + 2𝑧 −1 + 𝑧 −2 0 𝑇 (1 + 𝑧 −1 )
𝐺𝑃𝑅 (𝑧) =
4 1 − 2𝑧 −1 + 𝑧 −2
( 2) ( ) + 𝜔02
𝑇 1 + 2𝑧 −1 + 𝑧 −2
4𝐾𝑃 1 − 2𝑧 −1 + 𝑧 −2 2 2𝐾𝑅 1 − 𝑧 −1
( ) ( ) + 𝐾 𝜔 +
𝑇 2 1 + 2𝑧 −1 + 𝑧 −2 𝑃 0 𝑇 (1 + 𝑧 −1 )
4(1 − 2𝑧 −1 + 𝑧 −2 ) + 𝑇 2 𝜔02 (1 + 2𝑧 −1 + 𝑧 −2 )
𝑇 2 (1 + 2𝑧 −1 + 𝑧 −2 )
(Eq. 3-4)
In order to get a simpler transfer function, some adjustments are made to the (Eq. 3-4)
above. This is shown as in the next page.
35
Methodology
4𝐾𝑃 1 − 2𝑧 −1 + 𝑧 −2 𝑇 2 (1 + 2𝑧 −1 + 𝑧 −2 ) 2𝐾 𝑇 1 − 𝑧 −2
( 2 )( −1 −2 ) + 𝐾𝑃 𝜔02 ( 2 −1 −2 ) + 𝑅2 ( )
𝑇 1 + 2𝑧 + 𝑧 𝑇 (1 + 2𝑧 + 𝑧 ) 𝑇 1 + 2𝑧 −1 + 𝑧 −2
4(1 − 2𝑧 −1 + 𝑧 −2 ) + 𝑇 2 𝜔02 (1 + 2𝑧 −1 + 𝑧 −2 )
𝑇 2 (1 + 2𝑧 −1 + 𝑧 −2 )
4𝐾𝑃 (1 − 2𝑧 −1 + 𝑧 −2 ) + 𝐾𝑃 𝜔02 𝑇 2 (1 + 2𝑧 −1 + 𝑧 −2 ) + 2𝐾𝑅 𝑇(1 − 𝑧 −2 )

=
4(1 − 2𝑧 −1 + 𝑧 −2 ) + 𝑇 2 𝜔02 (1 + 2𝑧 −1 + 𝑧 −2 )
4𝐾𝑃 − 8𝐾𝑃 𝑧 −1 + 4𝐾𝑃 𝑧 −2 + 𝐾𝑃 𝜔02 𝑇 2 + 2𝐾𝑃 𝜔02 𝑇 2 𝑧 −1 + 𝐾𝑃 𝜔02 𝑇 2 𝑧 −2 + 2𝐾𝑅 𝑇 − 2𝐾𝑅 𝑇𝑧 −2

=
4 − 8𝑧 −1 + 4𝑧 −2 + 𝑇 2 𝜔02 + 2𝑇 2 𝜔02 𝑧 −1 + 𝑇 2 𝜔02 𝑧 −2
(4𝐾𝑃 + 𝐾𝑃 𝜔02 𝑇 2 + 2𝐾𝑅 𝑇) + (2𝐾𝑃 𝜔02 𝑇 2 − 8𝐾𝑃 )𝑧 −1 + (4𝐾𝑃 + 𝐾𝑃 𝜔02 𝑇 2 − 2𝐾𝑅 𝑇)𝑧 −2
=
(4 + 𝑇 2 𝜔02 ) + (2𝑇 2 𝜔02 − 8)𝑧 −1 + (4 + 𝑇 2 𝜔02 )𝑧 −2
(Eq. 3-5)
By dividing the nominator and the denominator of the transfer function in (Eq. 3-5)
2
above by (4 + 𝑇2 𝜔0 ), the discrete form becomes:
(4𝐾𝑃 + 𝐾𝑃 𝜔02 𝑇 2 + 2𝐾𝑅 𝑇) (2𝐾𝑃 𝜔02 𝑇 2 − 8𝐾𝑃 ) −1 (4𝐾𝑃 + 𝐾𝑃 𝜔02 𝑇 2 − 2𝐾𝑅 𝑇) −2

+ 𝑧 + 𝑧
(4 + 𝑇 2 𝜔02 ) (4 + 𝑇 2 𝜔02 ) (4 + 𝑇 2 𝜔02 )
(2𝑇 2 𝜔02 − 8) −1
1+ 𝑧 + 𝑧 −2
(4 + 𝑇 2 𝜔02 )
(Eq. 3-6)
For simplification, the transfer function is re-written as:
𝑏𝑜 + 𝑏1 𝑧 −1 + 𝑏2 𝑧 −2
1 + 𝑎1 𝑧 −1 + 𝑎2 𝑧 −2
(Eq. 3-7)
where
(4𝐾𝑃 + 𝐾𝑃 𝜔20𝑇2 + 2𝐾𝑅 𝑇)

𝑏𝑜 = 2
( 4 + 𝑇2 𝜔0 )
(2𝐾𝑃 𝜔20𝑇2 − 8𝐾𝑃 )

𝑏1 = 2
( 4 + 𝑇 2 𝜔0 )
(4𝐾𝑃 + 𝐾𝑃 𝜔20 𝑇2 − 2𝐾𝑅 𝑇)

𝑏2 = 2
( 4 + 𝑇2 𝜔0 )
36
Methodology
2
(2𝑇2 𝜔0 − 8)
𝑎1 = 2
( 4 + 𝑇2 𝜔 0 )
𝑎2 = 1
𝐾𝑃 and 𝐾𝑅 are the proportional gain and the resonance gain, 𝑇 is the sampling time and
𝜔𝑜 is the fundamental frequency of the system in rad/sec. With the resonance part added
to the proportional controller, the steady state error of the system is nearly eliminated.
Substituting 𝑇 with 50 µs and 𝜔𝑜 with 2π x 50 Hz, the optimum proportional and
resonance gains of the PR current controller technique can be achieved by trial and
error tuning.
Figure 3-9 below shows the controller block diagram. The resulting output signal will
be used in the inverter switching process.
Figure 3-9: A proportional resonance (PR) current controller block diagram.
37
Methodology
3.3.3 Modified Proportional Resonance (PR) Technique

The last and improved technique proposed in this research project is the modification
of the proportional resonance technique above. Instead of using a fixed proportional
gain, KP, a random signal is added to generate a new randomly varying proportional
signal for the controller. This technique is expected to have an effect on the overall
harmonic orders that will result in further decrement of the grid current THD both on
single inverter system and parallel inverter system. Based on (Eq. 3-7), the discrete
transfer function of the modified PR current controller is;
𝑏𝑜 + 𝑏1 𝑧 −1 + 𝑏2 𝑧 −2
𝐺𝑃𝑅𝑚𝑜𝑑𝑖𝑓𝑖𝑒𝑑 (𝑧) =
1 + 𝑎1 𝑧 −1 + 𝑎2 𝑧 −2
(Eq. 3-8)
Although the transfer function above looks no different with the proportional resonance
control technique, modification is made to the 𝑏𝑜 , 𝑏1 , and 𝑏2 where;
(4𝑅𝐾𝑃 + 𝑅𝐾𝑃 𝜔02 𝑇 2 + 2𝐾𝑅 𝑇)

𝑏𝑜 =
(4 + 𝑇 2 𝜔02 )
(2𝑅𝐾𝑃 𝜔02 𝑇 2 − 8𝑅𝐾𝑃 )

𝑏1 =
(4 + 𝑇 2 𝜔02 )
(4𝑅𝐾𝑃 + 𝑅𝐾𝑃 𝜔02 𝑇 2 − 2𝐾𝑅 𝑇)

𝑏2 =
(4 + 𝑇 2 𝜔02 )
𝑎1 and 𝑎2 remain unchanged thus the values are identical to those obtained for the
proportional resonance control method. In order to get the correct value for the
randomized proportional gain, 𝑅𝐾𝑃 , a limit need to be set for both upper and lower
value so as to maintain the stability of the output current result. This limitation is
obtained by trial and error tuning where the edge where the output current starts to
become unstable is taken as the upper and lower limit for the randomized proportional
gain, 𝑅𝐾𝑃 . With the gain keeps changing within the limitation set each time, the output
current is expected to maintain its shape and stability whilst at the same time its
harmonic spectrum will results in the reduction of the THD.
38
Methodology
Figure 3-10: A modified PR current controller block diagram.
3.4 Chapter Summary
This chapter describes the fundamental operating principles of a single phase H-Bridge
inverter. For this inverter topology, two PWM techniques are briefly discussed; bipolar
and unipolar switching. The literature review shows that the Unipolar PWM switching
scheme offers a number of advantages, including improved harmonic performance, less
dv/dt stress on the semiconductor switches, and a reduction in the DC link voltage
ripple which allows for a smaller DC link capacitor to be used [74] [76]. For these
reasons, in this project, unipolar switching is selected as the preferred PWM control
technique.
This chapter also introduces three control techniques that can be applied in grid
connected inverter systems. The PI controller is described, as it is one of the most
commonly understood control methods and is often used as a benchmark for assessing
research results. A review of the PR controller is also presented, as this is becoming
another popular control technique in inverter based systems. Compared to the PI
controller, the literature suggests that the PR controller can be tuned to achieve better
harmonic performance. Furthermore, it does not exhibit the steady state error
associated with PI controller when tracking an AC reference signal.
Following this, the concept of PI parameter randomisation is introduced. Literature

suggests that this is a good technique to apply in order to reduce the harmonic content
of the inverter output; this is one of the main research objectives of this work. Based
on these findings, a new control strategy is described which extracts the benefits of the
39
Methodology
aforementioned techniques. Here, a PR controller with randomisation of the

proportional gain is proposed. The remainder of this thesis will focus on testing the
hypothesis that this proposed controller will yield better harmonic performance than
conventional PI and PR control techniques.
40
Modelling of Grid Connected, PV Inverter System
Chapter 4: Modelling of Grid Connected, PV Inverter System
4.1 Introduction
This chapter is written to model the closed loop system of grid-connected PV inverter
system. It begins with modelling the system impedances of the low pass filter, followed
by modelling the current controllers and the PWM. Three types of current controllers
are modelled here; the conventional PI current controller, the PR current controller, and
the PR current controller with added harmonic compensators (PR+HC). At the end, the
closed loop response is plotted, the system stability is analysed using the s-plane and
the critical part is the range of controller gains that can be varied without making the
system unstable can be determined.
4.2 Modelling of Grid Connected, Single PV Inverter System
Zg
DC-AC Low Pass

VDC Inverter Filter VG
Current
Controller
Figure 4-1: Basic arrangement of grid connected inverter system.
Figure 4-1 above shows the basic arrangement of a grid connected inverter system. In
order to model the whole system, it is divided into three parts. First, the low pass filter
is modelled using the system impedance. Second, the current controllers are modelled
and last, the PWM switching is modelled. They are discussed and explained in the next
sub-sections.
41
4.2.1 Modelling of Low Pass Filter with System Impedance
Ignoring the current control loop, the low pass filter consists of an inductance and
capacitance with an equivalent resistance value in series (ESR). To model the system
impedance, this will include the grid impedance at the output side of the filter thus
forming a LCL circuit arrangement as in
Figure 4-2. Next, the relation of the impedance is explained and discussed based on the
diagram.
IDC IINV Lf RLf IO Lg Rg

ICf
IG
RCf
VDC DC-AC VX VPCC VG
Inverter Cf
Figure 4-2: Equivalent circuit for low pass filter and the grid.
From the figure, 𝐿𝑓 is the filter inductance, 𝐿𝑐 is the filter capacitance, 𝐿𝑔 is the grid
inductance, and the three resistance; 𝑅𝐿𝑓 , 𝑅𝐶𝑓 , and 𝑅𝑔 , are the respective ESR for the
system. The relationship in time domain between the voltage at point of common
coupling, 𝑉𝑃𝐶𝐶 , and the voltage at inverter output, 𝑉𝑋 can be expressed as:
𝑑𝐼𝑋 (𝑡)
𝑉𝑋 (𝑡) − 𝑉𝑃𝐶𝐶 (𝑡) = 𝐿𝑓 + 𝑅𝐿𝑓 𝐼𝑥 (𝑡)
𝑑𝑡
(Eq. 4-1)
Whereas, in frequency domain, Laplace transform is made and given as:
𝑉𝑋 (𝑠) − 𝑉𝑃𝐶𝐶 (𝑠) = 𝐿𝑓 𝑠𝐼𝑋 (𝑠) + 𝑅𝐿𝑓 𝐼𝑋 (𝑠)
(Eq. 4-2)
In order to gain the Laplace transfer function of the low pass filter induction, it has to
be represented by its output/input characteristics. Considering the equation above, the
relationship becomes:
42
𝑉𝑋 (𝑠) − 𝑉𝑃𝐶𝐶 (𝑠) = 𝐼𝑋 (𝑠)(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 )
𝐼𝑋 (𝑠) 1
𝐻𝐿𝑓 (𝑠) = =
𝑉𝑋 (𝑠) − 𝑉𝑃𝐶𝐶 (𝑠) 𝐿𝑓 𝑠 + 𝑅𝐿𝑓
(Eq. 4-3)
Next, the same steps are taken in order to get the Laplace transfer function for the filter
capacitance branch. Equations below derived the steps.
1
𝑉𝑃𝐶𝐶 (𝑡) = 𝑅𝐶𝑓 𝐼𝐶𝑓 (𝑡) + ∫ 𝐼𝐶𝑓 𝑑𝑡
𝐶𝑓
(Eq. 4-4)
1
𝑉𝑃𝐶𝐶 (𝑠) = 𝑅𝐶𝑓 𝐼𝐶𝑓 (𝑠) + 𝐼 (𝑠)
𝐶𝑓 𝑠 𝐶𝑓
(Eq. 4-5)
1
𝑉𝑃𝐶𝐶 (𝑠) = 𝐼𝐶𝑓 (𝑠)(𝑅𝐶𝑓 + )
𝐶𝑓 𝑠
(Eq. 4-6)
Note that the current flowing through the filter capacitance branch is the current
difference between the inverter output current, 𝐼𝑋 and the grid current, 𝐼𝑔 . Therefore,
since 𝐼𝐶𝑓 (𝑠) = 𝐼𝑋 (𝑠) − 𝐼𝑂 (𝑠), the equation above becomes
𝑅𝐶𝑓 𝐶𝑓 𝑠 + 1
𝑉𝑃𝐶𝐶 (𝑠) = (𝐼𝑋 (𝑠) − 𝐼𝑂 (𝑠))( )
𝐶𝑓 𝑠
(Eq. 4-7)
Finally, the Laplace transfer function is:
𝑉𝑃𝐶𝐶 (𝑠) 𝑅𝐶𝑓 𝐶𝑓 𝑠 + 1

𝐻𝐶𝑓 (𝑠) = =
𝐼𝑋 (𝑠) − 𝐼𝑂 (𝑠) 𝐶𝑓 𝑠
(Eq. 4-8)
For the grid impedance branch, the same steps are again used to express its Laplace
transfer function. These are described in detail as in the following equations.
43
In time domain, the relationship is:
𝑑𝐼𝑂 (𝑡)
𝑉𝑃𝐶𝐶 (𝑡) − 𝑉𝐺 (𝑡) = 𝐿𝑔 + 𝑅𝑔 𝐼𝑂 (𝑡)
𝑑𝑡
(Eq. 4-9)
In frequency domain, the equation above becomes:
𝑉𝑃𝐶𝐶 (𝑠) − 𝑉𝐺 (𝑠) = 𝐿𝑔 𝑠𝐼𝑂 (𝑠) + 𝑅𝑔 𝐼𝑂 (𝑠)
𝑉𝑃𝐶𝐶 (𝑠) − 𝑉𝐺 (𝑠) = 𝐼𝑂 (𝑠)(𝐿𝑔 𝑠 + 𝑅𝑔 )
(Eq. 4-10)
To obtain the Laplace transfer function for the grid impedance branch, the output/input
characteristic is:
𝐼𝑂 (𝑠) 1
𝐻𝐿𝑔 (𝑠) = =
𝑉𝑃𝐶𝐶 (𝑠) − 𝑉𝐺 (𝑠) 𝐿𝑔 𝑠 + 𝑅𝑔
(Eq. 4-11)
By combining the three transfer functions from (Eq. 4-3), (Eq. 4-8) and (Eq. 4-11), the
relationship between the input and the output of the low pass filter and the grid can be
made with the output current, 𝐼𝑂 become the output and the inverter output voltage, 𝑉𝑋
become the input. Figure 4-3 is the block diagram when the three transfer functions
from the filter inductance branch, the filter capacitance branch and the grid impedance
branch are linked together.
Figure 4-3: Block diagram of the low pass filter and grid impedance branch.
44
𝐼𝑂 (𝑠)
Based on Figure 4-3, a single Laplace transfer function of ⁄𝑉 (𝑠) can be made.
𝑋
It is derived as followed [45].
1 𝑅𝐶𝑓 𝐶𝑓 𝑠 + 1 1
𝐼𝑂 (𝑠) = {{[(𝑉𝑋 (𝑠) − 𝑉𝑃𝐶𝐶 (𝑠)) ] − 𝐼𝑂 (𝑠)} − 𝑉𝐺 (𝑠)}
𝐿𝑓 𝑠 + 𝑅𝐿𝑓 𝐶𝑓 𝑠 𝐿𝑔 𝑠 + 𝑅𝑔
𝑉𝑋 (𝑠) − 𝑉𝑃𝐶𝐶 (𝑠) − 𝐼𝑂 (𝑠)(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 ) 𝑅𝐶𝑓 𝐶𝑓 𝑠 + 1 1

𝐼𝑂 (𝑠) = {{ } − 𝑉𝐺 (𝑠)}
𝐿𝑓 𝑠 + 𝑅𝐿𝑓 𝐶𝑓 𝑠 𝐿𝑔 𝑠 + 𝑅𝑔
[𝑉𝑋 (𝑠) − 𝑉𝑃𝐶𝐶 (𝑠) − 𝐼𝑂 (𝑠)(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 )](𝑅𝐶𝑓 𝐶𝑓 𝑠 + 1) 1

