Modeling, Control and Stability Analysis of

VSC-HVDC Links Embedded in a Weak

Multi-Machine AC System

by

Liying WANG

B.E. and M.E., North China Electrical Power University

Thesis submitted for the degree of

Doctor of Philosophy
in
School of Electrical and Electronic Engineering
Faculty of Engineering,
Computer and Mathematical Sciences
The University of Adelaide, Australia
August 2013

Copyright 2013
Liying WANG
All rights Reserved

THE UNIVERSITY
of ADELAIDE
Typeset in Word2010
Liying WANG

Dedicated to my family, my husband, Xike, and my beloved Yaoyi

for their constant support and unconditional love.
I love you all dearly.

ABSTRACT

Abstract
The primary aim of this thesis is to investigate the small-signal dynamic performance
of high voltage direct current (HVDC) transmission links based on voltage source
converter (VSC) technology operating in parallel with the existing longitudinal Australian
power system. This thesis presents the principle design methodology to achieve robust
controllers for VSCs including inner current controller, outer power and voltage controllers
as well as the supplementary damping controllers for enhancing the small-disturbance
rotor-angle stability of a weak multi-machine power system with embedded VSC-HVDC
links.
Three types of linear current controller schemes (proportional-integral, proportionalresonant and Dead-Beat schemes) are investigated and discussed in detail to identify the
most suitable control method. Due to its wider bandwidth and superior performance under
unbalanced operating conditions, the Dead Beat current controller is set as the inner current
controller that has not been analysed in detail in the literature.
A new methodology for the selection and optimization of the parameters of the
proportional-integral compensators in the various control loops of a VSC-HVDC
transmission system using a decoupled control strategy is also proposed in this thesis. It
was found that the new methodology is effective in a relatively strong system. However,
since the method did not take various operating conditions and system disturbances into
account, it will not be effective in a relatively weak system. The analysis shows that the

ABSTRACT

design of robust outer loop controllers is challenging due to the limited bandwidth of the
inner current controller in a weak AC system. Therefore, the second primary objective of
the project was to develop a simple fixed parameter controller, which can perform well
over a wide range of operating points within the active/reactive power (PQ) capability
chart of the VSCs. To achieve this second objective, various grid conditions including
various Short Circuit Ratios (SCRs), different X/R ratios and PQ capabilities of the VSC
system were studied.
To support the primary objectives, a detailed higher order small-signal model of the
DB controlled VSC is developed and systematically verified. As an original contribution,
the study developed a new methodology to linearize the modulator/demodulator blocks
which are used to develop the small signal models for several key components such as the
sampling block, the delay block and the DB inner current controller.
The initial values of the PI/PID compensator parameters are obtained by applying the
classical frequency response design methods to a set of detailed linear models of the openloop transfer functions of the VSC-HVDC control system. It was concluded that an
iterative process may be required after examining the co-operation performance of these
controllers designed.
In the final chapter of this thesis, the small-signal rotor-angle stability of a model of
the Australian power system with embedded VSC based HVDC links was examined. For
the analytical purposes of this thesis a simplified model of the Australian power system is
used to connect the high capacity, but as yet undeveloped, geothermal resource in the
region of Innamincka in northern South Australia via a 1,100 km HVDC link to Armidale
in northern New South Wales. It is observed that the introduction of the new source of
geothermal power generation has an adverse impact on the damping performance of the
system. Therefore, two forms of stabilization are examined: (i) generator power system
stabilisers (PSS) fitted to the synchronous machines which are used to convert geothermal
energy to electrical power; and (ii) power oscillation damping controllers (PODs) fitted to
the VSC-HVDC link. In the case of the PODs two types of stabilizing input signals are
considered: (i) local signals such as power flow in adjacent AC lines and (ii) wide-area
signals such as bus voltage angles at key nodes in the various regions of the system. It was
concluded that the small-signal rotor-angle stability of the interconnected AC/DC system
has been greatly enhanced by employing the designed damping controllers.

ii

STATEMENT OF ORIGINALITY

Statement of Originality

This work contains no material which has been accepted for the award of any other
degree or diploma in any university or other tertiary institution and, to the best of my
knowledge and belief, contains no material previously published or written by another
person, except where due reference has been made in the text.
I give consent to this copy of my thesis when deposited in the University Library, being
made available for loan and photocopying, subject to the provisions of the Copyright Act
1968.
I also give permission for the digital version of my thesis to be made available on the
web, via the Universitys digital research repository, the Library catalogue and also
through web search engines, unless permission has been granted by the University to
restrict access for a period of time.

______________________________

______________________________

Signed

Date

iii

ACKNOWLEDGEMENTS

Acknowledgements
Firstly, I would like to express my sincere appreciation to my supervisor, Associate
Professor Nesimi Ertugrul for all your help, support, guidance, time, and, most of all, the
lessons you have taught me. Special thanks to Mr David Vowles, my co-supervisor, for his
excellent guidance, insightful conversations and endless encouragement throughout the
duration of the research. Working side-by-side with him was an honour and a privilege. I
also extend my gratitude to Professor Boon-Teck Ooi, Mr Jian Hu and Dr Lianxiang Tang
for their assistance in understanding and implementing the Dead-Beat control algorithm.
I am also grateful to the China Scholarship Council (CSC) and the University of
Adelaide (UA) for their financial support of this work. In addition, my thanks go also
towards all the members in the school of Electrical and Electronic Engineering of the
University of Adelaide, in particular Dr Nicolangelo Iannella, Ahmed Abdolkhalig and
Qiming Zhang for their great help with thesis writing, interesting discussions and
providing data analysis tool for producing the technical eigenvector compass plots.
Finally, I would like to thank my parents for the invaluable support they provided
during the period of research. Special thanks to my beloved husband and daughter, Xike.
Thank you so much for your endless love, support and understanding. Thank you, Yaoyi,
for the joy and encouragement you have brought to my life.

iv

TABLE OF CONTENTS

Table of Contents
Abstract ................................................................................................................................. i
Statement of Originality...................................................................................................... iii
Acknowledgements ............................................................................................................. iv
Table of Contents ................................................................................................................. v
List of Figures ................................................................................................................... viii
List of Tables .................................................................................................................... xix
List of Publications ........................................................................................................... xxi
Symbols ........................................................................................................................... xxii
Acronyms ........................................................................................................................ xxiv
Chapter 1:

Introduction ........................................................................................................................ 1

1.1

Background ................................................................................................................ 1

1.2

VSC-HVDC System................................................................................................... 2

1.2.1

VSC Topology ................................................................................................... 2

1.2.2

Operation Principle of VSC ............................................................................... 3

1.2.3

Characteristics and Applications of VSC-HVDC .............................................. 4

1.3

Literature Review ....................................................................................................... 6

1.3.1

Control Strategies of VSC-HVDC ..................................................................... 6

1.3.2

Small Signal Modeling ..................................................................................... 11

1.3.3

Grid Interconnection ........................................................................................ 13

1.4

Thesis Overview....................................................................................................... 15

1.4.1

Research Gap ................................................................................................... 15

TABLE OF CONTENTS

1.4.2
Chapter 2:

VSC-HVDC Control Structures ...................................................................................... 18

2.1

Introduction .............................................................................................................. 18

2.2

Background Study for System Modeling ................................................................. 20

2.2.1

Frame Transformation ...................................................................................... 20

2.2.2

Transient Mathematical System Model ............................................................ 22

2.3

Description of Controllers and Determination of Controller Parameters ................. 24

2.3.1

Proportional Integral Control ............................................................................ 24

2.3.2

Proportional Resonant Control ......................................................................... 39

2.3.3

Dead Beat Control ............................................................................................ 42

2.4

Conclusion ................................................................................................................ 56

Chapter 3:

Analytical Modeling of DB Controlled VSC .................................................................. 58

3.1

Diagram of DB model .............................................................................................. 59

3.2

Characterizing the Components of VSC Small Signal Model .................................. 60

3.2.1

Phase Locked Loop .......................................................................................... 61

3.2.2

Frame Transformation ...................................................................................... 72

3.2.3

Grid Voltage Signal Filter ................................................................................ 76

3.2.4

Reference Current Phase Compensation and Voltage Prediction Block .......... 77

3.2.5

Discrete DB Current Feedback Control Block and Current Input Generation . 78

3.2.6

Characterizing ZOH Block, Sampling Block and Inner Current Control......... 79

3.2.7

Reference Voltage Generation of the Converter............................................. 102

3.2.8

Voltage Source Converter............................................................................... 103

3.2.9

Grid Model ..................................................................................................... 107

3.3

Comparison of the Small Signal Linear Model and PSCAD Simulation ............... 109

3.4

Extension of the DB Controlled VSC Together with DC Link .............................. 112

3.4.1

Model of DC Link .......................................................................................... 112

3.4.2

Interconnection of Subsystems ....................................................................... 115

3.4.3

Model of the DB Controlled VSC Including a DC Link ................................ 117

3.5

Conclusion .............................................................................................................. 118

Chapter 4:

vi

Thesis Outline ................................................................................................... 16

Controller Design for VSC-HVDC Connected to a Weak AC Grid ............................. 120

4.1

Introduction ............................................................................................................ 120

4.2

Issues for VSC-HVDC Embedded in a Weak AC Grid ......................................... 121

4.2.1

Circuit Analysis .............................................................................................. 122

4.2.2

System Stability Analysis based on A Simplified Model ............................... 123

TABLE OF CONTENTS

4.3

Controller Design Methodology............................................................................. 124

4.3.1

Scenarios Considered in Controllers Design .................................................. 125

4.3.2

Calculation of the Steady State Operating Points .......................................... 127

4.3.3

Controller Design for Rectifier ...................................................................... 129

4.3.4

Controller Design for Inverter ........................................................................ 148

4.3.5

Cross-Coupling Effects Examination ............................................................. 154

4.4

Discussions and Conclusions ................................................................................. 163

Chapter 5:

Stability of VSC-HVDC Links Embedded with the Weak Australian Grid .................. 165

5.1

Introduction ............................................................................................................ 165

5.2

Preparation Tasks for Interconnection ................................................................... 167

5.2.1

Admittance Matrix Representation of the Integrated Grid and Filter ............ 168

5.2.2

Scaling the Existing System ........................................................................... 173

5.2.3

DC Link Parameter Sensitivity ...................................................................... 178

5.3

Integrating with the Simplified Australian Grid .................................................... 179

5.3.1 Accuracy Evaluation of Designing VSC Controllers Based on Simplified Grid

Admittance Model.................................................................................................................. 180
5.3.2

Stability Analysis ........................................................................................... 186

5.3.3

Damping Controllers Design .......................................................................... 194

5.4

Conclusion and Discussion .................................................................................... 210

Chapter 6:

Conclusion...................................................................................................................... 212

6.1

Summary ................................................................................................................ 212

6.2

Original Contributions and associated Key Results ............................................... 214

6.2.1

Original Contributions ................................................................................... 214

6.2.2

Associated Key Results of the Thesis ............................................................ 214

6.3

Suggestions for Further Research .......................................................................... 217

Appendix A: VSC-HVDC System Parameters ................................................................ 219

Appendix B: Method (II) for Elimination of Modulator/De-modulator System ............. 221
Appendix C: Method (I) for Elimination of Modulator/Demodulator System ................ 225
Appendix D: Lead-Lag Block.......................................................................................... 231
Appendix E: Regional Boundaries for the National Electricity Market & Committed
Developments............................................................................................................................. 233
Appendix F: Performance Evaluation of all the Power and Voltage Controllers with a New
set of DC Link Parameters ......................................................................................................... 235
Appendix G: VSC-HVDC System (II) Parameters ......................................................... 243
Reference ......................................................................................................................... 245

vii

LIST OF FIGURES

List of Figures
Figure 1-1 A typical VSC topology connected to the AC grid (a) simplified diagram,
(b) the details of the circuit including three phase two-level VSC using IGBTs................... 3
Figure 2-1 Current control classifications .................................................................. 19
Figure 2-2 Representation of rotating vector in converter including stationary
reference frame and abc natural reference frame. .......................................................... 21
Figure 2-3 Representation of rotating vector in dq reference frame and eference
frame .................................................................................................................................... 22
Figure 2-4 Single line diagram representation of VSC-HVDC .................................. 22
Figure 2-5 (a) The diagram of the inner and outer controllers; (b) Inner current
control loop. ......................................................................................................................... 25
Figure 2-6 Flow chart of the process for determining the PI compensator parameters
for VSC-HVDC control system ........................................................................................... 27
Figure 2-7 Detailed inner current control loop in pu system ...................................... 28
Figure 2-8 Block diagram of DC voltage control scheme in pu system..................... 29
Figure 2-9 (a) Block diagram of active power control scheme in pu system; (b) block
diagram of reactive power control scheme in pu system. .................................................... 30
Figure 2-10 VSC-HVDC system diagram .................................................................. 32

viii

LIST OF FIGURES

Figure 2-11 Open-loop Bode plots of the current controller transfer function (i)
without PI control (blue solid line); (ii) with PI compensation using initial parameters
(green dashed line). .............................................................................................................. 33
Figure 2-12 Step responses of current controller transfer function (i) without PI
control (blue solid line); (ii) with PI compensation using initial parameters (green dashed
line) ...................................................................................................................................... 34
Figure 2-13 Open-loop Bode plots of the outer DC controller transfer-function (i)
without PI compensation (solid blue line); and (ii) with PI compensation using the initial
parameters (dashed green line). ........................................................................................... 35
Figure 2-14 Step responses of the outer DC controller transfer-function (i) without PI
compensation (solid blue line) and (ii) with PI compensation using the initial parameters
(dashed green line). .............................................................................................................. 35
Figure 2-15 Open-loop Bode plots of the outer active/reactive power controller
transfer-function (i) without PI compensation (blue solid line) and (ii) with PI
compensation using the initial parameters (dashed green line) ........................................... 36
Figure 2-16 Responses of VSC-HVDC following a reversal of the power order for (i)
the initial set of PI compensator parameters (red line); (ii) the parameters obtained with the
optimization function (2-34) (blue line) and (iii) the parameters obtained with the refined
optimization function (2-35) (cyan line). The reference values of the variables are shown in
green line.............................................................................................................................. 37
Figure 2-17 Structure of the paralleled harmonic compensators ............................... 40
Figure 2-18 Diagram of reference frame mathematical model of VSC system .... 41
Figure 2-19 The simplified model of PR current controlled VSC system in
reference frame .................................................................................................................... 41
Figure 2-20 The categories of DB current controllers ............................................... 42
Figure 2-21 Digital DB current controlled VSC system ............................................ 43
Figure 2-22 The closed-loop DB current Control ...................................................... 44
Figure 2-23 Pole-zero map of the one sample delay system: red cross: one sample
delay DB without considering the computation delay time; blue cross: represents one
sample delay DB considering the computation delay time; green cross: represents reducing
the proportional gain; ........................................................................................................... 45
Figure 2-24 One sample delay DB control with Smith Predictor ............................ 46

ix

LIST OF FIGURES

Figure 2-25 Internal model control design for DB implementation block ................. 47
Figure 2-26 The block diagram of solving feedback transfer function DB current
control .................................................................................................................................. 48
Figure 2-27 The structure of the solving feedback transfer function DB current
controller .............................................................................................................................. 48
Figure 2-28 Frequency response and step response of various DB current controllers
.............................................................................................................................................. 49
Figure 2-29 PSCAD step response simulation results of the DB current controller .. 54
Figure 3-1 Small signal model of VSC with the discrete DB current controller........ 60
Figure 3-2 Structure of phase locked loop.................................................................. 62
Figure 3-3 Non-linear model for PLL ........................................................................ 65
Figure 3-4 Block diagram of PLL linearized model................................................... 66
Figure 3-5 Frequency response of PLL in the case study........................................... 68
Figure 3-6 Linearized model verification, 1% step change in phase.......................... 68
Figure 3-7 50% step (180o) change in phase .............................................................. 69
Figure 3-8 1% step change in frequency .................................................................... 70
Figure 3-9 PLL responses to 90% magnitude step change of input voltage .............. 70
Figure 3-10 PLL output phase angle in comparison with input phase angle when
being subject to a sudden 90% magnitude step change of input voltage ............................. 70
Figure 3-11 PLL output frequency in comparison with input frequency when being
subject to a sudden 90% magnitude step change of input voltage ....................................... 71
Figure 3-12 PLL output phase angle when being subject to an A-phase to ground
fault ...................................................................................................................................... 71
Figure 3-13 PLL output frequency when being subject to an A-phase to ground fault
.............................................................................................................................................. 72
Figure 3-14 Relationship between converter reference frame and grid RI reference
frame .................................................................................................................................... 73
Figure 3-15 Controlled voltage source representation with an inductance ................ 74
Figure 3-16 Test results for grid RI reference frame transformation ......................... 74
Figure 3-17 The test circuit for small signal mathematical equation ......................... 75
Figure 3-18 Small signal test results for the grid frame transformation ..................... 75
Figure 3-19 Current compensation block verification ................................................ 77

LIST OF FIGURES

Figure 3-20 Frequency responses of pure one sample delay...................................... 80

Figure 3-21 Step responses of pure one sample delay ............................................... 81
Figure 3-22 Demonstration of ZOH function............................................................. 82
Figure 3-23 A typical sampled data system structure ................................................ 83
Figure 3-24 Composition of a rectangular impulse .................................................... 83
Figure 3-25 Frequency responses of ZOH block and its approximation ................... 84
Figure 3-26 Step responses of ZOH block and its approximation ............................. 85
Figure 3-27 Verification of responses to the sinusoid input for ZOH by Matlab ...... 86
Figure 3-28 Responses to the sinusoid input for ZOH by PSCAD ............................ 86
Figure 3-29 Comparison of the simulation results ..................................................... 87
Figure 3-30 Frequency response of inner current controller and its approximations 88
Figure 3-31 Step response of inner DB current controller and its approximation ..... 89
Figure 3-32 Implementation of the inner DB current controller scheme in continuous
domain and in the z domain ................................................................................................. 90
Figure 3-33: Comparison of the responses of (i) the discrete DB current controller
and (ii) its first order PadApproximation to a sinusoidal input signal. ............................. 91
Figure 3-34 The abc natural reference frame DB current control block .................... 91
Figure 3-35 The simplified abc natural reference frame DB current control block ... 92
Figure 3-36 The simplified reference frame DB current control block ................ 92
Figure 3-37 Simplified dq reference frame DB current control block including the
time-varying terms (modulator/de-modulator) .................................................................... 92
Figure 3-38 Modulator/de-modulator system............................................................. 93
Figure 3-39 Transfer function representation of the modulator/de-modulator system
transform from

reference frame to dq reference frame .................................................. 96

Figure 3-40 Small signal model for DB current controller in equivalent dq reference
frame .................................................................................................................................... 98
Figure 3-41 Compare Yd/Yq outputs from (i) the modulator/de-modulator system; (ii)
the small-signal equivalent transfer function and iii) the small-signal equivalent state space.
(a)UdSTEP = 0, UqSTEP = -0.02 and (b) UdSTEP = -0.02, UqSTEP = 0; ....................................... 99
Figure 3-42 Difference between Yd/Yq outputs from (i) the modulator/de-modulator
system versus the small-signal equivalent transfer function; (ii) the modulator/de-

xi

LIST OF FIGURES

modulator system versus the small-signal equivalent state space. (a) UdSTEP = 0, UqSTEP = 0.02 and (b) UdSTEP = -0.02, UqSTEP = 0; .............................................................................. 99
Figure 3-43 Compare Yd/Yq outputs from the (i) modulator/de-modulator system; (ii)
the small-signal equivalent transfer function and iii) the small-signal equivalent state space
............................................................................................................................................ 102
Figure 3-44 Test circuit for the detailed VSC model ............................................... 106
Figure 3-45 The equivalent circuit diagram of grid ................................................. 107
Figure 3-46 Small signal model of the grid .............................................................. 108
Figure 3-47 (a) Grid side current and (b) filter bus voltage responses of i) large signal
model ii) small signal model following a input voltage step change order on VcqC = 0.56kV. ............................................................................................................................... 109
Figure 3-48 Comparison of the converter output current ICcd , ICcq using the (i) large
signal model simulating with VSC simplified as a controlled voltage source; (ii) the smallsignal model for AC system with different SCRs. ............................................................. 111
Figure 3-49 Comparison of the converter output currents ICcd , ICcq from the (i) large
signal model using the detailed VSC model, and (ii) the small-signal model for the weak
AC system .......................................................................................................................... 112
Figure 3-50 The model DC transmission link .......................................................... 113
Figure 3-51 Test circuit for DC link test .................................................................. 114
Figure 3-52 The converter currents behind the DC capacitor from both of the rectifier
side and inverter side (circle: Large signal model; rectangular: Small signal model) ....... 114
Figure 3-53 Demonstration of sub-modular interconnection ................................... 115
Figure 3-54 The base model for the converter side controller ................................. 117
Figure 3-55 Comparison of the converter output currents ICcd , ICcq using the (i)
large-signal model simulated in the detailed VSC model including the DC link, and (ii) the
small-signal model for a weak AC system together with the DC link ............................... 118
Figure 4-1 Single phase equivalent circuit ............................................................... 122
Figure 4-2 Simplified DB current controlled VSC ................................................... 123
Figure 4-3 Zeros/poles in z-domain for (a) weak system (b) strong system of a DB
current controlled VSC system .......................................................................................... 124

xii

LIST OF FIGURES

Figure 4-4 (a) Bode plots and (b) step responses of a DB current controlled VSC
system ................................................................................................................................ 124
Figure 4-5 Flow chart of the controller design methodology ................................... 125
Figure 4-6 PQ capability chart (a) demonstration chart; (b) defined operating points
........................................................................................................................................... 126
Figure 4-7 Single-line diagram of AC side converter including a high-pass filter .. 129
Figure 4-8 Influence of the DC link model: (a) pole-zero map of Iq(s)/Iqref(s); (b)
Bode plots of Iq(s)/I qref(s) .................................................................................................. 129
Figure 4-9 Basic structure for outer power controller design .................................. 130
Figure 4-10 (a) In the DB current controlled VSC with P=Pmax (a) eigenvalue maps
with SCR reduced from 7.5 to 0.5 at a step of -1 (i.e. A to G); (b) Bode plots of the TF
(blue: SCR=7.5, red: SCR=4.5, cyan: SCR=2). ................................................................ 133
Figure 4-11 The time-domain simulation results: PSCAD SCR=7.5 (solid cyan line);
Matlab SCR=7.5 (dash-dot blue line); PSCAD SCR=2 (solid black line); Matlab
SCR=2(dash-dot red line). ................................................................................................. 134
Figure 4-12 Four main low frequency modes for A: rectifier mode with a 90o angle;
B: rectifier mode with a 75o angle; C: rectifier mode with a 60o angle; a: inverter mode
under 90o angle; b: inverter mode under 75o angle; c: inverter mode under 60o angle; .... 135
Figure 4-13 The Bode plots of the DB current controlled VSC. blue: rectifier mode
under 90o angle; cyan: Rectifier mode under 75o angle; red: rectifier mode under 60o angle;
magenta: inverter mode under 90o angle; black: inverter mode under 75o angle; green:
inverter mode under 60o angle; .......................................................................................... 135
Figure 4-14 Step responses of DB current controlled VSC for the rectifier (left) and
the inverter (right). rectifier mode at angle 90o (blue); rectifier mode at angle 75o (cyan);
rectifier mode at angle 60o (red); inverter mode at angle 90o (magenta); inverter mode at
angle 75o (black); inverter mode at angle 60o (green); ...................................................... 136
Figure 4-15 For DB controlled VSC with SCR=2 (a) the eigenvalue map with P
equals to be 1pu (A), 0.5pu (B) and 0pu (C); (b) Bode plots of the TF with different
loading conditions. blue: P=1pu, cyan: P=0.5pu, red: P=0pu). ......................................... 137
Figure 4-16 (a) The eigen-value map and (b) Bode plot of the transfer function for
the DB current controlled VSC at SCR=2 and different power factors (Q=Qmax, Q=0 and

xiii

LIST OF FIGURES

Q=Qmin) under full load condition. (dark blue: lagging power factor, light blue: unity
power factor, red: leading power factor). ........................................................................... 138
Figure 4-17 The analysis of the transfer functions using Bode plots and at 36
operating conditions in total (a) SCR=2 at 90o (b) SCR=2 at 75o; (c) SCR=7.5 at 90o; (d)
SCR=7.5 at 75o; ................................................................................................................. 139
Figure 4-18 Right half-plane zeros for the transfer function of the open loop power
controller ............................................................................................................................ 141
Figure 4-19 Bode plots of the selected cases for the power controller design ......... 142
Figure 4-20 The Bode plots with (a) low-pass filter; (b) DB current controlled VSC
with a low-pass filter; (c) designed PI compensator; (d) open loop transfer function of
P(s)/Pref(s). .......................................................................................................................... 143
Figure 4-21 Step responses of P(s)/Pref(s) ................................................................ 143
Figure 4-22 Control block for the power controller ................................................. 144
Figure 4-23 A typical MIMO system, in which the hidden feedback loop is shown in
red lines .............................................................................................................................. 144
Figure 4-24 Bode plots of selected cases in the AC voltage controller design ........ 146
Figure 4-25 The Bode plots of (a) low-pass filter; (b) Uf(s)/Idref (s) of DB current
controlled VSC with a low-pass filter; (c) the designed PI compensator and (d) open loop
transfer function of Uf(s)/Ufref(s) at the rectifier side. ........................................................ 147
Figure 4-26 Step responses of Uf(s)/Ufref(s) ............................................................. 148
Figure 4-27 Control block for AC voltage controller ............................................... 148
Figure 4-28 The frequency responses of (a) Udc(s)/Iqref(s); (b) Udc(s)/Idref(s); (c)
Uf_inv(s)/ Iq_ref_inv(s) and (d) Uf_inv(s)/Id_ref_inv(s)

being subject to the operating point

variation of the rectifier side converter (36 scenarios), but with a constant operating point
at the inverter side converter that is PmaxQ0, SCR=2 and with angle equals to 90o. .......... 149
Figure 4-29 (a) The Bode plots of the selected cases of the DC voltage controller
design Udc(s)/Iqref_inv(s); (b) designed PID compensator (Kp_dc+Ki_dc/s+Kd_dcs/(1+TfUdcs)=0.024-0.088/s+0.0009s/(1+0.06s)); (c) frequency responses of open loop transfer
functions of Udc(s)/ Udc_ref (s) and (d) step responses of the closed-loop Udc(s)/ Udc_ref(s).
............................................................................................................................................ 151
Figure 4-30 The control block diagram of the DC voltage controller ...................... 151

xiv

LIST OF FIGURES

Figure 4-31(a) Bode plots of selected cases for inverter end AC voltage controller
design Uf_inv(s)/Idref_inv(s); (b) designed filter+PI compensator; (c) frequency responses of
open loop transfer functions of Uf_inv(s)/Uf_inv_ref(s); (d) step responses of closed-loop
Uf_inv(s)/ Uf_inv_ref(s)............................................................................................................ 153
Figure 4-32 1% step of AC voltage reference at the rectifier end; (a) responses of
power at the rectifier end; (b) responses of the filter bus voltage Uf_rec at the rectifier end;
(c) responses of DC voltage Udc_inv at the inverter end; (d) responses of the filter bus
voltage Uf_inv at the inverter end. ....................................................................................... 155
Figure 4-33 1% step of power reference at the rectifier end; (a) responses of the
power at the rectifier end ; (b) responses of the filter bus voltage Uf_rec at the rectifier end;
(c) responses of the DC voltage Udc_inv at the inverter end; (d) responses of the filter bus
voltage Uf_inv at the inverter end. ....................................................................................... 156
Figure 4-34 1% Step of the AC voltage reference at the inverter end; (a) response of
the power at the rectifier end; (b) response of filter bus voltage Uf_rec at the rectifier end; (c)
response of the DC voltage Udc_inv at the inverter end and (d) response of the filter bus
voltage Uf_inv at the inverter end. ....................................................................................... 157
Figure 4-35 1% step of the DC reference voltage at the rectifier side; (a) the
responses of power at the rectifier; (b) the responses of the filter bus voltage Uf_rec at the
rectifier; (c) the responses of the DC voltage Udc_inv at the inverter end; (d) the responses of
the filter bus voltage Uf_inv at the inverter end. .................................................................. 158
Figure 4-36 The step responses of Ufinv_ref (a) and Udcinv_ref (b) and their induced
performances in their cross-coupling loops as the system working at an operating point
PmaxQmax, SCR=2 with =90o. red: Matlab 1%; cyan: PSCAD 1%; black: PSCAD 2%/2;
magenta: PSCAD 4%/4; green: PSCAD 8%/8. ................................................................. 161
Figure 4-37 The step responses of Udcinv_ref (a) and Ufinv_ref (b) and their induced
performances in their cross-coupling loops as the system working at an operating point
P0Qmin, SCR=7.5 with =75o. red: Matlab 1%; cyan: PSCAD 1%; black: PSCAD 2%/2;
magenta: PSCAD 4%/4; green: PSCAD 8%/8. ................................................................. 162
Figure 5-1 The diagram of the extended Simplified South-East Australian power grid
with VSC-HVDC links. ..................................................................................................... 167
Figure 5-2 Grid model as an admittance .................................................................. 168

xv

LIST OF FIGURES

Figure 5-3 1% DC voltage step responses of the dynamic and grid admittance models
(a) for rectifier side; (b) for inverter side. .......................................................................... 170
Figure 5-4 The grid and filter model as an admittance............................................. 171
Figure 5-5 1% DC voltage step response comparison of the admittance representation
of grid model only, and the gird and filter models adopting admittance representation: (a)
the rectifier side; (b) the inverter side. ............................................................................... 172
Figure 5-6 System parameter scaling scheme ......................................................... 173
Figure 5-7 Scaling of inner current controllers ........................................................ 174
Figure 5-8 Scaling of PI controller with additional filter ......................................... 176
Figure 5-9 Scaling of the PID controller .................................................................. 176
Figure 5-10 Flow chart of the verification methodology of modeling ..................... 177
Figure 5-11 Eigenvalue map of the original and scaled systems ............................. 178
Figure 5-12 The comparison of the system eigenvalues between the new updated DC
link system and the original system ................................................................................... 179
Figure 5-13 The frequency responses for the rectifier side (a) and for inverter side (b)
for the higher order grid impedance models (I to V). ........................................................ 181
Figure 5-14 The open loop frequency responses, P_rec/P_ref ..................................... 182
Figure 5-15 The step responses test of the power reference at the rectifier side ...... 183
Figure 5-16 The open loop frequency responses, Uf_ref_rec/Uf_rec .............................. 183
Figure 5-17 The step response test of the AC voltage reference at the rectifier side
............................................................................................................................................ 184
Figure 5-18 The open loop frequency responses, U_dc_ref_rec/U_dc _rec ....................... 184
Figure 5-19 The step response test of the DC voltage reference at the inverter side
............................................................................................................................................ 185
Figure 5-20 The open loop frequency responses, Uf_ref_inv/Uf_inv ............................ 185
Figure 5-21 The step response of the AC voltage reference at the inverter side ...... 186
Figure 5-22 Eigenvalue map of the interconnected system ...................................... 187
Figure 5-23 Right eigenvector prototype of mode I40 ............................................. 188
Figure 5-24 Participation factor for mode I40 .......................................................... 189
Figure 5-25 Right eigenvector prototype of mode I35 ............................................. 190
Figure 5-26 Participation factor for mode I35 .......................................................... 190
Figure 5-27 Right eigenvector prototype of mode I25 ............................................. 191

xvi

LIST OF FIGURES

Figure 5-28 Participation factor for mode I25 .......................................................... 192