𝐼𝑂 (𝑠) = { − 𝑉𝐺 (𝑠)}
𝐶𝑓 𝑠(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 ) 𝐿𝑔 𝑠 + 𝑅𝑔
[𝑉𝑋 (𝑠) − 𝑉𝑃𝐶𝐶 (𝑠) − 𝐼𝑂 (𝑠)(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 )](𝑅𝐶𝑓 𝐶𝑓 𝑠 + 1) − 𝑉𝐺 (𝑠)𝐶𝑓 𝑠(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 ) 1
𝐼𝑂 (𝑠) = { }
𝐶𝑓 𝑠(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 ) 𝐿𝑔 + 𝑅𝑔
𝑠
[𝑉𝑋 (𝑠) − 𝑉𝑃𝐶𝐶 (𝑠) − 𝐼𝑂 (𝑠)(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 )](𝑅𝐶𝑓 𝐶𝑓 𝑠 + 1) − 𝑉𝐺 (𝑠)𝐶𝑓 𝑠(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 )
𝐼𝑂 (𝑠) =
𝐶𝑓 𝑠(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 )(𝐿𝑔 𝑠 + 𝑅𝑔 )
(Eq. 4-12)
(Eq. 4-11) can be rearranged to 𝑉𝑃𝐶𝐶 (𝑠) − 𝑉𝐺 (𝑠) = 𝐼𝑂 (𝑠)(𝐿𝑔 𝑠 + 𝑅𝑔 ). This then gives:
𝑉𝑃𝐶𝐶 (𝑠) = 𝐼𝑂 (𝑠)(𝐿𝑔 𝑠 + 𝑅𝑔 ) + 𝑉𝐺 (𝑠)
(Eq. 4-13)
Substituting both (Eq. 4-12) and (Eq. 4-13) gives:
[𝑉𝑋 (𝑠) − 𝐼𝑂 (𝑠)(𝐿𝑔 𝑠 + 𝑅𝑔 ) − 𝑉𝐺 (𝑠) − 𝐼𝑂 (𝑠)(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 )](𝑅𝐶𝑓 𝐶𝑓 𝑠 + 1) − 𝑉𝐺 (𝑠)𝐶𝑓 𝑠(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 )
𝐼𝑂 (𝑠) =
𝐶𝑓 𝑠(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 )(𝐿𝑔 𝑠 + 𝑅𝑔 )
1=
𝐼𝑂 (𝑠)𝐶𝑓 𝑠(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 )(𝐿𝑔 𝑠 + 𝑅𝑔 )
0= −1
0
[𝑉𝑋 (𝑠) − 𝐼𝑂 (𝑠)(𝐿𝑔 𝑠 + 𝑅𝑔 ) − 𝑉𝐺 (𝑠) − 𝐼𝑂 (𝑠)(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 )](𝑅𝐶𝑓 𝐶𝑓 𝑠 + 1) − 𝑉𝐺 (𝑠)𝐶𝑓 𝑠(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 ) − 𝐼𝑂 (𝑠)𝐶𝑓 𝑠(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 )(𝐿𝑔 𝑠 + 𝑅𝑔 )
=
(Eq. 4-14)
45
0 = [𝑉𝑋 (𝑠) − 𝐼𝑂 (𝑠)(𝐿𝑔 𝑠 + 𝑅𝑔 ) − 𝑉𝐺 (𝑠) − 𝐼𝑂 (𝑠)(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 )](𝑅𝐶𝑓 𝐶𝑓 𝑠 + 1) − 𝑉𝐺 (𝑠)𝐶𝑓 𝑠(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 ) −
0 = 𝑉𝑋 (𝑠)(𝑅𝐶𝑓 𝐶𝑓 𝑠 + 1) − 𝐼𝑂 (𝑠)(𝐿𝑔 𝑠 + 𝑅𝑔 )(𝑅𝐶𝑓 𝐶𝑓 𝑠 + 1) − 𝑉𝐺 (𝑠)(𝑅𝐶𝑓 𝐶𝑓 𝑠 + 1) − 𝐼𝑂 (𝑠)(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 )(𝑅𝐶𝑓 𝐶𝑓 𝑠 +

1) − 𝑉𝐺 (𝑠)𝐶𝑓 𝑠(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 ) − 𝐼𝑂 (𝑠)𝐶𝑓 𝑠(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 )(𝐿𝑔 𝑠 + 𝑅𝑔 )
0 = 𝑉𝑋 (𝑠)(𝑅𝐶𝑓 𝐶𝑓 𝑠 + 1) − 𝐼𝑂 (𝑠)[(𝐿𝑔 𝑠 + 𝑅𝑔 )(𝑅𝐶𝑓 𝐶𝑓 𝑠 + 1) + (𝐿𝑓 𝑠 + 𝑅𝐿𝑓 )(𝑅𝐶𝑓 𝐶𝑓 𝑠 + 1)

+ 𝐶𝑓 𝑠(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 )(𝐿𝑔 𝑠 + 𝑅𝑔 )] − 𝑉𝐺 (𝑠)[(𝑅𝐶𝑓 𝐶𝑓 𝑠 + 1) + 𝐶𝑓 𝑠(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 )]
(Eq. 4-15)
𝐼𝑂 (𝑠)[(𝐿𝑔 𝑠 + 𝑅𝑔 )(𝑅𝐶𝑓 𝐶𝑓 𝑠 + 1) + (𝐿𝑓 𝑠 + 𝑅𝐿𝑓 )(𝑅𝐶𝑓 𝐶𝑓 𝑠 + 1) + 𝐶𝑓 𝑠(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 )(𝐿𝑔 𝑠 + 𝑅𝑔 )]

= 𝑉𝑋 (𝑠)(𝑅𝐶𝑓 𝐶𝑓 𝑠 + 1) − 𝑉𝐺 (𝑠)[(𝑅𝐶𝑓 𝐶𝑓 𝑠 + 1) + 𝐶𝑓 𝑠(𝐿𝑓 𝑠 + 𝑅𝐿𝑓 )]
𝐼𝑂 (𝑠)[(𝐿𝑔 𝑅𝐶𝑓 𝐶𝑓 𝑠 2 + (𝐿𝑔 + 𝑅𝑔 𝑅𝐶𝑓 𝐶𝑓 )𝑠 + 𝑅𝑔 ) + (𝐿𝑓 𝑅𝐶𝑓 𝐶𝑓 𝑠 2 + (𝐿𝑓 + 𝑅𝐿𝑓 𝑅𝐶𝑓 𝐶𝑓 )𝑠 + 𝑅𝐿𝑓 )
+ 𝐶𝑓 𝑠(𝐿𝑓 𝐿𝑔 𝑠 2 + (𝐿𝑓 𝑅𝑔 + 𝑅𝐿𝑓 𝐿𝑔 )𝑠 + 𝑅𝐿𝑓 𝑅𝑔 )]
= 𝑉𝑋 (𝑠)(𝑅𝐶𝑓 𝐶𝑓 𝑠 + 1) − 𝑉𝐺 (𝑠)[𝐶𝑓 𝐿𝑓 𝑠 2 + (𝑅𝐶𝑓 𝐶𝑓 + 𝐶𝑓 𝑅𝐿 𝑓)𝑠 + 1]
(Eq. 4-16)
𝐼𝑂 (𝑠)[𝐶𝑓 𝐿𝑓 𝐿𝑔 𝑠 3 + (𝐿𝑔 𝑅𝐶𝑓 𝐶𝑓 + 𝐿𝑓 𝑅𝐶𝑓 𝐶𝑓 + 𝐶𝑓 𝐿𝑓 𝑅𝑔 + 𝐶𝑓 𝑅𝐿𝑓 𝐿𝑔 )𝑠 2

+ (𝐿𝑔 + 𝑅𝑔 𝑅𝐶𝑓 𝐶𝑓 + 𝐿𝑓 + 𝑅𝐿𝑓 𝑅𝐶𝑓 𝐶𝑓 + 𝐶𝑓 𝑅𝐿𝑓 𝑅𝑔 )𝑠 + 𝑅𝑔 + 𝑅𝐿𝑓 ]
= 𝑉𝑋 (𝑠)(𝑅𝐶𝑓 𝐶𝑓 𝑠 + 1) − 𝑉𝐺 (𝑠)[𝐶𝑓 𝐿𝑓 𝑠 2 + (𝑅𝐶𝑓 𝐶𝑓 + 𝐶𝑓 𝑅𝐿 𝑓)𝑠 + 1]
(Eq. 4-17)
From (Eq. 4-17), it is clearly seen that 𝐼𝑂 (𝑠) is the input and both 𝑉𝑋 (𝑠) and 𝑉𝐺 (𝑠) are
the output of the low pass filter and grid impedance branch. In order to obtain the
transfer function for the branch, the equation above is arranged to be as equation below:
𝐼𝑂 (𝑠)
𝐺𝐿𝐶𝐿 (𝑠) =
𝑉𝑋 (𝑠)(𝑅𝐶𝑓 𝐶𝑓 𝑠 + 1) − 𝑉𝐺 (𝑠)[𝐶𝑓 𝐿𝑓 𝑠 2 + (𝑅𝐶𝑓 𝐶𝑓 + 𝐶𝑓 𝑅𝐿 𝑓)𝑠 + 1]
1
=
𝐶𝑓 𝐿𝑓 𝐿𝑔 𝑠 3 + (𝐿𝑔 𝑅𝐶𝑓 𝐶𝑓 + 𝐿𝑓 𝑅𝐶𝑓 𝐶𝑓 + 𝐶𝑓 𝐿𝑓 𝑅𝑔 + 𝐶𝑓 𝑅𝐿𝑓 𝐿𝑔 )𝑠 2 + (𝐿𝑔 + 𝑅𝑔 𝑅𝐶𝑓 𝐶𝑓 + 𝐿𝑓 + 𝑅𝐿𝑓 𝑅𝐶𝑓 𝐶𝑓 + 𝐶𝑓 𝑅𝐿𝑓 𝑅𝑔 )𝑠 + 𝑅𝑔 + 𝑅𝐿𝑓
(Eq. 4-18)
46
4.2.2 Modelling of PWM
Whilst there are benefits associated with Unipolar PWM switching, both bipolar and
unipolar PWM techniques results in the same average fundamental output voltage
regardless of the switching behaviour. For control loop analysis, it has been shown that
modelling the bipolar PWM strategy is significantly easier than that of the unipolar
technique [44]. Therefore, for the control loop analysis only, the bipolar PWM
technique is applied here. This is considered a valid “approximation” to the actual
situation, provided we are only concerned with low frequency analysis below the PWM
switching frequency (typically 20kHz); fortunately, this is the case here.
Figure 4.4 shows the single update PWM mode [79]. Here, the switching process is
updated for each duty cycle. Based on the Figure 4-4, 𝑚𝑐 (𝑡) is the output signal
generated from the current controllers, 𝑚𝑠 (𝑡) is the modulating signal and 𝑐(𝑡) is the
carrier signal. The duty cycle 𝑑, is the fraction of time, which happens each time the
modulating signal is higher than the carrier signal. This can be expressed by the value
of a single update modulating signal in a cycle, 𝑚𝑠 𝑛𝑇𝑠 , divided by the peak of the
carrier signal, 𝑐𝑝𝑘 . The sample and hold effects need also be considered in the
modelling.
47
Figure 4-4: Single update PWM with triangular carrier waveform.
Equations below show the expression without and with the delay effects of sample
and hold [79].
𝑚𝑠 𝑛𝑇𝑠
𝑑= , 𝑛 = 1,2,3,4, ….
𝑐𝑝𝑘
(Eq. 4-19)
𝑇𝑠
𝑚𝑐 (𝑠) 1 − 𝑠 4
𝑑=
𝑐𝑝𝑘 1 + 𝑠 𝑇𝑠
4
(Eq. 4-20)
In order to produce the output voltage of the H-bridge inverter, which goes from −𝑉𝐷𝐶
to +𝑉𝐷𝐶 , it must depend on the PWM duty cycle. This is shown in equation below. In
48
order to simplify the analysis, it should be noted here that the voltage drop of the
switches is ignored.
𝑉𝑋 (𝑠) = 2𝑑𝑉𝐷𝐶
𝑇𝑠
𝑚𝑐 (𝑠) 1 − 𝑠 4
𝑉𝑋 (𝑠) = 2𝑉
𝑐𝑝𝑘 1 + 𝑠 𝑇𝑠 𝐷𝐶
4
(Eq. 4-21)
Re-arranging (Eq. 4-21), the output/input relationship of the PWM model is

expressed as:
𝑇𝑠
𝑉𝑋 (𝑠) 1 1−𝑠 4
𝐺𝑃𝑊𝑀 (𝑠) = =2 𝑉
𝑚𝑐 (𝑠) 𝑐𝑝𝑘 1 + 𝑠 𝑇𝑠 𝐷𝐶
4
(Eq. 4-22)
4.2.3 Modelling of Current Controllers
Chapter 3 has mentioned the transfer function for the current controllers used in this
research project. They are again expressed as below:
a) Proportional Integral (PI) Current Controller
The Laplace transfer function for the output/input relationship of PI current controller
is;
𝑚𝑐 (𝑠) 𝐾𝑃 𝑠 + 𝐾𝐼
𝐺𝑃𝐼 (𝑠) = =
𝜀(𝑠) 𝑠
𝐾𝑃 𝑠 + 𝐾𝐼
=
𝑠
(Eq. 4-23)
𝐾𝑃 is the proportional gain and 𝐾𝐼 is the integral gain of the controller system. Figure
4-5 shows the open loop Bode plot for the respective controller system. The chosen
value used to develop the graph is 0.017 for 𝐾𝑃 and 0.035 for 𝐾𝐼 . These values are
determined by a manual tuning approach in which the PI controller is initially
49
configured as a proportional only controller. The value of proportional gain is increased

until the impact of continuing to increase the controller gain diminishes. The integral
gain is then increased to reduce the steady state error. A small amount of fine tuning is
applied until the best response is observed. At this point, the controller gains are
recorded.
Figure 4-5: Open loop Bode plot for PI current controller scheme shows flat response
at 50 Hz indicating no harmonic rejection at fundamental frequency.
b) Proportional Resonant (PR) Current Controller
The Laplace transfer function for PR current controller is;
𝑚𝑐 (𝑠) 2𝐾𝑅 𝑠
𝐺𝑃𝑅 (𝑠) = = 𝐾𝑃 + 2
𝜀(𝑠) 𝑠 + 𝜔𝑜2
𝐾𝑃 (𝑠 2 + 𝜔𝑜2 ) + 2𝐾𝑅 𝑠
=
𝑠 2 + 𝜔𝑜2
𝐾𝑃 𝑠 2 + 2𝐾𝑅 𝑠 + 𝐾𝑃 𝜔𝑜2
=
𝑠 2 + 𝜔𝑜2
(Eq. 4-24)
50
However, as the ideal equation above is hard to implement in reality because of the
round off and quantization error in digital transformation, a non ideal controller is used
where a controller cut off frequency, 𝜔𝑐 is added to the system [58]. Furthermore, an
increased 𝜔𝑐 will ensure the system stability and yield to a less sensitive controller [80].
This then makes the equation as:
𝑚𝑐 (𝑠) 2𝐾𝑅 𝜔𝑐 𝑠
𝐺𝑃𝑅 (𝑠) = = 𝐾𝑃 + 2
𝜀(𝑠) 𝑠 + 2𝜔𝑐 𝑠 + 𝜔𝑜2
𝐾𝑃 (𝑠 2 + 2𝜔𝑐 𝑠 + 𝜔𝑜2 ) + 2𝐾𝑅 𝜔𝑐 𝑠

=
𝑠 2 + 2𝜔𝑐 𝑠 + 𝜔𝑜2
𝐾𝑃 𝑠 2 + (2𝐾𝑃 𝜔𝑐 + 2𝐾𝑅 𝜔𝑐 )𝑠 + 𝐾𝑃 𝜔𝑜2

=
𝑠 2 + 2𝜔𝑐 𝑠 + 𝜔𝑜2
(Eq. 4-25)
𝐾𝑃 is the controller proportional gain, 𝐾𝑅 is the controller resonance gain of the system
and 𝜔𝑜 on the other hand is equal to 2 times 𝜋 times the fundamental frequency which
is 315.159 rad/s.
Figure 4-6: Open loop Bode plot of PR current controller scheme shows high
magnitude at 50 Hz indicating good harmonic rejection at fundamental frequency.
The open loop Bode plot of the PR current controller without the cut off frequency 𝜔𝑐
is shown in Figure 4-6 with 0.046 and 580 for the 𝐾𝑃 and 𝐾𝑅 value respectively. Again,
51
the gain values are obtained in the simulation. Whereas the next following figure shows
the open loop Bode plot of the PR current controller with the 𝜔𝑐 added and its effect
if the value varies.
(a)
(b)
Figure 4-7: Open loop Bode plot of PR current controller scheme when
(a) 𝜔𝑐 =1: wider magnitude response at 50 Hz shows less sensitive controller,
and (b) 𝜔𝑐 =1, 5, 20, 50: sensitivity of the controller is reduced with increasing
value of 𝜔𝑐 .
52
c) PR+HC Current Controller
The PR+HC scheme is the combination of the PR controller scheme but with harmonic
compensators added. This depends on the number of harmonics that one needs to use
in order to reduce the harmonics of interest. In this project, three harmonic
compensators are modelled in the simulation; the 3rd harmonic order, the 5th harmonic
order, and the 7th harmonic order. This is because these lower order harmonics are the
most dominant harmonics and difficult to reduced. Therefore, the Laplace transfer
function for the PR+HC current controller scheme with added 𝜔𝑐 is:
𝑚𝑐 (𝑠)
𝐺𝑃𝑅+𝐻𝐶 (𝑠) =
𝜀(𝑠)
2𝐾𝑅 𝜔𝑐 𝑠 2𝐾𝑐3 𝜔𝑐3𝑠 2𝐾𝑐5 𝜔𝑐5 𝑠 2𝐾𝑐7 𝜔𝑐7 𝑠

= 𝐾𝑃 + + 2 + 2 + 2
𝑠2 + 2𝜔𝑐 𝑠 + 𝜔𝑜 𝑠 + 2𝜔𝑐3 𝑠 + 9𝜔𝑜 𝑠 + 2𝜔𝑐5 𝑠 + 25𝜔𝑜 𝑠 + 2𝜔𝑐7 𝑠 + 49𝜔𝑜2
2 2 2
(Eq. 4-26)
Figure 4-8: Open loop Bode plot of PR+HC current controller scheme with 𝝎𝒄 =10.
Based on (Eq. 4-26), 𝐾𝑃 is the proportional gain, 𝐾𝑅 is the resonant gain at fundamental
frequency, 𝐾𝑐3 is the resonant gain at three times the fundamental frequency, 𝐾𝑐5 is the
resonant gain at five times the fundamental frequency, and 𝐾𝑐7 is the resonant gain at
53
seven times the fundamental frequency. The chosen values for all the gains mentioned
are 0.046, 580, 70, 70, and 70 respectively which are obtained in the simulation earlier.
The overall open loop Bode plot for PR+HC current controller is as in Figure 4-8.
4.2.4 Model of Grid Connected, Single PV Inverter System
By cascading (Eq. 4-18) and (Eq. 4-22) above with the chosen current controller ((Eq.
4-23), (Eq. 4-25) or (Eq. 4-26)), the complete model of a grid connected, PV inverter
system can be developed. This is illustrated in Figure 4-9.
Figure 4-9: Model of grid connected inverter system with PI current controller.
From the figure, two input variables are observed which produce two different transfer
function responses. One is the response when 𝑉𝐺 (𝑠) is set to zero and the other one is
the response when 𝐼𝑟𝑒𝑓 (𝑠) is set to zero. The overall closed loop response of the grid
connected, single PV inverter system model is discussed further in the next sub-section.
4.3 Closed Loop Frequency Response
Earlier, two input variables are observed from the model in Figure 4-9. In order to get
the overall closed loop response of the system model, each input is set to zero value for
each time to develop two transfer functions. The mathematical expressions for the grid
connected, inverter model with PI current controller are explained next.
54
When 𝑉𝐺 (𝑠) = 0;
𝐼𝑜 (𝑠) 𝐺𝑃𝐼 (𝑠)𝐺𝑃𝑊𝑀 (𝑠)𝐺𝑉𝑋 (𝑠)𝐺𝐿𝐶𝐿 (𝑠)

=
𝐼𝑟𝑒𝑓 (𝑠) 1 + 𝐺𝑃𝐼 (𝑠)𝐺𝑃𝑊𝑀 (𝑠)𝐺𝑉𝑋 (𝑠)𝐺𝐿𝐶𝐿 (𝑠)
𝐺𝑃𝐼 (𝑠)𝐺𝑃𝑊𝑀 (𝑠)𝐺𝑉𝑋 (𝑠)𝐺𝐿𝐶𝐿 (𝑠)

𝐼𝑜 (𝑠) = 𝐼𝑟𝑒𝑓 (𝑠)
1 + 𝐺𝑃𝐼 (𝑠)𝐺𝑃𝑊𝑀 (𝑠)𝐺𝑉𝑋 (𝑠)𝐺𝐿𝐶𝐿 (𝑠)
(Eq. 4-27)
When 𝐼𝑟𝑒𝑓 (𝑠) = 0;
𝐼𝑜 (𝑠) 𝐺𝑉𝐺 (𝑠)𝐺𝐿𝐶𝐿 (𝑠)