Figure 5-29 Right eigenvector prototype of mode I15 ............................................. 193
Figure 5-30 Participation factor for mode I15 .......................................................... 193
Figure 5-31Eigenvalue evolution for the PSS damping gain on each generator is
increased from zero (no PSS in service) to 30 pu on machine base (750MVA) in 5 pu steps.
........................................................................................................................................... 194
Figure 5-32 Bode plot for DAMP(s) with DPSS=20pu on Mbase 750 MVA, Tw=3s and
Tp=0.05s. ............................................................................................................................ 196
Figure 5-33 PVr characteristic and curve fitting for Innamincka generator #1 ....... 197
Figure 5-34 The comparison of the rotor modes of the system with the equivalent
damping torque (blue star) and the system fitted with PSS (red dot) under the high loading
condition and light loading condition. The damping gain in Innamincka generators are
increased from zero (no PSS in service) to 30 pu in 5 pu steps. ........................................ 198
Figure 5-35 Comparison on the power output of generator #1 from Innamincka under
the condition with and without PSS in service by applying a small disturbance 0.01pu on
the reference voltage of Innamincka generator #1 ............................................................. 198
Figure 5-36 (a) Damping torque coefficients introduced by the PSS in the
Innamincka generator #1 (red curve) compared with the specified damping torque
coefficient (blue curve) (b) induced synchronizing torque (De equals to be 20 pu on Mbase)
........................................................................................................................................... 199
Figure 5-37 The schematic diagram of the VSC-HVDC damping system .............. 200
Figure 5-38 Frequency response of lead compensator Q(s) for the local POD ....... 203
Figure 5-39 The eigenvalue map of system fitted with POD, where the letters
represent the gains of the POD and KSS is increased from 3.642 10-3 to 7.03642 10-2 with a
step size of 0.01. ................................................................................................................ 204
Figure 5-40 The eigenvalue map of system fitted with POD and PSS. Here the letters
are the gains of the POD and KSS is increased from 3.642 10-3 to 7.03642 10-2 with a step
size of 0.01. ........................................................................................................................ 205
Figure 5-41 Power output of Innamincka generator #1 following a 0.01 pu step
change on the voltage reference of Innamincka generators. .............................................. 205

xvii

LIST OF FIGURES

Figure 5-42 The performance evaluation of the VSC controllers with the
supplementary controllers in service following a step change of 0.01 pu on power reference
of the rectifier side converter. ............................................................................................ 206
Figure 5-43 Frequency response of lead compensator Q(s) of the WAPOD ........... 207
Figure 5-44 The eigenvalue map of system fitted with WAPOD under high loading
condition, where the letters represent the gains of the WAPOD, as KSS is increased from
3.64210-4 to 0.12 with a step size of 0.02. ........................................................................ 207
Figure 5-45 The eigenvalue map of system fitted with WAPOD under light loading
condition, where the letters represent the gains of the WAPOD, while KSS is increased
from 3.64210-4 to 0.12 with a step size of 0.02. ................................................................ 208
Figure 5-46 Eigenvalue analysis of WAPOD with PSS under high loading condition,
where KSS is increased from 3.642 10-4 to 0.12 with a step size of 0.02 .......................... 209
Figure 5-47 Power outputs of Innamincka generator #1 following a step change of
0.01 pu on voltage reference of Innamincka generators. ................................................... 209
Figure 5-48 The performance evaluation of the VSC controllers with PSS and
WAPOD in service following a step change of 0.01 pu on power reference of the rectifier
side converter. .................................................................................................................... 210

xviii

LIST OF TABLES

List of Tables
Table 1-1 Categories of SCR........................................................................................ 8
Table 2-1 PI parameters comparison .......................................................................... 38
Table 2-2 Open loop transfer function of DB current controllers .............................. 49
Table 2-3 Sensitivity to plant parameters ................................................................... 50
Table 2-4 Simulation parameter ................................................................................. 51
Table 2-5 Summary of specific data obtained from Figure 2-29 ............................... 54
Table 2-6 Mean derivation of simulation results........................................................ 56
Table 3-1 PLL parameters .......................................................................................... 67
Table 3-2 Characteristics of the linearized model ...................................................... 69
Table 3-3 Characteristics of the step response corresponding to the pure one sample
delay ..................................................................................................................................... 81
Table 3-4 Characteristics of the step response corresponding to ZOH ...................... 85
Table 3-5 Summary of transfer functions for DB current controller, sampling block
and delay block .................................................................................................................... 97
Table 3-6 Equivalent dq reference frame transfer function for the lead-lag block .... 98
Table 3-7 State space equations for DB controller in equivalent dq reference frame
........................................................................................................................................... 100
Table 3-8 Simulation results for VSC model verification ........................................ 107

xix

LIST OF TABLES

Table 3-9 Summary of the large signal model of the AC grid in abc reference frame
and small signal model in dq reference frame ................................................................... 108
Table 3-10 Large and small signal for DC link ........................................................ 113
Table 3-11 Parameters for the DC link test .............................................................. 114
Table 4-1 Equivalent Thevenin impedance of the AC grid ...................................... 127
Table 4-2 Basic power flow equation ....................................................................... 127
Table 4-3 Equations for the power flow study ......................................................... 128
Table 4-4 Modes of the inner current controller (SCR=2) ....................................... 131
Table 4-5 Participation factors of dominant state variables for selected modes of the
inner DB current controller ................................................................................................ 132
Table 5-1 The units derivation for the PID coefficients ........................................... 177
Table 5-2 Inter-area modes of the integrating system .............................................. 186
Table 5-3 Approximate improvements on system damping: comparison of the results
obtained from adding with equivalent damping torque (Figure 5-31) and the analysis for
participation factor. ............................................................................................................ 194
Table 5-4 Residue analysis of the system ................................................................. 202

xx

LIST OF PUBLICATIONS

List of Publications
Published
[1] W. Liying and N. Ertugrul, "Selection of PI compensator parameters for VSC-HVDC
system using decoupled control strategy," in Universities Power Engineering Conference
(AUPEC), 2010 20th Australasian, pp. 1-7.
[2] W. Liying, N. Ertugrul, and M. Kolhe, "Evaluation of dead beat current controllers for
grid connected converters," in Innovative Smart Grid Technologies - Asia (ISGT Asia),
2012 IEEE, 2012, pp. 1-7.

Papers in Preparation
[1] W. Liying, David J. Vowles and N. Ertugrul, "Reference Frame Transformation
Approach for Small Signal Modeling of VSC with Stationary Frame Controllers"
[2] W. Liying, David J. Vowles and N. Ertugrul, " Generalized small signal modeling of
DB controlled VSC"
[3] W. Liying, David J. Vowles and N. Ertugrul, "Robust controllers design for DB
controlled VSC linked to a weak AC system"
[4] W. Liying, David J. Vowles and N. Ertugrul, "Damping controller design of DB
controlled VSC operating in parallel with a Weak Multi-machine AC Power System "

xxi

SYMBOLS

Symbols

xxii

system matrix

filter-bus values of the VSC

active power

ac

alternating current quantity

base value

capacitance

converter terminal values of the VSC

current

d component in the dq frame

De

damping torque

Kd

derivative gain

dc

direct current quantity

feedthrough matrix

imaginary component in the grid RI frame

inductance

input matrix

SYMBOLS

Ki

integral gain

Laplace factor

max

maximum value

min

minimum value

operating point value

output matrix

peak

peak value

phase angle

abc

phase quantities

PLL

PLL quantity

Kp

proportional gain

q component in the dq frame

quantity in converter dq frame

quantity in grid RI frame

reactive power

real component in the grid RI frame

ref

reference value

Ts

sampling time constant

source values of the interconnected grid

values of controlled plant

db

values of DB current controller

Kv

gain of the voltage controlled oscillator

voltage

component in the frame

component in the frame

xxiii

ACRONYMS

Acronyms

xxiv

AC

Alternating Current

AVM

Average Value Modeling

AVR

Automatic Voltage Regulator

DB

Dead Beat

DC

Direct Current

DG

Distributed Generation

EMT

Electro-Magnetic Transient

FIR

Finite Impulse Response

GM

Gain Margin

GPS

Global Positioning System

GTOs

Gate Turn-off Devices

H Infinity

HSV

Hankel Singular Value

IGBT

Insulated Gate Bipolar Transistors

Im

Imaginary

IMC

Internal Model Control

ITAE

Integral of the Time Absolute-Error Products

ACRONYMS

KCL

Kirchhoff's Current Law

KVL

Kirchhoff's Voltage Law

LCC

Line-Commutated Control

LL

Lead-Lag

LMI

linear Matrix Inequality

LTI

Linear Time Invariant

Max

Maximum

MIMO

Multi-Input Multi-Output

Min

Minimum

MTDC

Multi-Terminal Direct Current

PCC

Point of Common Coupling

PM

Phase Margin

PI

Proportional-Integral

PLL

Phase Lock Loop

POD

Power Oscillation Damping

PQ

Active-/Reactive- Power

PR

Proportional-Resonant

PSS

Power System Stabilizer

pu

Per Unit

PWM

Pulse Width Modulation

Re

Real

RHP

Right Half Plane

SCR

Short Circuit Ratio

SISO

Single-In Single-Out

SVCs

Static Var Compensators

TF

Transfer Function

VCO

Voltage Controlled Oscillator

VSC-HVDC

Voltage Source Converter- High Voltage Direct Current

WAPOD

Wide Area Power Oscillation Damping

ZOH

Zero Order Holding

xxv

Chapter 1: Introduction
1.1 Background
As is known, the limited availability and environmental concern of fossil fuels, as well
as the continuous growing demand of electricity, have caused renewable energy to become
commercially attractive. However, the integration of large scale renewable energy sources
into the power grid changes the characteristics of the supply in terms of the composition of
energy generation, the transmission network, the technology and the economics of the
electricity industry. This will be very different from the current situation in which fossilfuel generation occupies the dominant position [1].
To allow the large scale integration of renewable energy sources into the grid, VSCHVDC is a commonly developed power electronic converter. This converter topology is
particularly suitable for long distance power transmission, such as in offshore wind farms.
Since VSCs do not require commutating voltage from the connected AC grid, they are also
highly effective in supplying power to isolated and remote loads and convenient when
interconnecting distributed sources.

CHAPTER 1: INTRODUCTION

The existing transmission networks and power stations will not be able to meet the
growing challenges of renewable energy transmission. The increasing penetration of power
electronic converters into existing power systems have also caused several power system
stability issues [2, 3]. The modern VSC-HVDC technology can help to solve the adverse
impact on system stability and power quality issues. Furthermore, it can also significantly
increase the transmission capacity and can provide flexible control of power flow.
Therefore, it is envisaged that there is a strong need to investigate the design, operation and
control of VSC-HVDC transmission links. This includes the links integration into the
existing power network which may be notably weak as in the Australian case which is
investigated in this thesis.

1.2 VSC-HVDC System

1.2.1 VSC Topology
A typical VSC converter station consists of a voltage source converter, a reactor, a
transformer, a filter and a DC capacitor as shown in Figure 1-1a.
The detailed VSC circuit diagram is also given in Figure 1-1b which includes a
conventional three phase six-switch converter. A number of series switching devices such
as insulated gate bipolar transistors (IGBTs) or gate turn off devices (GTOs) can be used to
increase the voltage blocking capability of the converter and hence increase the voltage
rating of the HVDC transmission system. Note that the anti-parallel diodes are required in
the converter to allow possible voltage reversals caused by changing external circuit
conditions, thus facilitating four-quadrant operation of the converter. Note also that such
voltage source converter can be treated as a voltage source from the AC side. The VSC
controls the angle and voltage difference across the reactor and possibly transformer so as
to regulate the AC power and reactive power delivered by the converter. The reactor also
filters the AC harmonics. Reactive power compensation devices are not required as the
converter can control the reactive power generation. Note that only a small AC filter will
be required to eliminate higher order harmonics. The VSC acts as a constant current source
on the DC side. The DC bus capacitor is required to store energy to facilitate the control of
power flow and to reduce the DC-link harmonics.

1.2. VSC-HVDC SYSTEM

Constant Voltage
T

Constant Current

L
C

VSC
S2

Filter

S1

(a)

Pdc
U c c

P,Q
Ra

Usa
n

T1

T3

idc +

T5

La

Usb

Rb

Usc

Rc

ia

Lb

Uca

ib

U dc

Ucb

ic

Lc

Ucc
T4

U s s

DC Line

T6

T2

C
DC Line-

System Control

Converter
Control

Firing Control

Inner Current
Control

(Power, power factor reactive power or AC/DC voltage)

Outer Control
Other Control

(b)
Figure 1-1 A typical VSC topology connected to the AC grid (a) simplified diagram, (b)
the details of the circuit including three phase two-level VSC using IGBTs.

1.2.2 Operation Principle of VSC

The operational principle of VSC can be explained using the following well-known
power equations.
P

U sU c
sin
XL

U (U U c cos )
Q s s
XL

(1-1)

CHAPTER 1: INTRODUCTION

Where Uc is the fundamental frequency positive-phase sequence component of the

converter output voltage magnitude; Us is the fundamental frequency positive-phase
sequence component of the AC source voltage magnitude (see Figure 1-1a); is the phase
angle difference between the converter and source voltage phasors (=s-c) and XL is the
inductance between the source and converter.
It can be easily seen from the above equations that the real power is primarily a
function of , whereas the reactive power is mainly determined by the difference (Us-Uc)
since cos approaches 1. This indicates that the reactive power flows from the voltage with
higher amplitude to the voltage with lower amplitude. This allows us to independently
control the magnitude and the direction of real and reactive power synchronously by an
appropriate control of the amplitude and the phase angle of the converter voltage.
Normally, the pulse width modulation (PWM) control scheme is employed to realize the
control requirement. However, due to the switched characteristics of c, the modulation
index M of PWM (where M=2Ucpeak/Udc) should also be considered to determine its
amplitude [2]. Here, the phase angle
Therefore, the desired

is determined by the reference modulation voltage.

and Uc can be obtained by appropriately adjusting the PWM

outputs, which consequently regulate P and Q. This allows us to operate the VSC in four
quadrants.

1.2.3 Characteristics and Applications of VSC-HVDC

The VSC-HVDC transmission technology provides an effective alternative to
conventional power transmission and distribution. The unique features of this technology
can be listed as below [4-9]:
1) Active and reactive power exchange can be controlled flexibly and independently.
2) The power quality and power system stability can be improved by means of fast
continuously acting power oscillation damper; and/or power transfer controller.
3) This technology can supply AC systems with low short circuit capacity or even
passive networks with no local power generation.
4) It can be used for the black start-up of a power grid.
5) There is no communication required between multiple converters, which can also
be used to form a multi-terminal DC system (MTDC).
6) It has a fast recovery of its control capabilities after a grid fault.

1.2. VSC-HVDC SYSTEM

7) It can reduce harmonics due to the adoption of higher switching frequencies,

although this has an adverse effect of creating losses.
8) There is no need to reverse the DC voltage polarity when reversing the power
direction.
9) Possible to develop small converters to reduce the space requirement.
10) It requires only compact filters.

The VSC-HVDC technology can also be applied in a wide range of applications due to
the above mentioned characteristics [4, 5, 10-13]. For example, it can

connect distributed small-scale power plants, small-medium sized hydropower

plants, wind farms, tidal power stations, solar power stations and so on;

operate asynchronously between AC systems with the same or different

nominal frequencies.

increase power capacity of urban centres.

supply power to remote areas or even passive networks;

be integrated in a multi-terminal system;

improve the power quality of a distribution network;

ensure the security and efficiency of power transmission;

be applied to a deregulated power system.

The VSC-HVDC projects around the world are summarized in [8] and [14]. Among
these the Hellsjn-Sweden prototype project 3MW, 10kV was the first VSC-HVDC
transmission project, which was commissioned in 1997. The worlds first HVDC project
employing VSCs in a modular multilevel converter (MMC) topology was the Trans Bay
Cable Project (TBC) of USA. It has a transmission capacity 400MW and 200kV of DC
voltage rating which was the highest rating for this type of technology until 2010 [15].
After that, the Caprivi link interconnector was commissioned in Oct 2010. This was the
first ABBs HVDC-Light transmission system (350kV, 300MW) to employ overhead line
which is 950km long. The latest milestone is the Skagerrak 4 link from Denmark to
Norway, for which the voltage rating is 500kV, the highest to date [16]. Note that it will be
used as a reference for determining the DC voltage rating in this project.

CHAPTER 1: INTRODUCTION

1.3 Literature Review

1.3.1 Control Strategies of VSC-HVDC
The performance of a VSC-HVDC system depends on its control system and the
system parameters. Therefore, it is very important to investigate the VSC-HVDC control
system. In the following paragraphs the history on the development of VSC control
strategy and a literature survey on controllers design for VSC operating under both strong
and weak AC system conditions are reviewed and summarized.
1.3.1.1 History on Development of Control Strategy
The PWM control strategy in VSC-HVDC system accommodating IGBTs was initially
proposed in the early 1990s [17-20]. However, in these studies only the phase shift angle
between output fundamental frequency positive-phase sequence voltage and AC bus
voltage was set as a control parameter, ignoring the amplitude of fundamental frequency
positive-phase sequence voltage. Therefore, independent control of active and reactive
power was not realized at that time. In these early schemes separate facilities were required
to control reactive power. Therefore these early studies did not fully demonstrate the
technological superiority of the VSC-HVDC technology. Later, in the mid to late 1990s a
simpler and straight-forward control strategy based on the power control concept
mentioned in section 1.2.2 is developed [21-23]. These control schemes were characterized
by relatively low bandwidth and consequently these schemes are unable to damp various
resonances that exist in the AC systems. Furthermore, these schemes did not have effective
capabilities of limiting over current. Consequently, this control strategy is undesirable.
Vector current control is now widely employed worldwide. It has the characteristics
that inherent converter current protection, fast dynamic responses and decoupled active
power and reactive power control. Conventionally, it is realized through a hierarchical
control structure including outer-loop controllers and inner current loop controllers (as
shown in Figure 1-1b), within which a dq decoupling technique is applied [24].
Note that, the outer-loop controllers produce reference values for the faster acting
inner-loop current-controllers, and typically, the sending-end converter controls the real
power and the receiving-end converter regulates the DC voltage. However, it is sometimes
the case that one end is designated for power flow control and the other end for DC voltage

1.3. LITERATURE REVIEW

control irrespective of the direction of power flow. The reactive power at either end of the
link is controlled separately by the respective converters. The control of reactive power is
used to control reactive power directly or indirectly as a means of controlling the powerfactor or AC voltage at a designated bus. Due to the simplicity and robustness, double
closed-loop vector oriented PI controllers have been utilized in compensating the system to
achieve the desired performance. The inner-loop controllers employ feed-forward
decoupled control to make the active- and reactive-current track the reference values
produced by the outer-loop controllers [25].
1.3.1.2 Controller Design for VSC Operating Under Strong AC System Condition
The selection of the parameters of the PI compensators of the various VSC control
systems is a key issue to ensure an adequate dynamic performance (including a fast
response, and a sufficient stability margin for the VSC-HVDC transmission system). Note
that the dynamic performance criteria typically results in conflicting requirements when
tuning the PI controller parameters. Consequently, a compromise between the speed of
response and stability for small disturbances is needed. Further compromises may also be
needed to achieve adequate performance in response to large disturbances. The need to
coordinate the tuning of several PI compensators is particularly challenging. Typically, the
PI controller parameters of the various compensators are obtained by trial-and-error which
relies heavily on the experiences and skills of the design engineers.
Very few publications were identified that quantify the selection and optimization of PI
parameters associated with VSC control systems. The approach adopted in [26] applies a
trial and error method with the objective of ensuring the transient stability of the system
without using any theoretical evaluation criteria. In [27] frequency response analysis is
used to obtain an envelope of PI compensator parameter values which satisfy specified
stability criteria. Then an objective function is evaluated for each set of PI parameters in
the above envelope using a detailed electro-magnetic transients (EMT) model. The set of
PI parameters which minimizes the objective function is selected. The approach proposed
in [27] was found to be promising but requires a significant amount of computation time
because there is no mathematically derived procedure for selecting the next set of PI
parameters on the basis of accumulated experience gained from the previous trials [28]. In
addition, it does not provide any insight on how to adjust several PI parameters
simultaneously. Although the papers in [29] and [30] apply optimization techniques based

CHAPTER 1: INTRODUCTION

on the Simplex Algorithm to different VSC-HVDC control systems, the selection of the
initial set of parameters for input to the optimization algorithm is ad-hoc. In addition, the
method tends to find a local optimum rather than the global optimum set of PI compensator
parameters. References [31] and [32] propose approaches for determining the PI
compensator parameters, but these methods do not include the simultaneous adjustment of
several PI parameters. Therefore, it is desirable to develop a new methodology to
determine these parameters.
1.3.1.3 Controller Design for VSC Operating Under Weak AC System Condition
It is noted that there are only a few research papers that consider the impacts of VSCHVDC transmission systems connected to a weak grid. It is identified that the outer power
loop, the inner current loop, the synchronization method and the input-output impedance
can be studied in such grids that accommodate the VSC-HVDC system. The strength of the
AC system to which the link is connected is determined by the short circuit ratio (SCR).
The SCR is a ratio of the AC-system short-circuit capacity divided by the rated power of
the HVDC link. The AC system strength is classified in [33] and given below in Table 1-1.

Table 1-1 Categories of SCR

Strong system

SCR3

Weak System

3 > SCR 2

Very weak system

SCR<2

A. Outer Loop Control

Controller design for VSC-HVDC converters connected to a strong AC system
condition is simple and accurate enough to be designed based on a simplified model
without considering the dynamics of the PLL and bus filter as in refs [31, 32, 34].
However, a robust controller design which is suitable for a wide range of equivalent grid
impedances is challenging due to the resonances in both the low (1.5-15 rad/s) to high
frequency range. In order to suppress the higher order harmonics caused by the converter
switching, a third order LCL filter is conventionally adopted. In this approach, a higher
order model is required to adequately represent the grid behaviour. This complicates the
dynamics of VSC-HVDC control system particularly when a wide range of grid impedance

1.3. LITERATURE REVIEW

is considered. In a weak system, the voltage is sensitive to converter current, whereas the
sensitivity can be ignored when the grid is strong. Moreover, the dynamics of the PLL and
the bus filter also become significant in the dynamic performance of the system.
The literature survey has also considered how to improve the design and performance
characteristics of the various outer loop controllers. It was reported in [35] that the flexible
AC voltage control can be realized through optimum power and reactive power
management to offer additional voltage support to reduce the impact of fast varying filter
bus voltage. A non-linear eight state model of the VSC was developed in [36] by taking
into account the dynamics of the PLL and the bus filter to accommodate the connection to
weak AC grids. Based on this multi-input multi-output (MIMO) model, a robust H infinity
(H) controller over a range of operating points is designed by employing linear matrix
inequality (LMI) techniques [36, 37]. Two genetic algorithms based on direct search
methods are also applied in [38] to optimize this type of controller.
B. Inner Loop Control
A challenging feature for the design of controllers associated with distributed
generation (DG) systems is high grid impedance and resistance. Hence, the advanced
techniques developed in the DG research area provide useful insights to controller design
for VSC-HVDC system [39-42]. As discussed below it was suggested in previous studies
that the instability problems associated with the connection of VSC-HVDC systems to
weak AC grids can be divided into two categories: i) high frequency instability which can
be yielded due to grid-converter resonance (e.g. LCL filter resonance) and interaction
between the inner current control and the active damping controller [41, 43]; ii) low
frequency instability which is highly influenced by the inner current control loop.
a) High Frequency Instability
Mitigation of the high frequency instability can be achieved by a number of alternative
active damping control schemes which include: i) the utilization of poorly damped
complex poles [43-45], ii) split capacitor [46, 47], iii) full state feedback control [48] and
iv) virtual resistance [39-41, 49]. With the employment of active damping control, the open
loop gain and bandwidth of the inner current controller are consequently increased and
current harmonic distortion is reduced. As a result, the increased bandwidth facilitates the
design and tuning of the outer-loop controller parameters.

CHAPTER 1: INTRODUCTION

b) Low Frequency Instability

In the following paragraphs the relative effectiveness of the PI, PR and DB innercurrent controllers on mitigating the low frequency instability are summarized.
The conventionally adopted PI inner loop control is ineffective in weak AC
application irrespective of the reference frame used. This is largely due to the weakness in
the low bandwidth of the PI control and the disability to suppress the grid induced
distortion and imbalance [50].
The PR controller developed in [50] , which is tuned for the selective harmonics
elimination, can relieve the above mentioned problem to some extent. However, this
controller is very sensitive to variation in harmonic frequencies since it is tuned for pre-set
harmonic frequencies. In addition, the high sensitivity to the grid impedance contributes to
another drawback. This is due to the bandwidth of the current controller tends to be lower
than the resonant frequencies with the increase of the grid impedance [43]. Consequently,
the PR controller becomes failure.
Thanks to the high bandwidth of the DB controller, it does not only minimize the crosscoupling effects among the controller loops, but also maximizes the disturbance rejection
ability against the grid distortion and the parametric uncertainties. Several DB control
algorithms are reported in [51-53]. Therefore, it is necessary to review the various DB
algorithms and explore their relative advantages and disadvantages. The current control
schemes are also evaluated and verified in [54] and [55], which concluded that the DB
control scheme is the best tailored current control scheme for a grid with large impedance.
A more advanced and robust DB control is developed in the DG field which is designed
based on a higher order augmented plant model by taking into account the filter and the
active damping resistance during the control process [40]. However, in this thesis the
former conventional DB control scheme is adopted, while the latter scheme is left for
subsequent investigation.
C. Synchronization Method
The conventional d-q-z type PLL [52, 56, 57] for extracting the phase angle of the grid
voltages is employed in this thesis. However, there are also alternative approaches in the
literature as discussed below. A small number of investigations on the synchronization
methods as virtual flux estimation and power synchronization control are identified.

10

1.3. LITERATURE REVIEW

The virtual flux estimation method investigated in [58-60] is a synchronization method

based on virtual flux concept. It is found in [59] that this scheme can be used to improve
the operation of VSC in weak grids with high impedance seen from the converter. The
mechanism adopted here is to synchronize to a voltage at a relatively stable point,
consequently achieve a more robust control system. However, normally, the closer to the
converter terminal in terms of choosing a synchronization point, the more accurate of the
control system performance can be achieved. Hence, a compromise needs to be made to
achieve a satisfactory performance. Nevertheless, a contrary viewpoint is presented in [61]
that virtual flux estimation is robust for distributed grid but cannot outperform
conventional PLL.
The power synchronization might be viewed as a combination of power-angle control
and vector current control as reported in [62, 63]. Reference to the power synchronization
control, two types of complex outer controllers are developed in [64] which are based on
H and IMC control theories respectively. It was concluded that both of these two
controllers perform similarly.
D. Input/output Impedance
To investigate the overall stability of the interconnected VSC-HVDC link and power
system, a new method based on the converter output impedance together with grid input
impedance is developed in [65]. The primary criteria used here is to evaluate whether the
ratio between the grid impedance and the converter output impedance satisfies the Nyquist
stability criterion, which is also discussed in [66, 67]. It was concluded that the admittance
shaping technique provides an effective method in optimizing the control, the bus filter and
on the design of the controllers.

1.3.2 Small Signal Modeling

As indicated in [68], there is a need to develop a detailed linear time-invariant (i.e.
small-signal) model of VSC-HVDC links. Such modelling is needed to understand how the
sub-components impact on the dynamic performance of both the VSCs and the wider
interconnected power system to which the link is connected. Detailed modelling of the
links is particularly required when the converters are connected to weak parts of the grid.
Clearly, the pre-requisite for such a small-signal model is a detailed three-phase non-linear
model including comprehensive representation of the switching controls. Linearization of

11

CHAPTER 1: INTRODUCTION

such a non-linear model facilitates the application of well-established linear systems

analysis and controller design methods. The applications include such as the design of the
VSC controls to power systems incorporating VSC-HVDC links and other technologies
based on VSCs. These techniques include but are not limited to eigen-analysis, frequency
response analysis, residue analysis, classical controller design methods as well as modern
controller design methods. It is well known that the conventional eigen-analysis provides a
powerful tool for the system stability study only if the analytical model is sufficiently
accurate to represent the target plant.
The small signal modeling of a conventional line-commutated converter HVDC (LCCHVDC) is reported in [56, 69, 70]. A relatively early investigation of the small signal
model of VSC is also given in [71]. Due to the development of the off-shore wind industry
specifically in Europe, the multilevel VSC technology with cascaded bridge is considered a
common solution by main system manufactures. Due to this impetus, a good number of
studies on small signal modeling are given in [72-76]. However, the latter models have
inner current controllers that are based on the PI control scheme rather than the DB
controllers that are employed in this thesis. In reference [77], two types of inner current
controllers: i) predictive controller and ii) fast PID control scheme are investigated. It can
be noted here that the predictive controller mentioned in ref [77], is in fact similar to the
conventional PI control with additional current feed-forward technique.
As an alternative to dq-frame controllers in the grid connected converters, PR control is
also widely used owing its superiority to the PI controller under unbalanced condition.
However, due to the presence of the time-varying components, the small signal model of
PR controlled VSC is very challenging. Using the method proposed in [78], a small signal
model of the PR controlled VSC is developed in [79, 80], which also studied the dynamics
of a DC voltage control loop. All these works facilitate the ongoing research studies
with regards to the area of PR controlled VSCs.
As stated earlier, the DB control scheme is the best tailored inner current scheme for a
weak AC grid connected VSC. However, no studies were reported previously on the small
signal model of such controller, which will also be investigated in this thesis.

12

1.3. LITERATURE REVIEW

1.3.3 Grid Interconnection

Modern large interconnected power systems are characterized by the requirement to
transmit large amounts of power over long distances, large-scale of integration of
renewable energy, and deregulation of power system operations. These and other factors
are significant driving forces behind the development of power systems with a hybrid
AC/DC structure. The hybrid AC/DC grid is not only capable of enhancing the system
stability through relieving the congestion and transmission bottlenecks, it also acts as a
firewall against cascade disturbances, further preventing blackouts [4, 81]. However, the
integration of VSC-HVDC links may deteriorate the damping of the inter area modes [82,
83]. In addition, the sub synchronous resonance is also a common problem since the first
adverse interaction between the turbine-generator torsional modes of the vibration and a
conventional HVDC transmission system was observed in Square Butte Electric Coop
Project in North Dakota [84] dating back to 1977. The solutions for enhancing the damping
of electromechanical modes of oscillation in a system can be divided into two categories: i)
PSSs fitted to synchronous generators [85]; ii) Power Oscillation Dampers (PODs) fitted to
flexible alternating current transmission systems (FACTs) [3, 86-89], HVDC links [4, 82,
83, 90-94]; and potentially asynchronous generators such as wind turbines [95]. Note that
since the VSC-HVDC is the primary research topic in this thesis, the majority of the
review has been performed about the HVDC damping controller.
It was reported in [83] that the VSC-HVDC link has four controllable quantities,
within which the inter-area modes are able to be decoupled by two Ps and the local
modes within the area where the converter located can be simultaneously damped by each
V (or Q). However, it is investigated in [90] that active power control is more effective
in damping electromechanical modes of oscillation than reactive power control. Therefore,
in this thesis active power and the DC voltage are considered as the most suitable inputs to
PODs fitted to VSC-HVDC links.
Depending on whether the input signal to the POD is local or remote it is classified as
either: i) a local POD or ii) a wide area power oscillation damper (WAPOD). Due to the
limitations in signal communication, the local POD is adopted initially. However, with the
developments in Global Position System (GPS) and satellite communication technologies,
the WAPOD is may now be technically feasible. It is conceivable that the combination of
these two types of PODs will become increasingly important as power systems evolve.

13

CHAPTER 1: INTRODUCTION

The effectiveness of the POD design depends on a number of factors including the
location of controlled plant (relates to controllability); the modal content of the selected
stabilizing signal (relates to observability); and the measurement accuracy of the selected
signal. In [96], to damp out the inter-area mode in the Hydro-Quebec network, two
different approaches on feedback signal selection were investigated. They were i) the
geometric approach (also known as controllability/observability index) and ii) the residue
approach. It was concluded that the geometric approach is more reliable and useful in this
application, rather than the residues approach which is widely used. Moreover, it was also
concluded that the global signal input is more effective than the local signal. However, it
should be noted that, this is not always true. Differences in performance between local
POD and WAPOD vary depending on several factors including: i) the actual tuning of
those PODs, ii) the availability of measurements with good modal content, iii) the power
system configuration and disturbances, iv) the location of VSC and v) the specific
operating point [89].
Later, in addition to the aforementioned controllability/observability and residue
methods, the Hankel singular value (HSV) approach is added to complete the comparison
in [97]. It was concluded that based on a small two-area system, the residue and Hankel
singular value methods perform in a similar way. However, in a medium scale system, the
system damping performance was better when signal selection was based on the HSV
method. There are also a number of studies on investigations of the design approaches of
PSS and POD, which are documented in [98-103].
Recently, several investigations on enhancing system stability through POD are
published based on the Hydro-Qubec network [96], the Nordic power system [89], the
reduced Great Britain model [104], the China Southern Power Grid (CSG) [105], and the
New England test system integrated with the New York power system [94]. In fact,
Australia also has an important track record in utilizing PODs for damping inter-area
modes of oscillation. References [98] and [100] formed the basis of PODs fitted to SVCs
in Brisbane that were required to stabilize the interconnection between NSW and
Queensland [106]. Furthermore, improved damping controls fitted to the SVCs in the
Brisbane area were investigated in [107] in the mid to late 2000s. The new controls were
based on local bus frequency measurements.

14

1.4. THESIS OVERVIEW

In Australia, significant amount of geothermal energy is available in the area of Cooper

Basin near South Australia and Queensland. It can be foreseen that the energy production
will be economically harvested by the geothermal resources in the future, although this has
not yet been established, only a number of pilot projects are being undertaken to establish
the technical and economic feasibility. However, such resources are located far away from
the major cities and loads. Therefore, this thesis considers as a case study the possible
future development of parallel VSC-HVDC and AC links to transmit power generated
from large geothermal sources in the far north east of South Australia to northern New
South Wales and the mid north of South Australia. Although there are some pilot studies
on utilizing these geo-thermal energy [108, 109], the study undertaken in this thesis are
different, which is characterized by a more suitable power transmission option and
employing a detailed DB controlled VSC model. Furthermore, the focus of the case study
is the design of the VSC-HVDC link controllers and the damping performance of the
system, including the design of damping controllers.

1.4 Thesis Overview

1.4.1 Research Gap
Based on the literature review conducted above, several significant limitations are
identified which form the basis of the main original research topics in this thesis.

A simple and effective methodology for the controller design and optimization of
VSC-HVDC systems connected to weak grids is not available.

A comparison of various types of DB current controllers for the VSCs in terms of

principle, parameter sensitivity as well as robustness against large disturbances has
not been studied previously.

Stability studies involving the DB controlled VSC-HVDC links are only based on
time-domain analysis in the previous studies. The small signal analysis, which is
essential to characterize the system characteristics, of such discrete power system
has not yet been reported due to the difficulty that the time varying nature prevents
the direct application of small signal studies.

15

CHAPTER 1: INTRODUCTION

Controller design in a weak multi-machine AC system integrated with VSC-HVDC

links by taking into account the robustness of the controllers for a wide range of
operating conditions and system disturbances has not been studied in the literature.

The possible interactions between VSC-HVDC links and the extended weak
Australian grid have not been reported.

Finally, the compensation methodologies (PSS, POD and WAPOD) for a secure
operation of VSC-HVDC links embedded in an extended weak Australian grid also
require further investigation.

This thesis aims to address the above research gaps.

1.4.2 Thesis Outline

The thesis is divided into six main chapters which have the following contents.
Chapter 2: Investigates three types of linear inner-current loop controllers in terms of
principle and controller design. A new methodology for determining controller parameters
based on simplified linear models of VSC-HVDC links and simultaneously optimizing
several control loops are presented.
Chapter 3: VSC control systems typically involve transformations between
fundamental frequency positive-phase sequence signals in a stationary reference frame
(either the abc or reference frame) and the corresponding DC signals in the
synchronous dq reference frame. A feature of the proposed transformation method in this
chapter is that it facilitates the development of an equivalent linear time-invariant
representation of controllers that in practice are time-varying. In addition, a small signal
model for the discrete DB controlled VSC is also developed and verified systematically
against the detailed PSCAD model from which it is derived.
Chapter 4: The methodology for the robust outer loop controller design of a VSCHVDC transmission system is developed which considers the operating range of the
converters and the grid characteristics as represented by the SCR and X/R ratio.
Chapter 5: An in-depth investigation of the small signal stability of a simplified model
of the Australian power system incorporating a VSCHVDC link is undertaken. The case
study also includes supplementary damping controllers design to achieve adequate system
damping performance.

16

1.4. THESIS OVERVIEW

Chapter 6: The chapter summarizes the conclusions drawn in each chapter and also
reiterates the original contributions.

17

Chapter 2: VSC-HVDC Control

Structures
This chapter aims to evaluate various inner current controllers widely employed by the
VSC-HVDC systems. The primary focus is mainly set on three linear control structures,
i.e. i) the conventional decoupled PI-controller in the synchronously rotating reference
frame, ii) Proportional Resonant (PR) controller and iii) discrete Dead Beat (DB) controller
in the natural stationary reference frame. The controller design methodologies are also
investigated in this chapter.

2.1 Introduction
In a typical VSC-HVDC system, the hierarchical control structure is conventionally
adopted including system control, converter control and firing control from top to the
bottom, as shown in Figure 1-1b. Here, only two levels of the converter control that the
outer power-, voltage- control and the inner current control are discussed. The main
function of the outer loop controller is to control voltage or power of the three-phase
converter. Yet, the inner current control is mainly responsible for current control according
to current references that are generated by the outer loop controllers. For example, to
18

2.1. INTRODUCTION

realize sinusoidal current wave tracking with a unity power factor by using inner current
control. Since the inner current loop control is much faster than the outer loop control,
hence, the inner current control loop plays a significant role in improving the overall
performance of the control system.
Usually the adopted outer loop control contains a) constant DC voltage control; b)
constant real power control; c) constant reactive power control; d) constant frequency
control and e) constant AC voltage control. Generally speaking, a, b and d are a group of
incompatible control objects, which means a VSC station can only work in either of the
three control modes. In order to maintain the real power balance and DC voltage stability,
there must be one VSC station adopts constant DC voltage control, and others can work in
b or d mode. For the same reason, the c and e are also two incompatible control modes.
Note that one VSC station can only realize one control mode.
Converter control is the core strategy of the VSC-HVDC system, and it is therefore
being set as the research emphasis of this chapter.
In accordance with the linearity properties, the control method can be classified into
two categories: i) linear control scheme and ii) non-linear control scheme [61]. The linear
controllers can also be classified as the conventional decoupled PI-controller in the
synchronously rotating reference frame, the PR controller in the stationary reference frame,
and the predictive controller involving feed-forward and DB control scheme, as shown in
Figure 2-1. All these linear controllers are the focus of this chapter and will be explained
further in the following section.

PI
Linear

Predictive
Proportional
Resonant

Current Control
Method

Feed-Forward
Dead-Beat(DB)

Fuzzy
Non-Linear

Passivity
Hysterisis

Figure 2-1 Current control classifications

19

CHAPTER 2: VSC-HVDC CONTROL STRUCTURES

2.2 Background Study for System Modeling

Frame transformation plays a significant role in developing the control algorithm.
Therefore, the frame transformation criteria will be explained first.

2.2.1 Frame Transformation

2.2.1.1 Clark Transformation
Let us assume that a symmetrical sinusoidal three phase voltage uabc with a constant
angular frequency

is under investigation. Hence, the voltage equations can be given by

equation (2-1), where up and PLL denote the peak value and initial phase angle of the
rotating vector respectively.

ua (t ) u p sin[0 (t )t PLL (t )]
2
ub (t ) u p sin[0 (t )t PLL (t ) ]
3
2
uc (t ) u p sin[0 (t )t PLL (t ) ]
3

(2-1)

The adopted Clark Transformation matrix for transformation from the converter
stationary abc natural reference frame to

reference frame is given in equation (2-2),

and the time domain expression of the resultant

components is given in equation (2-3)

[85].
1
1

1 2 2

ua
u
3
3
u 2 0
ub
3
2
2

uc
u0
1
1
1

2 2
2

(2-2)

u
sin[0 (t )t PLL (t )]
u u p

cos[0 (t )t PLL (t )]

(2-3)

Therefore, the positive sequence voltage for the balanced three phase supply voltage
can be represented using the stationary reference frame and abc natural reference frame
as given in equation (2-4) and shown in Figure 2-2 [110], where the scaling of the factor K

20

2.2. BACKGROUND STUDY FOR SYSTEM MODELING

can be equal to 2/3 to ensure power invariance, alternatively it can also be chosen to

2/3

to ensure the voltage invariance [111].

2

j
j
U u (t ) ju (t ) K (ua (t ) ub (t ) e 3 uc (t ) e 3 )

(2-4)

The inverse voltage transformation equations are also given below in equation (2-5),

0
1
ua
3
u 1
b 2
2
uc
3
1
2 2

1
2
u
1
u
2
u0
1
2

0 t
PLL

(2-5)

'

Figure 2-2 Representation of rotating vector in converter including stationary

frame and abc natural reference frame.

2.2.1.2 Transformation from Stationary

reference

Reference Frame to Rotating dq

Reference Frame
Reference to the relative motion theory, it can be noted that the dq reference frame
rotates at the same frequency 0 with the rotating vector . Therefore observing from
vector it will appear stationary, and will contain a set of DC terms which will facilitate
the controller design and modeling of VSC system. As shown in Figure 2-3, the q-axis is
chosen aligning with the opposite direction of the rotating vector, and d-axis is orthogonal
to the q-axis, lagging it by 90o. This transformation can be given in a vector form as in
equation (2-6), and in matrix form as in equation (2-7). Equation (2-8) and equation (2-9)
provide the reverse transformation from dq reference frame to

reference frame written

in a vector form and matrix form respectively.

21

CHAPTER 2: VSC-HVDC CONTROL STRUCTURES

udq (t ) u (t )e j (0 (t )t PLL (t ))

(2-6)

d cos(0 (t )t PLL (t )) sin(0 (t )t PLL (t ))

q sin( (t )t (t )) cos( (t )t (t ))

0
PLL
0
PLL

u (t ) udq (t )e j (0 (t )t PLL (t ))

(2-7)
(2-8)

cos(0 (t )t PLL (t )) sin(0 (t )t PLL (t )) d

sin( (t )t (t )) cos( (t )t (t )) q

0
PLL
0
PLL

(2-9)

d
0 t

PLL

0 t

'
q

Figure 2-3 Representation of rotating vector in dq reference frame and reference frame

2.2.2 Transient Mathematical System Model

From the perspective of controller design, it is important to understand the plant to be
controlled. The single line diagram of VSC-HVDC is given below in Figure 2-4, where R
and L represent the internal resistance and summation of the inductance of VSC
transformer and converter reactor respectively.

iL

idc
L A

B R

Us

is

Udc

Uc
-

Figure 2-4 Single line diagram representation of VSC-HVDC

Assuming that the system is operating under a balanced condition, considering only the
fundamental frequency positive-phase sequence components of us_abc and is_abc (i.e. point B
in Figure 2-4), the fundamental frequency positive-phase sequence components of the

22

2.2. BACKGROUND STUDY FOR SYSTEM MODELING

three-phase AC voltage at the VSC terminals (i.e. point A in Figure 2-4) uc_abc are able to
be given by equation (2-10) [112-114].

disa

L
Risa
sa
ca

dt

disa

Risb
U sb U cb L
dt

disc

U sc U cc L dt Risc

(2-10)

Due to the DC link capacitor C, the converter DC voltage (Udc) and currents (idc and iL)
are related, which can be given by equation (2-11),
C

dU dc
idc iL
dt

(2-11)

If we apply Clark Transformation (2-5), the equation (2-10) can be transformed to

reference frame as in equation (2-12),
di

U s U c L s Ris

dt

U U L dis Ri
s
c
s

dt

(2-12)

Since the fact that the relationship between and dq reference frames is determined
by equation (2-8) .Then, the voltage and current equations in reference frame becomes,

U s U sdq e jt

jt
U c U cdq e
i i e jt
s sdq

(2-13)

Substituting equation (2-13) into equation (2-12) yields to the following set of equations,
U s dq e

jt

U cdq e

jt

d (isdq e jt )

U s dq e jt U cdq e jt e jt L
U s dq e jt U cdq e jt e jt L

dt
disdq
dt
disdq

Risdq e jt

Lisdq

d (e jt )
Risdq e jt
dt

(2-14)

j Lisdq e jt Risdq e jt

dt
If we extend these equations to the matrix form and apply the Laplace Transformation,

then VSC systems mathematical model can be given in the dq reference frame by equation
(2-15).
usd sL R L isd ucd
u

sL R isq ucq
sq L

(2-15)

23

CHAPTER 2: VSC-HVDC CONTROL STRUCTURES

2.3 Description of Controllers and Determination of

Controller Parameters
Three main linear inner current controllers are discussed in this sub-section. First, the
mechanisms are introduced and explained in detail. Then, a methodology concerning the
selection of controller parameters is proposed for each type of current controllers, and the
simulation models established on a PSCAD/EMTDC platform are also employed for
verification purpose.

2.3.1 Proportional Integral Control

In steady state, the dq components of the variables are typically transformed by using
abc-dq (i.e. Parks) transformation modulation, which synchronously rotate with the
voltage vector, but behave as DC quantities as given earlier. A PI control is normally
associated with the control structure (see Figure 2-5b), since it has the capability to reduce
the steady-state error.
The reason to present a more detailed explanation of PI control here is to provide a
better understanding on the algorithm and its operation principle.
Note that equation (2-15) can be rearranged for the currents as given below,

R
1

sisd isd isq (usd ucd )

L
L

si R i i 1 (u u )
sq
sq
sd
sq
cq

L
L

(2-16)

Let us assume that the VSC output voltage is determined by the following feed-forward
decoupled PI controller [113],

ucd ( Kip iI )(isd* isd ) Lisq usd

u ( K KiI )(i* i ) Li u
cq
ip
sq
sq
sd
sq

(2-17)

where K ip and K iI are proportional and integral gains, i*sd and i*sq are the current
references of the d,q axes respectively, and isd and isq are the measured converter output
currents in terms of the d-, q- axis respectively.
Substituting equation (2-17) into equation (2-16),

24

2.3. DESCRIPTION OF CONTROLLERS AND DETERMINATION OF CONTROLLER

PARAMETERS
K

[ R ( Ki p iI )] / L
0
*
isd 1
sisd
KiI isd
s
)
( Kip
si
i
K
L
s isq*
sq
0
[ R ( Ki p iI )] / L sq
s

(2-18)

From the above equation, it can be easily seen that, d- and q- axis current id and iq are
successfully decoupled. Hence, the control strategy structure using the above discussions
can be illustrated as in Figure 2-5b.
Converter

is_abc

P,Q
R

icap
L

u s_abc

PI

PI

U dc_ref /Pref

abc-dq

Q
+-

isd_ref
isq_ref

isd

PWM

isq

u c_abc _ref

cd _ref
Inner
Current
dq-abc
Control u cq _ref
Loop

u sq

u sd

U dc /P

dc

u c_abc
is_abc

Q ref

iL

idc

abc-dq

us _ abc

PLL

(a)
u sd

isd_ref

+
+
isd

isq
+

isq_ref

PI

u cd_ref

u cq_ref

PI

+
u sq

(b)
Figure 2-5 (a) The diagram of the inner and outer controllers; (b) Inner current control loop.

25

CHAPTER 2: VSC-HVDC CONTROL STRUCTURES

Figure 2-5a shows a three-phase, two-level VSC adopting a voltage oriented vector
control structure implemented in the synchronously rotating dq-reference frame, where the
three-phase AC voltage at the network node of the VSC and the three-phase current
flowing from the network into the VSC are denoted by us_abc and is_abc respectively, C is the
capacitance of the DC filter, and the entire control system including the outer- power and
voltage loop, the inner current control loop together with phase locked loop (PLL) control
are also given. The error of Pref and P is the input to the sending end controller, while the
error of Udc_ref and Udc is the input to the receiving end controller. Note that, only the
positive sequence current is the control target considered. The reference value of the
converter terminal voltage uc_abc_ref is generated from the control system then fed into the
VSC-HVDC system through the firing control that PWM control. Lastly, it should be
noted that the reference phase angle of control system is extracted from the filter bus
voltage by the PLL control.
2.3.1.1 Controller Design and Optimization for PI Controlled VSC-HVDC System
As is well known, the performance of a system highly depends on the performance of
its controllers, which was emphasised here about the importance of controller design. This
section proposes an approach to the selection and optimization of the parameters of the PI
compensators in the various control loops of a VSC-HVDC transmission system using a
decoupled control scheme.
In this study, an optimization algorithm based on the simplex method is adopted. The
main objective of this method is to simultaneously minimize the weighted sum of the
integral of the time absolute-error products (ITAE) of the active power, the reactivepower, the DC voltage and the inner current controllers of the respective VSCs. The initial
values of the PI compensator parameters for input to the optimization algorithm are
obtained by the application of classical frequency response design methods to simplified
linear models of the open-loop transfer functions (TFs) of VSC-HVDC control system.
The optimization process is applied to a detailed electromagnetic transient (EMT) model of
the VSC-HVDC system to which a large disturbance is applied.
The effectiveness of the optimized PI compensator parameter settings are assessed in
terms of rise-time, overshoot, and settling times for a range of disturbances applied to the
detailed EMT model. On the basis of these assessments, modifications to the weightings of

26

2.3. DESCRIPTION OF CONTROLLERS AND DETERMINATION OF CONTROLLER

PARAMETERS
the error signals are found to be necessary to obtain good performance according to the
specific control design requirements.
Figure 2-6 shows the flow chart of the process for determining the PI compensator
parameters for VSC-HVDC control system.

Transfer function analysis

Frequency response analysis
(system stability margin)
give the initial value of PI parameters

Input values of PI parameters into

VSC-HVDC system built in PSCAD
New parameters set X

Objective function Of(X)

Simplex Algrithm

Convergence ?
Y
End

Figure 2-6 Flow chart of the process for determining the PI compensator parameters for
VSC-HVDC control system
In the below paragraphs the design parameters of the controllers are explained
systematically.
A. Transfer Function of Control Loops
a) Inner Current Control loop
The inner current loop control shown in Figure 2-7 is composed of 4 parts: i) PI
compensator

(Hc1(s)=(Kp1+Ki1/s));

ii)

equivalent

simplified

filter

model

(G1(s)=1/(R(pu)+sXL(pu)/)); iii) converter circuit (G2(s) =KPWM/ (1+1/2Tss)) and iv)

measurement circuit (G3(s)=1/(1+Tss)). Hence, the compensated open loop transfer
function of the inner current control loop can then be written as

27

CHAPTER 2: VSC-HVDC CONTROL STRUCTURES

F1 (s) H c1 s Gp1 s

(2-19)

where the plant transfer function G p1 s is

G p1 s G1 ( s) G2 (s) G3 (s )

K PWM
1
1

1
X ( pu )
1 Ts s 1 Ts s R( pu ) s L
2

(2-20)

Here, Ts is the sampling interval of the inner current control loop, KPWM is the
equivalent gain of the PWM, which is normally defined by KPWM= 2 uac/Udc, where uac is
the peak value of the terminal line-to-ground voltage of the converter.

isd.max/sq.max
i dref/qref

-+

PI Controller

PWM

u dis

PI (1,2)

K PWM
T
1+s s
2

-+

-isd.max/sq.max

Load

i d/q

1
R(pu)+s

X L (pu)

Sample & hold

1
1+sTs

Figure 2-7 Detailed inner current control loop in pu system

b) Outer DC Voltage Control Loop

The outer DC voltage control loop shown in Figure 2-8 is composed of 5 parts: (i) PI
compensator

(Hc2 (s)=(Kp3 +Ki3 / s)); (ii)

equivalent

inner

current

control

loop

(G 4 (s)=1/(1+Teq s)) , where Teq is the equivalent time constant of first order approximation

of current control loop [31, 32]; (iii) line circuit (G5 (s)=1/(s Cpu )) ; (iv) current
transmission relationship according to power balance between AC side and DC side

(G6 (s)=3usq.pu /2Udc.pu ) and (v) measurement circuit (G7 (s)=1/(1+Ts s)) [115]. The
compensated open loop transfer function of the outer DC voltage control loop can then be
written as

F2 (s) H c 2 s Gp 2 s
where the plant transfer function G p2 (s) is

28

(2-21)

2.3. DESCRIPTION OF CONTROLLERS AND DETERMINATION OF CONTROLLER

PARAMETERS

G p 2 ( s) G4 ( s) G5 ( s) G6 ( s) G7 ( s)

2 U dc.pu

I L.pu Feed forward

3 u sq.pu

isq.max
PI controller

Vdcref

PI(3)

+-

-isq.max

usq. pu
1
1
1

Teq s 1 U dc. pu s C pu Ts s 1

+-

i qref

1
Teq s+1

isq.pu

(2-22)

I L.pu

3 u sq.pu Idc.pu - ICap.pu 1

+
s Cpu
2 U dc.pu

Vdc

Equivalent inner current controller

Sample & hold

1
1+sTs

Figure 2-8 Block diagram of DC voltage control scheme in pu system

c) Outer Active/Reactive Power Control Loop
Neglecting the power loss of the transformer, grid filter and converter and using the
two-axes theory, the injected real power Ps and reactive power Qs can then be expressed as,
Ps 3 / 2 (usq isq usd isd )

Qs 3 / 2 (usq isd usd isq )

(2-23)

According to Figure 2-3, the q-axis is aligned with the opposite direction of Us phasor.
Therefore, usd equals to be 0. The power equation can be simplified as,
Ps 3 / 2 usq isq

Qs 3 / 2 usq isd

(2-24)

The above equation shows that the active power and reactive power can be controlled
independently by controlling the q axis current isq and the d axis current isd.
The outer- active and reactive power control loops are shown in Figure 2-9. For initial
design purposes the network voltage can be assumed to be fixed with a value of one perunit, i.e. u sq =1.0 .
The outer- active/reactive power control loops is composed of 4 parts: i) PI
compensator

(Hc3 (s)=(K p5 +Ki5 / s)); (ii)

equivalent

inner

current

control

loop

(G8 (s)=1/(1+Teq s)) ; (iii) power balance transmission relationship (G9 (s)=3/2u sq (pu) and
(iv) measurement circuit (G10 (s)=1/(1+Ts s)) . The compensated open loop transfer function
of the outer active/reactive power control loop is shown in equation (2-25),

F3 (s) H c3 s Gp3 s

(2-25)

29

CHAPTER 2: VSC-HVDC CONTROL STRUCTURES

isq.max
isqref (pu)

Pref (pu)

+-

PI(4)

-isq.max

isq (pu)
1
1+Teq s

P(pu)
3
u sq (pu)
2

Sample & hold

1
1+s Ts
(a)

isd.max
Q ref (pu)
+-

isdref.pu

PI(5)

-isd.max

1
isd.pu 3
u sq (pu)
1+Teq s
2

Q pu

Sample & hold

1
1+s Ts
(b)

Figure 2-9 (a) Block diagram of active power control scheme in pu system; (b) block
diagram of reactive power control scheme in pu system.
where the plant transfer function G p3 s is

Gp 3 (s) G8 (s) G9 (s) G10 (s)

1
3
1
usq ( pu)
Teq s 1 2
Ts s 1

(2-26)

Simplifying equation (2-26) by eliminating third order term, equation (2-27) is

obtained

F3 (s) ( K p Ki / s)

1
3
usq ( pu)
4Ts s 1 2

(2-27)

B. Initial Estimates of PI Compensator Parameters

The initial values of the PI compensator parameters are selected based on the
application of frequency response design methods to simplified linearized models of the
various control loops. The PI compensator parameters are chosen such that the gain margin
(GM) and phase margin (PM) of the compensated open-loop system are respectively
within the ranges of GM >= 6dB and 40 <= PM <= 60 deg. The following design criteria
are applied to the closed-loop step responses:

30

2.3. DESCRIPTION OF CONTROLLERS AND DETERMINATION OF CONTROLLER

PARAMETERS

the rise time must be less than 30 ms for underdamped system or less than 70
ms for a damped system;

the overshoot must be no more than 30% [27].

C. Optimization Consideration and Simplex Algorithm

Due to the adoption of hierarchical control strategy, the VSC at each end of the link has
four double closed-loop PI controllers which must be integrated. Therefore the
coordination of the controller parameters is a challenging problem for the designer. In total,
12 parameters must be determined (note that since the respective d- and q-axis inner-loop
current controllers are identical the number of controller parameters is reduced from 16 to
12). Note also that the selection of an appropriate objective function for the optimization
procedure is also critical.
The Simplex Algorithm is considered as comparably faster optimization technique in
terms of reaching the optimal solution compared with other optimization techniques [29,
30, 116]. The algorithm is a geometric object formed by N+1 vertices in the N-dimensional
space (number of parameters) and suitable for cases where the number of variables
ranges from 2 to 20, which makes it the most suitable optimization method here. It
successively compares the values of the objective function at the N+1 vertices of a simplex
and discards the worst one then another new value of vertex is selected to be calculated in
the same procedure until the optimum point is reached.
D. System Objective Functions
The ITAE index is a simplified form of the integral time absolute error and is one of
the error-integral indices used to evaluate the dynamic response of a control system [27,
28, 117]. The ITAE index JITAE is given by,
T

J ITAE t edt
0

(2-28)

Where t is the time since the disturbance is applied, |e| is the absolute value of the
control system error and T is a finite time chosen so that the integral approaches to a
steady-state value and is usually chosen as the settling time Tf.
For a VSC-HVDC transmission system which includes multiple control loops, the
objective is to minimize the weighted sum of ITAE indices Of (X) associated with each of
the control systems [28].

31

CHAPTER 2: VSC-HVDC CONTROL STRUCTURES

O f ( X ) mi fi ( X )

(2-29)

fi ( X ) j t e j (t ) dt

(2-30)

where fi is the performance index for evaluating controller operation performance; mi is

the weighting factor applied to the ith objective function (i is the number of controllers);
ej(t) is the error between the real value of the jth controlled variable and its reference value;
and j is the weighting factor applied to the ITAE of the jth controlled variable. The vector
X=(X1 X2 Xn) in the equations is the set of control system parameters, i.e. the
parameters of the PI compensators.
2.3.1.2 Computer Simulation of the PI controlled VSC-HVDC system
The approach has been tested on a detailed EMT model of a VSC-HVDC system using
the control strategy described in 2.3.1.1. The system block diagram is given in Figure 2-10.
The model data is given in the Appendix A. The process of initializing the parameter
values of all the PI compensators is listed below.

AC power grid 1

Converter 1

Pc1 Qc1

Pc1 Qc1

U s11

U c10
R1

Converter 2

DC Cable

I dc1 I L1

I L 2 I dc 2

L1

AC power grid 2

Pc 2 Qc 2

Qc 2 Pc 2

U c 2 0

U s 2 2
L2

R2

I cap 2

I cap1
Phase Reactor

Phase Reactor
AC filter 1

U dc1

U dc 2

Figure 2-10 VSC-HVDC system diagram

A. Initial Setting of PI Parameters by Classical Frequency Response Methods
a) Inner Current Control Loop
By substituting the system parameters that are given in Appendix A into equation
(2-19) , equation (2-31) is obtained.

32

2.3. DESCRIPTION OF CONTROLLERS AND DETERMINATION OF CONTROLLER

PARAMETERS

F1 ( s) ( K p1 Ki1 / s)

0.78509
1
1

4
4
1 3.7 10 s 1 7.4 10 s 9.5493 104 s 0.0015
0.78509K p1 s 0.78509Ki1

(2-31)

2.615 1010 s 4 1.06 106 s 3 0.0009566 s 2 0.0015s

The frequency response of the plant transfer function Gp1(s) in equation (2-20) is
calculated and displayed in Bode plot as shown in Figure 2-11 (solid blue line). According
to the requirements listed in section 2.3.1.1 B, it can be seen that the phase margin of 39.4
deg is smaller than the minimum requirement of 40 deg. For the uncompensated system the
gain margin is 11.4dB, the overshoot of the step response is 30%, and the settling time is
0.01s, all of which are acceptable (see solid blue line in Figure 2-12). PI compensation is
applied (Kp1=0.62, Ki1= 53) to reduce the gain cross-over frequency thereby increasing the
phase margin to 50.4 deg and the gain margin to 17.1dB (dashed green line in Figure
2-11). The overshoot of 23.6% (dashed green line in Figure 2-12) is lower than the original
30%. Although the settling time has been increased to 0.025s, it is still acceptable.

Magnitude(dB)

200

without PI
with PI

100
0
-100
-200
-4
10

10

-2

10

10

10

10

Phase(deg)

0
-100
-200
-300
-4
10

10

-2

10

10

10

10

Frequency (rad/sec)

Figure 2-11 Open-loop Bode plots of the current controller transfer function (i) without PI
control (blue solid line); (ii) with PI compensation using initial parameters (green dashed
line).

33

CHAPTER 2: VSC-HVDC CONTROL STRUCTURES

Step Response Based on Simplified Linear Model

1.4

without PI
with PI

Converter Current (pu)

1.2

0.8

0.6

0.4

0.2

0.01

0.02
0.03
time (sceonds)

0.04

0.05

Figure 2-12 Step responses of current controller transfer function (i) without PI control
(blue solid line); (ii) with PI compensation using initial parameters (green dashed line)
PI compensation is applied (Kp1=0.62, Ki1= 53) to reduce the gain cross-over frequency
thereby increasing the phase margin to 50.4 deg and the gain margin to 17.1dB (dashed
green line in Figure 2-11). The overshoot of 23.6% (dashed green line in Figure 2-12) is
lower than the original 30%. Although the settling time has been increased to 0.025s, it is
still acceptable.
b) Outer Constant DC Voltage Control
Substituting numerical values of the system parameters in equation (2-22) yields the
following compensated open-loop transfer-function for the DC voltage controller,
F2 ( s)

Ki 3 K p 3 s
3.33592 104 s3 0.1127 s 2

(2-32)

The frequency response of the plant transfer function Gp2(s) in equation (2-22) is
calculated and displayed in Bode plot as shown by the solid blue line in Figure 2-13. The
phase margin of 88.5 deg is significantly higher than the specified maximum of 60 deg,
and the settling time of 0.432 (solid blue line in Figure 2-14) is considered excessive. The
bandwidth of the controller is increased with the PI compensation (KP=20, Ki=1000). The
compensated open-loop transfer-function (dashed green line in Figure 2-13) has a phase
margin of 47.1 deg, decreased settling time 0.046s (dashed green line in Figure 2-14). The
overshoot of 30.2% (dashed green line in Figure 2-14) is relatively high, but it will be
revised in the next optimization procedure.

34

2.3. DESCRIPTION OF CONTROLLERS AND DETERMINATION OF CONTROLLER

PARAMETERS

Magnitude(dB)

100

without PI
with PI

50
0
-50
-100
0
10

10

10

10

10

Phase(deg)

-50
-100
-150
-200
0
10

10

10
Frequency (rad/sec)

10

10

Figure 2-13 Open-loop Bode plots of the outer DC controller transfer-function (i) without
PI compensation (solid blue line); and (ii) with PI compensation using the initial
parameters (dashed green line).
Step Response Based on Simplified Linear Model
1.4

without PI
with PI

1.2

DC voltage (pu)

1
0.8
0.6
0.4
0.2
0

0.2

0.4
0.6
time(sceonds)

0.8

Figure 2-14 Step responses of the outer DC controller transfer-function (i) without PI
compensation (solid blue line) and (ii) with PI compensation using the initial parameters
(dashed green line).
c) Outer Active/Reactive Power Control
Similarly, inserting the numerical values into equation (2-25), the following
compensated open-loop transfer-function of the active/reactive power controller is
obtained.

Ki 4 K p 4 s
3
F3 ( s)
2 29.6 104 s 3 s 2

(2-33)

35

CHAPTER 2: VSC-HVDC CONTROL STRUCTURES

The frequency responses of the active/reactive power controller without PI

compensator are shown in Figure 2-15 (solid blue line). The phase margin is 89.7 deg. The
PI compensator (Kp = 3.65, Ki = 20.0) is used to increase the gain cross-over frequency,
thereby increasing the speed of response. The phase margin is reduced to 50.6 deg (dashed
green line in Figure 2-15), which is still acceptable.

Magnitude(dB)

100

Step Response Based on Simplified Linear Model

without PI
with PI

0
-100
-200
-1
10

10

10

10

10

10

Phase(deg)

-50
-100
-150
-200
-1
10

10

10

10

10

10

Frequency (rad/sec)

Figure 2-15 Open-loop Bode plots of the outer active/reactive power controller transferfunction (i) without PI compensation (blue solid line) and (ii) with PI compensation using
the initial parameters (dashed green line)

B. Simplex Algorithm for Multi-objective Optimization of VSC-HVDC System

The objective function of multi-objective optimization for VSC-HVDC system is
expressed as in equation (2-34) which is obtained from equation (2-30)

O f ( X ) t eQ _ rec dt t eP _ rec dt t eQ _ inv dt t edc _ inv dt

0

t eid _inner _ rec dt t eiq _inner _ rec dt t eid _inner _ inv dt t eiq _inner _ inv dt

(2-34)

Where ej(t) is the difference between the jth controlled variable and its reference value.
The value of j is set to unity indicating that all of the controlled variables have the same
importance.
The vector X= (Kp1 Ki1 Kp2 Ki2 Kp3 Ki3 Kp4 Ki4 Kp5 Ki5 Kp6 Ki6) represents the
variables.