=−
𝑉𝐺 (𝑠) 1 + 𝐺𝑃𝐼 (𝑠)𝐺𝑃𝑊𝑀 (𝑠)𝐺𝑉𝑋 (𝑠)𝐺𝐿𝐶𝐿 (𝑠)
𝐺𝑉𝐺 (𝑠)𝐺𝐿𝐶𝐿 (𝑠)

𝐼𝑜 (𝑠) = −𝑉𝐺 (𝑠)
1 + 𝐺𝑃𝐼 (𝑠)𝐺𝑃𝑊𝑀 (𝑠)𝐺𝑉𝑋 (𝑠)𝐺𝐿𝐶𝐿 (𝑠)
(Eq. 4-28)
By adding (Eq. 4-27) and (Eq. 4-28), the closed loop response can be obtained.
𝐼𝑜 (𝑠) = 𝐼𝑜 (𝑠)|𝑉𝐺(𝑠)=0 + 𝐼𝑜 (𝑠)|𝐼𝑟𝑒𝑓 (𝑠)=0
𝐺𝑃𝐼 (𝑠)𝐺𝑃𝑊𝑀 (𝑠)𝐺𝑉𝑋 (𝑠)𝐺𝐿𝐶𝐿 (𝑠) 𝐺𝑉𝐺 (𝑠)𝐺𝐿𝐶𝐿 (𝑠)

𝐼𝑜 (𝑠) = 𝐼𝑟𝑒𝑓 (𝑠) − 𝑉𝐺 (𝑠)
1 + 𝐺𝑃𝐼 (𝑠)𝐺𝑃𝑊𝑀 (𝑠)𝐺𝑉𝑋 (𝑠)𝐺𝐿𝐶𝐿 (𝑠) 1 + 𝐺𝑃𝐼 (𝑠)𝐺𝑃𝑊𝑀 (𝑠)𝐺𝑉𝑋 (𝑠)𝐺𝐿𝐶𝐿 (𝑠)
(Eq. 4-29)
Re-arranging (Eq. 4-29),
𝑉𝐺 (𝑠)
𝐺𝑃𝐼 (𝑠)𝐺𝑃𝑊𝑀 (𝑠)𝐺𝑉𝑋 (𝑠)𝐺𝐿𝐶𝐿 (𝑠) − [𝐺 (𝑠)𝐺𝐿𝐶𝐿 (𝑠)]
𝐼𝑜 (𝑠) 𝐼𝑟𝑒𝑓 (𝑠) 𝑉𝐺
𝐺𝐶𝐿𝑃𝐼 (𝑠) = =
𝐼𝑟𝑒𝑓 (𝑠) 1 + 𝐺𝑃𝐼 (𝑠)𝐺𝑃𝑊𝑀 (𝑠)𝐺𝑉𝑋 (𝑠)𝐺𝐿𝐶𝐿 (𝑠)
(Eq. 4-30)
55
Using the same steps, the closed loop response for the grid connected, inverter model
with PR and PR+HC current controller can be obtained as in (Eq. 4-31) and (Eq. 4-32).
𝑉𝐺 (𝑠)
𝐺𝑃𝑅 (𝑠)𝐺𝑃𝑊𝑀 (𝑠)𝐺𝑉𝑋 (𝑠)𝐺𝐿𝐶𝐿 (𝑠) − [𝐺 (𝑠)𝐺𝐿𝐶𝐿 (𝑠)]
𝐺𝐶𝐿𝑃𝑅 (𝑠) = =
𝐼𝑟𝑒𝑓 (𝑠) 1 + 𝐺𝑃𝑅 (𝑠)𝐺𝑃𝑊𝑀 (𝑠)𝐺𝑉𝑋 (𝑠)𝐺𝐿𝐶𝐿 (𝑠)
(Eq. 4-31)
𝑉𝐺 (𝑠)
𝐺𝑃𝑅+𝐻𝐶 (𝑠)𝐺𝑃𝑊𝑀 (𝑠)𝐺𝑉𝑋 (𝑠)𝐺𝐿𝐶𝐿 (𝑠) − [𝐺 (𝑠)𝐺𝐿𝐶𝐿 (𝑠)]
𝐺𝐶𝐿𝑃𝑅+𝐻𝐶 (𝑠) = =
𝐼𝑟𝑒𝑓 (𝑠) 1 + 𝐺𝑃𝑅+𝐻𝐶3 (𝑠)𝐺𝑃𝑊𝑀 (𝑠)𝐺𝑉𝑋 (𝑠)𝐺𝐿𝐶𝐿 (𝑠)
(Eq. 4-32)
Table 4.1 shows the list of parameters used in the system. These values are typical of a
commercial scale single phase inverter system, and have been successfully applied in
previous work [ref Hong]. For consistency, they are applied here also for the purposes
of the Simulink simulation.
56
Table 4-1: List of parameters of grid connected model.
Parameter Value
𝑉𝐷𝐶 400 V
𝐶𝑝𝑘 2
𝑇𝑠 50 µs
𝐿𝑓 1.6 mH
𝑅𝐿𝑓 0.15 Ω
𝐶𝑓 12 µF
𝑅𝐶𝑓 0.0566 Ω
𝐿𝑔 0.05 mH
𝑅𝑔 0.1 Ω
𝑉𝑔 240√2 Vpeak
𝐼𝑟𝑒𝑓 20 A
𝐾𝑃 (PI) 0.017
𝐾𝐼 0.035
𝐾𝑃 (PR and PR+HC) 0.046
𝐾𝑅 580
HC3, HC5, HC7 70
The following three figures below the Bode plot of the closed loop system response
with PI, PR and PR+HC current controller together with its root locus.
57
(a)
(b)
Figure 4-10: Closed loop response of the system with PI current controller. (a) Bode
plot shows the attenuation of the low order harmonics, and controller resonant point
at approximately 7 kHz, and (b) Root locus confirms the stability of the closed loop
controller for the applied gains.
58
(a)
(b)
Figure 4-11: Closed loop response of the system with PR current controller. (a) Bode
plot shows improved attenuation of low order harmonics, and lower impact of
resonance (now at 10 kHz), and (b) Root locus confirms the stability of the closed
loop controller for the applied gains.
59
(a)
(b)
Figure 4-12: Close loop response of the system with PR+HC current controller.
(a) Bode plot shows attenuation of low order harmonics, and further improvement of
the resonant effect at 10 kHz, and (b) Root locus confirms the stability of the closed
loop controller for the applied gains.
60
4.4 Discussion
Analysis of the PI controller bode plot shows that output gain of the controller is fairly
uniform across the frequency range. For the PR controller, it is apparent that the
controller can be tuned so that the control system has low impedance to the fundamental
frequency of interest; in the case of a grid connected inverter system this is the network
frequency – typically 50/60 Hz. This is one of the reasons that PR controller are
becoming increasingly popular in distributed generation systems; it is well suited for
readily injecting mains frequency signals into the grid, whilst presenting a higher
impedance to other frequencies such as harmonics where it is preferential to minimise
these components. The PR+HC takes this one step further, This controller includes
additional harmonic compensation terms, which are tuned to specific harmonics which
are desirable to mitigate. The example in this chapter has shown a typical application,
which might be seen in a grid connected inverter system, where the HC terms are
targeted towards the low order harmonics of the system. These are particularly
troublesome in grid connected applications, since they are close to the fundamental
frequency of the system, and thus immensely difficult to remove via traditional passive
low pass filters. A typical passive low pass filter in a grid connected inverter might be
designed around a cut of frequency in the range of 500 Hz – 1500 Hz, therefore 3rd, 5th,
7th harmonic components are particularly troublesome. In the Bode of plot of the PR-
HC, it is clear that the impedance to these components has been significantly increased
(shown by the three “spikes” in the bode plot).
4.5 Chapter Summary
This chapter has explained the modelling process of a grid connected single inverter
system employing three different types of controller. The approach to the modelling
has been fully described, with full details on the development of the closed loop
Laplace transfer function of each system. Based on an example set of gains used in
each controller, it is clear to see that a robust control solution can be developed with
each method. The key attributes of each controller have been shown via the frequency
domain responses, and discussed. Whilst the PI controller produces a robust solution,
the PR controller produces a solution which is better optimised to the injection of the
fundamental frequency into the grid. Troublesome low order harmonic components can
be mitigated further by the introduction of harmonic compensation (HC) terms. Care
61
must be taken when introducing these terms, since each adds complexity to the control
structure and will require DSP resources when applied in practice.
62
Simulations Of Grid Connected Inverter System With PI, PR, and Modified PR Controller
Chapter 5: Simulations of Grid-Connected Inverter System with

PI, PR and Modified PR Control
Simulations are the assumptions of representations of the real model. Before any
experimental hardware is developed, simulations are normally done in order to test the
effectiveness of the proposed idea. Hence, the parameters used in the simulation
process will be referred to in the future when developing the experimental hardware.
In this chapter, simulations of a grid connected system are done using three control
techniques that are the PI control, PR control and the modified PR control. At first, a
single inverter system is simulated and followed by two inverter systems in parallel and
connected to the same point in the supply grid.
5.2 Simulation of a single grid connected inverter

Before parallel inverter systems are simulated, a single connected inverter system is
modelled initially using Matlab as the simulation tool. It comprises of PWM circuit, an
H bridge inverter that consists of 4 mosfets with internal diodes, a low pass filter,
current controller and the supply grid with grid impedance. The low pass filter consists
of an inductance and a capacitor with values as listed in Table 5-1. It is designed so
that the magnitude of frequencies lower than approximately of 1 kHz can pass through
and magnitude of the frequencies higher than an approximate of 1 kHz is filtered. This
is to prevent the unwanted switching frequency harmonics being injected to the grid.
63
Figure 5-1: Simulink model of grid-connected inverter system.
Figure 5-1 shows a single inverter system model. The parameters are listed in Table
5-1.
64
Table 5-1: System parameters.
DC Voltage Source 400 V

Frequency 50 Hz
Filter Inductance and Resistance L=1.6 mH, R=0.15 Ω
Filter Capacitance and Resistance C=12 µF, R=0.0566 Ω
Filter Cut-Off Frequency 1148 Hz
Grid Voltage 240 𝑉𝑅𝑀𝑆
Grid Impedance R=0.1 Ω
L=0.15 mH
Figure 5-2 is the grid model of 240 𝑉𝑅𝑀𝑆 which contains different magnitude and phase
angle for each particular frequency of up to the 50th. In total, this will create an
approximate background THD of 2.45% in the grid supply voltage. The harmonic data
is based on experimental data acquired by a previous student in the Power Electronics,
Drives and Machines laboratory at Newcastle University [44]. This data is readily
available for simulation use by the research group.
Figure 5-2: Grid model which contains a background THD of 2.45% [44].
65
Table 5-2: Grid model harmonic data [44].
Freq Amplitude Freq Amplitude Phase

(Hz) (RMS) Phase (deg) (Hz) (RMS) (deg)
50 241.72 320.29 1050 0.05 265.06
100 0.07 340.9 1100 0.01 277.76
150 3.56 90.01 1150 0.12 281.76
200 0.02 296.59 1200 0.01 49.23
250 3.45 98.5 1250 0.13 285.01
300 0.01 326.46 1300 0.01 151.49
350 2.45 253.58 1350 0.15 105.57
400 0.01 33.74 1400 0.01 103.03
450 1.09 303.98 1450 0.03 295.9
500 0.01 14.62 1500 0.01 98.5
550 0.5 91.26 1550 0.08 157.67
600 0 168.4 1600 0.01 175.79
650 1.37 13.7 1650 0.02 147.72
700 0.01 43.99 1700 0 335.9
750 0.73 313.33 1750 0.07 310.1
800 0.01 112.09 1800 0 204.46
850 0.7 350.93 1850 0.01 283.66
900 0.01 208.84 1900 0.01 20.7
950 0.2 167.34 1950 0.02 144.62
1000 0.01 316.74 2000 0 272.65
As there are three different digital current control methods used in this project, each is
tested independently. The first simulation uses the conventional PI current controller,
followed by the PR+HC and the modified PR+HC. The reference current is chosen to
be 20 A peak as the system is assumed to work in a perfect sunny day with 3 kW output
power. The simulation model samples the currents and voltages every 50 us. In
addition, the PWM output is updated at the same rate. This simulates the basic
operation of a real digital control system.
Such a set of conditions assumes close to full power operating conditions. Under these
circumstances, the modulation index within the PWM is set to be very high. It is well
known that the harmonic performance of grid connected inverters is typically best at
full power, and that low modulation index operation can result in in greater harmonic
distortion. This is because at low modulation indices, dead-time is a greater proportion
of the pulse width. Therefore, the impact on low order harmonic distortion is greater.
66
Good quality dead-time compensation can mitigate the impact of this to some extent,
but will not complete eliminate the phenomena. In a grid connected inverter PV system,
such a scenario can potentially occur when there is a prolonged period of low
irradiation. It is important to recognise, however, that such characteristics will apply
regardless of the control strategy chosen. In most of the results presented in this thesis,
a high modulation index is observed in the PWM unit, implying that the inverter is
typically operating at high/full power conditions which is typically desirable where
possible.
5.2.1 Conventional PI control technique
Figure 5-3 shows the PI controller model used in the simulation. The measured inverter
output current will be In1. It is then compared with the 𝐼𝑟𝑒𝑓 , resulting an error which is
used in the control process. The output of the current controller is then compared with
a 20 kHz triangular wave signal to generate the PWM switching pattern to control the
state of the H-Bridge power converter. Both gains; proportional gain and integral gain
are tuned using a simple a manual tuning method until the output current matches the
reference signal. Further fine tuning is then carried out to achieve an optimised current
output, with good power quality (low harmonics). In this simulation, the gains used are
0.03 for 𝐾𝑃 and 0.03 for 𝐾𝐼 . The model is run for 1.0 s. Following this, FFT analysis is
performed and the harmonic content of the output current waveform is recorded, up to
the 20th harmonic (1 kHz). Above this, the harmonics are significantly attenuated by
the low pass filter. However, the overall THD of the current waveform is also recorded
as an additional measure of power quality.
67
Figure 5-3: Conventional PI current controller digital Simulink model.
Figure 5-4 in the next page shows the screen shot of the reference current signal, the
inverter output current signal, and the observed current error signal applied to the PI
controller.
68
Current (A)
Time (s)
a)
Current (A)
Time (s)
b)
Current (A)
Time (s)
c)
Figure 5-4: Screen shot of (a) Reference current of PI controller with 20 A peak
value, (b) Inverter output current of PI controller which shows a distorted waveform,
and (c) Large magnitude error between reference current and inverter output current
of PI controller.
With the chosen 𝐾𝑃 and 𝐾𝐼 value, when the model is run, the error between both the
inverter output current and reference current is the smallest which is approximately
1.5 A peak. The harmonic spectrum for the inverter output current is measured using
FFT Analysis. The sample taken for the analysis is 10 cycles starting from 0.6 s to 0.8 s.
The low order harmonic data is then exported to Excel and post processed to produce
the result illustrated in Figure 5-5 .
69
0.8
0.7
Mag (% of fundamental)
0.6
0.5
0.4
0.3
0.2
0.1
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Harmonic Number
Figure 5-5: Low order harmonic spectrum of inverter output current with PI control
(𝐾𝑃 = 0.03 , 𝐾𝐼 = 0.03)
The inverter output current shows a THD of 2.06 %, with high harmonics
appear between the 3rd and 17th order. These low order harmonics are the prominent
harmonics which are the focus to be eliminated or reduced in this research work. As
mentioned previously, higher order harmonics beyond the 20th are less severe due to
the 1 kHz cut off frequency of the low pass filter.
5.2.2 Proportional resonance (PR) control technique

Next, the same simulation model of a single inverter grid connected system is once
again simulated. This time, instead of using the PI current control technique, the
proportional resonance (PR) current control technique is implemented. Figure 5-6
shows the controller model of the PR technique. In2 is the measured inverter output
current whilst Out2 is the output signal after the controlling process that will be used
for PWM switching purposes. Based on Figure 5-6, several coefficients need to be
determined carefully in targeting a reduction or elimination of the low order harmonics
thus achieving a low THD value of the inverter output current. Chapter 3 has shown
the equation for 𝑏0 , 𝑏1 , 𝑏2 , 𝑎1 and 𝑎2 . For ease of simulation, the denominator from the
equation is strategically replaced with a constant value for the controller, const. This
leaves the 𝑏0 , 𝑏1 , 𝑏2 , 𝑎1 and 𝑎2 as one single value with only the proportional gain, 𝐾𝑃
and the resonance gain, 𝐾𝑅 to be tuned. Therefore
70
1
𝑐𝑜𝑛𝑠𝑡 =
4 + 𝑇 2 𝜔𝑜 2
𝑏𝑜 = (4 + 𝑇 2 𝜔𝑜 2 )𝐾𝑃 + 2𝑇𝐾𝑅
𝑏1 = (2𝑇 2 𝜔𝑜 2 + 8)𝐾𝑃
𝑏2 = (4 + 𝑇 2 𝜔𝑜 2 )𝐾𝑃 − 2𝑇𝐾𝑅
2𝑇 2 𝜔𝑜 2 − 8
𝑎1 =
4 + 𝑇 2 𝜔𝑜 2
𝑎2 = 1
By inserting 50 µs for 𝑇 (sampling time) and 2π times the fundamental

frequency (50 Hz) for 𝜔𝑜 , all values, with the exception of 𝐾𝑃 and 𝐾𝑅 are obtained.
Then, 𝐾𝑃 and 𝐾𝑅 need to be tuned so that the lowest inverter output current THD is
achieved. Following a manual tuning method, the optimum output result is achieved
when 𝐾𝑃 is 0.04 and 𝐾𝑅 is 700.
Figure 5-6: Simulink model of current controller using PR technique.
The reference current, measured inverter output current and the error can be seen in
Figure 5-7. Based on the figure, the error current for the system using the PR control
71
technique is minimised when compared with the error current for the system using the
PI control technique (Figure 5-4). This shows an advantage of the PR control system.
Current (A)
Time (s)
a)
Current (A)
Time (s)
b)
Current (A)
Time (s)
c)
Figure 5-7: Screen shot of (a) Reference current of PR controller with 20 A peak
value, (b) Inverter output current of PR controller which shows smoother waveform
compared to PI controller, and (c) Reduced error between reference current and
inverter output current of PR controller.
72
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Harmonic order
Figure 5-8: Low order harmonic spectrum of inverter output current with PR current
controller.
Figure 5-8 shows the low order harmonic spectrum of the inverter output current after
the data from FFT analysis in Simulink has been transferred to Excel. When run in the
same length time as the system using PI current controller and sample time taken for
the analysis made is also the same as the system using PI current controller, the THD
achieved is 1.80% which is a reduction of approximately 13% from using the
conventional PI control technique.
73
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Harmonic order
PI (THD=2.06%) PR (THD=1.80%)
Figure 5-9: Comparison in the harmonic profile of PI and PR current controllers

shows better reduction of harmonics magnitude with PR current controller.
Figure 5-9 illustrates the spectrums of Figure 5-5 and Figure 5-8 when they are set in a
single bar graph. Besides the improvement in the THD, Figure 5-9 clearly shows that
the harmonics magnitude between the 3rd and the 17th harmonic orders have also been
reduced except for the 11th and 19th harmonic order. However, when observing the
magnitude of the 11th harmonic, it is not a big concern here as it is the similar range of
magnitude value that is between 0.1 and 0.2.
5.2.3 Modified PR control technique

As a novel approach, a slight modification is made to the proportional resonance (PR)
current controller parameter. Instead of using the controller model as in Figure 5-6, the
model as in Figure 5-10 is used in the simulation.
74
Figure 5-10: Simulink model of current controller using the modified PR control
technique.
75
Based on Figure 5-10, it can be seen that the terms 𝑏𝑜 , 𝑏1 , and 𝑏2 that are used in the
previous PR controller technique has been divided into several other fraction. The main
thing to know is that all values in the 𝑏𝑜 , 𝑏1 , and 𝑏2 terms remain unchanged except for
the proportional gain, 𝐾𝑃 . This is because, the proportional part for this novel approach,
𝑅𝐾𝑃 is tuned randomly within a set limit. As illustrated in Figure 5-10, a random
number block is used and added to the proportional gain, 𝐾𝑃 , to become a new
randomized proportional gain, 𝑅𝐾𝑃 for the controller process; a gain that varies within
𝐾𝑃 ± 0.001. The limit is chosen so that the THD as well as the harmonic spectrum of
the inverter output current is in acceptable value and shape. Similar to the system with
PR control technique, the value for 𝐾𝑃 is 0.04 and 𝐾𝑅 is 700. Below are the screen shots
of the random signal generation.
a)
b)
Figure 5-11: Screen shot of (a) Random signal generation of the modified PR
controller, and (b) The randomized proportional gain, 𝑅𝐾𝑃 signal which varies within
a set limit.
If a fixed gain is used as for the system with PR current controller, the signal for 𝐾𝑃 as
in Figure 5-11(b) is maintained at 0.04 at all times. However, with this technique, the
76
proportional signal is randomly varies between 0.0395 to 0.0405. Beyond this limit, the
signal of the inverter output current becomes unstable.
After the simulation of the modified control technique is run, FFT analysis of the
inverter output current is once again captured and transferred using Excel. Result can
be observed in Figure 5-12. From the figure, some addition and cancellation have
occurred to the current harmonic profile as the effect of the randomly varying
proportional gain. Most importantly, this addition and cancellation have further
reduced the grid current THD from 1.80% to 1.78%. Though it is small, it is a good
result where it clearly proves and shows that the modified PR control technique by
adjusting the proportional gain randomly within a range that has been set is one
important contribution to the knowledge and as a novel current controller for grid
connected inverters. There are no extra circuits that need to be added thus no extra
burden to the system when using this new technique.
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Conventional PR Modified PR
Figure 5-12: Harmonic profile of inverter output current with modified PR current
controller shows better outcome compared to the conventional PR current controller.
5.3 Simulation of two grid connected inverters in parallel

The next step is to simulate two inverters in parallel with the grid. This is to prove that
with the addition and cancellation of the harmonic trend, the proposed technique can
further improved the THD of grid current. The simulation model of the parallel inverter
system can be seen in Figure 5-13. The model is first run using the conventional PI
77
current controller and then followed by the proportional resonance (PR) current
controller and the modified proportional resonance current controller. Details of the
results are explained next.
Figure 5-13: Parallel inverter system model.
78
5.3.1 Conventional PI Current Controller

In the simulation, all parameters for the electronic components in both inverter systems
remain the same as single inverter system simulation. However, the current controller
gains are adjusted by a small amount to optimise the output inverter current. Based on
Kirchoff`s Current Law, the current injected into the grid is the summation of inverter
1 output current and inverter 2 output current. Therefore, the the grid current is 40 A
peak. With greater grid current and fixed grid voltage, the total output power to the
supply grid is increased.
In the following results, the simulation model is run for 1.0 s. The time domain
waveforms of the signals are captured and downloaded to a personal computer. From
this, waveforms and low order harmonic spectra of each individual inverter, and the
grid, are post processed using excel. Figure 5-14 shows the error signal between the
reference and the actual inverter current for both inverter systems. The output signals
for the inverters and the grid together with the low order harmonic profiles are
illustrated in Figure 5-15 and Figure 5-16.
(a)
(b)
Figure 5-14: Current error signal of (a) Inverter 1, and (b) Inverter 2, with PI current
controller which shows big magnitude errors.
79
(a)
(b)
(c)
and (c) Grid current with PI controller.
From figure 5-15, it can be seen that:
𝐼𝑔𝑟𝑖𝑑 = 𝐼𝑖𝑛𝑣1 + 𝐼𝑖𝑛𝑣2
= 20 A peak + 20 A peak
= 40 A peak
The low order harmonic magnitude for each inverter current, and overall grid current,
is measured using FFT analysis. 10 cycles are taken as a sample for the analysis,
between 0.6 s to 0.8 s. Results are shown in figure 5-16 where the harmonic profiles
show significant harmonic emissions between the 3rd and 19th order. It is also apparent
that a very similar harmonic spectrum exists in each profile.
80
1.2
1
0.8
0.6
0.4
0.2
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Harmonic order
Inv 1 (Thd = 3.38%)
(a)
1.2
0.8
0.6
0.4
0.2
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Harmonic order
Inv 2 (Thd = 3.36%)
(b)
81
1.2
1
0.8
0.6
0.4
0.2
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Harmonic order
Grid (Thd = 3.38%)
(c)
Figure 5-16: Harmonic profiles of (a) Inverter 1 output current, (b) Inverter 2 output
current, and (c) Grid current. All shows significant harmonic emissions between the
3rd and 19th harmonic orders.
1.2
1
0.8
0.6
0.4
0.2
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Harmonic order
Inv 1 Inv 2 Grid
Figure 5-17: Harmonic profiles of inverter 1, inverter 2, and grid current with PI
current controller which shows similar harmonic spectrum exists except for the 15th,
17th and 19th order.
82
5.3.2 Proportional resonance (PR) control technique

The next simulation is to run the parallel inverter system using the proportional
resonance (PR) control technique. In the simulation model, the parameters for the
proportional gain, 𝐾𝑃 and the resonant gain, 𝐾𝑅 of the controller system are again
slightly re-tuned from the single inverter settings to ensure optimum performance is
achieved. All waveforms and results are recorded and post processed in Excel. Figure
5-18 shows the error signal between the reference and the actual inverter current for
both systems.
(a)
(b)
Figure 5-18: The error signal between reference and the actual current of
(a) Inverter 1, and (b) Inverter 2 with PR current controller.
The output current waveforms for each inverter and overall grid current, together with
the low order harmonic profiles, can be seen in Figure 5-19.
83
(a)
(b)
(c)
and (c) Grid current with PR current controller.
The low harmonic orders for each inverter current and grid current is again measured
using FFT analysis and 10 cycles are taken as a sample for the analysis, starting from
0.6 s to 0.8 s. The graphs are then transferred to Excel as shown in Figure 5-20.
84
0.8
Mag (% of fundamental) 0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Harmonic order
Inv 1 (Thd = 3.27%)
(a)
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Harmonic order
Inv 2 (Thd = 3.30%)
(b)
85
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Harmonic order
Grid (Thd = 3.27%)
(c)
Figure 5-20: Harmonic profiles with the respective per unit magnitude of
(a) Inverter 1 output current, (b) Inverter 2 output current, and (c) Grid current.
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Harmonic order
Inv 1 Inv 2 Grid
Figure 5-21: Harmonic profiles of Inverter 1, Inverter 2, and Grid current with PR
controller in one graph shows harmonics addition and cancellation occurred.
86
From Figure 5-21, it can be clearly seen that there is some addition as well as
cancellation of the harmonic magnitude. For instance, the 3rd and 5th harmonic order
magnitudes of inverter 1 output current are 0.34 and 0.59 respectively but the same
harmonic order magnitudes of inverter 2 output current are 0.35 and 0.58 respectively.
The effect of this controller technique can also be observed on all the low harmonic
orders. For that reason, the harmonic profile of the grid current is also affected resulting
in an improvement to the grid current THD that is 3.27%; a reduction of approximately
3.25% from the previous parallel inverters simulation when using the conventional PI
current controller.
5.3.3 Modified PR Control Technique

For this simulation, the modified proportional resonance control technique is applied.
In the simulation model, the parameters for the proportional gain, 𝐾𝑃 and the resonant
gain, 𝐾𝑅 of the controller system remain the same as the value used in the second
technique. However, the proportional gain is randomly adjusted within a set limit, as
explained in section 5.2.3. The model is again run for 1.0 s and all waveforms and
results are recorded once again. Figure 5-22 shows the error signal between the
reference and the actual inverter current for both inverter systems.
(a)
(b)
Figure 5-22: The error signal between reference and the actual current of
(a) Inverter 1, and (b) Inverter 2 with modified PR current controller.
87
The output current waveforms for each inverter, and the grid current, are shown in
Figure 5-23.
(a)
(b)
(c)
and (c) Grid current with modified PR current controller.
The low harmonic orders for each inverter current and grid current is again measured
using FFT Analysis. 10 cycles are taken as a window for the analysis, between 0.6 s to
0.8 s. The data is then transferred to Excel, as shown in Figure 5-24.
88
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Harmonic order
Inv 1 (Thd = 3.06%)
(a)
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Harmonic order
Inv 2 (Thd = 3.05%)
(b)
89
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Harmonic order
Grid (Thd = 3.03%)
(c)
Figure 5-24: Harmonic profiles with the respective per unit magnitude of (a) Inverter
1 output current, (b) Inverter 2 output current, and (c) Grid current.
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Harmonic order
Inv 1 Inv 2 Grid
Figure 5-25: Harmonic profiles of inverter 1, inverter 2, and grid current with
modified PR controller shows harmonic cancellation and addition leading to a further
THD reduction compared to the PR current controller technique.
The grid current THD observed is 3.03% which shows a further improvement
compared to the previous simulation when using the proportional resonance controller
90
technique alone. This also proves that the latter technique works even better when two
inverters are connected in parallel to the grid. This is because when a single inverter
system is run using the PR and the modified PR control technique which involves the
randomized gain in the controller system, the THD improvement observed is
approximately 1.1%. However, when two inverter systems are run in parallel using the
same techniques, the THD improvement observed is approximately 7.3%. The next
following figure compares the harmonic spectrum of the grid current when each
controller technique is applied to the parallel inverter system.
1.2
0.8
0.6
0.4
0.2
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Harmonic order
Grid PI (Thd = 3.36%) Grid PR (Thd = 3.27%) Grid Mod PR (Thd = 3.03%)
Figure 5-26: Current harmonic profile using PI, PR and modified PR controllers.
THD of grid current with modified PR controller shows the lowest value.
From Figure 5-26, it is noticed that a reduction in the overall harmonic content occurs
when using the proportional resonance and modified proportional resonance technique,
compared to using the conventional PI control technique. When a random signal is
added to the proportional gain of the PR controller, a random interaction occurs
between the inverters, resulted in a time-varying harmonic spectrum. The simulation
results provide a basis for verifying the control technique experimentally.
91
5.4 Chapter Summary

This chapter compares simulation results for a single, and parallel, operated grid
connected inverter system. The impact of three different control schemes have been
compared; conventional PI, conventional PR and random parameter variation PR.
Results show that in a single inverter system, a THD reduction by 1.1% in the grid
current can be achieved with the random parameter PR scheme. Further improvement
is observed in the parallel connected inverter system, where a significant 7.3% THD
reduction is observed. The research in this chapter has shown for the first time that, in
simulation, it is possible to reduce the low order harmonic content of a parallel
connected inverter system by utilising a random parameter PR control scheme.
92
Experimental Results of Stand Alone Inverter
Chapter 6: Experimental Results of Stand Alone Inverter
This chapter starts by describing the software and hardware used in this research
project. Following this, experimental results from applying the three different control
methods described in Chapter 4 to a stand-alone inverter system is shown. Finally, a
comparison of the results is presented, and conclusions are drawn.
6.2 General Hardware Description

Different tools i.e. the software and hardware are used for the purpose of this research
project. The software includes the Multisim and Ultiboard for designing the circuits
and the layout boards as well as the Code Composer Studio (CCS) for programming.
In total, five PCB boards need to be designed and made. They are the inverter board,
the current measurement board, the voltage measurement board, the gate driver board
and the DSP interface board. The screen shot of the boards can be seen in Appendixes.
To cut a bit of time for this research project, only the inverter board need to be designed
in the Multisim and Ultiboard. All the other four boards were designed by a previous
student who had the same area of research and they are allowed to be used in this
project.
6.2.1 Inverter Board
The power inverter board uses four 100 V TO-220 case style Mosfets. The inverter
board is then connected to a low pass output filter. The schematic diagram of the power
inverter board and filter is shown in Figure 6-1.
93
Galvanic
Lf /2 Isolator
L
T1 T3 O
A
Cf D
VDC C2
/
C1
G
R
T2 T4 I
D
Lf /2
PWM1 PWM2 PWM3 PWM4
GATE DRIVER
&
TMS320F2812
DSP
I ref
Figure 6-1: Schematic diagram of power inverter board.
6.2.2 TMS320F2812 Digital Signal Processor
A Texas Instruments ezDSP, TMS320F2812, processor board is used for this work.
This system has four major interface blocks; the parallel port controller interface, the
external JTAG interface, the analogue expansion interface and the I/O expansion
interface. The block diagram of the eZdspTM F2812 chip can be seen in Figure 6-2 [81].
The description of all the blocks functions are as follows:
a) Parallel port controller interface – connects to the communication environment

such as Code Composer Studio (CCS).
b) External JTAG interface – the emulators interface to the DSPs.
c) Analogue expansion interface – connects to the analogue input for digital conversion
purposes.
d) I/O expansion interface – used to get the input and output signals from the DSP.
94
Figure 6-2: Block diagram of the eZdspTM F2812 [81].
6.3 PI Experiments and Results
As stated in the earlier chapter, this research is concerned with two inverter systems
operating in parallel. However, before any further experiments are done, each inverter
system need to be tested with a resistive load at the output. This is also known as a
stand-alone inverter where no connection with the grid is made. Initially, one of the
inverters (Inverter 1) is tested with a conventional PI current control loop. Output
waveforms are taken from a normal oscilloscope and the FFT of the harmonic current
is taken from a power analyzer which is then transferred to Excel to get the graph. All
tests in this stand-alone mode used a 25 V DC supply with a 3 A peak current demand
and 5 Ohms resistive load at the end. In order to get the best proportional and integral
gains for the system, the software is run with different gain values each time. The
purpose is to observe which gain values can get the best output waveform with the best
total harmonic distortion (THD) value. Using trial and error method, the optimum value
for the proportional gain 𝐾𝑃 is 1.6 and the optimum value for the integral gain 𝐾𝐼 is
0.275. Figure 6-3 and Figure 6-4 are the screen shots of the current output waveform
and the respective FFT for the gains stated with the scope set to 1 A per division.
95
Figure 6-3: Screen shot of current output waveform.
2.5
1.5
0.5
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Harmonic order
PI (THD=3.186% _ I=2.993Ap)
Figure 6-4: FFT of output current.
6.3.1 Dead time Effect and Its Compensation
From Figure 6-3, it can be seen that whenever the current crosses the zero line, a small
distortion occurred which is known as the PWM dead time effect. Dead time is a short
time delay that is applied in the PWM switching to avoid the switches from turning on
or off at the same time. It is discussed briefly in Appendix. The effect of dead time is
96
normally seen at the output waveform as can be seen above. To compensate the effect
of dead time to the output results, an adjustment in the software is made. This is done
by adding a particular value based on the dead time value used in the PWM switching.
In this project a dead time used in the PWM switching is 1.7 µs which is around 3% of
the switching time. For the purpose of the compensation, this time delay in per unit
need to be multiplied by the maximum magnitude of the modulating signal and then be
added to the current controller output. To observe the effect of the dead time
compensation to the output waveforms, a test is done with a variable dead time
compensation unit of 0.00, 0.01, 0.02, and 0.03 using the same 𝐾𝑃 and 𝐾𝐼 gain for all
tests. Below are the screen shot of the current output waveforms when each of the dead
time compensation unit is used in the test.
(a) dead time compensation unit of 0.00.
97
(b) dead time compensation unit of 0.01.
(c) dead time compensation unit of 0.02.
98
(d) dead time compensation unit of 0.03.
Figure 6-5: Screen shot of the output current waveform with (a) dead time 0.00,
(b) dead time 0.01, (c) dead time 0.02, and (d) dead time 0.03.
2.500
2.000
% of fundamental
1.500
1.000
0.500
0.000
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Harmonic order
dt comp unit 0.00 (THD=3.19%) dt comp unit 0.01 (THD=3.02%)
dt comp unit 0.02 (THD=2.89%) dt comp unit 0.03 (THD=2.9%)
Figure 6-6: Output current FFT for different dead time compensation unit.
Comparing the output waveforms shown in Figure 6-5 above, it is apparent that the
output waveform with a dead time compensation unit of 0.03 gives the best waveform
of the four. It is clearly seen when the output current reaches zero value where the
99
distortion is much reduced. Following the figures is the FFT of all four states in one
graph for easy comparison. The observed reduction of the distorted waveforms in
Figure 6-5 is agreed by the finding results from the FFT analysis shown. As the dead
time compensation unit is increased by 0.01, the THD of the output current can be
decreased. In addition, the 3rd and the 5th harmonic orders which is known as the hardest
order harmonics magnitude to be reduced is reduced in this case. Although the output
current THD when using the dead time compensation unit of 0.03 is slightly bigger by
0.01% than output current THD when using the dead time compensation unit of 0.02,
the 3rd as well as the 5th harmonic orders of the former are well reduced. These findings
conclude that the dead time compensation unit of 0.03 will be used in other test of PI
current controller later on.
6.3.2 Zero Crossing Detection (ZCD)
In real life, the inverter system is connected to the supply grid to support users demand.
For that reason, the current injected from the inverter system must be in phase and
synchronised with the current in the supply grid. Moreover, in order to connect several
inverter systems in parallel to the same point in the supply grid, the same condition
must be met. This is achieved by applying a zero crossing detection (ZCD) in the
overall system. For the purpose of completing this research project, a readily available
ZCD circuit is used. This circuit consists of step down transformer (240V: 9V); to
reduce the grid voltage to a significantly lower voltage to interface with digital interface
electronics. This interface comprises of a low pass filter; to filter out noise that can
cause false reading of the ZCD, a transceiver; to create rectangular pulse, and a
comparator; to compare the filtered voltage with the DC offset. Using an oscilloscope,
the output waveform from the ZCD circuit together with the inverter output current is
observed and can be seen in Figure 6-7. This output is then interfaced to the DSP. In
this research work, each power electronic inverter under investigation is synchronised
using its own Zero Crossing Detector Unit and DSP. In this way, each inverter unit
operates fully independently as would be the case in a commercial system. For this
reason, there are no abnormal synchronisation issues created due to the experimental
setup.
100
Figure 6-7: Output waveforms of ZCD and inverter system in phase.
6.3.3 Results on Inverter 1 system
Before a parallel inverter system is tested, experiments are done for each separate
inverter. In this research, Inverter system 1 is tested first and results are recorded.
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Harmonic order
Optimum PI gain (Thd=5.650%)
Figure 6-8: Harmonic data of inverter 1 output current.
101
Figure 6-9 shows the screen shot of the output current and the error between the current
demand and the actual current.
(a)
(b)
Figure 6-9: Screen shot of (a) Inverter 1 output current, and (b) Inverter 1 current
error.
102
The software is set to have a dead time compensation unit of 0.03 and a ZCD is
connected into the system. The DC input voltage is fixed to 30 V and the resistive load
value used is approximately 5 Ohm. The power analyzer is set so that the output current
FFT analysis is done by taking an average of 16 fundamental current cycles. The
proportional gain, 𝐾𝑃 and the integral gain, 𝐾𝐼 are tuned once again to acquire the best
possible output current with the best THD. For that reason, the chosen 𝐾𝑃 and 𝐾𝐼 value
is 2 and 0.35 respectively. During the tuning process, from the observation of the
waveforms and the FFTs, a further increment as well as decrement in the proportional
gain, 𝐾𝑃 will push the waveform to instability. For the tuning of the integral part, a
further increment of the gain, 𝐾𝐼 will also push the waveform to instability whilst a
further decrement of the gain, 𝐾𝐼 will reduced the output current magnitude. Four
readings are taken and the average data is calculated and shown in Figure 6-8.
With the chosen proportional and integral gains, the THD achieved is 5.65%.
Furthermore, it can be observed from Figure 6-8 that the most dominant low order
harmonic are of the 3rd and 5th orders which are approximately 2.5% and 2% of the
fundamental value. Other than that, a problem spotted in the 2nd harmonic order.
Despite that it should be in a low magnitude value, it shows a definite high value which
in fact higher than the other harmonic orders. The reason for this is believed to be
because of the resistive load used which has some inductive value in it. At the time of
the experiments are done, this is the only type of resistor (linear resistor) that is
available in the lab. Next, the results from Inverter 2 are demonstrated, all with the
same DC input voltage and resistive load value.
With the same DC input voltage of 30V and resistive load value of 5 Ohm, a second
inverter system, Inverter 2 is tested, all using the same type and value of electronic
components. The proportional and integral gains chosen are 2 and 0.11 respectively.
Four readings are taken in the experiment of Inverter 2 stand alone system. The average
data of the harmonic spectrum is then calculated and shown in Figure 6-10.
103
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Harmonic order
Optimum PI gain (Thd=5.86%)
Figure 6-10: Harmonic data of inverter 2 output current using PI control.
As can be seen in Figure 6-10, the output current THD observed is 5.86% with high
harmonics magnitude in the lower order. Figure 6-11 below combines the harmonic
spectrums in Figure 6-8 and Figure 6-10 in the same graph.
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Harmonic order
PI inv 1 PI inv 2
Figure 6-11: Harmonic spectrums of Inv 1 and Inv 2 with PI current control.
Although theoretically the spectrum as in Figure 6-11 should resemble each other
because of the same values used for the components in both inverters, this is not the
case in the research. After much effort has been done to the experimental area; such as
104
shorten the wire connector, reduce noise by wrapping wires with a tape and placing the
DSP further with the inverter output point, the inequality to the output current harmonic
spectrum could not be resolved. Therefore, the only reason for that to happen is because
of the components used are physically different even though the same value of
components are used for both inverters.
6.4 PR Experiments and Results
The next step in this research is implementing the Proportional Resonant (PR)
controller in the inverter system. Once again, before the experiment is done with a grid
connected system, a stand-alone system with the same value of resistive load as in PI
experiments is used. It has to be noted here that the DC input voltage used and the
current demand are also the same as when experimenting with the conventional PI
current controller. For the purpose of this test, a new programming code based on the
digitized PR transfer function is added to the software. After the code is loaded and
run, two gains; the proportional gain, 𝐾𝑃 and the resonant gain, 𝐾𝑅 need to be tuned to
get the best output current waveform with the best THD possible.
There are a number of papers [80, 82, 83] mentioning that using a PR controller without
adding a cut off frequency in the system will make the system unstable. Based on a
paper by [84], after a cut-off frequency is added to the system, the analogue transfer
function becomes as follows.
2𝐾𝑅 𝜔𝑐 𝑠
𝐻𝑃𝑅 (𝑠) = 𝐾𝑃 +
𝑠2 + 2𝜔𝑐 𝑠 + 𝜔𝜊2
𝐾𝑃 (𝑠 2 + 2𝜔𝑐 𝑠 + 𝜔𝑜2 ) + 2𝐾𝑅 𝜔𝑐 𝑠