36

2.3. DESCRIPTION OF CONTROLLERS AND DETERMINATION OF CONTROLLER

PARAMETERS
As it is well known, the control system is required to operate satisfactorily when being
subject to both large and small disturbances. For the purpose of parameter optimization the
disturbance is the reversal of the power-order from +1 pu to -1 pu at time t = 2s. The value
of the optimization function is computed and is then passed to the Simplex Optimization
Algorithm which computes an update set of the PI compensator parameters, X. The final
dynamic response of the system with the optimized set of PI compensator parameters
obtained with the objective function equation (2-34) is given in Figure 2-16. Table 2-1
shows the initial and optimized values of the control parameters and the resultant values of

pu

pu

the objective function.

Time (seconds)
Active Power at Rectifier end

Time (seconds)
(b) Reactive Power at Rectifier end

pu

pu

(a)

Time (seconds)
(c) DC voltage at inverter end

Time (seconds)
(d)Reactive Power at Inverter end

Figure 2-16 Responses of VSC-HVDC modelled in PSACD following a reversal of the

power order for (i) the initial set of PI compensator parameters (red line); (ii) the
parameters obtained with the optimization function (2-34) (blue line) and (iii) the
parameters obtained with the refined optimization function (2-35) (cyan line). The
reference values of the variables are shown in green line.
As it can be seen in Figure 2-16a, the optimization does greatly improve the tracking
capability of power order, where the settling time is reduced from 500ms to 40ms. It can be
reported that the value of the objective function is also reduced from 3.41 to 3.29 within
duration of 3s. Moreover, the resulting transient performance including the relatively long

37

CHAPTER 2: VSC-HVDC CONTROL STRUCTURES

settling times and the relatively high overshoot of the other reactive power controllers and
DC voltage controller are considered unsatisfactory (see Figure 2-16c). Therefore the
definition of the objective function is refined according to the controller design
requirement as illustrated in Figure 2-16. As an example, an increased weighting of 4 is
applied to the DC voltage controller and the active power controller which results in the
following objective function as equation (2-35).
Table 2-1 PI parameters comparison
Optimized Values Initial Values of PI

Of(X)

of PI Parameters

Parameters

3.2939 (3s)

3.4139 (3s)

Refined Values of
PI Parameters

4.5148 (4s)
Kp
Ti(1/Ki)

4.4717 (4s)
Kp
Ti(1/Ki)

Kp

Ti(1/Ki)

Q(rec)

0.4809
3.6540

0.0436
0.0512

0.6200
3.6500

0.0189
0.0500

0.4850
3.6577

0.0484
0.0508

P(rec)

3.6467

0.0145

3.6500

0.0500

3.6500

0.0139

Q(inv)

3.6569

0.0436

3.6500

0.0500

3.6569

0.0427

Udc(inv)

20.0061

0.0032

20.0000

0.0010

20.0107

0.0019

Inner Current(inv)

0.4765

0.0312

0.6200

0.0189

0.4688

0.0346

Inner current (rec)

O f ( X ) t eQ _ rec dt 4 t eP _ rec dt t eQ _ inv dt 4 t edc _ inv dt

T

t eid _inner _ rec dt t eiq _inner _ rec dt t eid _inner _ inv dt t eiq _inner _ inv dt

(2-35)

It can be seen from Figure 2-16 that the performance of the active and reactive power
controllers at the sending end are barely altered as a result of the refined optimization
function. However, the dynamic performance of the DC voltage controller and the
receiving end reactive power controller are improved significantly with refined objective
function. The values of the new objective function have been reduced from 4.51 to 4.4729
within duration of 4s after optimization. To conclude, the dynamic behaviour of the VSCHVDC system is improved with the refined set of PI parameters.

38

2.3. DESCRIPTION OF CONTROLLERS AND DETERMINATION OF CONTROLLER

PARAMETERS
C. Summary of Proposed Controller Design Methodology
A methodology for the design of the control system of a VSC-HVDC transmission
system is developed in this section. An initial set of PI compensator parameters is
determined according to the frequency response analysis of the open loop transfer
functions of simplified linear models of the VSC-HVDC control system. This initial set of
controller parameters is then optimized by the application of the Simplex Algorithm. The
objective is to simultaneously minimize the weighted sum of the integral of the time
absolute-error products (ITAE) of the power, reactive-power, DC voltage and current
controllers of the respective VSCs. The methodology has been demonstrated by simulation
studies based on a detailed EMT model of a VSC-HVDC system. It is found that the
objective function for the optimization function must be carefully chosen to obtain
adequate performance.
It should be clarified that, in the proposed new design methodology, neither the design
of starting point for the simplex algorithm by using classical control technique nor the
simplex algorithm for optimizing the control parameters is new, but the combination of
these approaches has not been recorded in the literatures, which can be treated as new. In
addition, it also worth noting that this controller design methodology is only applicable for
relatively strong system, in other words, the system which is not operating point
dependent. For a weak AC system, the controller design method should take into account
the dynamics of the filter and PLL which play a significant role in the system stability.
This will be further discussed in Chapter 4 later.

2.3.2 Proportional Resonant Control

2.3.2.1 Principle
The PR control is developed based on the use of generalized integrator. Paper [118] has
mapped out how to derive PR controller through the synchronous frame PI control. In
addition, paper [79] provided a simpler method to obtain the relationship between the PR
control and the PI control based on a block diagram approach [78]. Furthermore, paper
[119] extends the previous work while developing a stationary frame controller PRX2 that
is an exact equivalence to the decoupled d-q frame PI controllers by adding another two
cross-coupling branches X-control and X-feedback. A set of stationary frame controllers
have been obtained based on this PRX2 controller by eliminating one of the cross-coupling

39

CHAPTER 2: VSC-HVDC CONTROL STRUCTURES

branches or all of them (which is the exact PR controller). Results showed that it is this
elimination that caused the increase of the frequency sensitiveness. This study offers a
systematic method to understand this type of control in detail, and explains the
discrepancies from the frequency analysis and experimental verification point of views.
A more complicated small signal model of PR controlled VSC was first proposed in
paper [79] and thesis [80], which provides researchers a valuable tool to understand how
such controller can affect the stability of an entire system.
One of the most preferable characteristics of PR controllers is it is easy to compensate
on the low-order harmonics. This is done simply by positioning the generalized integrators
that tuned to resonate at the harmonic frequencies to be compensated, in parallel with the
main controller without affecting the dynamics of the entire closed-loop system as shown
in Figure 2-17 [54, 120].
is_ref

Kp

sK i3
s +302

sK i5
s +502

sK i7
s +702

u*

is

Figure 2-17 Structure of the paralleled harmonic compensators

The typical transfer function of PR controller is given by,

GPR ( s) K p Ki

s
s 02
2

(2-36)

The harmonic compensator transfer function of PR controller is defined as,

GHC (s) Kih

h 3,5,7

s
s 2 (h0 )2

(2-37)

This is because the PR controller works at a very narrow band around the resonant
frequency 0. However the bandwidth of these compensators has to be smaller, lower
enough than the bandwidth of the system to prevent triggering the system low frequency

40

2.3. DESCRIPTION OF CONTROLLERS AND DETERMINATION OF CONTROLLER

PARAMETERS
instability [121]. Moreover, an extremely high gain (near the resonant frequency) can be
achieved offering a fast dynamic response. Therefore, it can realize the zero steady state
tracking error for sinusoidal signals [119, 122]. This control scheme offers a better
capability when regulating the negative-sequence control target since it has symmetrical
characteristics in the frequency response plot. In other words, it is superior to PI controller
under unbalanced situation [119]. However, the biggest drawback of this control structure
is its high sensitiveness to frequency fluctuation. In addition to this, it is hard to insert the
harmonic compensators in to the system when the interconnected system becomes weak,
which is characterized by a relatively narrow bandwidth. In this case, modified active
damping controller together with PR controller is required as investigated in paper [43].
2.3.2.2 Controller Design
Applying Laplace Transformation to the reference frame mathematical model of
VSC system equation (2-12), equation (2-38) can be obtained, which is also illustrated in
Figure 2-18.

( R sL) is us uc

u c

u s

(2-38)

is
1
R+sL

Figure 2-18 Diagram of reference frame mathematical model of VSC system

Figure 2-19 shows the diagram of the simplified reference frame model of PR
current controlled VSC system, where 0 represents the resonance frequency of the
controller, Kp1 is the proportional gain and Ki1 is the integrator of the PR controller
respectively.
reference frame current controller
is_ref

K p1

is

sK i1
s 2 +02

+
+

2 m
U dc

reference frame VSC system

mathematical model
u s
is
u c
2
1
+
U dc
R+sL

Figure 2-19 The simplified model of PR current controlled VSC system in reference
frame

41

CHAPTER 2: VSC-HVDC CONTROL STRUCTURES

Then the closed-loop transfer function of the system yields to equation (2-39),
GIc ( s)

I s
I s _ ref

K p1 ( s 2 02 ) sK s1

(2-39)

( R sL)( s 2 02 ) K p1 ( s 2 02 ) sK s1

The proportional gain of the controller Kp1 can be tuned by root locus theory provided
that the integrator gain K i1 is set to be zero first. Then the integrator of the controller Ks1
can be obtained by frequency response analysis of the open loop transfer function. As
suggested in [123], the controller parameters for PR scheme should be kept the same as the
parameters of PI controllers.

2.3.3 Dead Beat Control

The working principle of DB control is to increase the step-response speed of a system
by manipulating the input signal into two parts. In such control scheme, a proportionally
larger step signal (positive) drives the system response fast to reach the original constant
value, which is followed by a delayed smaller step signal (negative) to cancel the
remainder transient response of the former step [52]. Since the DB control itself is allowed
to reach its reference value at the end of next switching period, consequently one sample
time delay is introduced in this method of control. In addition, DB control can be divided
into two parts depending on whether or not the computational delay time being taken into
consideration as demonstrated in Figure 2-20. This sub-section of the study also
investigates the operation and control algorithm design of DB controllers, and presents the
results about the system stability and the sensitivity analyses of plant parameter variations
for the four types of DB current controllers.
Reduce Gain
One sample delay DB with compensation
(Achieve current command in one sampling period )

Smith Predictor

DB
Two samples delay DB with compensation
(Achieve current command in two sampling periods )

IMC
Solving Feedback TF

Figure 2-20 The categories of DB current controllers

42

2.3. DESCRIPTION OF CONTROLLERS AND DETERMINATION OF CONTROLLER

PARAMETERS
2.3.3.1 Model of the System
The principal block diagram of the digital DB current controlled VSC system is given
in the Figure 2-21, which includes three main blocks: the phase locked loop, the feedforward loop and the DB current control loop. In the figure, the fundamental frequency
positive-phase sequence components of three-phase AC voltage at the network node of the
VSC and the three-phase current flowing from the network into the VSC are denoted by
Usabc and Iabc respectively. Rac and Lac in the figure are the resistance and inductance of the
Thevenin equivalent circuit of the AC system. In addition, Rv and Lv represent the
resistance and inductance of converter; and C is the capacitance of the DC filter. The
modulation input voltage signal

of the SPWM (a block for generating pulse

signal) is composed of two parts: the output of voltage prediction block Uv_abc_2 and the
output of the DB current control block Uv_abc_1.

Uf_abc

Usabc
-

Uv_abc

Rac

Lac

Ia

Rv

Lv

Rac

Lac

Ib

Rv

Lv

Rac

Lac

Ic

Rv

Lv

C
-

U dc

Triangular
Carrier

SPWM

Iabc

Uv_abc_ref
Id_ref
Iq_ref

Iabc_ref2
Dead- beat
Control

d-q-0 to a-b-c
Transformation

Uf_a
Uf_b
Uf_c

Uv_abc1

Uv_abc2

U line_

Uf_abc

Phase
Forward

U line_
a-b-c to d-q-0
Transformation

Voltage
Prediction

Phase Lock
Loop

Figure 2-21 Digital DB current controlled VSC system

The measured PCC voltage Uf_abc in Figure 2-21 needs to be filtered out and then be
added with a 3Ts/2 time delay before sending to the modulation signal. This is due to the
fact that a time delay of Ts/2 is introduced by the zero order holding (ZOH) block and also
the sampling block contributes an additional time delay Ts, where Ts is the sampling

43

CHAPTER 2: VSC-HVDC CONTROL STRUCTURES

interval [53, 54, 124]. The DB current control block in Figure 2-21 generates a precalculated additional voltage that is equal and opposite to the voltage reduction caused by
the Rv and Lv. This allows the current in the inductance is not affected by the PCC filter
bus voltage. The PLL is used for synchronizing the PCC voltage. A

to abc reference

frame transformation block is also needed in the control system. The phase forward loop
block in the control system is employed for time delay compensation.
2.3.3.2 DB Current Control with One Sample Time Delay
The closed-loop DB current control block without considering the computation time
delay is illustrated in Figure 2-22. The internal model control for design of DB control can
be expressed by using the following set of equations [51],

G 1 ( z ) z 1
GDB ( z )
(
Lv Rv ) / ( z 1)
z 1
Ts

uva _1 ((k 1)Ts )

I (kTs )

uva ((k 1)Ts )

(2-40)

Lv
L
( Rv v ) z 1 ) / (1 z 1 )
Ts
Ts

(2-41)

Lv
L
I (kTs ) ( Rv v )I ((k 1)Ts )
Ts
Ts

(2-42)

uva _1 (kTs ) u fa ((k 1)Ts )

where GDB (z) is the transfer function of the current control, uva_1 is the first part of
VSC input voltage generated by DB current controller, and kTs represents the kth time
interval, uva is the total input voltage for VSC, ufa((k+1)Ts) is the predicted filter bus
voltage at (k+1)Ts time interval, and I represents the current change order. Hence the
control algorithm can be implemented by using equation (2-42).

Ia_ref

G DB (z)

GZOH(z)

G(z)

Ia

Ia
Figure 2-22 The closed-loop DB current Control
The closed-loop transfer-function of the one sample time delay DB current control with
the presence of the one sample computation time delay can be given by,

44

2.3. DESCRIPTION OF CONTROLLERS AND DETERMINATION OF CONTROLLER

PARAMETERS

GDB _ CL ( z )

z 1 / ( z 1)
z 1
1

1 z 1 / ( z 1) z 1 z 1 z 2 z 1

(2-43)

Therefore, one sample delay DB current control does work at the stability margin with

two poles which are

(in the presence of one sampling period

computational delay).
A. Reducing the Proportional Gain Method
This method used in papers [54, 124] does move the poles inside the unit circle, but at
the cost of changing the closed-loop gain, which increases the phase difference between
the reference and output current. For example, by changing the proportional gain part Lv/Ts
of the controller to Lv/(2Ts) and Rv to Rv/2, the system characteristic equation becomes
z2 -z+1/2=0 . This results in two new poles as z1,2 = 1/2 (1j), which are inside the unity

circle making the system stable as illustrated in Figure 2-23,

Pole-Zero Map
1
0.6/T

0.8

0.5/T

0.10.3/T

0.7/T

0.6

0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9

0.8/T

Imaginary Axis

0.4
0.9/T

0.2
0
-0.2

0.4/T

0.2/T

0.1/T

1/T
1/T

0.9/T

0.1/T

-0.4
-0.6
-0.8

0.8/T

0.2/T

0.7/T

0.3/T
0.6/T

-1
-1

-0.5

0.5/T

0
Real Axis

0.4/T

0.5

Figure 2-23 Pole-zero map of the one sample delay system: red cross: one sample delay
DB without considering the computation delay time; blue cross: represents one sample
delay DB considering the computation delay time; green cross: represents reducing the
proportional gain;

45

CHAPTER 2: VSC-HVDC CONTROL STRUCTURES

B. Smith Predictor Method

The one sample delay DB current controller with Smith Predictor by considering the
computation delay has the same performance with the two-sample delays DB current
controller [51], which can be verified by using the derivation process given below.
The principle of Smith Predictor is that the output of the delayed DB plant model
should ideally be cancelled through the feedback while the non-delayed process remains
unchanged [125].

Ia_ref

G DB (z)

VDB

z -1

Ia

G(z)

Ia

G(z)(1-z -1 )

(a) Block diagram of Smith Predictor structure

Gsp(z)
Ia_ref

G DB (z)

VDB

z -1

G(z)

Ia

Ia

G(z)(1-z -1 )

(b) Smith Predictor structure in classical control mode

Figure 2-24 One sample delay DB control with Smith Predictor
The transfer function of one sample delay DB current controller with the Smith
Predictor that Gsp (z) can be expressed by,
Gsp ( z )

GDB ( z )
G ( z)
DB 1
1
1 GDB G ( z )(1 z ) 1 z

G 1 ( z )

( z 1)(1 z 1 )

(2-44)

2.3.3.3 DB Current Control with Two Sample Time Delays

In the following paragraphs, the control algorithms of two sample delays DB current
control schemes based on internal model control (IMC) and solving feedback TF
approaches are analysed below.

46

2.3. DESCRIPTION OF CONTROLLERS AND DETERMINATION OF CONTROLLER

PARAMETERS
A. IMC Design for DB Current Controller
'
-1
The delayed plant model G (z)=z G(z) is a second order system [51] and therefore the

controller should be designed according to the following equation,

GDB 2 ( s)

2
s 2 2s

G '1 ( s)

(2-45)

where, represents the bandwidth of the low pass filter adopted for damping purpose.
After discretizing the equation (2-45), the following controller is obtained,
GDB 2 (z)=

G '1 (z)

((z 1)/Ts ) 2 (z 1)/Ts

2

1
Ts

(2-46)

G (z)
(1 z 1 ) (z 1)

Expanding and rearranging equation (2-46), we can obtain,

Lv
L
I ((k 1)Ts ) ( Rv v )I (kTs ) uva _1 ((k 1)Ts )
Ts
Ts

uva _1 ((k 1)Ts )

(2-47)

The developed control algorithm is very similar to the one used in [54], which clearly
demonstrate the derivation of the control algorithm from a different point of view (IMC).
The algorithm can be implemented in PSCAD according to Figure 2-25.

Sample&Hold
u va_ 2

Ia
Ia_ref

Sample&Hold

K1

+
++

-K 2

u va_1

u va_ref

Z-1

Gp(z)

Ia

e-sTs

e-2sTs

Figure 2-25 Internal model control design for DB implementation block

B. Solving Feedback Transfer Function of DB Current Controller

The algorithm is obtained by solving the feedback transfer function of the closed-loop
DB current controlled VSC model, also named as AC current tracking control, has been
studied in [52] firstly and then successfully applied in [53, 126]. The control block diagram
of solving feedback transfer function method in z domain is given below in Figure 2-26.

47

CHAPTER 2: VSC-HVDC CONTROL STRUCTURES

Ia_ref

u va_1
Ia_ref (k)

Execution
-1
delay z

G db (z)

Ia

Gp(z)

GZOH(z)

T
GF(Z)

Figure 2-26 The block diagram of solving feedback transfer function DB current control
In the above figure, the block Gdb(z) and the feedback loop transfer function GF(z) are
given by,

Gdb ( z) N1 / (1 N2 z 1 )

(2-48)

GF ( z ) N 2 2

(2-49)

where, N1 Rv / (1 N2 ) , N2 e( Rv / Lv )Ts
The control algorithm can be implemented in PSCAD using the following equation
(2-50) and is also illustrated in Figure 2-27.

N1I (kTs ) N2uva _1 (kTs ) uva _1 ((k 1)Ts )

Sample & Hold

(2-50)

G F (s)

Sample & Hold

Ia_ref (t)
Sample&Hold

u va_ 2

Ia

u va_1

N1

e-sTs

++

u va_ref

Ia

z-1

Gp(z)

N2

Figure 2-27 The structure of the solving feedback transfer function DB current controller

2.3.3.4 System Stability Analysis

The system stability analysis is done through frequency response analysis of the open
loop transfer function of the simplified model of each controller. Table 2-2 shows the open
loop transfer function of each DB current controller, and Figure 2-28 is given to
summarize the frequency and step response of various DB controllers.

48

2.3. DESCRIPTION OF CONTROLLERS AND DETERMINATION OF CONTROLLER

PARAMETERS
Table 2-2 Open loop transfer function of DB current controllers
Types of DB controller

Open loop transfer function

Reducing the proportional gain method

z-2 /(2(1- z-1))

Smith predictor and IMC

based two sample delays DB
Solving feedback transfer function
based two sample delays DB

z-2/[(1- z-1) (1+ z-1)]

z-2e -2RvTs/Lv/(1- z-2 e-2RvTs/Lv)

a) Reducing the proportional gain method

b) Smith Predictor and IMC based two sample delays DB

c) Solving feedback transfer function based two sample delays DB

Figure 2-28 Frequency response and step response of various DB current controllers

49

CHAPTER 2: VSC-HVDC CONTROL STRUCTURES

As they are shown and observed in Figure 2-28 that the Smith Predictor method, the
IMC method and the solving feedback TF method have nearly the same bandwidth, the
same cut-off frequency and the same settling time. The performance of the reducing gain
method is observed slightly poor specifically with a relatively narrower bandwidth, higher
overshoot and longer settling time. Note that, the results in Figure 2-28 are obatained based
on the simplified linear models under the assumption that the voltage applied across the
AC side L/R circuit is unlimited. It is done for analytical purpose only, and which is not
the authentic case in practice.
2.3.3.5 Sensitivity to Plant Parameters
The expressions of the sensitivity to plant parameters for each of the four controllers
are shown in the Table 2-3 which are derived according to the sensitivity formula defined
as the percentage change in overall transfer function divided by percentage change in the
plant transfer function [12], The sensitivity formula is depicted by,

S (s)

G ( s ) / Go ( s)
G ( s ) GPo ( s)

GP ( s ) / GPo ( s) GP ( s) Go ( s)

(Gdb ( s ) G p ( s) / (1 GF ( s)Gdb ( s) G p ( s))) GPo ( s )

GP ( s )
Go ( s )

(2-51)

(Gdb ( s ) G p ( s) / (1 GF ( s)Gdb ( s) G p ( s)))

GP ( s )
Gdb ( s ) / (1 GF ( s)Gdb ( s) G p ( s)) 2
S ( s)

1
1 GF ( s)Gdb ( s) G p ( s)

(2-52)

Table 2-3 Sensitivity to plant parameters

50

Sensitivity Expressions

Sensitivity Results

Reduce Gain

1 z 1
2 2 z 1 z 2

0.2239

Smith Predictor & IMC

1 z 2

0.6173

Solving Feedback

1 e2 aTs z 2

0.6152

2.3. DESCRIPTION OF CONTROLLERS AND DETERMINATION OF CONTROLLER

PARAMETERS
It has been observed through the simulation studies that the sensitivity of one sample
delay DB with reducing gain has the lowest sensitivity to plant parameters with respect to
the operational frequency of 50Hz. The remaining other three two sample delays DB
controllers demonstrate similar higher sensitivity values reference to the plant parameters.
2.3.3.6 Computer Simulation of DB Current Controller
The simulation model has been developed by using the equivalent circuit model shown
in Figure 2-21 and the block diagrams given in Figure 2-25 and Figure 2-27, and the
complete system was implemented in PSCAD. In addition, in order to improve the
dynamic performance of the controller, a disturbance observer [127] is introduced in each
control structure. The simulated circuit parameters are given in the Table 2-4.
Table 2-4 Simulation parameter
Parameters
Values
The rated power

75 MW

AC bus voltage

62.5kV

Inductance, Lv

0.15 pu

Resistance, Rv

0.0015 pu

Sampling time period, Ts

0.001s

A. Step Response Simulation of the DB Current Controllers

It is expected from a control system to operate satisfactorily when it is subject to both
large and small disturbances. For the purpose of evaluating the performance of the
transient operation, the simulation studies are carried out by varying the level of the active

(pu)

(pu)

current reference from -1 pu to -2 pu and followed by an increase from -2 pu to -1.8 pu.

Time (seconds)

51

CHAPTER 2: VSC-HVDC CONTROL STRUCTURES

Time (seconds)

Reduce gain results

(pu)

(pu)

a)

Time (seconds)

Time (seconds)

b) Smith Predictor results

52

(pu)

(pu)

2.3. DESCRIPTION OF CONTROLLERS AND DETERMINATION OF CONTROLLER

PARAMETERS

Time (seconds)

Time (seconds)

Internal model control results

(pu)

(pu)

c)

Time (seconds)

53

CHAPTER 2: VSC-HVDC CONTROL STRUCTURES

Time (seconds)

d) Results obtained by solving feedback transfer function

Figure 2-29 PSCAD step response simulation results of the DB current controller
It can be observed from Figure 2-29a that the reducing gain method has a frequently
altered current as seen from dq synchronous reference frame plots due to the existence of
phase difference caused by reducing the gain. The settling time of this controller is fast, but
its overshoot in reactive current is also the highest among these four controllers, which is
not desirable. The Smith Predictor method has approximately the same performance with
IMC based DB when being subject to the step response tests. However, the overshoot in
reactive current is observed slightly higher than IMC based DB controller. Compared with
the IMC based method, the solving feedback TF method has a much shorter settling time,
but with a slightly higher reactive current overshoot. The summary of the characteristics of
the simulation results are given in Table 2-5.
Table 2-5 Summary of specific data obtained from Figure 2-29
Overshoot
Controller
Step Test
Settling
Overshoot
Reactive
Type
Type
Time (ms)
(%)
Current (%)

Error
(pu)

Large Step

14.9

16

84.6

0.479

Small Step

2.5

0.2

Smith

Large Step

26

25

48.8

0.349

predictor

Small Step

20

5.46

0.110

IMC

Large Step

25

17.2

34

0.358

Designed

Small Step

21

4.9

0.067

Solving

Large Step

14.5

20

44.78

0.3348

Feedback

Small Step

14.9

16

84.6

0.479

Reduce Gain

54

Largest

2.3. DESCRIPTION OF CONTROLLERS AND DETERMINATION OF CONTROLLER

PARAMETERS
The simulation results presented here simply aim to verify the accuracy of the analysis
given. The observed discrepancies are highly induced by the model linearization. It can be
concluded that the entire performance of a controller is a trade-off between the response
speed and reactive current overshoot. Furthermore, the solving feedback transfer function
method proves to be superior to the other methods specifically during the large step
response with a fast settling time and reasonable reactive current overshoot.
B. DB Current Control System Performance Simulations under Plant Parameter
Variations
In this part of the simulation study, only the IMC based method and solving feedback
transfer function method have been considered, since the one sample delay with Smith
Predictor is approximately equal to the IMC based two sample delays DB current control.
Therefore, the reducing gain method can be analysed as one sample delay DB with
controller designed by halving the original plant parameters.
In order to examine the effect of parameter variation in each controller, the mean
deviation examination method [52] is used in this study. The different mean deviation
values equation (2-53) are used by varying the plant parameter value for three different Id
and Iq specifications, where Id and Iq represent for the current references in the dq reference
frame.
Mean

Deviation

( I d I d 0 )2 ( I q I q 0 )2
I d 02 I q 02

(2-53)

where, Id0 and Iq0 represent the resulted current under original plant parameter.
The simulation results shown in the Table 2-6 demonstrate that the resistance has little
effect on the sensitivity, which is mainly determined by the inductance. The sensitivity of
solving the feedback transfer function method is somewhat lower than the IMC based DB
controller in this particular case, and it is the same with the analysis given in section
2.3.3.4. However, the simulation result is significantly smaller than the results given in
section 2.3.3.4, which is likely due to the inaccuracy introduced during simplification. The
simulation results show that the output of control system have an approximate 3%
variation when with the 10% change in an inductance value.

55

CHAPTER 2: VSC-HVDC CONTROL STRUCTURES

Table 2-6 Mean derivation of simulation results

IMC(%) SF(%) IMC(%) SF(%) IMC(%) SF(%)

Original Plant

Iq

Id

Iq

Id

Iq

Id

-1

-1

-1

-1

R L

0.8382

0.2557

0.5

0.203

0.354

0.358

R L

0.559

0.487

0.65

0.254

0.703

R L

2.761

2.716

2.75

3.198

3.147

2.974

R L

3.25

2.795

3.238

3.758

3.522

R L

3.509

3.27

2.761

3.198

3.041

2.974

R L

3.25

2.795

3.5

3.238

3.487

3.48

2.4 Conclusion
This chapter has investigated several linear inner current controllers from operational
principles introduction and their controller design methodologies. Firstly, the operation
principle of the PI control is introduced and discussed. Furthermore, a methodology for the
design and optimization of the various loops of controllers of the PI controlled VSCHVDC transmission system is developed. This is adequately effective for a relatively
strong system. However, for the weak AC system application, the inaccurate feed-forward
compensation caused by low-order harmonics distortion in the case of PI control structures
stands for the major disadvantage [61, 120]. In addition, the cross-coupling terms also adds
to these shortcomings further. The theory and the simple control design method for the PR
control strategy are also introduced in this chapter. It was reported that this method has a
serious frequency sensitiveness problem, which makes it hard to implement when
interconnected with a weak AC grid characterized by large grid impedance.
As suggested in [54], the DB controller is the best-tailored scheme in a weak AC
application in terms of robustness against parameter variation and system disturbances.
Therefore, this chapter also studied four different types of DB controllers by including one
sample delay with the i) reducing gain and ii) Smith Predictor, two sample delays based on
iii) IMC control design and iv) solving feedback transfer function. It was also found that

56

2.4. CONCLUSION

the two sample delays DB current controller based on IMC derived and introduced in this
chapter demonstrates similar performance as the one employed by A. Timbus, M. Liserre,
R. Teodorescu, P. Rodriguez, and F. Blaabjerg in [54].
The analysis of the results showed similar step responses for both of two sample delays
DB current controllers. However, simulation results indicated that solving feedback
transfer function method proves to be superior to the other methods specifically during the
large step response with a fast settling time and with an acceptable reactive current
overshoot. In addition, it was observed that simulation results of the same method presents
smaller variations under parameter sensitivity tests compared with the IMC based DB
current controller. Therefore, the solving feedback TF based DB control method is chosen
as the definite inner current control scheme which is utilized in the remaining chapters of
this thesis.

57

Chapter 3: Analytical Modeling of DB

Controlled VSC
In the previous chapter, the digital DB current control which is implemented in abc
natural reference frame was introduced and evaluated. This control method has two major
benefits: i) fast response to set point change and ii) limiting the current magnitudes during
ac faults, which offers better sinusoidal signals tracking capability compared with the
conventional PI control and PR control. Until now, however, no small signal model for this
type of DB controlled VSC has been reported in the literature. A major difficulty withholds
the establishment of the small signal model for such DB controlled VSC is because it
contains non-linear time invariant (non-LTI) and time varying components. Hence,
conventional linear control design theory cannot be applied to it directly. Moreover, it is a
digital controller, which makes it more complicated to develop the small signal model.
Therefore, to understand and implement this control scheme, it is necessary to develop
a LTI model (small signal model) for the DB controlled VSC. In this chapter, Pad
Approximation is used first to transform the model from discrete domain to continuous
domain. Then, a methodology for model transforming from the reference frame to dq

58

3.1. DIAGRAM OF DB MODEL

synchronous reference frame was proposed as an extension to the Charles Saos and Peter
W Lehns works in [78-80].
This newly proposed reference frame transformation approach, which facilitates the
understanding the DB controlled VSC, is one of the original contributions of this thesis.
The study aims at developing a detailed higher order small signal model for the discrete
DB current controlled VSC, which can facilitate further linear controller design, system
parameterizations when implementing such type of VSC based applications, and
establishing the foundation for the system small-signal stability analysis that Modal
analysis (e.g. Eigen values, mode shapes, participation factor, etc.).

3.1 Diagram of DB model

The system structure of the proposed benchmark small signal model is given by Figure
3-1. Firstly, the input current references

and

for the inner current controller

are pre-set based on the specified active power Q and reactive power P, where Ic stands for
converter output current, the superscript C represents the converter frame, the subscript dq
implies the variable is in the dq synchronous reference frame, the subscript RI denotes
Real-Imaginary, and the subscript ref indicates it is a reference value. Note that the
similar format is used in the rest of the thesis. Then the input current references are
compensated with a two sample period ahead values, followed by a sampling block to
obtain the discrete form of the input current references

and

. In the

meantime, the measured converter output currents pass through a ZOH block and feedback
gain GF(s), which arrive at

and

respectively.

Based on the reference values and measured values of the input currents, the current
errors

and

as inputs to the discrete DB controller are generated. Using

these current errors, the voltage drop across the controlled plant can be predicted by the
DB control algorithm, including

and

. Moreover, the filter bus

voltage should also be accessed and compensated by a 1.5 sample period ahead values.
Then the second part of the converter input voltage references

are obtained. Finally, the converter input references

can be generated by computing the summation of

and
,

and

respectively.

59

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

PLL

c
cq _ q

I (k )
c
cd

(k )

(k )

Vcdc _ ref 1 (k )

I cdc _ e (k )

Dead- beat
Control
+
I cqc _ e (k )

Vcqc _ ref 1 (k )

I (k )

GF ( s )
+

c
cq

Vcdc _ ref (k )
Vcqc _ ref (k )

c
cd

Vcdc _ ref (k )'

d
Computation
Vcqc _ ref (k )'
q
Delay

VcR _ ref

R
I

VSC

Sampling
V

c
cq _ ref 2

Vcdc _ ref 2
Vcqc _ ref 2

(k )

I
I cqc _ ref (k )' Sampling

U dc E g

PLL

c
'
cd _ ref

I cqc _ ref '

I cIg
u gfR
u gfI

EIgg

Rg

c
fd

u cfq

d
q

R
I
PLL

u cfd '
cos(1.50Ts ) sin(1.50Ts )
sin(1.5 T ) cos(1.5 T )
c '
0 s
0 s u fq

Voltage Prediction

I cdc _ ref (k )'

Grid

g
cI

Filter

Vcdc _ ref 2 (k )

I cRg

VcRg

VcI _ ref
PLL

R
I

d
q

I cqc

Sampling

Filter

+
+

c
cd _ d

cos(20Ts ) sin(20Ts )
sin(2 T ) cos(2 T )
0 s
0 s

I cdc _ ref
I cqc _ ref

Reference Current Phase Compensation

Figure 3-1 Small signal model of VSC with the discrete DB current controller
The computation delay needs to be taken into account as well which is approximately
one sample period (Z-1). Before conducting the calculation in grid reference frame, the
VSC input references (

) in the converter reference frame should

be transformed into the grid reference frame, dq to RI block, as shown in Figure 3-1.
Similarly, the grid output current and voltage should also be back transformed from the
grid frame to the converter frame, RI to dq block, before conducting the converter frame
calculation.
By ignoring the dynamics of the DC voltage, the VSC is supposed to be ideal. This
means that the VSC ideally outputs voltage as required. Utilizing this voltage, the
converter reactance and the filter bus voltage, the converter output current can be easily
obtained.

3.2 Characterizing the Components of VSC Small Signal

Model

60

3.2. CHARACTERIZING THE COMPONENTS OF VSC SMALL SIGNAL MODEL

The procedure on how to develop the small signal model for each individual
component needs to be carefully addressed. Therefore, the block diagram and its
corresponding non-linear equations are established first, which is followed by a
linearization process to obtain the small signal model and the small signal state-space
equations. Finally, simulation results based upon tests setting on both of Simulink/Matlab
and PSCAD platforms are utilized to assess the accuracy of the linearized models. The
following sub-sections of this thesis will provide the details of these studies. It should be
clarified that in the remnent of the thesis the large signal model can also be termed as nonlinear model which is detailed model transient simulation (PSCAD). Similarly, the small
signal model can also be named as linear model which is simplified s-domain simulation
(Matlab).

3.2.1 Phase Locked Loop

The successful implementation of DB current control for a VSC highly depends on the
accuracy of the performance of a PLL which can extract the phase of the filter bus voltage
and system operating frequency, specifically in the case of the digital DB controlled VSC
integrating with a weak AC grid. A good PLL requires four desirable characteristics: i) fast
response speed, ii) no steady state error, iii) broad frequency acquisition range and iv) high
noise rejection capability. A fundamental review and assessment of research on the d-q-z
type PLL are given in reference [52, 56, 57] and which is currently used worldwide. This
section includes discussions concerning the operational principle of such type of PLL, the
derivation of a linearized model for the PLL and model verification based on PSCAD
simulation. Furthermore, the PLL dynamic performance for several disturbance scenarios
will also be examined and discussed.
As shown in Figure 3-2, the three-phase PLL is composed of three parts: i) phase
detector; ii) PI filter and iii) voltage controlled oscillator (VCO). In PLL, The phase
detector compares the phase difference between the grid phase voltages in the reference
frame ( u s , u s in per unit) and the outputs of VCO ( u v and u v ) to calculate the error d.
This is then passing through a filter outputs uf followed by a magnification by a gain Kv
and outputs the frequency difference . The output frequency is the summation of the
central frequency 0 and . The output phase angle is obtained by the integration of
frequency [52]. In the meantime, the phase angle is utilized by the two functional

61

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

blocks sin () and cos () of the VCO block, which are magnified by a gain KL, then output
uv, uv. Finally the close loop PLL is formed.
u v
u s

u s

1
u sb

1
u sb

K p_PLL +

K i_PLL

uf

Kv

1
s

Central frequency 0
u v

u sb =voltage base

cos

KL

Phase Detector

Low Pass
Filter

KL

sin

Voltage Controlled
Oscillator

Figure 3-2 Structure of phase locked loop

Using the arrangement in Figure 3-2, the output of the phase detector d can be easily
derived as,

d us uv us uv
usp sin(1t 1 ) K L cos usp cos(1t 1 ) K L sin

(3-1)

Since the PLL is assumed to be locked to its input signal, then the output of PLL can
be given by,

2t 2

(3-2)

Substituting equation (3-2) into equation (3-1) yields,

d usp sin(1t 1 ) K L cos(2t 2 ) usp cos(1t 1 ) K L sin(2t 2 )

d usp K L sin[et (1 2 )]

(3-3)

Since, e t=1t-2 t , the above equation can be expanded as

d usp K L sin[et (1 2 )]
d usp K L (sin et cos(1 2 ) cos et sin(1 2 ))

(3-4)

Note that if the et is small enough, then sin e t e t and cos e t 1 . However, this
is not strictly true since the grid frequency is a time dependent term. Therefore,
it is worth noting that under some condition, small variation in grid frequency may results
in large phase difference. Hence,

62

3.2. CHARACTERIZING THE COMPONENTS OF VSC SMALL SIGNAL MODEL

d usp K L (et cos(1 2 ) sin(1 2 ))

(3-5)

Assuming the variations in grid frequency and initial phase angle is small enough,
which means 1=2 and 1= 2,
d usp K L sin(1 2 )

(3-6)

d usp K L (1 2 )

Therefore, the double frequency components 21t are fully eliminated by this
arrangement. However, under unbalanced conditions, phase can be locked to the positive
sequence component of its input signal.
3.2.1.1 Derivation of Linearized Model of PLL
Linearized model of the PLL will be derived in this sub-section considering dynamic of
each individual component of PLL independently.
For facilitating the analysis, the nominal system frequency 0 of the VCO which is
assumed to be constant is involved linking the input phase angle 1 and the output phase 2.
Using Figure 3-2, the input reference frame voltages to PLL block can be given by
us
sin(0t 1 )
u usp

cos(0t 1 )
s

(3-7)

The outputs of VCO in reference frame ( u v and u v ), can be obtained by

uv
cos(0t 2 )
u K L

sin(0t 2 )
v

In which, the phase angle

and

(3-8)

can be given by

1 1 (t )t 1 (t ) 0t

(3-9)

2 (t ) 0t

(3-10)

A. Phase Detector
Note that using equation (3-9), one can derive that 1 (t)=1 +0 t-1 (t)t . If 2 =1 ,
using equation (3-2) and (3-10), 2 (t)=2 +0 t-1t can be obtained. If 1(t) and 2(t) are
substituted into equation (3-6), the controlled error can be obtained as

d usp K L sin(1 2 )
d usp K L sin(0t 1 1t (0t 2 1t ))
d usp K L sin(1 2 )

(3-11)

63

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

Assuming the PLL is initially locked, by definition, the phase error (1 - 2 ) is

approximately zero,

d usp K L (1 2 )

(3-12)

Therefore, the linearized model for phase detector becomes

d / (1 2 ) usp K L Kd

(3-13)

B. First Order Low Pass PI Filter

Using the first order low pass PI filter definition

u f ( s)

d ( s)

K p _ PLL

Ki _ PLL
s

(3-14)

the linearized filter output can be given by

u f ( K p _ PLL

Ki _ PLL
s

)d

(3-15)

C. Voltage Controlled Oscillator

The output angle of the VCO can be given by
t

(t ) [0 ]d 0t KV _ PLLu f ( )d
t

2 (t ) (t ) 0t 2 (t ) KV _ PLLu f ( )d

2 ( s)
u f ( s)

(3-16)
(3-17)

K v _ PLL
s

(3-18)

Hence, the linearized phase angle can be written as

2

K v _ PLL
s

u f

(3-19)

D. Formulation of Linearized State-Space Model of PLL under Balanced Condition

Since,

d us cos us sin
2
3

1
3

1
3

3
3
usb
usc ) sin
3
3
usa
2
2
2
d cos( ) cos( ) cos( ) usb
3
3
3
usc

d ( usa usb usc ) cos (

64

(3-20)

3.2. CHARACTERIZING THE COMPONENTS OF VSC SMALL SIGNAL MODEL

d usd

This means

d usd usp K L (1 2 )

Therefore,

(3-21)

since 10 =02 at nominal operating point A, the linearized error becomes

d usd usp K L (1 2 )

(3-22)

where, the subscript 0 stands for nominal operating point.

Therefore, the PLL model given in Figure 3-2 can be simplified as the model presented
in Figure 3-3, which forms the basis for deriving the small signal of PLL.

u sd

K p_PLL
K i_PLL

KV

1/s

Figure 3-3 Non-linear model for PLL

The mathematical equations of the non-linear model is given by [53],
0 0
d l Ki _ PLL

usd l

dt 2 KV K p _ PLL
KV 0

and,

0 KV K p _ PLLusd Kvl

(3-23)

(3-24)

When the above state space equations are linearized, the small signal model for PLL
can be obtained as,
0
d l Ki _ PLL

usd l

dt 2 KV K p _ PLL
KV

(3-25)

3.2.1.2 Analysis of linearized PLL model

In order to develop a PLL controller, it is necessary to obtain the transfer function of

2 (s)/1 (s) , where 2 (s) denotes the output of PLL and 1 (s) represents the input to
the PLL control block. The block diagram of the linearized model of the PLL is illustrated
below,

65

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

1 (s)

K d =u sp K L

d (s)=u sd

u f (s)
PI

Kv
s

2 (s)

Figure 3-4 Block diagram of PLL linearized model

The open and closed loop transfer functions of the model can be given by
Ki _ PLL KV
2 ( s)
K d ( K p _ PLL
)
1 ( s) open loop
s
s
K d KV ( Ki _ PLL K p _ PLL s )
2 ( s)
2
1 ( s) closeloop s K d KV K p _ PLL s K d KV Ki _ PLL

2n s n 2
2 ( s)
2
1 ( s) closeloop s 2n s n 2

K
n K d KV Ki _ PLL , n p _ PLL
2 Ki _ PLL

(3-26)

(3-27)

(3-28)

and the lock in range for PLL is [128]

L 2n Kd KV K p _ PLL

(3-29)

3.2.1.3 Parameter Design Procedure

The systematic design procedure for PLL is well documented in [52]. Therefore, this
sub-section aims to review the PLL controller design process briefly.
A. Determination of n
The lock in range is first determined by the application requirement. Using the pre-set
damping ratio , the nature frequency n can be determined first.

n L / 2

(3-30)

B. Determination of VCO Gain Kv_PLL

The gain Kv_PLL can be given by

Kv SM arg in (max min ) / (u f max u f min )

(3-31)

where, SMargin is the stability margin, max and min are the maximum and minimum
frequency that the PLL can lock to. In practice, the VCO control signal is usually limited to
a range smaller than the VCO supply voltage, which is mostly +5kV [52]. For example, the

66

3.2. CHARACTERIZING THE COMPONENTS OF VSC SMALL SIGNAL MODEL

values of ufmax and ufmin, with +5kV supply voltage, can be chosen as 90% 5kV and
10% 5kV respectively.
C. Let Kd = 1, then Determine Kp_PLL and Finally Determine Ki_PLL
It should be noted that the lock-in range and speed response to step change of PLL is
proportional to the characteristic frequency, but inversely proportional to the capability of
harmonics rejection. Decision on the value of n has to be made by trading off the
response speed and noise rejection capability of PLL according to the requirements of
applications.
3.2.1.4 Case study
Using the PLL parameters shown in Table 3-1, the transfer function of PLL can be
defined as equation (3-32).
K d K v ( Ki _ PLL K p _ PLL s )
2 (s)
2
1 ( s) closeloop s K d K v K p _ PLL s K d K v Ki _ PLL

(3-32)

2351.9s 4.509 105

2
s 2351.9s 4.509 105

Rated Line voltage (RMS)

Table 3-1 PLL parameters

Us_line_rated
62.5kV
PI Parameter

Phase Voltage Peak value

Us_phase_peak

VCO Functional Block

Gain
Equivalent Gain Phase
Detector
VCO Gain

51.031
kV

KL

Kd

51.031

KV

90

PI Parameter
Nature

Kp_PLL

0.163

Ki_PLL

31.25

213.74

Damping Ratio

1.751

Lock in Range

Frequency

748.61

Figure 3-5 shows the frequency response of the PLL, where the gain margin is infinite
and the phase margin is 84.5o. This indicates that the designed PLL is stable.

67

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

80
Magnitude (dB)

60
40
20
0

Phase (deg)

-20
-90

-135
Pm=85.4 deg (at 2.336e+003 rad/sec)
-180
1
10

10

10
Frequency (rad/s)

10

Figure 3-5 Frequency response of PLL in the case study

3.2.1.5 Verification by PSCAD simulation
To compare the performances of the linear and non-linear PLL model developed, two
step change tests on the phase of the source voltage Usa are employed using PSCAD. These
tests are summarized below.
A. 1% (3.6o) Step Change in Phase
This simulation is performed by setting a fixed frequency of 50Hz for the reference
voltage and introducing a 1% step in phase on the phase angle which is 3.6o. The output of
the PLL is a saw-tooth waveform as shown in Figure 3-9. To isolate the perturbation in the
measured phase angle, the angular rotation of the input signal, a constant periodic angle
0t is subtracted from measured phase angle as shown in Figure 3-6 (solid blue curve). The
results of 1% step phase change of PLL in PSCAD shows a good agreement with the
linearized model in MATLAB (dotted red curve), which successfully verify the accuracy
of the small signal model. The specific characteristic values of the linearized model are
summarized in Table 3-2.
0.08
0.06

Rad

0.04
0.02
0
-0.02
0

0.005

0.01
time(s)

PSCAD
MATLAB
0.015
0.02

Figure 3-6 Linearized model verification, 1% step change in phase

68

3.2. CHARACTERIZING THE COMPONENTS OF VSC SMALL SIGNAL MODEL

Rise
Time
0.00094

Table 3-2 Characteristics of the linearized model

Settling Settling Settling Over Under
Peak Peak Time
Time:
Min
Max: Shoot Shoot
0.0081

0.9553

1.059

5.925

1.059

0.0025

B. 50% (180o) Step Change in Phase

The conditions of the simulation study here are the same as in the previous text except
the size of the phase step change, 50%. The results presented in Figure 3-7 shows an
obvious difference at the initial step change stage in comparison with Figure 3-6.

Rad

2.5
2
1.5
1
0.5
0

0.01

0.02

0.03

PSCAD
Matlab
0.04

t(s)

Figure 3-7 50% step (180o) change in phase

The settling time in PSCAD is 9.9ms which is a little bit longer than the Matlab result
of 8.1ms. Hence, it can be easily concluded that the linear model is developed based on a
small signal and therefore 50% change cannot be considered as a small signal. However, it
is important to note that after about 20ms there is an exact match between two simulation
studies. This means that the PLL has a good tracking ability even though it experiences a
large phase perturbation.
C. 1% Step Change in Frequency
This test is performed by setting a fixed zero initial phase and a constant magnitude for
the reference voltage. Then a 1% step change in frequency (0.5Hz) is applied to the 50Hz
at t=5s which lasts for 0.02s. As shown in Figure 3-8, the settling time for the 1% step
change in frequency is 8ms.

69

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

Time (seconds)

Figure 3-8 1% step change in frequency simulated in PSCAD

D. 90% step change in Magnitude
In this test, a constant frequency of 50Hz and a zero initial phase angle and a sudden
90% step change for the magnitude are applied to the input reference voltage at t=5s.

Time (seconds)

Figure 3-9 PLL responses to 90% magnitude step change of input voltage simulated in
PSCAD

Time (seconds)

Figure 3-10 PLL output phase angle in comparison with input phase angle when being
subject to a sudden 90% magnitude step change of input voltage simulated in PSCAD

70

3.2. CHARACTERIZING THE COMPONENTS OF VSC SMALL SIGNAL MODEL

Time (seconds)

Figure 3-11 PLL output frequency in comparison with input frequency when being subject
to a sudden 90% magnitude step change of input voltage simulated in PSCAD
The results given in Figure 3-9 to Figure 3-11 show that the sudden magnitude
reduction has no effect on the PLL outputs.
E. Single Phase to Ground Fault
In this simulation study, while the initial phase is zero at a constant frequency of 50Hz
and at a constant magnitude, a ground fault is created at A-phase at t=1s.
Figure 3-12 shows the phase response of PLL under a phase to ground fault. As is
shown in the figure, the result of twice the fundamental frequency, which can be expressed
below,

y 0.00346cos(200 t )

(3-33)

Time (seconds)

Figure 3-12 PLL output phase angle when being subject to an A-phase to ground fault
simulated in PSCAD
Figure 3-13 shows the frequency response of PLL under the same fault, which can be
expressed by,

71

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

y 50.3475cos(200 t )

(3-34)

Time (seconds)

Figure 3-13 PLL output frequency when being subject to an A-phase to ground fault
simulated in PSCAD
3.2.1.6 Summary of PLL
The developed linearized model and non-linear model of PLL both have the fast
response and wide frequency range acquisition abilities. In addition, the linear model and
non-linear model match each other very well. Its performance for a large reduction in
voltage magnitude and for a single phase to ground fault are also examined in this section.

3.2.2 Frame Transformation

As shown in section 3.1, the small signal model requires calculations both in the grid
frame and converter frame. Therefore, the variables in the model need to be freely
transformed between these frames. In this sub-section, the small signal model for the
required frame transformation will be developed in terms of the VSC input voltage
reference Vc_ref, the converter output current Ic and the filter bus voltage Uf. The equations
(3-35) to (3-37) indicate the relationship between the grid RI reference frame and the
converter reference frame, as illustrated in Figure 3-1. The figure below is also given to
graphically demonstrate this relationship.
g
'
VcRI
_ ref (k ) e
c
I cdq
e

u cfdq e

72

j ( _ PLL /2)

c
'
Vcdq
_ ref (k )

j ( _ PLL /2)

j ( _ PLL /2)

(3-35)

g
I cRI

(3-36)

u gfRI

(3-37)

3.2. CHARACTERIZING THE COMPONENTS OF VSC SMALL SIGNAL MODEL

qc

0 t

_ PLL

Rg

dc

0 t

s t

Ig

U f

Figure 3-14 Relationship between converter reference frame and grid RI reference frame

3.2.2.1 Mathematical Equation of VSC Input Voltage Reference

If we expand equation (3-35), equation (3-38) can be obtained which then can be
utilized for linearization.
VcRg _ ref (k )' sin PLL Vcdc _ ref (k )' cos PLL Vcqc _ ref (k )'

(3-38)

VcIg _ ref (k )' cos PLL Vcdc _ ref (k )' sin PLL Vcqc _ ref (k )'

The phase angle PLL in the above set of equation is the phase difference between the
filter bus voltage and grid phase reference point, which is a constant value by excluding the
time varying component 0t.
The linearized form of equation (3-38) can be given as
VcRg _ ref (k )' sin PLL 0 Vcdc _ ref (k )' cos PLL 0 Vcqc _ ref (k )' VcIg _ ref (k )'0 PLL
VcIg _ ref (k )' cos PLL 0 Vcdc _ ref (k )' sin PLL 0 Vcqc _ ref (k )' VcRg _ ref (k )'0 PLL

(3-39)

3.2.2.2 Verification of the Large Signal Mathematical Equation

To verify the frame transformation, the test model that a controlled three-phase voltage
source with an inductance in series is used as shown in Figure 3-15. The main objective of
this approach is to measure the filter bus voltage Us in real-imaginary (RI) reference frame
(i.e. the grid frame). Therefore, the tests involved three steps: i) directly measuring Us by
locking the phase of the grid reference point voltage UG by PLL, and obtaining G to
calculate

; ii) measuring the phase of the filter bus voltage Us, and the phase

difference PLL between the grid source voltage and the filter bus voltage. Note that using
the angle , the representation of Us in the converter reference frame (

) can be

73

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

computed, which is then followed by a transformation to grid frame using equation (3-35).
This calculation utilizes PLL to obtain

. And iii) comparing the results

obtained in the previous two steps.

U G G

LG U s

Figure 3-15 Controlled voltage source representation with an inductance

The final results are compared and shown in Figure 3-16, which successfully verify the
accuracy of the mathematical formula of the frame transformation.

Time (seconds)

Figure 3-16 Test results for grid RI reference frame transformation

3.2.2.3 Testing for the Small Signal Mathematical Equation

In this section, only the steady state value is considered since the equations adopted to
describe the relationship of the frame transformation are two algebraic equations.
Therefore, the transient process can be ignored.
C
During the testing process a 1.0 pu large signal u sq0
is first applied as an input which

varied at t=0.5s on q-axis of the filter bus voltage. This is followed by a secondary 0.99 pu
C

large signal u sq0 that varied at t=0.5s imposed in another simulation before transforming to
the grid RI reference frame. The errors in the output quantities of both simulations are
G
G
given by u sR_Large
and u sI_Large
. In addition, a 0.01pu small step change on q-axis of the

filter bus voltage is applied simultaneously to the small signal model at t=0.5s. Then the
G
G
outputs u sR_small
and u sI_small
are compared with the results obtained from the non-linear

model (see Figure 3-17).

74

3.2. CHARACTERIZING THE COMPONENTS OF VSC SMALL SIGNAL MODEL

0pu
1.0pu
0pu

usdC 0
u

usqC 0

usqC 0

usdC

+0.99pu

G
dq to RI
usR
0
sin PLL cos PLL
G
usI 0
cos
sin PLL

PLL

C
sd 0

usqC +-

usdC

sin PLL
cos

PLL

usqC

usdC

sin PLL
cos

PLL

usqC

Non-Linear
g
usR
_ L arg e

-+
-

usIg _ L arg e

G
usR

cos PLL
G
sin PLL usI

g
usR
_ small
usIG 0
+-

cos PLL
sin PLL

++

G
usR
0

PLL
PLL

g
usR
_ small

Linear

Figure 3-17 The test circuit for small signal mathematical equation

The simulation results are given in Figure 3-18. Note that the exponentially slow rising
trend in the figure is actually caused by the combination of the voltage ramp up time which
is set as 0.02s in this test and the dynamics introduced by PLL utilized in the
transformation process of voltage from dq frame to RI frame. These results successfully
confirm the accuracy of the developed small signal model for the grid frame
transformation.

Time (seconds)

Figure 3-18 Small signal test results for the grid frame transformation
The large signal and the small signal algebraic equations for converter terminal current
Ic transforming from the grid RI reference frame to the converter dq reference frame are
given in equations (3-40) and (3-41).

I cdc sin PLL I cRg cos PLL I cIg

I cqc cos PLL I cRg sin PLL I cIg

(3-40)

75

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

g
I cdc sin PLL 0 I cR
cos PLL 0 I cIg I cqc 0 PLL
g
I cqc cos PLL 0 I cR
sin PLL 0 I cIg I cdc 0 PLL

(3-41)

Note that this transformation is similar to the VSC input voltage reference, which is
simply an opposite transforming direction. Therefore, no verification studies are needed.
The large signal and the small signal models for the filter bus voltage u Cf transforming
from grid RI reference frame to converter dq reference frame can be given by
u Cfd sin PLL u gfR cos PLL u gfI
u Cfq cos PLL u gfR sin PLL u gfI
u Cfd sin PLL 0 u gfR cos PLL 0 u gfI u Cfq 0 PLL
u Cfq cos PLL 0 u gfR sin PLL 0 u gfI u Cfd 0 PLL

(3-42)

(3-43)

3.2.3 Grid Voltage Signal Filter

The transfer function for voltage signal filter G Filter (s) is represented by,
GFilter ( s)

u cfd '
u

c
fd

u cfq '
u

c
fq

1
1 Tf s

(3-44)

The state-space equations can be established as in equation (3-45) and (3-46).

du cfd '

u fd
1
u cfd '
dt
Tf
Tf

du cfq '
dt

(3-45)

1 c ' u fq
u fq
Tf
Tf

(3-46)

After linearizing, the small-signal state-space model for the grid voltage signal filter
can be given by,
d u cfd '
dt
d u cfq '
dt

u fd
1
u cfd '
Tf
Tf
c

u fq
1
u cfq '
Tf
Tf

(3-47)

(3-48)

Note that there is no need to compare the large signal model and the small signal model
with respect to this component, since they do have the same formula. The only difference
is the step size to the input signals, which will naturally output proportionally results.

76

3.2. CHARACTERIZING THE COMPONENTS OF VSC SMALL SIGNAL MODEL

3.2.4 Reference Current Phase Compensation and Voltage

Prediction Block
The current compensation equation and inverse transformation of the compensated
current reference back to the converter dq reference frame are given in equation (3-49) and
equation (3-50) respectively.
c
I cc ref cos( _ PLL 0 20Ts ) sin( _ PLL 0 20Ts ) I cdref

c
I c ref sin( _ PLL 0 20Ts ) cos( _ PLL 0 20Ts ) I cqref

(3-49)

c
'
I cdref
cos _ PLL 0 sin _ PLL 0 I cc ref
c '
c
I cqref sin _ PLL 0 cos _ PLL 0 I c ref

(3-50)

If we substitute equation (3-49) into equation (3-50), the transformation relationship

between the compensated current and the original current in the converter dq reference
frame can be obtained,
c
'
c
c
I cdref
cos(20Ts ) I cdref
sin(20Ts ) I cqref
c
'
c
c
I cqref
sin(20Ts ) I cdref
cos(20Ts ) I cqref

(3-51)

The below set of equation denotes the linearized form of the above equations,
c
'
c
c
I cdref
cos(20Ts ) I cdref
sin(20Ts ) I cqref
c
'
c
c
I cqref
sin(20Ts ) I cdref
cos(20Ts ) I cqref

(3-52)

To verify these equations, a 0.01pu (1pu-0.99pu) step change is applied to the inputs of
both the large signal model and the small signal model, and the tests results are given in
Figure 3-19, which shows a good agreement.
-3

10

x 10

8
6

Icref(pu)

4
2
0

Icdref-matlab

-2

Icqref-matlab

-4

Icdref-PSCAD

-6
0

Icqref-PSCAD
0.005

0.01
Time(seconds)

0.015

0.02

Figure 3-19 Current compensation block verification

77

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

For the voltage prediction block in Figure 3-1, the relationship between the input
voltage u Cfd ' , u Cfq ' and the output predicted voltages in reference frame vCc_ref2 , vCc_ref2 can
be given by,

vcc _ ref 2 cos( PLL 0 1.50Ts ) sin( PLL 0 1.50Ts ) u cfd '
c

c '
vc _ ref 2 sin( PLL 0 1.50Ts ) cos( PLL 0 1.50Ts ) u fq

(3-53)

Similarly, the back transformation from the reference frame to the converter dq
reference frame for the predicted voltages can be written as,

vcdc _ ref 2' cos PLL 0 sin PLL 0 vcc _ ref 2

c
'
vcq _ ref 2 sin PLL 0 cos PLL 0 vc _ ref 2

(3-54)

Substituting equation (3-54) into equation (3-53), the direct link between the input
voltage and output predicted voltage are obtained as equation (3-55),
vcdc _ ref 2 cos(1.50Ts ) u cfd ' sin(1.50Ts ) u cfq '
vcqc _ ref 2 sin(1.50Ts ) u cfd ' cos(1.50Ts ) u cfq '

(3-55)

The linearized form of the set of equations is given by,

vcdc _ ref 2 cos(1.50Ts ) u cfd ' sin(1.50Ts ) u cfq '
vcqc _ ref 2 sin(1.50Ts ) u cfd ' cos(1.50Ts ) u cfq '

(3-56)

The verification process for the accuracy of the linearization of this component is
similar to the current compensation block. Hence, it will not be repeated.

3.2.5 Discrete DB Current Feedback Control Block and Current

Input Generation
The feedback transfer function for the discrete DB current controller yields to,

GF (s) e2( Rv / Lv )Ts

(3-57)

where, Rv and Lv represent the resistance and inductance of converter reactance, Ts

stands for the sampling time. The resultant large signal model and small signal model for
the feedback block are

I cdc d (k ) GF I cdc (k )
I cqc d (k ) GF I cqc (k )

78

(3-58)

3.2. CHARACTERIZING THE COMPONENTS OF VSC SMALL SIGNAL MODEL

I cdc d (k ) GF I cdc (k )
I cqc d (k ) GF I cqc (k )

(3-59)

Note that, there is also no need to verify the linearization of this feedback loop since it
is only a coefficient gain.
The large signal model and the small signal model for generation of the DB current
input errors can be given by
c
I cdc _ e (k ) I cdref
(k )' I cdc d (k )
c
I cqc _ e (k ) I cqref
(k )' I cqc d (k )
c
I cdc _ e (k ) I cdref
(k )' I cdc d (k )
c
I cqc _ e (k ) I cqref
(k )' I cqc d (k )

(3-60)

(3-61)

3.2.6 Characterizing ZOH Block, Sampling Block and Inner

Current Control
3.2.6.1 From Discrete to Continuous Approximation
VSCs are inherently discontinuous and they are typically equipped with digital
controllers. It is therefore necessary to develop the techniques to adequately approximate
the behaviour of these devices in the continuous domain within the bandwidth of interest.
The bandwidth of interest for small-signal rotor-angle stability analysis of power systems
is in the order of 0 to 5 Hz (31 rad/s). It is on this basis that the algebraic average responses
of the VSCs are used in power system analysis applications where rotor-angle dynamics
are of primary concern.
To develop the approximations of the digital controllers in the continuous domain it is
necessary to investigate linear transfer function approximations of the delay (i.e. e-sT).
Then equivalent continuous-time linear transfer function model for the zero order
holding (ZOH) block is developed utilizing the pure delay approximation. Then the
discrete time inner current controller model is converted to a continuous domain model
using the same approach. Finally, the characteristic of the ZOH block together with the
digital controller are investigated by its linear equivalence.

79

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

A. Pure One Sample Delay

The transfer function of the pure one-sample delay in the z-domain is F1 (z)=Z-1 and the
non-rational transfer function of the delay in the s-domain is F2 (s)=e-sTs . It is necessary to
develop a rational s-domain transfer-function approximation of the delay. Here, i) the first
order PadApproximation F3 (s)=(1-sTs /2)/(1+sTs /2) , ii) the third order PadApproximation

F4 (s)=(1-sTs /6)3 /(1+sTs /6)3 and iii) a comparably crude approximation given by a first
order lag F5 (s)=1/(1+sTs ) are investigated. The frequency responses of the delay are
compared with that of the above three approximations. Furthermore, the time-responses of
the exact delay and the approximations are compared for (i) a step signal; and (ii) a
sinusoidal input signal.
a) Frequency Response
Figure 3-20 shows the frequency responses of the pure one sample delay in s- and zdomain and the three s-domain approximations. It can be seen from the figure, the z -1 and
e-st are overlaid with each other as expected. The third order Pad approximation is the
most accurate equivalence among the three approximations. However, for the powersystem modeling purposes outlined in section 3.2.6.1, the first order PadApproximation
is more than adequate (where the sampling time Ts equals to 1ms) since it matches the
exact delay at least up to 389rad/s (62Hz), and our interested frequency range is within this
limit.