=
𝑠 2 + 2𝜔𝑐 𝑠 + 𝜔𝑜2
(Eq. 6-1)
By means of Tustin transformation, (Eq. 6-1) can be described as:
2
2 1 − 𝑧 −1 2 1 − 𝑧 −1 2 1 − 𝑧 −1
𝐾𝑃 [(𝑇 −1 ) + 2𝜔𝑐 (𝑇 −1 ) + 𝜔𝑜2 ] + 2𝐾𝑅 𝜔𝑐 (𝑇 )
1+𝑧 1+𝑧 1 + 𝑧 −1
𝐻𝑃𝑅 (𝑧) = 2
2 1 − 𝑧 −1 2 1 − 𝑧 −1
(𝑇 −1 ) + 2𝜔𝑐 (𝑇 ) + 𝜔𝑜2
1+𝑧 1 + 𝑧 −1
105
4 1 − 2𝑧 −1 + 𝑧 −2 4𝜔𝑐 1 − 𝑧 −1 2 4𝐾𝑅 𝜔𝑐 1 − 𝑧 −1
𝐾𝑃 [( ) ( ) + ( ) + 𝜔 ] +
𝑇 2 1 + 2𝑧 −1 + 𝑧 −2 𝑇 1 + 𝑧 −1 0 𝑇 (1 + 𝑧 −1 )
=
4 1 − 2𝑧 −1 + 𝑧 −2 4𝜔 1 − 𝑧 −1
( 2) ( −1 −2 ) + 𝑇𝑐 ( ) + 𝜔02
𝑇 1 + 2𝑧 + 𝑧 1 + 𝑧 −1
4(1 − 2𝑧 −1 + 𝑧 −2 ) 4𝑇𝜔𝑐 (1 − 𝑧 −2 ) 𝜔2 𝑇 2 (1 + 2𝑧 −1 + 𝑧 −2 ) 4𝑇𝐾 𝜔 (1 − 𝑧 −2 )

𝐾𝑃 [ 2 −1 −2 + 2 −1 −2 + 𝑜2 −1 −2 ] + 2 𝑅 𝑐 −1
𝑇 (1 + 2𝑧 + 𝑧 ) 𝑇 (1 + 2𝑧 + 𝑧 ) 𝑇 (1 + 2𝑧 + 𝑧 ) 𝑇 (1 + 2𝑧 + 𝑧 −2 )
= −1 −2 −2
4(1 − 2𝑧 + 𝑧 ) 4𝑇𝜔𝑐 (1 − 𝑧 ) 𝜔 𝑇 (1 + 2𝑧 + 𝑧 −2 )
2 2 −1
2 −1 −2 + 2 −1 −2 + 𝑜2
𝑇 (1 + 2𝑧 + 𝑧 ) 𝑇 (1 + 2𝑧 + 𝑧 ) 𝑇 (1 + 2𝑧 −1 + 𝑧 −2 )
(Eq. 6-2)
To simplify (Eq. 6-2) above, the transfer function becomes:
𝐻𝑃𝑅 (𝑧)
4𝐾𝑃 (1 − 2𝑧 −1 + 𝑧 −2 ) + 4𝑇𝐾𝑃 𝜔𝑐 (1 − 𝑧 −2 ) + 𝐾𝑃 𝜔𝑜2 𝑇 2 (1 + 2𝑧 −1 + 𝑧 −2 ) + 4𝑇𝐾𝑅 𝜔𝑐 (1 − 𝑧 −2 )
=
4 − 8𝑧 −1 + 4𝑧 −2 + 4𝑇𝜔𝑐 − 4𝑇𝜔𝑐 𝑧 −2 + 𝜔𝑜2 𝑇 2 + 2𝜔𝑜2 𝑇 2 𝑧 −1 + 𝜔𝑜2 𝑇 2 𝑧 −2
(Eq. 6-3)
Re-arranging (Eq. 6-3), the transfer function becomes:
𝐻𝑃𝑅 (𝑧)
(4𝐾𝑃 + 4𝐾𝑃 𝑇𝜔𝑐 + 𝐾𝑃 𝜔𝑜2 𝑇 2 + 4𝑇𝐾𝑅 𝜔𝑐 ) + (2𝐾𝑃 𝜔𝑜2 𝑇 2 − 8𝐾𝑃 )𝑧 −1 + (4𝐾𝑃 − 4𝐾𝑃 𝑇𝜔𝑐 + 𝐾𝑃 𝜔𝑜2 𝑇 2 − 4𝑇𝐾𝑅 𝜔𝑐 )𝑧 −2
=
(4 + 4𝑇𝜔𝑐 + 𝜔𝑜2 𝑇 2 ) + (2𝜔𝑜2 𝑇 2 − 8)𝑧 −1 + (4 − 4𝑇𝜔𝑐 + 𝜔𝑜2 𝑇 2 )𝑧 −2
(Eq. 6-4)
Based on (Eq. 6-4), the transfer function can be re-written as:
𝑏𝑜 + 𝑏1 𝑧 −1 + 𝑏2 𝑧 −2
𝐻𝑃𝑅 (𝑧) =
1 + 𝑎1 𝑧 −1 + 𝑎2 𝑧 −2
(Eq. 6-5)
Where;
(4𝐾𝑃 + 4𝐾𝑃 𝑇𝜔𝑐 + 𝐾𝑃 𝜔2 𝑇2 + 4𝐾𝑅 𝜔𝑐 𝑇)

0
𝑏𝑜 = 2
(4 + 4𝑇𝜔𝑐 + 𝑇2 𝜔0 )
(2𝐾𝑃 𝜔20 𝑇2 − 8𝐾𝑃 )

𝑏1 = 2
(4 + 4𝑇𝜔𝑐 + 𝑇2 𝜔0 )
(4𝐾𝑃 − 4𝐾𝑃 𝑇𝜔𝑐 + 𝐾𝑃 𝜔2 𝑇2 − 4𝐾𝑅 𝜔𝑐 𝑇)

0
𝑏2 = 2
(4 + 4𝑇𝜔𝑐 + 𝑇2 𝜔0 )
106
2
(2𝑇2 𝜔0 − 8)
𝑎1 = 2
(4 + 4𝑇𝜔𝑐 + 𝑇2 𝜔0 )
2
(4 + 4𝑇𝜔𝑐 + 𝑇2 𝜔0 )
𝑎2 = 2
(4 + 4𝑇𝜔𝑐 + 𝑇2 𝜔0 )
𝑇 is 50 µs, 𝜔𝑜 is 100 π, and 𝜔𝑐 is the controller cut off frequency. In a study by A.

Hasanzadeh et al [80], they identify a range of practical cut off frequency value that
can be used in the PR controller system. It is between 10 to 100 rad/s. Based on that
range, several different values of cut off frequency between 10 to 100 rad/s are applied
and tested in the experiment. Following that, it is noticed that the harmonic data of the
inverter output current waveform is much improved following the increased value of
cut off frequency. Besides, the output current waveform as well as the corresponding
THD is also improved. For the purpose of the research, a cut off frequency of 30 rad/s
is chosen as the output current harmonic spectrum shows an optimum performance of
all.
6.4.1 Results on Inverter 1 System
4
3.5
3
2.5
2
1.5
1
0.5
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Harmonic order
Optimum PR gain (Thd=5.764%)
Figure 6-12: Harmonic data of inverter 1 output current using PR control.
Similar to previous experiment with PI controller, the software is set to have a dead
time compensation unit of 0.03 and a ZCD is connected into the system. The power
107
analyzer is set so that the output current FFT analysis is done by taking an average of
16 fundamental current cycles.
(a)
(b)
Figure 6-13: Screen shot of (a) output current waveform with reference 3 A peak ;
and (b) current error waveform which shows approximately 0.2 A peak.
108
The proportional gain, 𝐾𝑃 and the integral gain, 𝐾𝑅 are tuned to acquire the best
possible output current with the best THD. For that reason, the chosen 𝐾𝑃 and 𝐾𝑅 value
is 2 and 410 respectively. During the tuning process, from the observation of the
waveforms and the harmonic data, a further increment as well as decrement in the
proportional gain, 𝐾𝑃 and the resonance gain, 𝐾𝑅 will push the waveform to instability.
Using the chosen gains, four readings are taken and the average data is calculated and
shown in Figure 6-12. It can be noticed that high harmonics appear between the 2nd to
6th orders. The reason for the high magnitude of the even harmonic numbers is similar
to the experiment when the inverter system is tested with PI control. Figure 6-13 shows
the screen shot of the output current waveform and the current error when the reference
current of 3 A peak is compared with the measured output current.
Following the testing of inverter 1 system using PR control, inverter 2 system with the
same type and parameter values is tested. The gains are tuned so that the lowest THD
is obtained. Again, it has to be noted that although inverter 2 system is of the same type
and use the same parameter values, the physical of components are different which
leads to different gains value. As a result, the proportional gain, 𝐾𝑃 and the resonant
gain, 𝐾𝑅 is 1.8 and 100 respectively.
6
5
4
3
1
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Harmonic order
Optimum PR gain (Thd=7.286%)
Figure 6-14: Harmonic data of inverter 2 output current using PR control
109
5
4
3
2
1
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Harmonic order
PR inv 1 PR inv 2
Figure 6-15: Harmonic spectrums of Inv 1 and Inv 2 with PR current control.
Figure 6-14 illustrates the averaged harmonic data of the inverter output current after
four readings are taken from the power analyzer. As can be seen, the output current
THD obtained is 7.286%. The highest harmonics magnitude are of the 2nd and 3rd order.
Figure 6-15 combines the harmonic spectrums in Figure 6-12 and Figure 6-14 in the
same graph. Once again, the dissimilarity of the harmonic spectrum between both
inverters are due to the components that are physically different even though the same
value of components are used for both inverters.
110
6.5 Comparison between the PI and PR Controller System
(a)
(b)
Figure 6-16: Compare output current waveform of two controller systems (a) Output
current from PI controller shows distortion at the top, and (b) output current from PR
controller shows smooth curve at the top.
111
Figure 6-16 compares the output current waveform of two controller systems. Each
current errors are illustrated in Figure 6-17. From observation, the reason for that is
because the PR controller system can reduce the steady state error between the
demand and the actual current.
Y-axis (1 A per div)
X-axis (40 ms per div)
(a)
Y-axis (1 A per div)
(b)
Figure 6-17: Compare error of two controller systems (a) current error with PI
controller system is approximately 0.5 A peak; and (b) current error with PR
controller system is reduced to approximately 0.2 A peak.
112
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Harmonic order
PI (Thd=5.65%) PR (Thd=5.764%)
Figure 6-18: Compare FFT of two controller systems.
Yet, when comparing the harmonic profile of inverter 1 system with the PI and PR
controller system, the harmonics of the 5th, 7th, 9th and 11th of the PR controller system
are reduced significantly, that is between 25% to 75%, lower than the harmonics of the
PI controller system. The results thus demonstrate that the same output in the
R.Teodorescu et al [59] study as explained in Chapter 2 is achieved. This can be seen
in Figure 6-18. The same figure also shows the output current Total Harmonic
Distortion (THD) of the two controller systems. It shows an output current THD of
5.65% when using PI control and 5.764% PR control system.
The THD results show that both current control methods yield similar THD
performance. However, with respect to the lowest harmonic orders (3rd 5th 7th), the PR
controller achieves lower individual harmonic components. This is an important result.
In previous works, the lowest order harmonics are often cited as being problematic in
grid connected inverter systems. Therefore, in terms of meeting the imposed
recommendations for individual harmonics (<2% for a commercial system), the PR
controller offers some performance advantages.
Based on the objective of this research project, a random gain is used in the controller
system instead of a fixed gain value. The modified controllers are then tested and results
are described in the following section.
113
6.6 Parameter Randomization Controller Technique
(a)
(b)
Figure 6-19: Random signal generation (a) generation of random signal more
scattered around before filtering process ; and (b) random signal becomes smoother
like a waveform after filtering process.
114
In order to improve the harmonic spectrum thus reduces the overall output current
THD, a random signal is added to the proportional and integral gain of the conventional
PI current controller. This random signal is generated using the software and is limited
to a certain range so that the output current instability is prevented by the adjusted gain,
𝑅𝑃𝐼 and 𝑅𝐾𝐼 tuning. The range of the signal is determined by testing the inverter system
with different values of gains. The maximum and minimum gain value before the
output current becomes unstable is then chosen as the random signal limit. Following
this, when the limit is added to the gains, the new randomized gains, 𝑅𝑃𝐼 and 𝑅𝐾𝐼 are
adjusted automatically while the inverter system is on test. A simple digital low pass
filter is also added after the random number generation for smooth variation signal.
This filter has a cut off frequency of approximately 560 Hz. It is observed that higher
cut off frequency will not make the random signal any smoother and lower cut off
frequency will limit the range of the random signal. The filtered random signal is then
added to the gains to form the newly randomized gains which in turn used in the current
control of the inverter system. Figure 6-19 shows the waveforms of the random signal
before and after filtering.
6.6.1 Results on Randomized PI on Inverter 1 System
Two tests were carried out in order to gain the best implementation of the randomized
signal. Firstly, the test of randomized signal being added to the proportional gain to
become the randomized proportional gain, 𝑅𝐾𝑃 whilst the integral gain, 𝐾𝐼 remain fixed
and secondly, the test of randomized signal being added to the integral gain to become
the randomized integral gain, 𝑅𝐾𝐼 whilst the proportional gain, 𝐾𝑃 remain fixed. In the
first test, the randomized proportional gain, 𝑅𝐾𝑃 is set to be in a range from 1.3 to 2.7
while the integral gain is set to be fixed at 0.35. After four readings are taken from the
power analyzer which has been set to calculate the output waveform of up to 16
fundamental cycles, an average in then calculated in Excel. The harmonic spectrum is
illustrates in Figure 6-20.
115
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Harmonic order
Random Kp (Thd=5.753%)

randomized proportional gain of PI control technique.
In the second test, the randomized integral gain, 𝑅𝐾𝐼 is set to be in a range from 0.2 to
0.5 while the integral gain is set to be fixed at 2. Again, four readings of harmonic data
are taken from the power analyzer and an average data value is then calculated in Excel.
The output current harmonic spectrum is illustrates in Figure 6-21.
3.5
3
2.5
2
1.5
1
0.5
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Harmonic order
Random Ki (Thd=4.816%)

randomized integral gain of PI control technique.
116
Figure 6-22 shows the comparison between the three graphs; Figure 6-8, Figure 6-20
and Figure 6-21, when inverter 1 system is tested using the conventional PI and the
randomized PI control technique.
Mag (% of fundamental) 4.5

4
3.5
3
2.5
2
1.5
1
0.5
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Harmonic order
Optimum PI gain (Thd=5.65%) Random Kp (Thd=5.753%)


randomized proportional, and randomized integral gain of PI control technique.
From the graph, it can be easily compared that when random gains are applied to the
inverter 1 system, the output current harmonic spectrum is slightly changed which in
turns affected the trend of the spectrum. As a result, the THD of the output current is
also affected. As seen in Figure 6-22, implementing the randomized integral gain to the
conventional PI current technique can improve the THD of inverter 1 system. The THD
reduced from 5.65% to 4.816% which is a reduction by approximately 15%. Most
importantly, the magnitude of the lower order harmonics that is from the 2nd to the 11th
harmonic orders are mostly decreased except for the 7th and 8th orders which have a
small increment. Harmonic orders beyond the 11th show a fair reduction as well as
increment. This behaviour of having a reduction and increment in harmonic orders
magnitude is the reason of the reduced Total Harmonic Distortion (THD) of inverter 1
system when using the randomized integral gain, 𝑅𝐾𝐼 .
Next, randomized proportional and integral gain of the conventional PI control

technique are applied and tested in inverter 2 system.
117
6.6.2 Results on Randomized PI on Inverter 2 System
The same test is then performed with the second inverter, inverter 2 system. As the
value of the gains when an optimum output current of inverter 2 result is achieved are
2 and 0.11 for 𝐾𝑃 and 𝐾𝐼 respectively, the limitation for the random signal need to be
set differently. Again, this is done by setting the gains slightly higher and slightly lower
to the point before the system goes unstable. After several gain changes, it is agreed
that when 𝐾𝐼 is fixed to 0.11, the lower limit for 𝐾𝑃 is 1.6 and the upper limit is 2.4.
Whereas when 𝐾𝑃 is fixed to 2, the lower limit for 𝐾𝐼 is 0.095 and the upper limit is
0.13. Figure 6-23 and Figure 6-24 shows the averaged data values of the output current
harmonic spectrum when the randomized proportional gain, 𝑅𝐾𝑃 and the randomized
integral gain, 𝑅𝐾𝐼 is applied to the conventional PI system in the inverter 2 system.
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Harmonic order

118
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Harmonic order

Figure 6-25 shows the comparison between the three graphs; Figure 6-10, Figure 6-23
and Figure 6-24.
5
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Harmonic order
Optimum Kp and Ki (Thd=5.86%) Random Kp (Thd=5.66%)


randomized proportional, and randomized integral gain of PI control technique.
119
While different trends are achieved when random gains of PI control are applied to the
inverter 1 system, this is not the case when the same techniques are tested on inverter
2. It can be seen in Figure 6-25 that only a small changes; increment and decrement are
observed to the output current harmonic spectrum which yields to almost no changes
to the THD. There is no exact explanation to the result as above. The only reason at the
time when the test is done is because of the difference in the physical of the components
used. Because of that, further experiment to answer about the question of why results
of inverter 2 system is not similar to the results of inverter 1 system is suggested in
future.
6.6.3 Results on Randomized PR in Inverter 1 System
Following the tests of randomized PI in both inverter systems, tests of randomized PR

are performed next. Again, randomized proportional gain of PR control is tested first
before the randomized integral gain is applied to the PR control. Results of the tests are
illustrated in Figure 6-26, Figure 6-27 and Figure 6-28.
4
3.5
3
2.5
2
1.5
1
0.5
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Harmonic order

randomized proportional gain of PR control technique.
120
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Harmonic order
Random Kr (Thd=6.048%)

randomized resonant gain of PR control technique.
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Harmonic order
Optimum PR gain (Thd=5.764%) Random Kp (Thd=5.58%)


randomized proportional and randomized resonant gain of PR control technique.
The first test is done when the resonant gain is fixed to 410 while the randomized
proportional gain limit is set to be in the range between 1.4 to 2.6. The second test is
done when the proportional gain is fixed to 2 while the randomized resonant gain limit
is set to be in the range between 370 to 450. From Figure 6-28, different trends are
observed in the harmonic spectrum of the inverter output current. The critical part to
121
notice is the improvement to the THD when randomized proportional gain, 𝑅𝐾𝑃 is
applied to the PR control. Although it shows a small THD reduction, it is believed that
the outcome can be improved when a higher DC voltage is used.
6.6.4 Results on Randomized PR in Inverter 2 System
For the tests of inverter 2 system, the randomized proportional gain is limited between
1.4 to 2.2 whilst the randomized resonant gain is limited between 85 to 115. Again, it
has to be noted that the limitation given is based in the test when different gain values
are in the beginning to observe the stability. Figure 6-29 and Figure 6-30 shows the
individual harmonic spectrum when the randomized proportional gain, 𝑅𝐾𝑃 and the
randomized resonant gain, 𝑅𝐾𝑅 are applied to the conventional PR control technique.
6
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Harmonic order

randomized proportional gain of PR control technique.
122
5
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Harmonic order

randomized resonant gain of PR control technique.
6
5
4
3
2
1
0
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Harmonic order
Optimum Kp and Kr (Thd=7.286%) Random Kp (Thd=7.08%)

Figure 6-31: spectrum of inverter 2 system using a conventional, randomized

proportional, and randomized resonant gain of PR control technique.
From Figure 6-31, it can be seen that increment as well as decrement of harmonics
magnitude are achieved especially to the harmonic orders between the 2nd to the 7th.
These in turn yield to the improvement in the total harmonic distortion, THD of the
inverter 2 system.
123
6.7 Chapter Summary
The overall of this chapter showed significant results using three different control
methods in a stand-alone system. It can be concluded here that by using a random signal
gain to the conventional controller, PI and PR, improvement in the output current THD
is achieved. From the results, it shows that using the randomized integral gain, 𝑅𝐾𝐼 in
the conventional PI controller is better than using a fix integral gain, 𝐾𝐼 alone. It also
shows that using the randomized proportional gain, 𝑅𝐾𝑃 in the conventional PR
controller is better than using a fix proportional gain, 𝐾𝑃 alone.
124
Experimental Results of Grid Connected Results
Chapter 7: Experimental Results of Grid Connected Inverters
7.1 Introduction
This chapter presents results for the parallel inverter system described in Chapter 6.
The chapter is split into three sections. The first section concentrates on the
performance of the parallel inverter system when operated with a conventional PI
current controller scheme. The second section considers the performance of the parallel
inverter system when run with the random parameter controller scheme. In both
sections, the performance of each inverter unit operating as an independent grid
connected unit is initially assessed. Specifically, the harmonic profile of each inverter
is shown, and performance comparisons are made between individual inverter units.
Attention is then given to assessing the performance of each inverter when operated in
parallel. The effect of each inverters harmonic profile on the overall harmonic
performance of the parallel inverter system is then considered. Finally, the performance
of the random parameter control technique is assessed via direct comparison of results
with the conventional PI control scheme.
7.2 Experimental results with conventional PI current control techniques
This sub-section concentrates on the performance of the parallel inverter system when
operated with a conventional PI current controller scheme first and followed by the
parallel system operated with a parameter randomisation of PI current controller
scheme. For each technique, a single inverter system is tested first followed by two
inverters working in parallel.
125
7.2.1 Conventional PI Current Control of Individual Inverter Units
Idc
Power Analyzer
Lx
Ix
Vdc
Vs
Cx
Figure 7-1: Experimental H-Bridge inverter. Single inverter grid connected system
Initially, each inverter is operated independently as a single grid connected unit with a
conventional current control loop (Figure 7-1). Each inverter operates from an
independent 50 V DC bus, and is connected to a 20 V AC grid voltage. Importantly,
each inverter is synchronised to the network via its own Zero Crossing Detector Unit
and Digital Control System. In this way, each inverter unit operates fully independently
as would be the case in a commercial system and there are no abnormal synchronisation
issues created due to the experimental setup. Harmonic data is acquired through the use
of a Yokogawa power analyser. The power analyser is set up to acquire the harmonic
data over 16 fundamental current cycles, utilising a Hanning sampling window. All
harmonic and THD results presented are acquired under these conditions. Each inverter
is tuned independently via the software parameters: proportional gain, 𝐾𝑃 and integral
gain, 𝐾𝐼 , to achieve the best output current fidelity possible. Each inverter is controlled
to operate at unity power factor with respect to the grid voltage. Via the power analyser,
the harmonic spectrum of the output current is recorded for each inverter. The results
obtained represent a snap shot in time of the performance of each inverter. To evaluate
the average performance of each inverter over time, the harmonic spectrum of each
inverter is recorded at three separate intervals. From this, an averaged harmonic
spectrum may be calculated and considered to be a typical measure of inverter
performance. The averaged harmonic data is imported into Microsoft® Excel® for
126
presentation in graphical format. Conclusions are drawn from the averaged harmonic
performance of each inverter.
(a)
(b)
conventional PI control: predominant harmonics appear between the 3rd and 15th
harmonic and (b) Inverter 2 output current with conventional PI control: predominant
harmonics appear between the 3rd and 15th harmonic.
The averaged harmonic performance of each individual inverter operating

independently is shown in Figure 7-2. This shows the per unit magnitude of all
harmonics between the 2nd and 22nd harmonic. For both inverters, the predominant
harmonics appear between the 3rd and the 15th harmonic. The harmonics become less
problematic beyond the 19th harmonic. This is to be expected, since the output inductor
and capacitor components yield a low pass filter with an approximate 800 Hz cut-off
frequency. Using the Microsoft® Excel® trend line function, a 6th order polynomial
trend line is imposed on each graph. This produces a harmonic profile for each inverter
127
over the averaged set of harmonic data. This approach is consistent with the work by
Armstrong [33], and allows for easy comparison between results. As reported by
Armstrong, the 6th order polynomial fit provides a good smooth average trend line
across the harmonic range. Lower order fits are more sensitive to individual harmonic
components [33]. The results show a strong correlation in harmonic profile for each of
the two inverters under test. The THD performance of the inverters is summarised in
Table 7-1 and indicates the average harmonic performance of each inverter when
operating independently is 5.74%.
Table 7-1: THD figures for independently operated, grid connected, inverters
using conventional PI current control
Operating Inverter THD SYSTEM OUTPUT
Inverter 1 5.82%
Inverter 2 5.65%
Single inverter average 5.74%.
7.2.2 Conventional PI Current Control of Two Inverters in Parallel

Id
c
Power Analyzer
L
x
Ix
Vdc
C
Vs
Id
c
L
x
Ix
Vdc
C
x
Figure 7-3: Experimental, parallel connected inverter system comprising of two

inverter units.
Having tuned and assessed the performance of the inverters operating independently,
the inverters are then configured to form a parallel, two inverter, grid connected system.
128
The circuit topology is shown in Figure 7-3. Current measurements are made by the
power analyser at the output of each individual inverter.
(a)
(b)
(c)
Figure 7-4: Low order harmonic spectra of parallel inverters with PI current control.
(c) Results of the parallel system output current. All shows strong correlation between
them with predominant harmonics appear between the 3rd and 15th harmonic.
129
To record the overall parallel system performance, an additional current measurement

is made at the point of common coupling. Each inverter is current controlled with the
same controller tuning as determined previously during independent operation. In each
case, the two applicable inverters are set to operate at unity power factor with respect
to the grid voltage. As before, three sets of harmonic data are collected and an averaged
harmonic spectrum is determined for each inverter (Figure 7-4). The recorded THD of
each individual inverter, along with the parallel inverter system output THD, is
summarised in Table 7-2. The results show that the average THD of the parallel system
output is 5.36%.
Table 7-2: THD Performance of Two Inverter Parallel System with conventional PI
control
PARALLEL THD THD THD

COMBINATION SYSTEM
INV 1 INV 2
OUTPUT
Inverter 1 & Inverter 2 5.64% 5.42% 5.36%
7.2.3 Summary: Conventional PI Current Control
For all the results presented with conventional PI current control, the same general
trend line is seen in the harmonic profile of the output current of each individual
inverter unit. In particular, the results show that the largest distortion components are
at the 3rd, 7th, 9th and 11th harmonic. When operated as a parallel inverter system, this
same harmonic trend is observed in the parallel system output current, and similar
individual harmonics dominate. With respect to THD, each inverter typically operates
with a THD around 5.5%. When operated as a parallel inverter system, a comparable
THD level is observed at the point of common coupling. This behaviour clearly
demonstrates the additive nature of the harmonics at the output from individual
inverters. It is this harmonic addition at the multiple inverter output which is
undesirable, as originally reported by Armstrong et al.
7.2.4 Experimental Results with Randomised PI Current Control Techniques
The second section of this chapter presents results for the parallel inverter system when
controlled via the randomised parameter PI control method. In order to assess the
130
performance of this proposed methodology, the same sequence of tests is carried out
as presented for the conventional PI current control method. Once again, the harmonic
performance of each inverter operating as an independent grid connected unit is
considered. This is followed by the harmonic assessment of each grid connected
inverter when operating in a parallel connected system. From this, the impact of the
randomised control scheme on the parallel inverter system output current is determined.
Once again, due to the inclusion of the controller randomisation, the harmonic
performance of each inverter varies with time. Therefore, a definitive assessment of the
randomised control strategy cannot be determined from one set of results taken at a
single instance in time. For this reason, the performance of each inverter is again
determined through averaging three sets of harmonic data. In this way, a statistical
conclusion is made possible, based on a number of experimental results taken over
time.
7.2.5 Parameter Randomisation PI Current Control of Individual Inverter Units
Figure 7-5 shows the averaged harmonic spectrum of each inverter when operated as a
single inverter system. Importantly, the harmonic profile of each inverter is now
different. This is shown through the differing trend line patterns and is in contrast to
the almost identical set of harmonic profiles observed with conventional current control
(Figure 7-2). This indicates that the random component in the current controller has
reduced the degree of correlation in the harmonic profile of each inverter. When the
inverter system is connected to the grid, the impact of the randomisation is much more
noticeable, due to the low order disturbance effect of the grid on the current control
loop. This effect was also observed by Armstrong et al [34]. It is this uncorrelated
performance which may be utilised to advantage in a parallel inverter system, since it
yields improved opportunities for harmonic cancellation at the system output. The
recorded THD at the parallel inverter system output, as measured by the power
analyser, is 5.15% (Table 7-3).
131
(a)
(b)
randomised PI control; and (b) Inverter 2 output current with randomised PI control;
where difference trend lines are observed between the two inverters.
Table 7-3: THD Performance of Two Inverter Parallel System with conventional
control
PARALLEL COMBINATION THD SYSTEM OUTPUT
Inverter 1 5.07%
Inverter 2 5.22%
Single Inverter Average 5.15%
132
7.2.6 Parameter Randomisation PI Current Control of Two Inverters in Parallel
(a)
(b)
(c)
Figure 7-6: Harmonic spectra of parallel inverters with randomised PI current control.
(c) Results of the parallel system output current. Different trend lines are observed.
133
Having assessed the performance of the inverters operating independently, the inverters
are then configured to form a parallel, two inverter, grid connected system. The circuit
topology remains the same as described for the conventional PI current control scheme;
whereby current measurements are made by the power analyser at the output of each
individual inverter, and the point of common coupling. Three sets of harmonic data are
collected, and an averaged harmonic spectrum is determined for each inverter Figure
7-6. The recorded THD of each individual inverter, along with the parallel inverter
system output THD, is summarised in Table 7-4. The results show that the average
THD of the parallel system output is 4.62%.
The risk of circulating current is always a potential problem in parallel connected

inverter systems. In this work, although possible circulating current was not directly
measured, there did not appear to be any practical evidence of it causing any problems
with the operation of the inverter units, or the control system. This is concluded from
knowledge that the characteristics of each inverter are similar regardless or individual
or parallel operation.
Table 7-4: THD Performance of Two Inverter Parallel System with random
parameter control
PARALLEL COMBINATION THD THD THD SYSTEM

INV 1 INV 2 OUTPUT
7.2.7 Summary, Parameter Randomisation PI Current Controller
The results presented for the random parameter PI current controller confirm that it is
possible to adjust the harmonic performance of each inverter unit. By using the random
parameter controller, the harmonic profile of each inverter unit now dynamically varies
with time. The fixed harmonic trend line observed with the conventional PI control
scheme is no longer typical of the behaviour of the system. Whilst the THD of each
inverter individual unit is still typically just over 5.3%, when operated as a parallel
inverter system a lower average THD is observed at the point of common coupling. For
a two inverter parallel system, the average THD is measured as 4.62%. This marks a
significant improvement in the net harmonic performance of the parallel inverter
system compared to the conventional PI current controller. However, as discussed by
134
Armstrong et al, there is still a significant contribution from the very low order
harmonics; 3rd to 9th, despite the introduction of the controller randomisation.
7.3 Experimental Results with Randomised PR Current Control Techniques
The experimental conditions in this section are identical to those presented for the PI
current controller. The only difference is the control algorithm where PR current
control is used instead of the PI current control.
7.3.1 Fixed PR Current Control of Individual Inverter Units
Again, initially, each inverter is operated independently as a single grid connected unit
with a conventional PR current control loop (Figure 7-1). Each inverter is tuned
independently via the software parameters: proportional gain, 𝐾𝑃 and resonant gain,
𝐾𝑅 , to achieve the best output current fidelity possible. Each inverter is controlled to
operate at unity power factor with respect to the grid voltage. Via the power analyser,
the harmonic spectrum of the output current is recorded for each inverter. To evaluate
the average performance of each inverter over time, the harmonic spectrum of each
inverter is recorded at three separate intervals. Conclusions are drawn from the
averaged harmonic performance of each inverter.
The averaged harmonic performance of each individual inverter operating

independently is shown in Figure 7-7. As anticipated, the results show some degree of
correlation in harmonic profile for each of the two inverters under test.
135
(a)
(b)
conventional PR control; and (b) Inverter 2 output current with conventional PI
control. Results show some degree of correlation in harmonic profile for each of the
two inverters.
using conventional PR current control
Inverter 1 4.98%
Inverter 2 4.92%
136
7.3.2 Fixed PR Current Control of Two Inverters in Parallel
(a)
(b)
(c)
Figure 7-8: Low order harmonic spectra of parallel inverters with PR current control.
(c) Results of the parallel system output current. Small cancellation and addition of
harmonics are observed at the system output current.
137
Table 7-6: THD Performance of Two Inverter Parallel System with conventional PR
control.
PARALLEL THD THD THD SYSTEM

COMBINATION INV 1 INV 2 OUTPUT
Each inverter is current controlled with the same controller tuning as determined
previously during independent operation. In each case, the two applicable inverters are
set to operate at unity power factor with respect to the grid voltage. As before, three
sets of harmonic data are collected and an averaged harmonic spectrum is determined
for each inverter (Figure 7-8). The recorded THD of each individual inverter, along
with the parallel inverter system output THD is summarised in Table 7-6. The results
show that the average THD of the parallel system output is 5.06% which is better than
the average THD of the parallel system when using the conventional PI control.
7.3.3 Summary: Fixed PR Current Controller
For the results presented with conventional PR current control, a typical harmonic
profile exists, although it differs that observed with the conventional PI controller.
Results show that the largest distortion components are at the 3rd, 7th, 9th and 11th
harmonic, although importantly these are less of a problem. Similar characteristics are
observed when the two converters are operated in parallel, and at the point of common
coupling there are still signs of harmonic addition.
138
7.3.4 Experimental Results with Randomised PR Current Control Techniques
This sub-section presents results for the parallel inverter system when controlled via
the randomised parameter of the PR control method. Each inverter is tuned
independently via the software parameters.
7.3.5 Parameter Randomisation PR Current Control of Individual Inverter

Units
(a)
(b)
randomised PR control; and (b) Inverter 2 output current with randomised PR control.
Different trend line patterns indicate the reduction in the degree of correlation in the
harmonic profile.
Figure 7-9 shows the averaged harmonic spectrum of each inverter when operated as a
single inverter system. The results demonstrate the success of the controller
randomisation in generating differing trend line patterns, thus indicating a reduction in
the degree of correlation in the harmonic profile of each inverter. This result suggests
139
that the controller randomisation technique is well suited to a grid connected inverter
system employing a PR controller. In terms of the aims and objectives of this work,
this is a successful outcome and one of the main contributions to knowledge.
using randomised PR current control.
Inverter 1 4.98%
Inverter 2 5.12%
7.3.6 Parameter Randomisation PR Current Control of Two Inverters in Parallel
Having assessed the performance of the inverters operating independently, the inverters
are again configured to form a parallel, two inverter, grid connected system. The circuit
topology remains the same as described for the conventional PI current control scheme;
whereby current measurements are made by the power analyser at the output of each
individual inverter, and the point of common coupling. Three sets of harmonic data are
collected, and an averaged harmonic spectrum is determined for each inverter Figure
7-10. The recorded THD of each individual inverter, along with the parallel inverter
system output THD, is summarised in Table 7-8. The results show that the average
THD of the parallel system output is 4.36%.
140
(a)
(b)
(c)
Figure 7-10: Low order harmonic spectra of parallel inverters with randomised PR
current control. (a) Results of inverter 1 output current; (b) Results of Inverter 2
output current; and (c) Results of the parallel system output current. Harmonics
magnitude reduction are observed at the parallel system output current between the
7th and 19th order.
141
Table 7-8: THD Performance of Two Inverter Parallel System with randomised PR
control.
PARALLEL THD THD THD SYSTEM