Magnitude(dB)

10
0
-10
-20
-30
0.1/Ts

e-sTs
First-Order Pade
Third-Order Pade
First Order Lag
Z-1
1/Ts

10/Ts

1/Ts
Frequency (rad/sec)

10/Ts

Phase(deg)

200
100
0
-100
-200
0.1/Ts

Figure 3-20 Frequency responses of pure one sample delay

80

3.2. CHARACTERIZING THE COMPONENTS OF VSC SMALL SIGNAL MODEL

b) Step Response
The unit step responses of the one sample delay and its approximations are shown
below in Figure 3-21.
1.5
1

pu

0.5
Z-1

0
-0.5
-1
0

Ts

2Ts

e-sTs
First-Order Pade
Third-Order Pade
First Order Lag
3Ts
4Ts
5Ts

Time in units of Ts

Figure 3-21 Step responses of pure one sample delay

Table 3-3 Characteristics of the step response corresponding to the pure one sample delay
Rise

Settling Settling Settling

Over

Under

Peak

Peak

Time

Time

Min

Max

Shoot

Shoot

e-st

8Ts

9.8Ts

100 Ts

z-1

Ts

Ts

1.1Ts

1.3Ts

0.8026

1.0138

1.3755

100

1.0138

1.7Ts

1.1Ts

2 Ts

0.8057

100

2.2Ts

3.9 Ts

0.9029

10.5 Ts

Third order
Pad
First order
Pad
First order lag

Time

Table 3-3 presents the characteristics of the step response. As it can be observed in
Figure 3-21, the one sample delay e-st and Z-1 performed very well as expected (exactly as
one sample delay). Among the three approximation methods, the first-order lag has the
longest settling time and rise time. The third order PadApproximation and the first-order
PadApproximation have the same rise time which is 1.1Ts. It was found that the third
order Pad Approximation has a comparably shorter settling time which is preferred.
However this approximation also has a higher overshoot which is undesirable. Note that,
both approximation methods have a common defect that imposes a positive zero (RHP

81

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

zero) to the system. Reflected on the unit step response performance, the plots start arising
from -1. Taking all the factors into consideration, the first-order Pad Approximation is
found to be the most suitable and sufficiently accurate to approximate the pure one sample
delay in this study.
B. Zero Order Holding Block
a) Transfer Function of Zero Order Holding Block in S Domain
The mathematical representation of zero order holding block is

e(nTs t ) e(nTs ) , where 0 dT Ts

(3-62)

If we input an ideal unit impulse signal to the ZOH block, a rectangular impulse with
magnitude 1 and the lasting time Ts will be obtained as shown in Figure 3-22, which can be
divided into a unit step followed by a delayed negative unit step with a time constant T s.
Note that, the ZOH output signal has two typical characteristics in Figure 3-22: i) contain
high frequency component (single frequency input) and ii) has the fundamental frequency
shifted by Ts/2.

Time (seconds)

Figure 3-22 Demonstration of ZOH function

For a better explanation, a typical sampled data system is shown below in Figure 3-23.
Note that the continuous signal is sampled to be a discrete signal with a sampling time Ts,
which then goes through a signal processing block, and being held by a ZOH block to be
transformed to a continuous signal again. It should be noted that the result of the processed
signal y[k] is kept constant during each sampling time.

82

3.2. CHARACTERIZING THE COMPONENTS OF VSC SMALL SIGNAL MODEL

r (t )

Ts

D( z )

r[k ] r[k Ts ]

ZOH

y[k ]

y (t )

y (t ) y[k ] , kTs t ( k 1)Ts ,

y[k ]:Constant

Figure 3-23 A typical sampled data system structure

The mathematical model of the processed continuous signal result y(t) can be described
as in equation (3-63), and can be illustrated as in Figure 3-24.
y(t ) y[k ](u(t kTs ) u(t (k 1)Ts ))

(3-63)

where, u (t) is a step function, which can be defined by

0, t 0

u (t ) 0.5, t 0
1.0, t 0

(3-64)

y
y(t)

y[k]
0

(k 1)Ts

kTs

-y[k]

Figure 3-24 Composition of a rectangular impulse

If we take the Laplace transform of equation (3-63),

1 skTs
s k 1 T
e
e s y k
s
1 esTs
y k e skTs
s
sTs
1 e
y(Ts )
s

y(t)

(3-65)

The transfer function of ZOH block can be obtained as

GZOH ( s)

1 e sTs
s

(3-66)

83

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

b) First Order PadApproximation for Transfer Function of ZOH Block

Substituting the one sample delay first-order Pad Approximation equation into
equation (3-66), the first-order PadApproximation of ZOH can be obtained as
'
ZOH

Ts
1 e sTs 1 (1 sTs / 2) / (1 sTs / 2)
( s)

s
s
1 sTs / 2

(3-67)

c) Frequency Response of the ZOH and its Approximation

Figure 3-25 shows the frequency response of ZOH and its first-order Pad
Approximation. It can be seen in the figure that the Bode plots of the exact one and the
approximate one are matched well up to 500 rad/s which is sufficient for the interested
frequency range. The attenuation at 50 Hz for the exact one is -0.0358dB and -0.108dB for
the approximate one. Furthermore, the phase shifts are found to be identical for both
models, which is 9 degree for half of the sampling period.

Magnitude(dB)

-50

-100
0.1/Ts

1/Ts

10/Ts

1/Ts

10/Ts

Phase(/deg)

0
-50
-100
-150
-200
0.1/Ts

Exact
First-Order Pade
Frequency (rad/sec)

Figure 3-25 Frequency responses of ZOH block and its approximation

d) Step Response of Zero Order Holding Block and its Approximation

As given in Figure 3-26 and Table 3-4, the ZOH block and its first-order Pad
Approximation are in a good agreement on the step responses. The reaction of the ZOH
block being subject to a step change is almost instantaneous.

84

3.2. CHARACTERIZING THE COMPONENTS OF VSC SMALL SIGNAL MODEL

1.4
1.2
1

pu

0.8
0.6
0.4
0.2
0
0

Ts

Matlab Exact
Matlab Pade
4Ts
5Ts

2Ts
3Ts
Time in units of Ts

Figure 3-26 Step responses of ZOH block and its approximation

Table 3-4 Characteristics of the step response corresponding to ZOH
Rise Settling
Settling Settling Over Under
Peak Time
Time
Time
Peak
Min
Max
Shoot Shoot
(s)
(s)
(s)
ZOH
First
order Pad

0.4605

0.5641

262.5

0.4605

0.5641

0.5756

e) Response to the Sinusoid Input for ZOH

i.

Verification by Matlab

According to the Bode plot of ZOH, the magnitude passing through the exact transfer
function should be 99.6% of the input magnitude and 98.8% for the approximate function.
The phase shift is 9 degree which is equal to half sampling time. Figure 3-27 shows the
Matlab simulation results, in which the amplitudes for the exact transfer function and the
first-order PadApproximation are 98.8% and 99.6% respectively. These results match the
desirable outputs in the frequency analysis.

85

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

Exact
First-Order Pade

pu

0.5

-0.5

-1
0.95

0.955

0.96

0.965

0.97
0.975
0.98
Time (s) (seconds)

0.985

0.99

0.995

Figure 3-27 Verification of responses to the sinusoid input for ZOH by Matlab
ii. Verification by PSCAD
PSCAD simulation studies are presented in Figure 3-28. The magnitude of the signal
passing through sampling block built in PSCAD is 1pu, which is ideal. The peak amplitude
of the output from the exact ZOH is 99.3% of the input signal and the output from the first
order PadApproximation is 98.8% of the input signal. The output amplitudes also match

(pu)

well with the attenuation value observed in the frequency response analysis.

Time (seconds)

Figure 3-28 Responses to the sinusoid input for ZOH by PSCAD

f) Comparison of the Results in Matlab and PSCAD

The combined simulation results from Matlab and PSCAD are given in Figure 3-29a,
which are acceptable. However, the Matlab results are found to be more accurate since
they match the predicted results well with the results obtained from the frequency response
analysis. The exact ZOH simulated results in PSCAD have a 0.3% error in output

86

3.2. CHARACTERIZING THE COMPONENTS OF VSC SMALL SIGNAL MODEL

amplitude and a period of oscillation (harmonics), which is acceptable. This conclusion

was observed in Figure 3-29 (b) that is enlarged from a portion of Figure 3-29 (a).
1

Matlab Pade
Matlab Exact
PSCAD Sin Input
PSCAD Pade
PSCAD Exact
PSCAD Sample

0.8
0.6
0.4

pu

0.2
0
-0.2
-0.4
-0.6
-0.8
-1
0.95

0.955

0.96

0.965

0.97
0.975
0.98
Time (s) (seconds)

0.985

0.99

0.995

(a) Combined simulation results

-0.75

-0.8

pu

-0.85

-0.9

-0.95

-1

0.973

0.974

0.975
0.976
Time (s) (seconds)

0.977

Matlab Pade
Matlab Exact
PSCAD Sin Input
PSCAD Pade
PSCAD Exact
PSCAD Sample
0.978

(b) Enlarged section of Figure 3-29 (a) for peak point

Figure 3-29 Comparison of the simulation results

C. Inner DB Current Controller

The algorithm of the inner current DB controller, also known as the AC current tracker,
has also discussed in depth in [52], and has been applied successfully in [53, 126]. The
control block diagram of the DB current control method in the z domain was given in
chapter 2.

87

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

a) Transfer function of Inner Current Controller First Order PadApproximation

Substituting the one sample delay first order PadApproximation into equation (2-40),
the transfer function of inner DB current control block based on the first order Pad
Approximation can be represented as,

Gdb ( s) N1 / (1 N 2 (1 sTs / 2) / (1 sTs / 2))

( N1 N1 Ts s / 2) / [(1 N 2 ) Ts (1 N 2 ) s / 2]

(3-68)

( a b s ) / (c d s )
a= N1 ; b= N1

where,

T
Ts
; c= (1 N 2 ) ; d= s (1 N 2 )
2
2

b) Frequency Response of Inner Current Controller and its Approximation

The z domain controller and its first order Padapproximation are the main focus of
this sub-section. However, for the comparison purpose, the first-order lag is also included

Magnitude(dB)

here.
80
60

Exact
First-Order Pade
First Order Lag

40
20
0.1/Ts

1/Ts

10/Ts

1/Ts
Time (s)

10/Ts

Phase(deg)

200

-200
0.1/Ts

Figure 3-30 Frequency response of inner current controller and its approximations
As it can be observed in Figure 3-30, the first-order lag approximation performs poorly
compared to the first-order Pad Approximation. Therefore, the first-order Pad
Approximation can still be considered as the most appropriate method. The attenuation
and phase shift for both the inner current DB controller and its first-order Pad
Approximation are found to be identical at 50 Hz.

88

3.2. CHARACTERIZING THE COMPONENTS OF VSC SMALL SIGNAL MODEL

c) Step Response of Inner DB Current Control and Its Approximation

Figure 3-31 shows the step responses of the inner DB current controller and its
approximation. It can be seen that the step response of Z-domain DB current controller is
an oscillatory pulse train, this can be explained as follows.
Since the transfer-function of the inner DB current controller is equation (2-48), the Z
transform of the unit step response is

Gdb ( z )

N1
N
z
z
A
B

1z
N1 z (

)
1
1 N 2 z z 1 N 2 ( z 1/ N 2 )( z 1)
z N2 z 1
A

where,

z
( z 1)

Gdb ( z ) N1 A

Hence,

B
z N 2

z
z N2

(3-69)
(3-70)

z 1

1
1
N1B
1
1 N2 z
1 z 1

(3-71)

Therefore, the unit step response of the z domain controller can be easily obtained in
time domain by using inverse of equation (3-71).
In this study, by substituting the parameter listed in Appendix A, the coefficient N1 and
N2 of the DB current controller are able to be obtained which are -24.904 and 0.9997
respectively. A and B are both 1/2. Therefore, the step response of the z domain DB
current controller is,

y (t ) 12.4695 12.4695(0.9997)

t
Ts

(3-72)
Ts 0.001s

0
-20

pu

-40
-60
-80
-100
-120
0

0.02

0.04

0.06

Discrete Inner Current Control

First-Order Pade
0.08
0.1
0.12
0.14
0.16
0.18
0.2
Time (s) (seconds)

Figure 3-31 Step response of inner DB current controller and its approximation

89

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

Hence, the step response of the inner current controller should be a pulse train (high
frequencies oscillation) with an exponentially reduced magnitude to which the time
constant is long, as shown in Figure 3-31.
d) Response to the Sinusoid Input for Inner Current Controller together with ZOH
The aim of this section is to compare the actual discrete system with the continuous
Pad Approximation model and also to examine the feasibility of replacing the discrete
inner current DB controller with the continuous approximation to be utilized in power
system simulation. Figure 3-32 shows the continuous and discrete implementation schemes
of the inner current control strategy.

Input
GZOH _ Pade ( s )

GDB ( s )

(a) Continuous approximation of inner DB current controller

Input

GZOH ( z )

Sampler

GDB ( z )
N1

N2

est

(b) Discrete implementation block of inner current DB current controller in PSCAD

Figure 3-32 Implementation of the inner DB current controller scheme in continuous
domain and in the z domain
The simulation results are given in Figure 3-33. The grey line in the figure is the
sinusoidal input to the system. The red line is the result of the discrete controller
implemented in PSCAD, and the green line shows the discrete controller output simulated
in Matlab. The blue line is the result of the equivalent continuous Pad Approximation
using Matlab. Note that the discrete controller has the same phase shift both in PSCAD and
MATLAB simulations and very small error in magnitude that can be ignored. The result

90

3.2. CHARACTERIZING THE COMPONENTS OF VSC SMALL SIGNAL MODEL

also shows that the continuous PadApproximation of the discrete DB current-controller is

an accurate equivalent for the type of power-system analysis considered in this research.
15

10

pu

-5

-10

-15

-20
0.95

0.955

0.96

0.965

0.97

0.975

0.98

0.985

PSCAD Discrete
Matlab Pade Approximation
Matlab Discrete
0.99
0.995
1

Time (s) (seconds)

Figure 3-33: Comparison of the responses of (i) the discrete DB current controller and (ii)
its first order PadApproximation to a sinusoidal input signal.
3.2.6.2 Transformation of DB Controller from Reference Frame to dq Reference
Frame
In the previous discussion, the discrete controller has been successfully transformed to
continuous domain. However, the DB current controller considered was defined in abc
natural reference frame as shown in Figure 3-34, which contains time-varying terms or
sinusoidal components which are inappropriate for developing the small signal model.
Therefore, it is necessary to transform the controller to dq reference frame. This will be
done in four steps as will be explained below.

VcC_ abc _ ref 2

C
c _ abc _ ref

C
c _ abc

VcC_ abc _ ref 2 ( s )

C
c _ abc _ ref

( s)

I cC_ abc ( s )
S

+_

a bs
c ds

VcC_ abc _ ref 1 ( s ) VcC_ abc _ ref ( s )

VcC_ abc _ ref _ d ( s )

e sTs

GF

Figure 3-34 The abc natural reference frame DB current control block

91

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

Step 1: The model is simplified by reducing all the sampling blocks (circled area in Figure
3-34), moving to behind the computation block, and leaving the rest of the model remained
unchanged, as illustrated Figure 3-35.

VcC_ abc _ ref 2

I cC_ abc _ ref

a bs
c ds

+_
I cC_ abc

C
VcC_ abc _ ref 1 ( s) Vc _ abc _ ref

VcC_ abc _ ref ( s)

VcC_ abc _ ref _ d ( s )

e sTs

GF

Figure 3-35 The simplified abc natural reference frame DB current control block
Step 2: The abc natural reference frame controller is transformed to reference frame by
using Clark Transformation (see equation (2-2)) as shown in Figure 3-36.

VcC_ _ ref 2
I cC_ _ ref

a bs
c ds

+_
I cC_

VcC_ _ ref 1 ( s ) VcC_ _ ref

VcC_ _ ref ( s)

VcC_ _ ref _ d ( s )

e sTs

GF

Figure 3-36 The simplified reference frame DB current control block

Step 3: The reference frame DB control block is modified by using inputs/outputs in dq
reference frame, which is done by adding with modulator/demodulator as shown in Figure
3-37.
Modulator

Demodulator

VcC_ _ ref 2

C
c _ dq _ ref 2

I cC_ dq _ ref

I cC_ dq

e j0t

I cC_ _ ref

e j 0 t

+_

C
c _

a bs
c ds

C
c _ _ ref

VcC_ _ ref ( s)

VcC_ _ ref 1 ( s) V

C
c _ _ ref _ d

e sTs

(s)

VcC_ dq _ ref _ d ( s)

GF

Figure 3-37 Simplified dq reference frame DB current control block including the timevarying terms (modulator/de-modulator)

92

3.2. CHARACTERIZING THE COMPONENTS OF VSC SMALL SIGNAL MODEL

Step 4: The modulator sectors are eliminated and transformed to dq reference frame which
do not contain time varying terms (modulator/de-modulator).
Note that after these transformations, each variable in reference frame going
through the modulator system is transformed to dq reference frame which is a linear time
invariant (LTI) system. The next step is to shift the rotary factor ejt to the right hand
through three blocks: i) DB block; ii) sampling block and iii) delay block step by step until
reaching to e-jt. During this process both of the terms are nullified without containing the
time-varying term since ejte-jt =1. Therefore, the controller then can be transformed to
dq synchronous reference frame.
A. Methodology Development for Cancelation of Modulator/De-modulator
The primary question is how to eliminate the modulator/de-modulator to achieve the
equivalent linearized small-signal model system. Figure 3-38 illustrates a generic
modulator/de-modulator system. The objective here is to derive the transfer function or
state-space equation for G(s) = Ydq(s)/Udq(s) of the system. It should be noted that the
derivation process with respect to the elimination of the system modulator/de-modulator is
one of the original contributions of this thesis. It should be emphasized that two
approaches are developed from different point of views which result in the same
conclusion. The first methodology accommodates the transfer matrix state-space equation,
which avoids the presence of the imaginary components during the derivation process but
is comparably more complicated, which will be discussed in the rest of this section. The
second approach is derived from the rotary viewpoint which is also based on the state
space equations. Due to the space consideration in this thesis, this approach is not included
in the main text but included in Appendix B.

Ud

Uq

e j0t
cos
sin

sin
cos

Modulator

X A X B U
Y C X D U

e j0t

cos
sin

sin
cos

Demodulator

Yd

Yq

Figure 3-38 Modulator/de-modulator system

93

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

It should be noted that the transmission path between the modulator and de-modulator
is represented by a linear set of time dependent state equations. Therefore, to facilitate the
development of the linearized system of equation, they can be defined as

Zdq Adq Z dq Bdqudq

(3-73)

ydq Cdq Zdq Ddqudq

(3-74)

where, Zdq [Zd1 , Zd 2 , Zdn , Zq1 , Zq 2 , Zqn ]T , udq [ud ,uq ] ydq [ yd , yq ]
T

Adq (2n 2n); Bdq (2n 2); Cdq (2 2n); Ddq (2 2)

Then, we can define the following state-variable transformation:

X i jX i (Zdi jZqi ) e j (t )
X i jX i ( Z di jZ qi )(cos (t ) j sin (t ))
X i jX i ( Z di cos Z qi sin ) j ( Z di sin Z qi cos )

(3-75)

If we rewrite above equation in matrix form,

X i cos sin Z di
X

i sin cos Z qi

In

the

below

paragraphs

the

complete

set

(3-76)
of

the

state

variables,

Xi ,Xi Zdi ,Zqi (i=1,....n) , developed in matrix form are given.

If we apply equation (3-76) to each Xi ,Xi Zdi ,Zqi pair, the following equation will
be obtained
X 1
0
0 sin
0
X cos
2 0
cos 0
0
sin

0
0
0

Xn 0
0 cos
0

sin
0
0
cos
0
X 1
sin 0
0
cos
X 0
2
0
0
0

0 sin
0
X 0
n

Z
d1
Zd 2
0

0
Z
0 sin dn

0 Z
q1

0
Zq2
0

0
cos Z
qn
0

(3-77)

Further, simplification of the above equation yields to

X cos In sin In Z d
X

sin In cos In Z q

where In=diag(1,1,,1) is identity matrix.

* Drop (t) and take it as implied in the sequel
94

(3-78)

3.2. CHARACTERIZING THE COMPONENTS OF VSC SMALL SIGNAL MODEL

If we take the partial derivative of this equation using

Zd
cos In sin In d X d
d Zd
Rn
, Rn Rn

dt Z q
sin In cos In dt X dt Z q
d cos
d d sin
d
sin
;
cos
dt
dt
dt
dt

(3-79)

We can obtain
d X sin In cos In d

dt X cos In sin In dt

Z d cos In sin In d Z d
Z

q sin In cos In dt Z q

(3-80)

In Figure 3-38, the modulator inputs can be rewritten as,

U jU (U d jU q )(cos (t ) j sin (t ))

u cos
u
sin

sin ud

cos uq

(a)

ud cos
u
q sin

sin u

cos u

(b)

(3-81)

X A 0 X B 0 u



X 0 A X 0 B u

(3-82)

Hence, substituting [u ,u ] in equation (3-82) from equation (3-81)a, we can obtain

X A 0 X B 0 cos

X 0 A X 0 B sin

sin ud

cos uq

(3-83)

Substituting [X ,X ]T in equation (3-80) into equation (3-83), substituting [X, X]T in

equation (3-78) into equation (3-83), substituting [X, X]T in equation (3-78) into equation
(3-84), substituting [u, u]T in equation (3-81)a into equation (3-85), and substituting [y,
y]T in equation (3-85), equation (3-86) can be obtained referring to Appendix C.
y C 0 X D 0 u
y


0 C X 0 D u
y cos
y
sin

sin yd

cos yq

Z dq Adq Z dq Bdqudq
ydq Cdq Z dq Ddqudq

(3-84)
(3-85)

(3-86)

95

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

where,

In0
B 0
C 0
D 0
, Bdq
, Cdq
, Ddq

A
0 B
0 C
0 D

A
Adq
In0

The equivalent transfer function of the modulator/de-modulator system in dq reference

frame can then be easily derived from the state-space equation (3-86) , which is a group of
TFs presented in equation (3-87). Figure 3-39 shows the block diagram for the transformed
dq reference frame modulator/de-modulator system. Although some details are given here,
the detailed derivation process for the equivalent dq frame state space equation or transfer
function can be found in Appendix C.

Gdd ( s) C ( sI A)[( sI A) 2 02 ]1 B D
Gdq ( s) C0 [( sI A) 2 02 ]1 B

(3-87)

Gqd ( s) Gdq ( s)
Gqq ( s) Gdd ( s)

ud
uq

cos 0t sin 0t
sin t cos t
0
0

ud ( s )

G ( s )
u

G ( s )

yd

cos 0t sin 0t
sin t cos t y
q
0
0

yd ( s )

Gdd ( s )

Gdq ( s )
uq ( s )

Gqd ( s )

yq ( s )

Gqq ( s )

Figure 3-39 Transfer function representation of the modulator/de-modulator system

transform from
reference frame to dq reference frame
Note that the equivalent dq reference frame transfer function can be given in a much
simpler format if it is written in a complex form as shown by

G(s) C0 [(s j) A]1 B0 D0

This equation indicates that the transfer function transforming from

(3-88)
reference frame

to dq reference frame can be done simply by adding a j term to the Laplace factor.
Although the above methodology is given for state-apace equation transforming from
reference frame to dq reference frame, similar relationship is also valid for abc natural
reference frame to dq reference frame transformation since the abc natural reference frame

96

3.2. CHARACTERIZING THE COMPONENTS OF VSC SMALL SIGNAL MODEL

and

reference frame is simply related by the Clark Transformation matrix which is a

constant matrix (see equation(2-2)). However, the dq reference frame rotates at a frequency
corresponding to the and abc reference frame variables. It can be noted that this
approach is valid for any linear component with transfer function derived in abc stationary
natural frame such as a grid filter, or linear controllers implemented in abc reference frame
or even in a physical grid model.
B. Small Signal Model for DB Current Controller, Sampling and Delay Blocks
As is discussed previously there is a common characteristic for the equivalent
continuous domain models references to the DB controller, sampling block and delay
block, since all of them can be written either in the lead-lag format or in the lead-lag-gain
format which are summarized in Table 3-5. Note that the third column in Table 3-5 is a
case example obtained by substituting the parameter listed in Appendix G and which will
be used for simulation verifications later in this chapter.

Table 3-5 Summary of transfer functions for DB current controller, sampling block and
delay block
Lead-Lag

a bs
c ds
DB
Block
Sampling
Block
Delay
Block

Lead-Lag-Gain

Case Study

1 Ta s
1 Tb s

T
1 s s
N1
2
(
)
1 N 2 1 Ts 1 N 2 s
2 1 N2

-0.01247s-24.94
0.000001566s 1.997

1
1 sTs / 2

1
1 sTs / 2

1
1 0.0005s

1 sTs / 2
1 sTs / 2

1 sTs / 2
1 sTs / 2

-0.0005s +1
0.0005s +1

N1 N1
(1 N 2 )

Ts
s
2

Ts
(1 N 2 ) s
2

Equivalent transfer function in dq reference frame

As is explained above, the equivalent dq reference frame transfer function was
simplified by replacing s term with s+j term. For a general lead-lag block, the
consequential shift for the reference frame is presented in Table 3-6. Substituting the
parameter listed in the case study column of Table 3-5, the equivalent dq frame transfer

97

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

functions for the target blocks can be obtained as shown in Figure 3-40. It should be noted
that so far, the transfer functions of the DB controllers are successfully transformed to dq
reference frame.
VcC_ d _ ref 2

VcC_ d _ ref 1 ( s )

I cC_ d _ ref

+_
C
c_d

GDB1

VcC_ d _ ref

VcC_ d _ ref ( s )

GSample1 ( s )

VcC_ d _ ref ' ( s )

GDelay1 ( s )

GF

GDB 2
GDB 2

I cC_ q _ ref
+_

I cC_ q

GDB1

C
c _ q _ ref 1

GF

GSample 2 ( s )

GDelay 2 ( s )

GSample 2 ( s )

GDelay 2 ( s )

GSample1 ( s )

( s ) V

C
c _ q _ ref

GDelay1 ( s )

VcC_ q _ ref ' ( s )

VcC_ q _ ref ( s )

VcC_ q _ ref 2

Figure 3-40 Small signal model for DB current controller in equivalent dq reference frame

Table 3-6 Equivalent dq reference frame transfer function for the lead-lag block
reference frame

dq reference frame transfer function

transfer function

Y ( s)
a b( s j )

G1 ( s) j G2 ( s)
X ( s) dq c d ( s j )

Lead lag
block

Y ( s)
a bs

X ( s) c ds

Main loop

Crosscoupling

G1 ( s)

bds 2 (ad bc) s (ac bd02 )

;
d 2 s 2 2dcs c 2 d 202

G2 ( s)

ad0 cb0
;
d s 2dcs c 2 d 202
2 2

C. Simulation Verification
To test the accuracy of the DB controller in terms of equivalent transfer function and
state space equations in the dq reference frame, two step tests are performed: i) -0.02pu
step change on Uq and ii) -0.02pu step change on Ud imposed on the three target blocks in
series (which included DB control block, sampling block and computation delay block).

98

3.2. CHARACTERIZING THE COMPONENTS OF VSC SMALL SIGNAL MODEL

(a)

(b)

Figure 3-41 Compare Yd/Yq outputs from (i) the modulator/de-modulator system; (ii) the
small-signal equivalent transfer function and iii) the small-signal equivalent state space.
(a)UdSTEP = 0, UqSTEP = -0.02 and (b) UdSTEP = -0.02, UqSTEP = 0;

(a)

(b)

Figure 3-42 Difference between Yd/Yq outputs from (i) the modulator/de-modulator
system versus the small-signal equivalent transfer function; (ii) the modulator/demodulator system versus the small-signal equivalent state space. (a) UdSTEP = 0, UqSTEP = 0.02 and (b) UdSTEP = -0.02, UqSTEP = 0;
The results presented in Figure 3-41 indicate that there is a good agreement on the test
results, since the difference between Yd/Yq outputs from the modulator/de-modulator
system compared with the small-signal equivalent transfer function is small. In addition,
the calculated errors in the modulator/de-modulator system compared to the small-signal

99

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

equivalent state-space model further confirm the conclusion drawn above, as shown in
Figure 3-42 .
Equivalent State-Space Equation in dq Frame
To obtain the equivalent state-space equations in dq reference frame, the state-space
equation for the reference frame transfer function should be defined first
which is written in the lead-lag-gain format (see Appendix D),

sz Az Bx

y Cz Dx
Where A

(3-89)

T
1
k k Ta
;B
; C 1; D k a
Tb
Tb Tb Tb
Tb

Table 3-7 State space equations for DB controller in equivalent dq reference frame
Large signal state space model
dwd _ db

Dead-Beat

Adb wd _ db wq _ db Bdb I cdc _ e

dt
Vcdc _ ref 1 Cdb wd _ db Ddb I cdc _ e

Block

dwq _ db
dt

wd _ db Adb wq _ db Bdb I cqc _ e

Vcqc _ ref 1 Cdb wq _ db Ddb I cqc _ e

dZ d _ sp
dt

Asp Z d _ sp Z q _ sp BspVcdc _ ref

Sampling
dZ q _ sp
dt

Z d _ sp Asp Z q _ sp BspVcqc _ ref

Vcqc _ ref (k ) Csp Z q _ sp DspVcqc _ ref

dM d _ dl
dt

d wd _ db
dt

Adb wd _ db wq _ db Bdb I cdc _ e

Vcdc _ ref 1 Cdb wd _ db Ddb I cdc _ e

d wq _ db
dt

wd _ db Adb wq _ db Bdb I cqc _ e

Vcqc _ ref 1 Cdb wq _ db Ddb I cqc _ e

Vcdc _ ref (k ) Csp Z d _ sp DspVcdc _ ref

block

Small signal state space model

Adl M d _ dl M q _ dl BdlVcdc _ ref (k )

Vcdc _ ref (k )' Cdl M d _ dl DdlVcdc _ ref (k )

d Z d _ sp

Asp Z d _ sp Z q _ sp Bsp Vcdc _ ref

dt
Vcdc _ ref (k ) Csp Z d _ sp Dsp Vcdc _ ref

d Z q _ sp
dt

Z d _ sp Asp Z q _ sp Bsp Vcqc _ ref

Vcqc _ ref (k ) Csp Z q _ sp Dsp Vcqc _ ref

d M d _ dl
dt

Adl M d _ dl M q _ dl Bdl Vcdc _ ref (k )

Vcdc _ ref (k )' Cdl M d _ dl Ddl Vcdc _ ref (k )

Delay block
dM q _ dl
dt
c
cq _ ref

100

M d _ dl Adl M q _ dl BdlVcqc _ ref (k )

(k ) Cdl M q _ dl D V
'

c
dl cq _ ref

(k )

d M q _ dl
dt

M d _ dl Adl M q _ dl Bdl Vcqc _ ref (k )

Vcqc _ ref (k )' Cdl M q _ dl Ddl Vcqc _ ref (k )

3.2. CHARACTERIZING THE COMPONENTS OF VSC SMALL SIGNAL MODEL

By adding with the cross-coupling term j , we can obtain the state space equation in
dq reference frame for lead-lag block as shown in equation (3-90). The state-space
equations for DB control block, sampling block and delay block in equivalent dq reference
frame are obtained reference to Table 3-5, and equations (3-89) and (3-90), which are
summarized in Table 3-7 (in which the second column stands for the large signal models,
while the third column denotes the small signal models).

d zd
ALL zd zq BLL xd

dt

yd CLL zd DLL xd
d zq
wd ALL zq BLL xq

dt

yq CLL zq DLL xq

(a)
(3-90)

(b)

where, the subscript LL denotes lead-lag.

D. Limitation of the Equivalent State Space Equation
It is important to understand the impact of the varying frequency to the modulator/demodulator elimination approach. Therefore, two types of tests have been performed: i)
small step on frequency (1%), (where = 0+2Ku(t), K=0.01, t=0.01s) and ii) sinusoidal
disturbance with small magnitude (where =0+Ksin(t), K=0.01, =10Hz, t=0.01s);
Before conducting these tests, a step change (ud=0, uq=-0.02pu, t=0.001s) is initially
applied to the system input. By examining the output differences between the frame TF
and dq frame TF, the limitation of the dq equivalence can be easily observed.

(a)

(b)

101

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

(c)
(d)
Figure 3-43 Compare Yd/Yq outputs from the (i) modulator/de-modulator system; (ii) the
small-signal equivalent transfer function and iii) the small-signal equivalent state space. (a)
Yd computes from UdSTEP = 0, UqSTEP = -0.02, (=0+2Ku(t), K=0.01, t=0.01s); (b) Yq
computes from UdSTEP = 0, UqSTEP = -0.02, (=0+2Ku(t), K=0.01, t=0.01s); c)Yd
computes from UdSTEP = 0, UqSTEP = -0.02, (=0+Ksin(t), K=0.01, =10Hz, t=0.01s);
(d) Yq computes from UdSTEP = 0, UqSTEP = -0.02, (=0+Ksin(t), K=0.01, =10Hz,
t=0.01s);
From the tests results shown in Figure 3-43a and b, it can be seen that the value of the
output is smaller than the exact one, which results in an under compensation on the
generated voltage reference. The problem can be solved by using the PLL output frequency
as a feedback to the TF coefficient calculation, or by increasing the controllers gain to
enhance the compensation. Therefore, it can be concluded that the increase or decrease in
frequency can cause the outputs to be under or over compensated in the equivalent dq
reference frame. Moreover, the results given in Figure 3-43c,d for sinusoidal input
frequency disturbance test reveal that the magnitude of the consequential oscillation
increases with the simulation time. Finally, it can be concluded that it is better to design the
linear controller in dq reference frame but implemented in abc natural reference frame as it
can offer better capability on frequency sensitiveness resistance.