COMBINATION INV 1 INV 2 OUTPUT
7.3.7 Summary: Parameter Randomisation PR Current Controller
The results presented for the random parameter PR current controller confirm that it is
possible to adjust the harmonic performance of each inverter unit. By using the random
parameter PR controller, the harmonic profile of each inverter unit now dynamically
varies with time in a similar way to that observed in the PI controller case. It is believed
that this is the first time that this approach has been applied to the PR controller in a
grid connected power converter system.
The THD of each inverter individual unit is still typically just under 5%, when operated
as a parallel inverter system a lower average THD is observed at the point of common
coupling. For a two inverter parallel system, the average THD is measured as 4.34%.
This marks a significant improvement over the individual inverter case. Furthermore,
it also represents a further reduction in harmonic content compared to the conventional
PI current controller approach. Whilst some caution has to be paid to the deliberately
induced random nature of the results, it is apparent that there is an improvement in the
low order harmonics around the 7th and 9th harmonic. This is an area which is difficult
to address with the PI controller approach.
Due to the randomised nature of the proposed controller, the harmonic profile tends to
become flatter once inverters are operating in parallel. This is to be expected, since
harmonic cancellation starts to take place once the randomised controller is applied.
This is in contrast to the fixed parameter controller, where the harmonics typically
reinforce due to the similar harmonic profiles.
As the number of parallel inverters increases, the opportunities for harmonic

cancellation will increase, hence the harmonic profile should tend to flatten out further.
However, as noted by Armstrong and in this work, the lowest order harmonics tend to
142
be dominant, hence these are the most difficult to eradicate. Therefore, a point is likely
to be reached where increasing the number of inverters in parallel brings a diminishing
return in terms of overall harmonic performance. The point of which this happens is
presently unknown. It would require complex statistical mathematics to be applied,
which is presently beyond the scope of this work. However, it does provide scope for
potential future research work for optimising the scale of the system.
7.4 Chapter Summary

This chapter has presented experimental results for a grid connected parallel inverter
system, implementing both conventional PI, and randomised parameter control
techniques. The results show that, on average, the current THD at the parallel inverter
system output is lower when using the randomised current control method. With
conventional current control, the harmonic spectrum of each inverter demonstrates a
high degree of correlation with the harmonic spectra of all other inverters in the parallel
system. This results in reinforcement of current harmonics at the parallel inverter
system output. By introducing randomised parameter current control, the harmonic
spectrum of each inverter is likely to differ from that of all other inverters. Therefore,
there is less chance of harmonic reinforcement at the parallel inverter system output.
The differing harmonic profiles enhance the opportunities for harmonic cancellation in
the system output current. This is reflected in the reduced current total harmonic
distortion of the parallel inverter system output.
143
Conclusions
Chapter 8: Conclusions
8.1 Conclusions
This thesis has presented research into power quality improvement of grid connected
inverter systems. In grid connected system, it is vital to inject a unity power factor
current into the grid with low levels of harmonic distortion. This can be achieved with
an appropriate choice of inverter topology, PWM switching scheme, passive filter
components, and current control scheme.
This work has specifically focused on improving the low order harmonic performance
of grid connected inverters by investigating the impact of different current control
methods. In this work, three techniques have been analysed, simulated, and
experimentally validated. They are the conventional PI control technique, the
proportional resonant (PR) control technique, and a novel modified proportional
resonant control technique.
The classic PI controller is very well understood and often applied by engineers in
industry. It has been successfully applied in grid connected inverter applications, and
shown it provide good overall performance. Literature shows that many different
tuning algorithms have been applied to the PI controller. Furthermore, novel
approaches such as gain randomisation have successfully been applied to achieve better
low order harmonic performance.
Despite the general merits of the PI controller, literature also shows a move away from
this control technique in grid connected applications. It is reported that the PI controller
suffers from steady state error when tracking AC signals. Such characteristics have
been confirmed in this research by considering the PI controller error signal.
For this reason, investigators are now leaning to alternative control methods such as
the Proportional Resonant (PR) controller. It is stated that this controller does not
exhibit the same steady state error when tracking AC signals. The work carried out in
this thesis appears to agree with conclusion, since the controller error signal is
improved compared to the PI controller. From this, it appears that there are distinct
advantages to applying the PR controller.
144
Conclusions
For this reason, following the same methodology as first applied to the PI controller by
Armstrong [33], this thesis set out to test the hypothesis that the gains for the PR
controller can also be randomised with a view to improving harmonic performance of
parallel connected inverters. Whilst the performance of each individual inverter may
not necessarily be improved directly, the overall harmonic injection into the grid may
be reduced if the controllers exhibit some degree of randomisation. The author believes
that this is the first time that such a technique has been applied to the PR controller in
a grid connected inverter system.
In this work it is shown, via simulation and experimental validation that using a
modified Proportional Resonant with random proportional gain can modestly improve
the THD of the injected grid current in a parallel connected inverters system. The
degree of improvement is admittedly only small, but it is believed that better results
may have been observed with a larger number of inverters operating in parallel.
However, due to limitation of resources for the project, this work was restricted to two
inverters only. In principle more inverters operating together should present greater
opportunities for harmonic cancellation. Verification of the scheme on a larger scale
system is one of the first suggested areas for future work and investigation.
However, the important outcome from this work is knowledge that gain randomisation
will work with the PR controller in grid connected inverters has been achieved by this
work. This is important given the shift towards this type of controller in grid connected
applications.
In Chapter 6, it is identified that the harmonic performance of the conventional PI

controller can also be improved by using a random integral gain, 𝑅𝐾𝐼 instead of a fixed
integral gain, 𝐾𝐼 . Such Integral gain randomisation was avoided by Armstrong due to
concerns over maintaining unity power factor. However, it has been shown in this
work, believed to be for the first time, that the integral gain can be modified to improve
harmonic performance. Provided the limits of variation are kept very small, the impact
on power factor seems minimal. For this reason, the author believes that further studies
into variation of all controller parameters may yield further improvement in harmonic
performance. At present, the level of randomisation applied is very much determined
by a trial and error approach. By observing the performance of the system, and applying
cautious limits, the integrity of the system is preserved and performance improvement
145
Conclusions
is simply recorded. It is therefore suggested that a more comprehensive analytical

approach to the problem may reveal more about methods of system optimisation.
Furthermore, it could be used to better quantify the robustness of the controller and its
system sensitivity to parameter changes. This would
Finally, it is worth noting that this work concentrates solely on small scale single phase
inverter systems– typically inverters designed for domestic installation (<3kW). In
theory, the controller principles investigated should be applicable to three phase
systems. However, this would require a significant amount of additional research work
to verify. A number of key points would need to be considered; the recommendations
on harmonic limits are different for larger scale inverter systems, the inverter
topologies are different, and the control algorithms are fundamentally different – for
example Space Vector Modulation techniques. Therefore, in the future, there is
potential for investigation into three phase applications.
146
Appendixes
Appendix A Multisim and Ultiboard Designing Tool
Before a PCB is made, the schematic diagrams of the circuit are designed in the circuit
design suite called Multisim.
Figure A-1: Screen shot of Multisim designing tool
After all the required components have been placed on the schematic capture space,
with point and click, the designs are then transferred to another tool in the circuit design
suite called Ultiboard 10.1 as shown in the Figure A-2. This tool will allow the user to
create and design the intended PCB.
147
Appendixes
Figure A-2: Screen shot of Ultiboard designing tool
There are five PCBs that were needed to be made in this project. They are the power
inverter board, the voltage measurement board, the current measurement board, the
gate driver board, and the extension board for the processor. As mentioned in earlier
chapter, a TMS320F2812 Digital Signal Processor (DSP) is used as the processor for
this project and has been briefly discussed in Chapter 6 . The next following figures are
the pictures of the PCBs. The gate driver layout, voltage measurement layout, current
measurement layout, and interface layout for the DSP are designed by the previous lab
members. However, they are permitted to be used in this research project.
148
Appendixes
Appendix B Screen Shot of the PCBs
The following pictures are the experimental boards that were used in this research
project.
(a) Current measurement board.
(b) Voltage measurement board.
149
Appendixes
(c) Gate driver and inverter board.
(d) Extension DSP board.
Figure B-1: Screen shot of the PCBs. (a) Current measurement board, (b) Voltage
measurement board, (c) Gate driver and inverter board, and (d) Extension DSP board.
150
Appendixes
Appendix C Schematic Diagram of Single Inverter System
Figure C-1 shows the full circuit diagram of the power inverter board that includes the
DC blocking circuit.
Galvanic
Lf /2 Isolator
L
T1 T3 O
A
Cf D
VDC C2
/
C1
G
R
T2 T4 I
D
Lf /2
PWM1 PWM2 PWM3 PWM4
GATE DRIVER
&
TMS320F2812
DSP
I ref
(a) power inverter with filter.
(b) DC blocking circuit.
Figure C-1: Circuit diagram of (a) power inverter board and (b) DC blocking circuit.
The low pass filter consists of two inductors, 𝐿𝑓 , of 160 µH each and one 20 µF
capacitor, 𝐶𝑓 . Besides the filter circuit, another small circuit that consists of capacitors
and diodes is connected in series with the output filter. This acts as an isolator to prevent
the flowing of DC current to the AC side of the inverter.
151
Appendixes
Appendix D PWM Generation
The generation of PWM signals can be done using the I/O interface of the DSP. There
are two event manager modules; the Event Manager A (EVA) and Event Manager B
(EVB). Only the EVA is used in this project where it contains pins of PWM1, PWM2,
PWM3, and PWM4 and are used to drive the inverter system. The General Purpose
Timer 1 in this module has a compare register T1CMPR and is used to compare the
PWM signals with the carrier signal. The following statements explain the step by step
way of generating a PWM signal. First, the carrier signal needs to be generated. In
order to do this, the period and frequency have to be set as in equation below
𝐺𝑃𝑇𝐼𝑀𝐸𝑅_𝐹𝑟𝑒𝑞
𝑇1𝑃𝑅 =
2 𝑥 𝑃𝑊𝑀_𝐹𝑟𝑒𝑞
(Eq. D.1)
𝑇1𝑃𝑅 is the amplitude of the carrier signal, 𝑃𝑊𝑀_𝐹𝑟𝑒𝑞 used in this project was
20 kHz and the 𝐺𝑃𝑇𝐼𝑀𝐸𝑅_𝐹𝑟𝑒𝑞 depends on the determination of a few other registers
such as the HISPCP bit 2-0 and T1CON bit 10-8. Below is the adapted diagram to help
with the calculation of 𝐺𝑃𝑇𝐼𝑀𝐸𝑅_𝐹𝑟𝑒𝑞 [85].
HISPCP [2-0]
Sysclkout GPTIMER_Freq
HSPCLK
T1CON [10-8]
Figure D-1: Diagram to determine the 𝑮𝑷𝑻𝑰𝑴𝑬𝑹_𝑭𝒓𝒆𝒒
From the figure, it can be seen that the system clock out, Sysclkout is scaled for two
times in order to get the 𝐺𝑃𝑇𝐼𝑀𝐸𝑅_𝐹𝑟𝑒𝑞. The first scaling depends on bit 2 to bit 0 in
152
Appendixes
High-Speed Peripheral Clock Prescaler (HISPCP). In this project, the corresponding

bits are 000 which is equal to the Sysclkout divided by 1. Sysclkout is equal to the
clock into the CPU that is 150 MHz. So, HSPCLK will be 150 MHz as well. The
second scaling depends on bit 10 to bit 8 in Timer 1 Control Register (T1CON) which
in this project was chosen as 000. This means that HSPCLK must be again divided by
1 and will then give the value of 150 MHz for the 𝐺𝑃𝑇𝐼𝑀𝐸𝑅_𝐹𝑟𝑒𝑞. According to
equation 1, the amplitude of the carrier signal, 𝑇1𝑃𝑅 is 3750.
To get signals from the PWM pins, the compare register T1CMPR need to be loaded
to a value in the range of 0 to T1PR. This will give the duty cycles of the PWM pulses.
For instance, a pulse of 90% duty cycle can be obtained when the value of T1CMPR is
3750 times 0.9. A pulse of 70% duty cycle can be obtained when the value of T1CMPR
is 3750 times 0.7. A pulse of 50% duty cycle can be obtained when the value of
T1CMPR is T1PR times 0.5 which is half of T1PR and so on. The following figure is
the output signal from pin PWM1 which is copied from the oscilloscope images.
(a) 90% duty cycle.
(b) 50% duty cycle.
153
Appendixes
(c) 10% duty cycle.
Figure D-2: Output pin PWM1 with (a) 90% duty cycle, (b) 50% duty cycle, and
(c) 10% duty cycle.
154
Appendixes
Appendix E Dead-time
Often, a short time delay must be applied between the turn on of one Mosfet and the
other in the same leg of the inverter. This is to ensure a complete turn off of the upper
Mosfet before the turning on the lower Mosfet. It prevents overlap current from each
other thus avoiding a shoot-through fault. Shoot-through occurs when both Mosfets are
in the on position at one same particular time. Therefore, a dead-time is introduced. In
order to set the time, the DBTCONA registers have to be properly setup. For simplicity,
below shows the calculation of the dead-time.
Dead-time = p x m x HSPCLK
(Eq. E.1)
where p will depend on bit 4 to bit 2 of DBTCONA registers and m will depend on bit
11 to bit 8 of DBTCONA registers. Previously, it is known that HSPCLK is equal to
150 MHz which will be 6 ns when it is converted to time. Based on the dead-time
formula stated above, a dead-time of approximately 1.7 µs can be set if bit 11 to bit 8
of DBTCONA registers is set to 1000, which is 8 in decimal value, and bit 4 to bit 2 of
DBTCONA registers is set to 110, which gives p of 32. Output of the PWM pins are
shown in Figure E-1.
1.7 µs
Figure E-1: Signals of PWM1 and PWM2 with approximately 1.7 µs dead-time.
155
Appendixes
Appendix F Code Composer Studio (CCS)
This software is the development software tool for this project. Here, the user can write
codes, edit, and debug. Figure below shows the screen shot of the software.
Figure F-1: CCS Software.
After a project has been created and source files have been added, they need to be
linked together before the program can be loaded to the DSP. The DSP must be
connected as a ‘target’ in order to run the program. This software has a window called
the ‘Watch Window’ to watch the program’s variables. There is a tutorial supplied with
the software to get a better understanding from creating to compiling a project. Figure
F-1 is the C programming code that is written in this software and used to generate the
PWM signals. In order to generate PWM signals, a modulating signal has to be created.
This modulating signal is a sine waveform and it has already been written into the
DSP`s memory originally. For the purpose of this research project, the tasks are; to get
a 50 Hz signal with a sampling rate of 20 kHz and to ensure that the peak to peak of
the sine waveform will not equal to or more than 3750, which is the peak value of the
triangular signal. Therefore, in order to change the sampling rate to 20 kHz or 50 µs,
156
Appendixes
and to create a 50 Hz sine wave, the table step size need to be increased to 150. This
works as follows;
i- the original table size is 60000
ii- 0.02 s ÷ 0.00005 s = 400 steps
iii- equation above means that it must have 400 steps from table size 0 to table
size 60000.
iv- that works out to 150 table size increment for each step.
It is known that the amplitude of the original sine wave is 1875. It oscillates between
+1875 to -1875. However the triangular waveform is between the values of 0 to 3750.
In order to get the PWM signals with modulating amplitude, 𝑚𝑎 not equal to or more
than 1, the modulating signal or the sine wave need to be in the same range as the
triangular waveform. To achieve that, a modulating amplitude, 𝑚𝑎 of 0.95 is set. An
offset of 1875 must then be added to the amplitude of the original sine wave and it is
then multiplied by the 𝑚𝑎. A limit is set to ensure that the sine wave will not go beyond
that value. This makes the sine wave oscillates between values of 188 to 3562.
For a unipolar PWM to be generated, two modulating signals of 180° phase difference
from one to another have to be compared with the triangular signal. This is done by
setting the CMPR1 register to the first modulating signal and setting the CMPR2
register to the inversion of the first modulating signal. Figure F-2 shows an example of
the output of PWM1 pin.
Figure F-2: Output of PWM1 pin after the triangular waveform is compared to the
modulating signal (𝒎𝒂 = 0.95 ).
157
Appendixes
Appendix G Testing an Open Loop System
Before a close loop with any controller system is applied, the power inverter board is
tested. A small DC power supply of 10 V is fed to the inverter system and the signals
from the PWM pins are connected to the gate driver board to switch the Mosfets.
Without connecting a load at the output, the voltage of the output filter is observed on
the oscilloscope. It can be seen in Figure G-1 that the amplitude of the waveform is
approximately 10 V, which is to be expected. This confirms that the power inverter
board connections are correct and can be used for further tests.
Figure G-1: Waveform of output filter without load
Following that, a resistive load of approximately 5 Ohms is connected at the output of

the inverter system. A current probe as in the Figure G-2 is used to measure the current
through the load and it is set to 100 mV/A. This setting is then need to be adjusted on
the oscilloscope setting so that the reading is in Ampere. This makes one easily read
the output current reading on the scope without the need to do any calculation later on.
158
Appendixes
Figure G-2: Current probe.
Figure G-3: Waveform of output filter after a resistive load is connected.
From Figure G-1 and Figure G-3, it is clearly seen that the latter has some distortion
on the line and magnitude is reduced. This distortion occurred as a result from using
dead time in the PWM switching. As for the magnitude reduction, it is expected from
the beginning because of the voltage drop.
159
Appendixes
Appendix H Current Sensor
Gain Adjustment
Aiming that this research project will not give an output current of more than 15 A
peak, the gain at the current sensor measurement board has to be adjusted corresponds
with the Analogue to Digital (ADC) channel of the DSP board. This is done by
connecting the current sensor board with a DC supply voltage and a fixed resistive load
at the end where in this case, a load of 2 Ohms is used. Several tests are done with
different DC voltages. For example, a DC voltage of 3 V will give a current of 1.5 A
and should therefore give a 1 V output at the end of the sensor board. In order to get
the 1 V output, the gain in the sensor board need to be adjusted. Following that, with
the adjusted gain, the DC supply voltage is increased to 6 V in order to give the output
current of 3 A. The voltage at the output of the sensor board is then measured and it
must be confirmed to give the value of 2 V. Same step is repeated with this time a DC
supply voltage of 9 V. When the measured output voltage of the sensor board is 3 V,
the current sensor board is ready to be used for this project.
Accuracy
Accuracy of any analogue to digital (ADC) channel measurement is a very important

aspect to look at. It is called the resolution. Using a low resolution of ADC basically
means that the measurement is not accurate and a high resolution ADC will give a very
high accuracy measurement. As the ADC used in this research project is of 14 bits and
in the range of ±10V, the resolution is then can be calculated using the formula as in
Eq. H.1.
160
Appendixes
𝑉𝐹𝑆𝑅
𝑄𝑣 =
2𝑁 − 1
(Eq. H.1)
20
=
214−1
= 1.2 𝑚𝑉
Earlier, it is known that 1 V output measured from the current sensor is equal to 1.5 A
current from the input. Equation above means that there need to be a change of at least
1.2 mV or 0.012 mA in the input for the output to change. This shows that the resolution
of the ADC is very high, accurate, and very acceptable to be used for this research
project.
161
References
References:
[1] R. DeGunther (2009). "Alternative Energy For Dummies." Wiley Publishing.