3.2.7 Reference Voltage Generation of the Converter

Based on the voltage prediction model developed in section 3.2.4 and the DB predicted
voltage in section 3.2.6, the input reference voltage of the converter can be easily obtained
as,

102

3.2. CHARACTERIZING THE COMPONENTS OF VSC SMALL SIGNAL MODEL

vcdc _ ref vcdc _ ref 1 vcdc _ ref 2

vcqc _ ref vcqc _ ref 1 vcqc _ ref 2

(3-91)

Then, linearized version of the above equation can be given by,

vcdc _ ref vcdc _ ref 1 vcdc _ ref 2
vcqc _ ref vcqc _ ref 1 vcqc _ ref 2

(3-92)

Since the small signal model is accurate, no simulation verification is required.

3.2.8 Voltage Source Converter

It is necessary to characterize the two-level three phase VSC which is also a non-linear
component of the system. Therefore, the mathematical model of the VSC is developed
using the average-value modeling (AVM) approach and the small signal model for VSC is
in dq0 reference frame. Note that, the AVM approach is appropriate enough when the
switching frequency of the PWM is much greater than the grid frequency. The simplified
linearized model is verified by comparing the outputs of the AVM model and the nonlinear model built in PSCAD/EMTDC.
3.2.8.1 Small Signal Model for VSC
Using the approach adopted in [129, 130], it is assumed that there is no energy storage
component in VSC, moreover, since the reaction speed is extremely fast compared to the
frequency range of interest, the relationship between the ac voltages/currents and the dc
side variables can be considered purely algebraic.
Therefore in a balanced system, the averaged model of the three phase two-level VSC
can be given by,
Vc _ abc (t )

Vdc (t )
mabc (t )
2

(3-93)

where, the Vc_abc (t) is the vector of VSC terminal voltages, Vdc (t) represents the dc
voltage, and the mabc (t) is the modulation signal, which can be given by,
ma (t ) M cos( )
mb (t ) M cos( 2 / 3 )

(3-94)

mc (t ) M cos( 2 / 3 )

Note that in the above equation M controls the magnitude of the AC output voltage,
is the angle of the bus voltage at PCC point that is provided by a PLL and the derivative of

103

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

(d/dt) is the angular frequency of the AC supply. In addition, is the relative angle
of VSC terminal voltage Vc_abc which is controlled by the control system of the converter.
Hence, the dq reference frame averaged model of three-phase two-level VSC can be
given by [2, 72],
Vdc
V
M q dc
2
2
Vdc
Vdc
Vcd M sin
Md
2
2
Vcq M cos

where, the magnitude M is equal to Vcd2 +Vcq2 /(

(3-95)

Vdc
) and the angle is equal to
2

tan -1 (Vcd /Vcq ) which is known as the modulation ratio or VSCs terminal voltage angle.
If we substitute M and into the above equation, we will obtain,
c
c
Vcd
Vcd
_ ref Vdc / Vdc 0
c
c
Vcq
Vcq
_ ref Vdc / Vdc 0

(3-96)

which can be written in grid RI reference frame as

g
g
VcR
VcR
_ ref Vdc / Vdc 0

VcIg VcIg _ ref Vdc / Vdc 0

(3-97)

If we linearize the above equations, we can obtain the small signal model that provides
the link between the AC and the DC voltages of VSC.
g
g
g
'
VcR
VcR
_ ref (k ) VcR _ ref 0 Vdc / Vdc 0

VcIg VcIg _ ref (k )' VcIg _ ref 0 Vdc / Vdc 0

(3-98)

Since the efficiency of the converters is high in such power applications, the
conduction losses can often be ignored. Therefore, the DC-side and AC-side terminal
quantities of the two-level VSC can be interconnected by the power balance equation as
Pdc(t)=Pac(t) , which can be written using terminal quantities as [72, 77].
3
U DC (t )iDC (t ) [Vcq (t )icq (t ) Vcd (t )icd (t )]
2

(3-99)

If we substitute equation (3-95) into the above equation, we can obtain the DC link
current as,
3
iDC (t ) [ M q (t )iq (t ) M d (t )id (t )]
4

104

(3-100)

3.2. CHARACTERIZING THE COMPONENTS OF VSC SMALL SIGNAL MODEL

Similarly, linearized version of the above equation can provide the small signal model
as [56],
3
iDC (icq M q M q icq icd M d M d icd )
4

(3-101)

The advantage of this model is that it does not include the complex switching states
and ignore the influence of the high frequency harmonics. Therefore, such model can run
faster than the detailed model simply acting as a voltage filtered model. However, ignoring
the high frequency harmonics will also ignore the impact of high harmonics on the grid
and the control. It should be noted that such model is only valid for frequencies up to about
one-third of the VSC switching frequency and the switched model operates in the linear
modulation region without over modulation, which is Vcd2 +Vcq2 Vdc /2 [131].
It should be reported that the above assumption will not represent the real operation.
However, a more accurate model is a very challenging task, since it depends on specific
operating point as well as the loading conditions. As reported in paper [130], a parametric
AVM model can be used for an accurate model. However, the establishment of the
parametric AVM model requires simulations of a wide range of loading conditions based
on a detailed model (closed loop simulation involving the control functions) to restore the
coefficients ((), (),

) as functions. Note that () represents the algebraic function

coefficient between the DC voltage and AC terminal voltage of the loading conditions; ()
represents the algebraic function coefficient between the AC terminal current and the DC
current of the loading conditions, and

is the power factor angle. This angle can be

expressed in terms of the dq components of the AC side voltage and the current as,
i
V
() tan 1 ( cd ) tan 1 ( cd )
icq
Vcq

(3-102)

If we replace the original constant value with these new algebraic coefficient functions
as in [130], we can obtain a more accurate AVM model suitable for various loading
conditions and operating points. However, it should be emphasized that these coefficients
which vary with the operating points and load conditions can only be evaluated for a
specific case, which cannot be formulated to represent all the operating conditions.

105

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

3.2.8.2 Simulation Verification

Since the small signal model is an ideal AVM model, it is expected to observe
inaccuracies using the detailed model simulation. This is also due to the fact that the rated
operating switching frequency in this study is 2000Hz which is not an ideal case. As it is
mentioned above, the coefficients will vary with the operating points and loading
conditions which cannot defined analytically. Note that the constant coefficient 1 used in
this study for

and

is an approximation.

Figure 3-44 illustrates the test circuit for the detailed VSC model. In this method a
small step change
input reference

(from

to

) is applied to the q-axis voltage

, while for d-axis voltage input reference

is set to be

Then the reference voltage is transformed to abc reference frame as an input to the
converter. To ensure the linearity of the VSC, the on-state resistance of the transistor and
diode in the converter has to be small enough, which is set as 0.001 in the case study. The
converter terminal output voltage

is in the abc natural reference frame needs to be

transformed to dq reference frame before connecting to the first order low-pass filter to
eliminate the high switching frequency harmonics. In the first stage, dq reference frame
converter terminal voltage

are obtained.

e jt
Vcq0_ ref

Vcq 0_ ref Vcq1 Vcd_ref

T0

Vcq_ref

d
q

Vca

Vca_ref

a
T b Vcb_ref
c Vcc_ref

Vcb

VSC

Vcc

a b c

T
d

Vcd

F
I
L
T
E
R

jt

Gf _ dq (s)
SIGNAL
FILTER

1
1 0.01s

Vcq

Figure 3-44 Test circuit for the detailed VSC model

The simulation results are given in Table 3-8. Note that the coefficients vary slightly
from case to case. This suggests that there are some cross-coupling effects between d-q
axes which are different from the AVM suggested coefficient 0, given earlier for Vcq/Vcdref
and Vcd/Vcqref. However, the discrepancies are found to be small once the loops of the

106

3.2. CHARACTERIZING THE COMPONENTS OF VSC SMALL SIGNAL MODEL

whole system are closed. But it should be noted that the AVM model causes an inaccurate
loop gain which should be considered in the controller design.

Table 3-8 Simulation results for VSC model verification

Test

Test for no Load Current Icd=0 Icq=0(Vcqref Step) G_qq & G_dq

_ / _
Vcqref(Vcqref=-8.11)
Vcq
Vcd

-49.6566

-49.7538

-8.1407

0.864

-0.02

-50.5129

-50.501

-8.1169

0.8173

0.0178

-50.67

-50.6294

-8.1197

--------

--------

-50.8271

-50.7472

-8.1183

0.7498

-0.0089

-51.6834

-51.7274

-8.1319

1.0835

0.012

Test for no Load Current Icd=0 Icq=0 (Vcdref Step) G_dd & G_qd

Test

Vcdref (Vcqref=-50.67)

Vcd

Vcq

-7.9478

-7.9607

-50.6363

0.9803

-0.0425

-8.0849

-8.0799

-50.6218

1.5857

0.302

-8.11

-8.1197

-50.6294

--------

--------

-8.1351

-8.1465

-50.6271

1.0678

-0.0916

-8.2722

-8.2757

-50.6166

0.9618

-0.079

3.2.9 Grid Model

Figure 3-45 is given to present the topology of the AC grid, in which U sg and i gg
represent the voltage and the current of the grid source respectively, while the R g and Lg
denote the resistance and the inductance of the grid. Note that, the capacitance Cf and the
resistance Rf in series in the figure represent the grid filter, and Rc and Lc are the resistance
and the inductance of the converter reactor. The converter terminal voltage and the current
are given by VCg and i gc .
igg

U sg

Rg

Lg
u gf

Lc
Cf

Rc

icg
VCg

Rf

Figure 3-45 The equivalent circuit diagram of grid

107

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

Using the circuit analysis (KVL, KCL) for the gird circuit given above, a set of
equations can be obtained as given in the second column of Table 3-9. As concluded in
section 3.2.6.2, the method used to obtain the equivalent linearized small-signal model of
modulator/de-modulator system is also suitable for the grid model. This can be done by
replacing the A matrix with a new Adq matrix which involves the cross coupling
components j, which is listed in the third column of Table 3-9. Figure 3-46 illustrates the
small signal model of the grid.
U sRg
VcRg

-+
+

1
Rv sLv

g
icR

-+
+

Lv

1
Rv sLv

1
Rg sLg

-+
+

Cf

Lv

VcIg

1
sC f

g
fR

Lg
Lg

Cf

icIg

1
sC f

g
igR

u gfI

1
Rg sLg

g
igI

U sIg

Figure 3-46 Small signal model of the grid

Converter Reactor

Table 3-9 Summary of the large signal model of the AC grid in abc reference frame and
small signal model in dq reference frame
abc reference frame large signal model
RI reference frame small signal model

dic _ abc
dt

Rc
1
ic _ abc (u f _ abc Vc _ abc )
Lc
Lc

Grid Model

Grid Filter

u g f _ abc Fftabc (i g g _ abc i g c _ abc ) R f

108

dFftabc
dt

dig _ abc
dt

1
(ig _ abc ic _ abc )
Cf

Rg
1
(U s _ abc u f _ abc ) ig _ abc
Lg
Lg

d icg_ R
R
1
c icg_ R 0 icg_ I (u g f _ R V g c _ R )

dt
L
L

c
c

g
d

i
Rc g
1

c_I
g
g
g
dt 0 ic _ R L ic _ I L (u f _ I V c _ I )
c
c

u g f _ R FftR (i g g _ R i g c _ R ) R f

g
g
g

u f _ I FftI (i g _ I i c _ I ) R f
d FftR
(1/ C f )i g g _ R (1/ C f )i g c _ R 0 FftI

dt

d FftI (1/ C )i g (1/ C )i g F

f
g_I
f
c_I
0
ftR
dt
d i g g _ R
Rg g
1

(U sg_ R u g f _ R )
i g _ R 0 i g g _ I

dt
L
L
g
g

g
Rg g
1
d i g _ I

(U sg_ I u g f _ I )
i g _ I 0 i g g _ R
dt
L
L
g
g

3.3. COMPARISON OF THE SMALL SIGNAL LINEAR MODEL AND PSCAD SIMULATION

Figure 3-47 shows the test results of the i) grid side current and ii) filter bus voltage
responses of the VSC model, which are obtained by applying a small step change -0.56kV
G
on the input of the grid model VcR
. Note that there is a good agreement on the results of

the large signal model and the small signal model.

(a)

(b)

Figure 3-47 (a) Grid side current and (b) filter bus voltage responses of i) large signal
model ii) small signal model following a input voltage step change order on VcqC = 0.56kV.
Finally, the characterization and validation tasks with regards to all the individual
component of the VSC small signal model are complete and prepared so far for the whole
system validation.

3.3 Comparison of the Small Signal Linear Model and

PSCAD Simulation
Before conducting the small signal analysis, the Newton Raphson method (selfdeveloped in m-file) is employed to solve the operating (equilibrium) point of the set of
non-linear algebraic equations. The results are compared with the outputs obtained from
the Matlab tool-box (f solve) which is also a functional tool for solving non-linear
equations.
Figure 3-48 shows the comparison between the small-signal model and detailed nonlinear PSCAD simulation following a step change in q axis of the converter current
reference for AC systems with two different SCRs. Figure 3-48a,b are the results for a

109

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

strong system with SCR equals to 20 following a -0.0095pu step change on Icq_ref, while
the results for a weak AC system with SCR equals to 2 following a 0.095pu step change on
Icq_ref are denoted by Figure 3-48c and d. Note that, for Figure 3-48a, the green line
denotes the reference value, the cyan line represents the output from the large signal
model, the purple line represents the and q-axis converter terminal current in the converter
reference frame

outputs from the small signal model; for Figure 3-48b, the blue line

denotes the reference value, the red line stands for the output from the large signal model,
the black line represents the d-axis converter terminal current in the converter reference
frame

outputs from the small signal model; for Figure 3-48c, the d- and q- axes

converter terminal currents in the converter reference frame,

and

, are the outputs

for an AC system with SCR=2; and a zoomed in figure for the initial section of Figure
3-48c is presented in Figure 3-48d.
The results presented in Figure 3-48a and b (where the VSC is simplified as a
controlled voltage source in PSCAD) for a strong AC system indicate that the method has
a limitation as it displays an extra error that is introduced when using the pad
Approximation. Note that the d-axis cost nearly 9 times sampling period (where the
sampling frequency is 4000Hz), while the q-axis settles within about 7 sampling periods.
However, the discrete controller itself only cost 2 sampling time to settle down as deigned.
Despite these limitations, the dynamic simulation results reveal a good agreement on the
system performance. However, in a weak AC system, Figure 3-48 (c) and (d), the dynamic
simulation results reveal a better agreement on the system performance. Both of the d and
q axes cost 4 times sampling periods to closely track the current output from the PSCAD
large signal model. In addition, the DB controller presents a slower current tracking ability
in the weak AC system compared to the strong AC system.

(a)
110

(b)

3.3. COMPARISON OF THE SMALL SIGNAL LINEAR MODEL AND PSCAD SIMULATION

(d)

(c)

Figure 3-48 Comparison of the converter output current ICcd , ICcq using the (i) large signal
model simulating with VSC simplified as a controlled voltage source; (ii) the small-signal
model for AC system with different SCRs.
It can be easily seen from the results shown in Figure 3-49 more harmonics with
switching frequency appeared in the plots (due to switched converter modeling) compared
with that of the results shown in Figure 3-48, However, it can be clearly concluded that the
linear small signal model for the DB controlled VSC demonstrates an accurate
approximation of the detailed model developed in PSCAD within the frequency range of

Converter Current Iccd,Iccq (pu)

interest.

Iccd-PSCAD(S)

0.1

Iccq-PSCAD(S)

0.08

Iccd-Matlab

0.06

Iccq-Matlab

0.04

Iccd-PSCAD(D)

0.02

Iccq-PSCAD(D)

Iccd-PSCAD(D-FIR)

-0.02
-0.04
0

Iccq-PSCAD(D-FIR)
0.02

0.04
Time (s)

0.06

0.08

(a)

111

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

Converter Current Iccd,Iccq (pu)

0.12

Iccd-PSCAD(S)

0.1

Iccq-PSCAD(S)

0.08

Iccd-Matlab

0.06

Iccq-Matlab

0.04

Iccd-PSCAD(D)

0.02

Iccq-PSCAD(D)

0
-0.02
-0.04
0

0.002

0.004
0.006
Time (s)

0.008

0.01

(b) The initial section of figure (a)

Figure 3-49 Comparison of the converter output currents ICcd , ICcq from the (i) large
signal model using the detailed VSC model, and (ii) the small-signal model for the weak
AC system, where i) S denotes the simplified VSC model; ii) D denotes the detailed VSC
model and iii) D-FIR denotes the results filtered by the finite impulse response (FIR) filter.

3.4 Extension of the DB Controlled VSC Together with

DC Link
3.4.1 Model of DC Link
The detailed model of the DC transmission system has already been studied in [53, 63,
71, 72, 74, 75, 77, 132, 133]. The model of the DC circuit is given in Figure 3-50, and
Table 3-10 summarises the relationship between the DC side current, voltage and load
current for both large and small signal models. As stated in section 3.2.8.1, the AC/DC side
can be linked through equation (3-101). Therefore, the equivalent circuit of the AC side
can be modelled as constant current sources Idc_rec and Idc_inv in the DC side modeling. If we
linearize these equations, the small signal model equations for DC link are obtained, which
are given in the second column of Table 3-10.

112

3.4. EXTENSION OF THE DB CONTROLLED VSC TOGETHER WITH DC LINK

I cdc _ rec
Rdc _ rec

I cdc _ inv
Ldc _ rec

Rdc _ inv

Ldc _ inv

I dc _ rec

I dc _ inv

Vdc _ rec
Cdc _ rec

Vcdc

Vdc _ inv

Cdc _ line

Cdc _ inv

Figure 3-50 The model DC transmission link

Table 3-10 Large and small signal for DC link
Large signal model for DC link

dVdc _ rec
dt
dI cdc _ rec
dt

Vdc _ rec
Ldc _ rec

I dc _ rec
Cdc _ rec

Small signal model for DC link

I cdc _ rec

d Vdc _ rec

Cdc _ rec

dt

Rdc _ rec I cdc _ rec

Vcdc

Ldc _ rec
Ldc _ rec

d I cdc _ rec
dt

dt

Vdc _ inv Rdc _ inv I cdc _ inv

Vcdc

Ldc _ inv Ldc _ inv

Ldc _ inv

dVdc _ inv
dt

I cdc _ inv
Cdc _ inv

Vdc _ rec
Ldc _ rec

I dc _ rec
Cdc _ rec

I cdc _ rec
Cdc _ rec

Vcdc Rdc _ rec I cdc _ rec

Ldc _ rec
Ldc _ rec

d Vcdc I cdc _ rec I cdc _ inv

dt
Cdc _ line
Cdc _ line

dVcdc I cdc _ rec I cdc _ inv

dt
Cdc _ line Cdc _ line

dI cdc _ inv

d I cdc _ inv
dt

Vcdc Vdc _ inv Rdc _ inv I cdc _ inv

Ldc _ inv Ldc _ inv

Ldc _ inv

I dc _ inv

d Vdc _ inv

Cdc _ inv

dt

I cdc _ inv
Cdc _ inv

I dc _ inv
Cdc _ inv

Note that, although the established DC link model appears to represent a mono-polar
link, it is also applicable to represent bi-polar links. But we need to apply the following
two rules to specify the line, cable and shunt capacitors parameters when representing a
bipolar link [134],

sum the respective quantities for each line and capacitance to specify the total
line, cable resistance, inductance and capacitance.

specify the shunt capacitance by summing the capacitance between the positive
polar to neutral point and the capacitance between the negative polar to neutral
point.

113

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

Figure 3-51 illustrates the block diagram of the small signal model for the DC link,
which will be verified against the PSCAD large signal model. In the verification study, the
parameters given Table 3-11 are used.

Table 3-11 Parameters for the DC link test

Rectifier Side Constant Current Source
75/130= 0.577 [kA]
Rectifier Side Capacitor

500.0 [uF]

Rectifier Side Transmission Line resistance

7[ohm]

Rectifier Side Transmission Line Inductance

0.5968 [H]

Transmission Line Capacitor

26.0 [uF]

Inverter Side Transmission Line resistance

7[ohm]

Inverter Side Transmission Line Inductance

0.5968 [H]

Inverter Side Capacitor

500.0 [uF]

Inverter Side Constant Current Source

[75-(75/130) 2]/130=0.574 [kA]

I cdc _ rec

I dc _ rec

Vcdc
Vdc _ rec

1
sCdc _ rec

1
sLdc _ rec Rdc _ rec

I cdc _ rec

I cdc _ inv

1
sCdc _ line

Vcdc

Vdc _ inv

I cdc _ inv
1
sLdc _ inv Rdc _ inv

I dc _ inv

1
sCdc _ inv

Figure 3-51 Test circuit for DC link test

Time (seconds)

Figure 3-52 The converter currents behind the DC capacitor from both of the rectifier side
and inverter side (circle: Large signal model; rectangular: Small signal model)

114

U dc _ inv

3.4. EXTENSION OF THE DB CONTROLLED VSC TOGETHER WITH DC LINK

The results shown in Figure 3-52 are obtained by applying a step change 0.577kA to
the input of the DC grid model Idc_rec, which indicate the accuracies of the small signal
model since the large signal model and the small signal model currents are very similar.

3.4.2 Interconnection of Subsystems

So far the models of all system components have been transformed into their state
space form. A single state-space model can be obtained by grouping these sub-models
appropriately and facilitating the application of the linear control design techniques.
Although this has been maintained in [80, 87, 135] before, they were all based upon the
matrix substitution technique. Therefore, this thesis offers an approach which is selfdeveloped and adopted as given below.

U iEXT
U j

X i J XXi

0 J ZXi

Y j X J XXj
j

0 J ZXj

J XZi X i J XUi
J

U i XUEi U iEXT

J ZZi Z i J ZUi
J ZUEi

Yi

U i
J XZj X j J XUj
J XUEj

U
J

j
jEXT
J ZZj Z j J ZUj
ZUEj
U jEXT

Figure 3-53 Demonstration of sub-modular interconnection

Figure 3-53 shows a set of state space equations and algebraic equations to demonstrate
the sub-modular interconnection approach.
The state space equations are defined by
X i J XXi

0 J ZXi

X j J XXj

0 J ZXj

J XZi X i J XUi
J XUEi

i
J
U iEXT
J ZZi Zi J ZUi
ZUEi

J XZj X j J XUj
J XUEj

J
U jEXT
j
J ZZj Z j J ZUj
ZUEj

(3-103)

(3-104)

115

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

Yi JYXi X i JYZi Zi JYUi U i ;

(3-105)

U j Yi ;
Y j JYXj X j JYZj Z j JYUj U j ;

(3-106)

U i Y j ;

The above equations can be written as,

X i J XXi

0 J ZXi

J XZi X i J XUi
J

U i XUEi U iEXT

J ZZi Zi J ZUi
J ZUEi

(3-107)

X
JYZi i JYUi U i Yi
Zi

(3-108)

J XZj X j J XUj
J XUEj

U j

U jEXT
J ZZj Z j J ZUj
J ZUEj

(3-109)

0 JYXi
X j J XXj

0 J ZXj
0 JYXj

X j
JYZj
JYUj U j Y j
Z j

(3-110)

0 U j Yi

(3-111)

0 Ui Yj

(3-112)

Then they can be combined in a matrix form as,

J XX (C )

0
xi J XXi
x 0
J XXj
j
0 J ZXi
0

0 0 J ZXj
0 J
0
YXi
J YXj
0 0
0 0
0

0 0
0
J ZX (C )

J XZ (C )

J XUE (C )

X i J XUei
0 J XZj 0 0 0
J XUj X j 0

J ZZi
0
0
0 J ZUi
0 Z i J ZUei

0 J ZZj 0
0
0
J ZUj Z j 0

J YZi
0 0 J YUi
0 Yi J YUei

0 J YZj
0 0
JYUj Y j 0
0
0 0
0
U i JUUei

0
0
0
0 U j 0
J XZi

0 0 J XUi

0
J XUej
0

J ZUej uiext

0 0

J YUei
0

JUUei

0
(3-113)
u jext

J ZUE (C )

J ZZ (C )

and
T

X (C ) X i

X j ;

Z (C ) Zi

Z j

Yi

Y j

U i

U j ;

Note that this allows us to obtain a single set of equation from two separate sets of
equations. Hence, using this block interconnection approach, all the sub-components of the

116

3.4. EXTENSION OF THE DB CONTROLLED VSC TOGETHER WITH DC LINK

system can be mathematically linked together, which completes the interconnection task of
the subsystems.

3.4.3 Model of the DB Controlled VSC Including a DC Link

Before studying the outer loop controller design, the accuracy of base model for the
controller design is determined as reference to the PSCAD large signal model. The circuit
topology of the converter side outer loop controller is given in Figure 3-54, which takes
into account the dynamics of the DC link.

Ic

u *f

RECTIFIER

C *f
INNER
CURRENT
CONTROLLER

AC Source
I cd _ ref

See Fig (3-21)

DC Link

DC Source

I cq _ ref

Figure 3-54 The base model for the converter side controller
Similar to the results given in Figure 3-49, Figure 3-55 also reveals a good agreement
on the system dynamic that includes the dynamics of the DC link, which is needed for the
outer loop controller design. The results are obtained by applying a 0.095pu small step
change on the q-axis of the input current reference of the converter.
0.14

Iccd-PSCAD(D)

Converter Current Iccd,Iccq (pu)

0.12

Iccq-PSCAD(D)

0.1

Iccd-Matlab

0.08

Iccq-Matlab

0.06

Iccd-PSCAD(D-FIR)

0.04

Iccq-PSCAD(D-FIR)

0.02
0
-0.02
-0.04
-0.06
0

0.05

0.1
Time (s)

0.15

0.2

(a)
117

CHAPTER 3: ANALYTICAL MODELING OF DB CONTROLLED VSC

Converter Current Iccd,Iccq (pu)

0.15

Iccd-PSCAD(D)
Iccq-PSCAD(D)

0.1

Iccd-Matlab
Iccq-Matlab

0.05

-0.05
0

0.01

0.02
Time (s)

0.03

0.04

(b)
Figure 3-55 Comparison of the converter output currents ICcd , ICcq using the (i) largesignal model simulated in the detailed VSC model including the DC link, and (ii) the
small-signal model for a weak AC system together with the DC link. The representations
of the abbreviation in terms of D and D-FIR are the same as in Figure 3-49.

3.5 Conclusion
This chapter developed a complete small signal model for the DB controlled VSC
including the DC link. In order to obtain the small signal model for the three phase discrete
controller, four steps are involved. In the first step, the PadApproximation is adopted to
transfer the discrete controller to the continuous domain controller. In the second step, the
abc three phase continuous domain controller is transformed to the

reference frame

controller, which is followed by transforming the controller from the

reference frame

to dq reference frame. In order to transform the controller from

reference frame to dq

reference frame, a frame transformation approach to eliminate the modulator and demodulator block was proposed in this chapter which has also been verified. It was noticed
that, although the implementation of the PadApproximation introduces a degree of error,
it can still capture the main properties of the system, which can be a valid tool for the
controller design. The small signal model development is presented systematically in the
chapter including the derivation of the entire system equations which are then verified
against the PSCAD simulations. In the simulation studies both of the large signal model

118

3.5. CONCLUSION

and small signal model are verified for every individual system components, as well as the
entire integrated VSC system. It is demonstrated that the contribution of this study can be
used as a solid foundation work that will be used in the remaining chapters of this thesis.

119

Chapter 4: Controller Design for VSCHVDC Connected to a Weak AC Grid

Chapter 3 has established the analytical base model for outer loop controller design,
which will be used in this chapter to demonstrate the outer loop controller design for VSCHVDC link integrated with a weak AC grid as it is a most likely scenario in planning the
future Australian Network. In addition, the selected parameters sensitivity analysis will
also be conducted. The small-signal stability analysis is performed to provide a powerful
tool to predict the dynamics and stabilities of the system in detail. Using the eigenvalue
analysis and participation factor analysis, the main causes of a system oscillation can be
isolated and studied.

4.1 Introduction
Integrating a VSC-HVDC into a weak AC system which is characterized by large grid
impedance and its corresponding control system design were a challenging task for the
researchers. Traditionally, a cascade control scheme involving outer loop control and inner
current loop control is utilized. In such an approach, the performance of the system
predominantly depends on the performance of the inner current control, since a fast inner
120

4.2. ISSUES FOR VSC-HVDC EMBEDDED IN A WEAK AC GRID

current controller can provide sufficient frequency space for the outer power loop
controller design.
It can be noted that the behavior of the inner current controller highly depends on the
interaction of the current control, PLL control and filter bus capacitor. Specifically, under
weak AC condition, the dynamics of PLL and filter bus capacitor play a critically
important role. This is due to the success of a control system is based on the accuracy and
speed response of PLL. However, the control of PLL is also highly governed by the
voltage that is being locked to. In such a case, however, the filter bus voltage fluctuates
dramatically, which is extremely sensitive to the converter current. This posed significant
challenges to the PLL controller design. Therefore, it can be concluded that it is a highly
complex control issue. Although two papers [54, 68] addressed this problem using
simulation studies, it has not been studied systematically and quantitatively so far.
One of the motivations of this thesis is to perform a complete analysis for VSCs
operating in weak AC grids by using a detailed accurate mathematical model from the
small signal stability point of view. Furthermore, another motivation was to develop a
fixed parameter controller with a simple structure (e.g. PI or PID). Although this is
constrained by the limited bandwidth of the inner current controller it can still perform
sufficiently well for a wide range of operating points.

4.2 Issues for VSC-HVDC Embedded in a Weak AC

Grid
In a strong power system, the impact of grid impedance and filter dynamics can be
g
ignored. Therefore, the filter bus voltage u f is independent of the converter current

change. In such a system, the plant to be controlled is only the converter reactor which is a
first-order system. However, the controlled plant becomes a third-order system since the
dynamics of the grid impedance and the filter capacitance are taken into account, such as in
a weak AC grid. The explicit circuit analysis in a weak AC grid is given in the following
section.

121

CHAPTER 4: CONTROLLER DESIGN FOR VSC-HVDC CONNECTED TO A WEAK AC GRID

4.2.1 Circuit Analysis

The single phase circuit diagram and its equivalent control circuit are given in Figure
4-1 [136, 137]. Note that the VSC is modelled by a controlled voltage source, in which the
resistance loss is ignored and the transformer leakage Ltr is added to the grid side
inductance Lg.
g
g

Egg

*
g

ugf

Lc
Cf

g
c

Vcg

1
sL*g

Egg

*1 L*g Lg Ltr

1
sC f

Rf

ugf

Rf

icg

1
sLc

Vcg

i gg

Figure 4-1 Single phase equivalent circuit

From the analysis of the above circuits, the input voltage can be given by
Egg z11
g
Vc z21
where, the impedance matrix is equal to

g
z12 ig

z22 icg

1
*
sLg sC R f
z11 z12
f

1
21 z22
Rf

sC f
Hence, the currents can be calculated as

igg Y11
g
ic Y21

1
sLc
Rf
sC f

1
Rf
sC f

Y12 Eg z11

Y22 Vcg z21
g

(4-1)

z12
z22

Egg
g
Vc

(4-2)

(4-3)

If we expand the above equation, the individual current functions can be given by
g
g
g
ig Y11 Eg Y12Vc
g
g
g
ic Y21 Vc Y22Vc

Hence, the admittance Y22 becomes

122

(4-4)

4.2. ISSUES FOR VSC-HVDC EMBEDDED IN A WEAK AC GRID

1 sC f R f s 2C f L*g
icg
Y22 3 *
Vcg
s Lg Lc C f s 2C f R f ( L*g Lc ) s( L*g Lc )

(4-5)

The above equation shows that the plant to be controlled is a third-order system by
taking into account the grid impedance L*g and filter capacitance Cf. If it is applied to a
strong AC grid, these two factors can be ignored and equation (4-5) is simply reduced to
1/sLc only, which is a first-order system.

4.2.2 System Stability Analysis based on A Simplified Model

To study the system stability, a simplified DB current controlled VSC block diagram is
given in Figure 4-2, in which Gdb(z) represents the DB current controller, GZOH (Z)
represents the ZOH block, Gp(z) is the full representation of the plant to be controlled (i.e.
the discrete form of equation (4-5)) and GF(z) represents the feedback co-efficient.