ISBN 9780470430620.
[2] P. Chiradeja, "Benefit of Distributed Generation: A Line Loss Reduction
Analysis," in Transmission and Distribution Conference and Exhibition: Asia
and Pacific, 2005 IEEE/PES, 2005, pp. 1-5.
[3] Y. Ma et al, "Distributed generation system development based on various
renewable energy resources," in Control Conference (CCC), 2011 30th
Chinese, 2011, pp. 6203-6207.
[4] D. Chen and L. Xu, "DC microgrid with variable generations and energy
storage," in Renewable Power Generation (RPG 2011), IET Conference on,
2011, pp. 1-6.
[5] B. Williams et al, "Using microgrids to integrate distributed renewables into
the grid," in Innovative Smart Grid Technologies Conference Europe (ISGT
Europe), 2010 IEEE PES, 2010, pp. 1-5.
[6] S. Jahdi et al, "Grid integration of wind-solar hybrid renewables using AC/DC
converters as DG power sources," in Sustainable Technologies (WCST), 2011
World Congress on, 2011, pp. 171-177.
[7] L. O. Williams (2002). "An end to global warming". Pergamon, ISBN
9780080440453.
[8] K. F. Katiraei et al, "Solar PV Integration Challenges," Power and Energy
Magazine, IEEE, vol. 9, pp. 62-71, 2011.
[9] REN21, "Renewables 2010 Global Status Report," ed, 2010.
[10] R. Foster et al (2009). "Solar Energy: Renewable Energy and the
Environment". CRC Press. ISBN 978-1420075663.
[11] R. A. Ewing and D. Pratt (2005), "Got sun? go solar: get free renewable energy
to power your grid-tied home". PixyJack Press. ISBN 9780977372461.
[12] F. C. Treble (1991). "Generating electricity from the sun". Pergamon Press.
ISBN 9780080409368.
[13] B. K. Hodge (2010). "Alternative energy systems and applications". Wiley.
ISBN 9780470142509.
162
References
[14] B. M. Hasaneen and A. A. Elbaset Mohammed, "Design and simulation of

DC/DC boost converter," in Power System Conference, 2008. MEPCON 2008.
12th International Middle-East, 2008, pp. 335-340.
[15] C.-T. Tsai and S.-H. Chen, "PV Power-Generation System with a Phase-Shift
PWM Technique for High Step-Up Voltage Applications," International Jurnal
of Photoenergy, vol. Volume 2012, Article ID 838231, 2012.
[16] T.Chaitanya et al, "Modelling and Simulation of PV Array and its Performance
Enhancement Using MPPT (P&O) Technique," International Journal of
Computer Science & Communication Networks, vol. Vol 1, September-October
2011.
[17] C. Hanju and T. K. Vu, "Comparative analysis of low-pass output filter for
single-phase grid-connected Photovoltaic inverter," in Applied Power
Electronics Conference and Exposition (APEC), 2010 Twenty-Fifth Annual
IEEE, 2010, pp. 1659-1665.
[18] J. Selvaraj and N. A. Rahim, "Multilevel Inverter For Grid-Connected PV
System Employing Digital PI Controller," Industrial Electronics, IEEE
Transactions on, vol. 56, pp. 149-158, 2009.
[19] K. Kang-Hoon et al, "The power factor control system of photovoltaic power
generation system," in Power Conversion Conference, 2002. PCC Osaka 2002.
Proceedings of the, 2002, pp. 643-646 vol.2.
[20] J. Selvaraj et al, "Digital PI current control for grid connected PV inverter," in
Industrial Electronics and Applications, 2008. ICIEA 2008. 3rd IEEE
Conference on, 2008, pp. 742-746.
[21] M. Katiamoorthy et al, "A new single-phase PV fed five-level inverter topology
connected to the grid," in Communication Control and Computing
Technologies (ICCCCT), 2010 IEEE International Conference on, 2010, pp.
196-203.
[22] W. Xiaojin et al, "Optimal Design of Structures of PV Array in Photovoltaic
Systems," in Intelligent System Design and Engineering Application (ISDEA),
2010 International Conference on, 2010, pp. 9-12.
[23] M. Ciobotaru et al, "Control of single-stage single-phase PV inverter," in Power
Electronics and Applications, 2005 European Conference on, 2005, pp. 10 pp.-
P.10.
163
References
[24] Z. Zhe et al, "An optimal control method for grid-connected photovoltaic
micro-inverter to improve the efficiency at light-load condition," in Energy
Conversion Congress and Exposition (ECCE), 2011 IEEE, 2011, pp. 219-224.
[25] M. Johns, et al. Grid-Connected Solar Electronics. [Contemporary Energy
Issues].
[26] K. Jung-Min et al, "Grid-Connected Photovoltaic Multistring PCS With PV
Current Variation Reduction Control," Industrial Electronics, IEEE
[27] R. H. Wills et al, "The AC photovoltaic module," in Photovoltaic Specialists
Conference, 1996., Conference Record of the Twenty Fifth IEEE, 1996, pp.
1231-1234.
[28] K. S.K. et al, "Power Quality in Grid connected Renewable Energy Systems:
Role of Custom Power Devices," International Conference on Renewable
Energies and Power Quality (ICREPQ'10), 2010.
[29] M. Calais et al, "Inverters for single-phase grid connected photovoltaic
systems-an overview," in Power Electronics Specialists Conference, 2002. pesc
02. 2002 IEEE 33rd Annual, 2002, pp. 1995-2000.
[30] G. T. Heydt, "Electric power quality: a tutorial introduction," Computer
Applications in Power, IEEE, vol. 11, pp. 15-19, 1998.
[31] "IEEE Recommended Practice for Utility Interface of Photovoltaic (PV)
Systems," IEEE Std 929-2000, 2000.
[32] E. Caamano et al, "State-of -the-art on dispersed PV power generation:
Publications review on the impacts of PV distributed generation and electricity
networks," Intelligent Energy Europe 2007-07-23 2007.
[33] M. Armstrong et al, "Low order harmonic cancellation in a grid connected
multiple inverter system via current control parameter randomization," IEEE
Transactions on Power Electronics, vol. 20, pp. 885-892, 2005.
[34] M. Cespedes and J. Sun, "Renewable Energy Systems Instability Involving
Grid-Parallel Inverters," in Applied Power Electronics Conference and
Exposition, 2009. APEC 2009. Twenty-Fourth Annual IEEE, 2009, pp. 1971-
1977.
[35] George J. Wakileh (2010). "Power Systems Harmonics: Fundamentals,
Analysis and Filter Design". Springer. ISBN 9783642075933.
164
References
[36] J. Schlabbach et al, "Influence of Harmonic System Voltages on the Harmonic

Current Emission of Photovoltaic Inverters," in Power Engineering, Energy
and Electrical Drives, 2007. POWERENG 2007. International Conference on,
2007, pp. 545-550.
[37] G. Chicco et al, "Characterisation and assessment of the harmonic emission of
grid-connected photovoltaic systems," in Power Tech, 2005 IEEE Russia, 2005,
pp. 1-7.
[38] A. Celebi and M. Colak, "The effects of harmonics produced by grid connected
photovoltaic systems on electrical networks."
[39] J. S. Subjak, Jr. and J. S. McQuilkin, "Harmonics-causes, effects,
measurements, and analysis: an update," Industry Applications, IEEE
[40] V. E. Wagner et al, "Effects of harmonics on equipment," Power Delivery,
IEEE Transactions on, vol. 8, pp. 672-680, 1993.
[41] D. G. Infield et al, "Power quality from multiple grid-connected single-phase
inverters," Power Delivery, IEEE Transactions on, vol. 19, pp. 1983-1989,
2004.
[42] I. T. Papaioannou et al, "Harmonic impact of small photovoltaic systems
connected to the LV distribution network," in Electricity Market, 2008. EEM
2008. 5th International Conference on European, 2008, pp. 1-6.
[43] A. Kotsopoulos et al, "Zero-crossing distortion in grid-connected PV inverters,"
Industrial Electronics, IEEE Transactions on, vol. 52, pp. 558-565, 2005.
[44] G. Hong Soo et al, "The effect of grid operating conditions on the current
controller performance of grid connected photovoltaic inverters," in Power
Electronics and Applications, 2009. EPE '09. 13th European Conference on,
2009, pp. 1-8.
[45] "IEEE Recommended Practices and Requirements for Harmonic Control in
Electrical Power Systems," IEEE Std 519-1992, 1993.
[46] M. Armstrong et al, "Auto-calibrating dc link current sensing technique for
transformerless, grid connected, H-bridge inverter systems," IEEE
Transactions on Power Electronics, vol. 21, pp. 1385-1393, 2006.
[47] A. Testa et al, "Interharmonics: Theory and Modeling," Power Delivery, IEEE
165
References
[48] T. Abeyasekera et al, "Suppression of line voltage related distortion in current

controlled grid connected inverters," Power Electronics, IEEE Transactions on,
vol. 20, pp. 1393-1401, 2005.
[49] H. Zhang et al, "Switching harmonics in the DC link current in a PWM AC-
DC-AC converter," in Industry Applications Conference, 1995. Thirtieth IAS
Annual Meeting, IAS '95., Conference Record of the 1995 IEEE, 1995, pp.
2649-2655 vol.3.
[50] S. Ma et al, "On suppression technologies of switching harmonics based on
VVVF," in Control Conference (CCC), 2011 30th Chinese, 2011, pp. 5137-
5141.
[51] I. E. P. Copper Development Association. Power Quality Application Guide:
Harmonics-Interharmonics.
[52] C. Li et al, "Interharmonics: Basic concepts and techniques for their detection
and measurement," Electric Power Systems Research, vol. 66, pp. 39-48, 2003.
[53] Z. Leonowicz, "Analysis of sub-harmonics in power systems," in Environment
and Electrical Engineering (EEEIC), 2010 9th International Conference on,
2010, pp. 125-127.
[54] V. Salas et al, "DC Current Injection Into the Network from PV Grid Inverters,"
in Photovoltaic Energy Conversion, Conference Record of the 2006 IEEE 4th
World Conference on, 2006, pp. 2371-2374.
[55] L. Gertmar et al, "On DC injection to AC grids from distributed generation," in
Power Electronics and Applications, 2005 European Conference on, 2005, pp.
10 pp.-P.10.
[56] R. Gonzalez et al, "High-Efficiency Transformerless Single-phase Photovoltaic
Inverter," in Power Electronics and Motion Control Conference, 2006. EPE-
PEMC 2006. 12th International, 2006, pp. 1895-1900.
[57] H. S. Patel and R. G. Hoft, "Generalized Techniques of Harmonic Elimination
and Voltage Control in Thyristor Inverters: Part I--Harmonic Elimination,"
Industry Applications, IEEE Transactions on, vol. IA-9, pp. 310-317, 1973.
[58] R. Teodorescu et al, "A new control structure for grid-connected LCL PV
inverters with zero steady-state error and selective harmonic compensation," in
Applied Power Electronics Conference and Exposition, 2004. APEC '04.
Nineteenth Annual IEEE, 2004, pp. 580-586 Vol.1.
166
References
[59] S. Xiu'e et al, "Proportional-resonant current control of single-phase grid-tied

PV inverter system," in Electricity Distribution (CICED), 2010 China
International Conference on, 2010, pp. 1-4.
[60] S. Guoqiao et al, "A New Current Feedback PR Control Strategy for Grid-
Connected VSI with an LCL Filter," in Applied Power Electronics Conference
and Exposition, 2009. APEC 2009. Twenty-Fourth Annual IEEE, 2009, pp.
1564-1569.
[61] S. Guoqiao et al, "An Improved Control Strategy for Grid-Connected Voltage
Source Inverters With an LCL Filter," Power Electronics, IEEE Transactions
on, vol. 23, pp. 1899-1906, 2008.
[62] M. Castilla et al, "Linear Current Control Scheme With Series Resonant
Harmonic Compensator for Single-Phase Grid-Connected Photovoltaic
Inverters," Industrial Electronics, IEEE Transactions on, vol. 55, pp. 2724-
2733, 2008.
[63] R. A. Villarreal-Ortiz et al, "Centroid PWM technique for inverter harmonics
elimination," Power Delivery, IEEE Transactions on, vol. 20, pp. 1209-1210,
2005.
[64] L. G. Franquelo et al, "A Flexible Selective Harmonic Mitigation Technique to
Meet Grid Codes in Three-Level PWM Converters," Industrial Electronics,
IEEE Transactions on, vol. 54, pp. 3022-3029, 2007.
[65] H. Qunhai et al, "A new method for the photovoltaic grid-connected inverter
control," in Electric Utility Deregulation and Restructuring and Power
Technologies, 2008. DRPT 2008. Third International Conference on, 2008, pp.
2626-2629.
[66] M. A. A. Younis et al, "Distributed generation with parallel connected
inverter," in Industrial Electronics and Applications, 2009. ICIEA 2009. 4th
IEEE Conference on, 2009, pp. 2935-2940.
[67] M. A. A. Younis et al, "High efficiency THIPWM three-phase inverter for grid
connected system," in Industrial Electronics & Applications (ISIEA), 2010
IEEE Symposium on, 2010, pp. 88-93.
[68] A. M. Roslan et al, "Improved Instantaneous Average Current-Sharing Control
Scheme for Parallel-Connected Inverter Considering Line Impedance Impact in
Microgrid Networks," Power Electronics, IEEE Transactions on, vol. 26, pp.
702-716, 2011.
167
References
[69] M. Jafari et al, "Robust controller design for parallel multi-inverter systems-
synthesis," in Power and Energy Conference, 2008. PECon 2008. IEEE 2nd
International, 2008, pp. 619-624.
[70] M. Khodsuz and A. Sheikholeslami, "Voltage regulation and harmonic
compensation in stand-alone inverter-based distributed generation using non-
sinusoidal hysteresis voltage control," in Power Electronics and Intelligent
Transportation System (PEITS), 2009 2nd International Conference on, 2009,
pp. 44-49.
[71] B. Geethalakshmi et al, "Harmonic compensation using multilevel inverter
based shunt active power filter," in Power Electronics, Drives and Energy
Systems (PEDES) & 2010 Power India, 2010 Joint International Conference
on, 2010, pp. 1-6.
[72] N. Mahamad et al, "Application of LC filter in harmonics reduction," in Power
and Energy Conference, 2004. PECon 2004. Proceedings. National, 2004, pp.
268-271.
[73] A. Zabihinejad and J. S. Moghani, "Implementation of a novel vector-
controlled drive by Direct Injection of Random Signal," in Vehicle Power and
Propulsion Conference, 2008. VPPC '08. IEEE, 2008, pp. 1-5.
[74] L. Bowtell and A. Ahfock, "Comparison between unipolar and bipolar single
phase gridconnected inverters for PV applications," in Power Engineering
Conference, 2007. AUPEC 2007. Australasian Universities, 2007, pp. 1-5.
[75] R. Sharma and J. A. R. Ball, "Unipolar switched inverter low-frequency
harmonics caused by switching delay," Power Electronics, IET, vol. 2, pp. 508-
516, 2009.
[76] T. Abeyasekera, "Improvements to grid connected photovoltaic inverters,"
Newcastle University, Newcastle upon Tyne, 2005.
[77] J. D. P. Gene F. Franklin, Michael L. Workman (1997). "Digital Control of
Dynamic Systems". ISBN 9780201820546.
[78] R. Teodorescu et al, "Proportional-resonant controllers and filters for grid-
connected voltage-source converters," Electric Power Applications, IEE
Proceedings vol. 153, pp. 750-762, 2006.
[79] S. Buso and P. Mattavelli (2006). "Digital Control in Power Electronics".
Morgan & Claypool Publishers. ISBN 9781598291124.
168
References
[80] A. Hasanzadeh et al, "A Proportional-Resonant Controller-Based Wireless

Control Strategy With a Reduced Number of Sensors for Parallel-Operated
UPSs," Power Delivery, IEEE Transactions on, vol. 25, pp. 468-478, 2010.
[81] S. D. Inc., "eZdsp F2812 Technical Reference," 2003.
[82] Z. Ye et al, "PR and harmonic compensation control scheme with current state
feedback for an single-phase Active Voltage Quality Regulator," in Power
Electronics and ECCE Asia (ICPE & ECCE), 2011 IEEE 8th International
Conference on, 2011, pp. 2749-2756.
[83] D. Sera et al, "Low-cost digital implementation of proportional-resonant
current controllers for PV inverter applications using delta operator," in
Industrial Electronics Society, 2005. IECON 2005. 31st Annual Conference of
IEEE, 2005, p. 6 pp.
[84] X. Guo et al, "A Single-Phase Grid-Connected Inverter System With Zero
Steady-State Error," in Power Electronics and Motion Control Conference,
2006. IPEMC 2006. CES/IEEE 5th International, 2006, pp. 1-5.
[85] K. C. B. Mehmet Tumay, Mehmet Ugras CUMA, Ahmet TEKE, "Experimental
Setup For a DSP Based Single-Phase PWM Inverter," 2005.
169

Salimin, S 2014

Uploaded by

Copyright:

Available Formats

Salimin, S 2014

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Salimin, S 2014

Uploaded by

Copyright:

Available Formats

LOW ORDER HARMONICS MITIGATION IN GRID CONNECTED,

Suriana Binti Salimin

A Thesis Submitted for the Degree of

School of Electrical, Electronics, and Computer Engineering

Keywords: Harmonics, current controller, parallel, PV inverter, grid connected system.

Special thanks to my husband, Khairul, as well as my precious forever babies, Danish,

Thank you very much.

Table of Contents........................................................................................................ iii

List of Figures........................................................................................................... viii

List of Abbreviations................................................................................................. xvi

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

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

1.2 Overview on Distributed Generation (DG) ....................................................... 1

1.2.1 Electricity from Solar ................................................................................ 3

1.3 Background of grid connected PV Inverter ...................................................... 4

1.3.1 Parallel Inverter System ............................................................................ 9

1.4 Aims and Objectives ....................................................................................... 10

1.5 Thesis Outline ................................................................................................. 11

1.6 Chapter Summary............................................................................................ 11

Chapter 2: Literature Review .................................................................................... 13

2.1 Chapter Introduction ....................................................................................... 13

2.2 Reviews on Harmonic Injection ...................................................................... 13

2.2.1 Types of Harmonic .................................................................................. 15

2.3 Reviews of Previous Works ............................................................................ 17

2.3.1 Single Connected PV System ................................................................. 17

2.3.2 Parallel Connected PV System ............................................................... 21

2.3.3 General Review on Harmonic Performance Improvements ................... 23

2.4 Research Idea .................................................................................................. 24

2.5 Chapter Summary............................................................................................ 25

Chapter 3: Methodology ........................................................................................... 26

3.1 Chapter Introduction ....................................................................................... 26

3.2 Identification of the system to be studied ....................................................... 26

3.2.1 PWM Technique ..................................................................................... 26

3.2.2 Digital Control System ............................................................................ 32

3.3 Current Controllers.......................................................................................... 33

3.3.1 Conventional PI Control Technique........................................................ 33

3.3.2 Proportional Resonance Controller Scheme (PR) ................................... 35

3.3.3 Modified Proportional Resonance (PR) Technique ................................ 38

3.4 Chapter Summary............................................................................................ 39

Chapter 4: Modelling of Grid Connected, PV Inverter System ................................ 41

4.1 Introduction ..................................................................................................... 41

4.2 Modelling of Grid Connected, Single PV Inverter System............................. 41

4.2.1 Modelling of Low Pass Filter with System Impedance .......................... 42

4.2.2 Modelling of PWM ................................................................................. 47

4.2.3 Modelling of Current Controllers............................................................ 49

4.2.4 Model of Grid Connected, Single PV Inverter System ........................... 54

4.3 Closed Loop Frequency Response .................................................................. 54

4.4 Discussion ....................................................................................................... 61

4.5 Chapter Summary............................................................................................ 61

Chapter 5: Simulations of Grid-Connected Inverter System with PI, PR and

5.1 Chapter Introduction ....................................................................................... 63

5.2 Simulation of a single grid connected inverter ............................................... 63

5.2.1 Conventional PI control technique .......................................................... 67

5.2.2 Proportional resonance (PR) control technique ...................................... 70

5.2.3 Modified PR control technique ............................................................... 74

5.3 Simulation of two grid connected inverters in parallel ................................... 77

5.3.1 Conventional PI Current Controller ........................................................ 79

5.3.2 Proportional resonance (PR) control technique ...................................... 83

5.3.3 Modified PR Control Technique ............................................................. 87

5.4 Chapter Summary............................................................................................ 92