Ia_ref

Ia_ref (k)

u va_1

G db (z)

Execution
-1
delay z

GZOH(z)

Ia

Gp(z)
T

GF(Z)

Figure 4-2 Simplified DB current controlled VSC

As shown in Figure 4-3 and Figure 4-4a, the bandwidth of the system becomes
narrower as the SCR decreases which corresponds to increase of the grid impedance. This
results in a relatively long time to track the current reference as shown in Figure 4-4b (blue
line). This narrow inner loop bandwidth poses further challenges to outer loop controller
design, which also inevitably results in relatively slow responses of the control system
when being subject to disturbances.

123

CHAPTER 4: CONTROLLER DESIGN FOR VSC-HVDC CONNECTED TO A WEAK AC GRID

(a)

(b)

Figure 4-3 Zeros/poles in z-domain for (a) weak system (b) strong system of a DB current
controlled VSC system
Frequency response of Ia(s)/Iaref(s)-open loop

Step response of Ia(s)/Iaref(s)-close loop

(a)

(b)

Figure 4-4 (a) Bode plots and (b) step responses of a DB current controlled VSC system

4.3 Controller Design Methodology

In the controller design, a set of detailed linear models of the open-loop transfer
functions of VSC-HVDC control system are employed due to their simple and robust
characteristics. The PI/PID compensator parameters are chosen such that the gain margin
(GM) and phase margin (PM) of the compensated open-loop systems are within the ranges

124

4.3. CONTROLLER DESIGN METHODOLOGY

of GM >= 6dB and PM >= 40o respectively. Figure 4-5 shows the flow chart of the process
for designing such controllers.
Determine OP points
(PQ,SCR,R/X Ratio)

Newton Raphson
(Initial Condition Calculation)
System Matrices
(A,B,C,D)

i>=36?
Y

A sets of systems
U
V ( s) U f _ INV ( s)
P( s )
f _ REC ( s )
(
;
; dc
;
)
I cq _ REC ( s) I cd _ REC ( s) I cq _ INV ( s) I cd _ INV ( s)
PI/PID Compensation
(GM>=6dB;PM>=40deg)

EMTDC Verification

End

Figure 4-5 Flow chart of the controller design methodology

4.3.1 Scenarios Considered in Controllers Design

The primary objective of control system design is to design robust controllers that can
allow VSC operating in a wide range of operating conditions while considering various
grid states and PQ capability of a VSC system. Two factors in the grid: i) SCRs and ii) X/R
ratios are considered in this section.
The valid operating points of VSC-HVDC links can be defined by a PQ capability
chart. Three major different limits are applied to VSC-HVDC links: i) maximum current
through the converter ii) DC voltage and iii) Australian automatic access standard (see
Figure 4-6a). By varying the rated voltages 1.02pu, 1.0pu and 0.98pu, three PQ charts are

125

CHAPTER 4: CONTROLLER DESIGN FOR VSC-HVDC CONNECTED TO A WEAK AC GRID

obtained with different radiuses. This can help us to understand how the VSC terminal
voltage will affect the PQ capability chart of VSC system.
The Automatic Access Standard in Australia states that [138]:
Any level of active power and any voltage at the connection point within the limits
established under cause S5.1a.4 without a contingency event, must be capable of supplying
and absorbing continuously at its connection point an amount of reactive power of at least
the amount equal to the product of the rated active power of the generating system and
0.395.
To meet the above requirement, an additional vertical purple line in Figure 4-6a is
added. In this study, the DC voltage is chosen reasonably high, which is not a limiting
factor. Therefore, only two limits are imposed on the system as shown in Figure 4-6b. The
operating scenarios in terms of VSC PQ capability considered in this section are shown in
Figure 4-6b which involves different loading conditions (Pmax, Phalf and P0) and various
level of power factors (leading, unity and lagging). In addition, the grid conditions that
SCR and X/R ratios are also considered as operating scenarios which are listed in Table
4-1.
For weak AC system, the SCR is chosen as SCR=2, while for a strong system it is
chosen as SCR=7.5. For X/R ratios, the pure impedance condition (at 90o) and line resistor
Rg with an impedance angle 75o are considered. The combination of these conditions
allows us to study 36 different operating scenarios, which covers a wide range of operating
points in such grid-connected VSC system.

DC Voltage Limit

Australia automatic access standard

Maximum Converter Current limitation

(a)
(b)
Figure 4-6 PQ capability chart (a) demonstration chart; (b) defined operating points

126

4.3. CONTROLLER DESIGN METHODOLOGY

Table 4-1 Equivalent Thevenin impedance of the AC grid

Uf=62.5 ;
SCR=
2

7.5

: (|Z|cos)( )

: (|Z|sin/2 f) (H)

|Z|

Angle of |Z|

26.042

90o

0.083

75o

6.740

0.080

90o

0.022

75o

1.797

0.021

6.944

4.3.2 Calculation of the Steady State Operating Points

As it was stated before, Newton-Raphson method is used to solve the set of non-linear
equations. The output results can be further compared with the results output from the
fsolve functions in MATLAB. It should be emphasized that a set of operating scenarios
are taken into account for solutions during the calculation processes.
In this section, a set of functions F are defined over the unknown variables X that start
with an initial guess of X0 which is reasonably close to the roots of the set of function F. A
better approximation X1 is obtained using the below equation which iterates many times
until convergence is reached [139].

X n 1 X n

n 0,1,2...

F(Xn)
F'(Xn)

Ic

Table 4-2 Basic power flow equation

Vc U f
2 P jQ
Ic
c c
Zc
3
Vc

Ig

U f E g 2 Pgc jQgc
Ig

Zg
3
U f

Ig Ig I f

2 Pgc jQgc 2 Pc jQc

jB U f
3
3
U f
Vc

(4-6)

Figure 4-7 shows a single line of the AC side of the converter that includes a high-pass
filter. Using the Kirchhoffs voltage law and the power balance relationship, the converter
current, the filter current and the grid side current can easily be obtained as shown in Table

127

CHAPTER 4: CONTROLLER DESIGN FOR VSC-HVDC CONNECTED TO A WEAK AC GRID

4-2. If we expand these equations and apply linearization, the F(X) and its Jacobian Matrix
equations F(X) can be defined which are summarized in Table 4-3, which form the basis
for steady-state operating point calculation.

No.

Table 4-3 Equations for the power flow study

Power Flow Study Equations
Jacobian Matrix Equations F(X)
F(X)

Pc

3
(VCR I CR VCI I CI )
2

3
Pc (VCR ICR _ 0 ICR VCR _ 0 VCI I CI _ 0 I CI VCI _ 0 )
2

Qc

3
(VCI I CR VCR I CI )
2

3
Qc (VCI ICR _ 0 I CR VCI _ 0 VCR I CI _ 0 I CI VCR _ 0 )
2

I CR

Rc
(U _ fR V_ cR ) ...
R X c2
2
c

ICR

Xc
(U _ fI V_ cI )
R X c2

Rc
X
(U _ fR V_ cR ) 2 c 2 (U _ fI V_ cI )
2
R Xc
Rc X c
2
c

2
c

I CI

Xc
(U _ fR V_ cR ) ...
Rc2 X c2

ICI

Rc
(U _ fI V_ cI )
R X c2

Xc
R
(U _ fR V_ cR ) 2 c 2 (U _ fI V_ cI )
2
R Xc
Rc X c
2
c

2
c

Pgc

3
(U fR I gR U fI I gI )
2

3
Pgc (U fR I gR _ 0 I gR U fR _ 0 U fI I gI _ 0 I gI U fI _ 0 )
2

Qgc

3
(U fI I gR U fR I gI )
2

3
Qgc (U FI I gR _ 0 I gR U fI _ 0 U fR I gI _ 0 I gI U fR _ 0 )
2

I gR

Rg
R X g2
2
g

Xg
R X g2
2
g

I gI

( E_ gR U _ fR ) ...

( E_ gI U _ fI )
Xg

Rg2 X g2

Rg
Rg2 X g2

I gR

( E_ gR U _ fR ) ...

( E_ gI U _ fI )

I gI

Rg
R X
2
g

2
g

Xg
R X
2
g

2
g

(E_ gR U _ fR )

(E_ gR U _ fR )

Xg
R X g2
2
g

Rg
R X g2
2
g

(E_ gI U _ fI )

(E_ gI U _ fI )

I gR I cR B U fI

I gR I cR B U fI

10

I gI I cR B U fR

I gI I cR B U fR

11

U fR U f 0 cos( _ PLL )

U fR U f 0 sin( _ PLL _ 0 ) ( _ PLL )

12

U fI U f 0 sin( _ PLL )

U fI U f 0 cos( _ PLL _ 0 ) ( _ PLL )

128

4.3. CONTROLLER DESIGN METHODOLOGY

Pgc jQgc
Eg 0o

Z g

Pfc jQ fc

U f _ PLL

Pc jQc
Vc

Zc

AC network Side

VSC Side

Ig

If

Ic

jB_ REC
Figure 4-7 Single-line diagram of AC side converter including a high-pass filter

4.3.3 Controller Design for Rectifier

It can be noted that the bandwidth of the outer loop is highly influenced by the subsystems. Therefore, the complete VSC-HVDC model is developed as a base study for outer
loop controller design. However, the model of the long transmission line is also involved in
this model, which has to be included in the design procedure for the accuracy purpose.
This is particularly suitable for the Australian grid, where the renewable energy sources
such as geothermal and wind energy resources are far away from the major load centres.

(a)

(b)

Figure 4-8 Influence of the DC link model: (a) pole-zero map of Iq(s)/Iqref(s); (b) Bode
plots of Iq(s)/I qref(s)
Figure 4-8 shows the pole-zero map and Bode plots of the inner DB current controller
with and without the DC link model. As it can be seen in the Figure 4-8a that there are two
more pole-zero pairs presented in the model with DC link than the model without DC link
at very low frequency, which are nearly cancelled with each other. However, at frequency

129

CHAPTER 4: CONTROLLER DESIGN FOR VSC-HVDC CONNECTED TO A WEAK AC GRID

around 30rad/s, a pair of cancellable pole-zero pair exists in the model without DC link,
while in the model with DC link there is only one pole pair which is un-cancellable (see
the circled area in the Bode plot). In addition, small magnitude difference was observed in
the Bode plots of two systems. The results reveal that it is possible to develop more
accurate and realistic controllers when the dynamics of DC link are being considered.
4.3.3.1 Controller Design Circuit for Rectifier
The model adopted for the outer power controller design at the rectifier side is shown
in Figure 4-9. The DC side transmission line is modelled as a T-section ended with a
constant DC voltage source.

u *f
RECTIFIER

PREC

C *f
INNER
CURRENT
CONTROLLER

AC Source

See Fig (3-21)

DC Link

DC Source

I d _ ref
Pref

isq _ max

I q _ ref

PI
isq _ max

Figure 4-9 Basic structure for outer power controller design

Note that the control scheme used here is a cascade control strategy. The DB control
scheme is adopted as the inner current control for both of the rectifier and inverter sides,
while for the outer loops, the rectifier is responsible for controlling the active power and
the constant filter bus AC voltage, and the inverter controls the DC voltage and the filter
bus AC voltage to be constant.
The above circuit diagram is the basis for rectifier side controllers design. Due to the
fact that the objective of the controller design is to control the DC voltage at the inverter
end, the inverter can be simplified as a constant DC voltage source. Although the
assumptions made here that the constant DC voltage controller is implemented ideally and
the response speed of the DC voltage controller is sufficiently fast does not 100 precent
hold it is still adequate to be used as a start point in the controller design.

130

4.3. CONTROLLER DESIGN METHODOLOGY

4.3.3.2 Inner Current Controller Analysis

The eigenvalues and their sensitivities are investigated to determine how the system
parameters and loading conditions contribute to the small signal stabilities of the entire
system.
A. Eigen-value Analysis
Table 4-4 shows all of the eigenvalues associated with the base loading scenario (at
maximum power, unity power factor and for SCR equals to 2). As listed in the Table 4-4,
all the eigenvalues have the negative real parts, which imply that the base loading scenario
is stable. The dominant modes of the system are the modes 15 and 16 without adequate
damping. The participation factors shown in Table 5-5 indicate that the DC link states have
a dominant impact on the conjugate complex pair 15 and 16 with a damping ratio of 0.033
and a natural frequency of 28.949Hz. The modes 19 and 20 are the root causes of the
small dip appeared in the frequency response plot of the weak AC system locating around
27.37 rad/s (see circled area in Figure 4-8b).

Table 4-4 Modes of the inner current controller (SCR=2)

Mode

Eigenvalue

Damping Ratio Nature Frequency (Hz)

1,2

-2.04107314j

50

3,4

-7.91103307j

48.834

5,6

-1.6251033.03103j

0.473

482.18

7,8

-1.621032.48103j

0.547

394.78

9,10

-31033.04102j

0.994

48.345

11,12 -2.791026.64102j

0.387

105.69

13,14

-78.7263.6j

0.286

41.955

15,16

-6.094182j

0.033

28.949

17,18

-25.37918.384j

0.81

2.926

19,20

-12.2127.37j

0.407

4.3558

As shown in Table 4-5, the above mentioned modes 15, 16, 19 and 20 are also greatly
influenced by the DC link states. Therefore, any attempts to improve these modes must

131

CHAPTER 4: CONTROLLER DESIGN FOR VSC-HVDC CONNECTED TO A WEAK AC GRID

take into accounts these DC link states. PLL is mainly associated with the modes 17 and 18.
While, the time constant of the AC voltage signal filter and the AC grid states are the main
contributors to modes 11-14. The participation factor table also suggests that the states
with respect to the circuit in adjacent area of VSC terminal (converter-reactance and grid
voltage filter) and the DB controller have dominant impacts on modes 5-10. Lastly, the
modes 1-4 are solely determined by the DB control block. In the following section, the
eigen-sensitivity analysis in terms of these states will be explicitly discussed.
Table 4-5 Participation factors of dominant state variables for selected modes of the inner
DB current controller
Sub-System State
Grid Controller
DB control

AC network

DC grid

Mode

PLL
Tf
DeadBeat
Sampling
Delay
Con-Reactance
Cf
AC grid
C_dc_rec
Ldc_rec
Cdc_line
Ldc_inv

1
2

100

3
4

5
6

7
8

9
10

1.139 2.37 3.22 5.19

93.31 14 13.8 19.7
2.24 14.2 14.2 17.1
2.253 22.6 22.6 25
33.7 33.4 22.5
1.054 12.1 12 9.62
0.97 0.92 0.87

11
12
0.892
19.1
3.079
2.772
5.352
10.87
11.55
46.39

13
15
14
16
6.24
36.2
1.54
2.58
5.1
5.32
12.3
29.3
1.49 1.455
26.33
48.58
23.63

17
19
18
20
86.4 12.75
1.48 0.609
0.1
0.19
0.34
2.68
3.35
2.68
0.09
2.67

1.003
39.9
21.28
1.087
23.37

B. Selected Parameter Sensitivity Effects on Main and Cross-Coupling Control Loops

In the below paragraphs, an in-depth analysis of the impact of the system parameters,
the loading conditions and the power factor on system small-signal stability is given.
a) SCR with Pure Impedance at Full Loading Condition
The impact of SCR on system small-signal stability is investigated in this section. It is
assumed that only the inner DB current loop is in operation, and operating under the full
loading condition Pmax. For simplicity, the loss of the transmission line is ignored.
Therefore, the AC grid model is represented by an AC voltage source behind impedance
through the Thevenins equivalence. The eigenvalue evolution map of the system with
SCR reducing from 7.5 to 0.5 at a step -1 is shown in Figure 4-10a. As it can be seen in the
figure, two modes move to the right half plane (RHP) which makes the system tending to
be unstable as SCR reduces. Moreover, the bandwidth of the inner current loop is narrowed

132

4.3. CONTROLLER DESIGN METHODOLOGY

down as the consequence of the high frequency modes evolving to the low frequency
region. This makes the design of outer loop controller design even more complex and
challenging.

(b)
(a)
Figure 4-10 (a) In the DB current controlled VSC with P=Pmax (a) eigenvalue maps with
SCR reduced from 7.5 to 0.5 at a step of -1 (i.e. A to G); (b) Bode plots of the TF (blue:
SCR=7.5, red: SCR=4.5, cyan: SCR=2).
Figure 4-10b shows the Bode plots of the transfer function for the DB current
controlled VSC. The upper left figure plot shows the magnitude plot for power. Note that,
only the magnitude plot is shown here for simplicity purpose. It can be observed that the
magnitude of the strong system has a higher gain compared with that of the weak AC
system. However, it should be noted that this is not a regular pattern for all loading
conditions. The test results of the zero loading condition (static synchronous compensator
(STATCOM) mode) show a reverse pattern. Nevertheless, the Bode plots of the filter bus
AC voltage (the lower plots of Figure 4-10b) confirm that it is more sensitive to the power
variation in a weak AC grid. Furthermore, the upper right and lower left plots also confirm
the fact that the cross-coupling effects increase as SCR reduces.
Figure 4-11 shows the time-domain simulation results in MATLAB and
PSCAD/EMTDC following a 0.01pu step change on iqref at t=0s. The PSCAD simulation
confirms the accuracy of the small signal model again. In addition, the results also suggest
that the time domain simulation agrees with the conclusion drawn from the frequency
response analysis. For example, the speed of the current tracking response of the weak AC
grid is slower compared with that of the strong AC grid, since the weak AC system has a

133

CHAPTER 4: CONTROLLER DESIGN FOR VSC-HVDC CONNECTED TO A WEAK AC GRID

narrower bandwidth. As it is can be seen in the middle power plot of Figure 4-11, the
system with a higher SCR has more reduction since the un-shown phase angle plot for the
power curve starts from -180o. Furthermore, the third plot in Figure 4-11 shows that a
much higher filter bus voltage (black and red plots) increase for weak AC system is

Icq(kA)

observed compared with that of the strong system (see blue and cyan plots).

0.02
0
-0.02
0

0.01

0.02
0.03
Time(Seconds)

0.04

0.05

0.01

0.02
0.03
Time(Seconds)

0.04

0.05

0.01

0.02
0.03
Time(Seconds)

0.04

0.05

P(MW)

1
0
-1
0

Uf(kV)

0.2
0
-0.2
0

Figure 4-11 The time-domain simulation results: PSCAD SCR=7.5 (solid cyan line);
Matlab SCR=7.5 (dash-dot blue line); PSCAD SCR=2 (solid black line); Matlab
SCR=2(dash-dot red line).

b) X /R Ratio for SCR=2 at Full Loading Condition

The second impact factor discussed is the equivalent grid impedance angle, which is
equals to X/R ratio. Four main low frequency modes are shown in Figure 4-12. The results
reveal that the higher the resistance of the grid impedance, the more damping is added to
the system regardless of the modes of operation (rectifier or inverter). This also implies
that the system is more stable by solely inspecting Mode I1 at relatively high frequency.
However, by observing Modes I3 and I4, a common trend can be summarized that the
damping effects for the inverter are less than that of the rectifier. This also means the
observed oscillation of the inverter should be a little higher than that of the rectifier when
system being subject to transient disturbances.

134

4.3. CONTROLLER DESIGN METHODOLOGY

I1

I3
I4

I2
I3
I4

Figure 4-12 Four main low frequency modes for A: rectifier mode with a 90o angle; B:
rectifier mode with a 75o angle; C: rectifier mode with a 60o angle; a: inverter mode under
90o angle; b: inverter mode under 75o angle; c: inverter mode under 60o angle;

40

|G PId(jw)|(dB)

|G PIq(jw)|(dB)

40
30
20
10 0
10

10

30
20
10 0
10

(rad/s)

(rad/s)

40

|G UfId(jw)|(dB)

30

|G UfIq(jw)|(dB)

10

20
10
0 0
10

10
(rad/s)

30
20
10 0
10

10

(rad/s)

Figure 4-13 The Bode plots of the DB current controlled VSC. blue: rectifier mode under
90o angle; cyan: Rectifier mode under 75o angle; red: rectifier mode under 60o angle;
magenta: inverter mode under 90o angle; black: inverter mode under 75o angle; green:
inverter mode under 60o angle;
The Bode plots of the DB current controlled VSC are shown in Figure 4-13. As it can
be observed in the top-left figure, the rectifier mode operation, the system with higher

135

CHAPTER 4: CONTROLLER DESIGN FOR VSC-HVDC CONNECTED TO A WEAK AC GRID

impedance has the highest gain in the active power loop. At the inverter side the system
with a higher resistance has the highest gain. It should also be emphasized here that a main
trend can be observed the inverter has a relatively higher gain in comparison with that of
the rectifier. Alternatively, for a rectifier, it can be interpreted as to transmit the same
amounts of power the system with a higher resistance experiences larger current variation
which results in the current level reaching its limit. Therefore, it can be concluded that the
minimum SCR required in such system is higher than the more inductive system, while
such trend reverses at the inverter reference to the Bode plots analysis. Moreover, the
inverter is more robust when embedded in a weak AC system since it generally has a much
higher gain than that of the rectifier side as shown in Figure 4-13a. The same conclusions
have also been reported in ref [140].
It was also observed that the rate of change in P/Id or Uf/Id has a similar trend as
in P/Iq. However, under above two conditions, the change in the relative gain is found to
be not as big as that of in the power loop. Similarly, the rate of change in Uf/Iq has an
inverse trend. The more resistive system has the highest gain reference to the rectifier side,
vice versa for the inverter side. In conclusion, the rectifier mode has a higher gain, but the
inverter mode has a wider relative gain change. The time domain simulation results shown
in Figure 4-14 also confirm the above conclusions.

Figure 4-14 Step responses of DB current controlled VSC for the rectifier (left) and the
inverter (right). rectifier mode at angle 90o (blue); rectifier mode at angle 75o (cyan);
rectifier mode at angle 60o (red); inverter mode at angle 90o (magenta); inverter mode at
angle 75o (black); inverter mode at angle 60o (green);

136

4.3. CONTROLLER DESIGN METHODOLOGY

c) Operation under Various Loading Condition at SCR=2

Figure 4-15 shows the eigen-value map and Bode plots of the DB current controlled
VSC system embedded in a weak AC system at SCR=2 and under varying loading
conditions. It is found that the loading condition has a significant impact on the crosscoupling behaviour of the system. As seen in Figure 4-15 (a), the mode I1 is relatively the
most sensitive mode to the loading condition since it moves to the left with smaller
damping ratio. However, it does not significantly affect the system stability.

I1

(a)

(b)

Figure 4-15 For DB controlled VSC with SCR=2 (a) the eigenvalue map with P equals to
be 1pu (A), 0.5pu (B) and 0pu (C); (b) Bode plots of the TF with different loading
conditions. blue: P=1pu, cyan: P=0.5pu, red: P=0pu).
It can also be observed that the magnitude of the transfer function of the cross-coupling
behaviors increased dramatically with loading conditions (see upper right and lower left
plots in Figure 4-15b). This indicates that more interaction between active and reactive
power loops could be observed once the outer power loops are closed.
d) Operation under Various Power Factors at SCR=2
As shown in Figure 4-16 modes I1 and I2 move to the opposite direction when varying
the operating modes of VSC from receiving reactive power from grid to sending reactive
power to grid. However, Mode I1 is the main limitation of the bandwidth of the GPIq(s).
Note that the system with a leading power factor (case C in Figure 4-16a) has the
narrowest bandwidth. Moreover, the impacts of changing power factor are more significant
on the active power related loop (left column of Figure 4-16b), whereas the impacts on the
reactive power loop are comparably small (right column of Figure 4-16).

137

CHAPTER 4: CONTROLLER DESIGN FOR VSC-HVDC CONNECTED TO A WEAK AC GRID

I1

I2

(a)

(b)

Figure 4-16 (a) The eigen-value map and (b) Bode plot of the transfer function for the DB
current controlled VSC at SCR=2 and different power factors (Q=Qmax, Q=0 and Q=Qmin)
under full load condition. (dark blue: lagging power factor, light blue: unity power factor,
red: leading power factor).
It can be concluded from these results that SCR, X/R ratios, loading conditions and
various power factors all have significant impacts on system small signal stabilities.
Therefore, it is expected that these analyses can provide a better understanding on the
characteristics of the plant to be controlled and facilitate ongoing controller design.
4.3.3.3 Power Controller Design for Rectifier
As it was mentioned at the beginning of this chapter, the primary aim of this chapter is
to design a linear controller with a set of fixed parameter that can work under a wide range
of operating conditions. However, it is important to note that the inner DB current
controller of the weak AC grid connected VSC has a limited bandwidth as mentioned
before.
A. Frequency Response Analysis of all the Covering Operating Points
It is important to examine the characteristics of the transfer functions for 36 scenarios
mentioned earlier before conducting the controller design.
As shown in Figure 4-17a, b, c and d, the gain varies from 31.97dB to 36.79dB, which
corresponds to a difference of 4.82dB (factor of 1.7418 which is acceptable under steady
state). Under the weak AC system condition, the gain varies from 31.97dB to 36.77dB,
which is much wider than the case of a strong system (in which the gain varies from 36.79
dB to 36.3 dB).

138

4.3. CONTROLLER DESIGN METHODOLOGY

(a)

(b)

(c)

(d)

Figure 4-17 The analysis of the transfer functions using Bode plots and at 36 operating
conditions in total (a) SCR=2 at 90o (b) SCR=2 at 75o; (c) SCR=7.5 at 90o; (d) SCR=7.5 at
75o;
It can be seen in Figure 4-17a that the operating points under the same loading levels
belong to the same group. In other words, the system delivering the same amounts of
power have similar gains. For example, the operating points 7-9 which transfer zero power
to the DC side have the highest set of gains. Similarly, the operating points 4-6 (half
loading condition) have a set of moderate gains in the middle group, and the remaining 3
operating points which are the full-loading modes have the lowest set of gains. Moreover,
the operating modes which absorb the maximum reactive power from grid has a
comparably higher gains compared with the modes where the maximum reactive power is

139

CHAPTER 4: CONTROLLER DESIGN FOR VSC-HVDC CONNECTED TO A WEAK AC GRID

sent to grid. Figure 4-17b has a similar trend with that of Figure 4-17a, while the rest of the
results c and d of Figure 4-17 have reverse trends compared with the pattern observed in
Figure 4-17a, b.
B. RHP Zero Limitation on Outer Loop Bandwidth
It is well known that the poles of a system have significant impact on system stability.
However, there are limited studies on how the zeros of a system limit the controller design
of a VSC-HVDC system. It was reported in [63] that the bandwidth of the closed-loop
system should be chosen at least lower than half position of the RHP zero. A more accurate
analysis was given in [141] indicated that the achievable bandwidth b of the outer loop
controller is significantly limited by the RHP zeros, which are
Re( z ) Im( z )
| z | /4

b c | z | / 2.8 Re( z ) Im( z )

| z |
Re( z ) Im( z )

(4-7)

where, z denotes the complex form of the system zero which is the closest to the origin.
It can be summarized that to have RHP zeros close to origin is inappropriate, and it is even
worse to have RHP zeros closer to the real axis rather than the imaginary axis [141]. It was
also found in [141] that the root causes resulting in the RHP zeros is the competing effects
among sub-components of a system.
As given in the figure below, there are 4 RHP zeros in Figure 4-18a and 1 RHP zero in
Figure 4-18b involved in case C and F locating in different frequency range. This caused
1080o change crossing low frequency to high frequency (see Figure 4-17a). It should be
noted that the cases A-I given in Figure 4-18 correspond to the cases 1-9 given earlier in
Figure 4-6b. Note that, these cases impose limits to the achievable bandwidth of the outer
loop controller. The case F has the closest zero to the origin, which is located at 44.28
Np/s+264.4 rad/s. This indicates that the designed outer loop controller should not exceed
the value |Z|/2 which corresponds to 134.04 rad/s. In other words, a controller with an
extremely fast responding speed is hard to achieve. Similarly, 3 RHP zeros included in the
cases A, B, D, E, and G, induce 720o change varying from low frequency to high
frequency. Similarly, only 1 RHP zero in the cases H and I produce 360o change in the
phase plot.

140

4.3. CONTROLLER DESIGN METHODOLOGY

(a)

(b)

Figure 4-18 Right half-plane zeros for the transfer function of the open loop power
controller
C. Power Controller Design and its Small Signal Model
As indicated previously, the main aim of this sub-section is to design a power
controller and then to develop its small signal model. Therefore, the design methodology is
proposed first, which is not just limited to the power controller design, followed by step
tests to evaluate the performance of the designed power controller.
a) Design Methodology
The proposed design methodology is composed of four steps:
Step 1: Observe the frequency responses of P(s)/Iqref(s), and then choose the cases with
the lowest bandwidth, the highest bandwidth and the bandwidth in the middle of those two.
Step 2: Add a low-pass filter to ensure the open-loop gain is substantially less than DC
gain at all frequencies (above the gain cross-over frequency, if necessary).
Step 3: Apply the classic loop-shaping technique to the system obtained in Step 2, and
obtain the controller parameters Kp, Ki or Kp, Ki, Kd depending on the type of controller
designed.
Step 4: Evaluate the controller performance thoroughly against the detailed model
established on PSCAD/EMTDC simulator considering different size steps, operating points
and SCRs.
In the first step, the frequency response plots presented in Figure 4-17 are analysed. In
this design, cases 4, 6 and 9 are chosen as a reference for the design as given in Figure
4-19.

141

CHAPTER 4: CONTROLLER DESIGN FOR VSC-HVDC CONNECTED TO A WEAK AC GRID

Phase(deg)

Magnitude(dB)

Frequency response of P(s)/Iqref(s)-open loop

40
20
0
-20 -1
10

10

10

10

10

500
0
-500 -1
10

PhalfQmax
PhalfQmin
P0Qmin
0

10

10

10

10

Frequency (rad/sec)

Figure 4-19 Bode plots of the selected cases for the power controller design
In step 2, a first-order low-pass filter (see Figure 4-20a) is added to the system. This
indicates that an additional attenuation (where 20dB/decay rolling off when exceeding
60rad/s) is added to the gain plot (see Figure 4-20b). To meet the stability criteria
mentioned in section 4.3 (in which the gain margin (GM) and the phase margin (PM) of
the compensated open-loop systems are required to be within the ranges GM >= 6dB and
PM >=

respectively), a set of PI parameters are selected. Using the PI compensator

designed in Figure 4-20c, these transfer functions yield a phase margin of better than about
62o at a gain cross-over frequency of about 7.46 rad/s (a little bit lower than the first dip in
the magnitude plot that is actually a zero caused by DC link) and a gain margin better than
about 27 dB (see Figure 4-20d). This suggests that the corner frequency may exceed the
frequency where zero occurs, which is caused by the dynamics of the DC link, and may
cause an inadequate performance or even instability due to the ignored dynamics, therefore
the dynamics of DC link should be considered in the controller design.

(a)
142

(b)

4.3. CONTROLLER DESIGN METHODOLOGY

(d)

(c)

Figure 4-20 The Bode plots with (a) low-pass filter; (b) DB current controlled VSC with a
low-pass filter; (c) designed PI compensator; (d) open loop transfer function of P(s)/Pref(s).
The closed loop responses of P(s)/Pref(s), including all 36 loading scenarios being
subject to a 1pu step change in the power reference are employed to evaluate the
performance of the designed power controller which is given in Figure 4-21.

1.4
1.2

Case 9: Overshoot(%):4.35
At time (seconds):0.219

P(pu)

1
Case 2:
Settling time (seconds):0.506

0.8
0.6
0.4
0.2
0
0

0.1

0.2

0.3
0.4
0.5
Time (s) (seconds)

0.6

0.7

0.8

Figure 4-21 Step responses of P(s)/Pref(s)

143

CHAPTER 4: CONTROLLER DESIGN FOR VSC-HVDC CONNECTED TO A WEAK AC GRID

Not that the responses are satisfactory as they settle to within 2% of the final value at
about 0.506s (the maximum settling time for the entire transfer functions). The overshoot
(height of peak relative to the final value) is about 4.35% which is within the acceptable
range (< 10%) suitable for the reference tracking.
b) Small Signal Model of Power Controller
Figure 4-22 shows the control block diagram of the power controller, where Pm
denotes the measured value of power and PX represents the power state. Tf_P, Kp_P and
Ki_P denote the control parameters derived from the previous sub-section. Hence, the
corresponding small signal model is given by the below equations.

d Pf
dt

Pref Pm Pf

d PX
Ki _ P Pf
dt

(4-9)

iqref K p _ P Pf PX

(4-10)

Pm

Pref

+-

(4-8)

Tf _ P

1/ (1 T f _ P s)

Pf

K p_P

iqref ( s )

+
Ki _ P / s

PX

Figure 4-22 Control block for the power controller

4.3.3.4 AC Voltage Controller Design for the Rectifier and its Small Signal Model
Note that the power control loop should be taken into account when designing the AC
voltage controller to improve the accuracy.
Ysp1 - E1
Gc1
+

U1

G p11

Y1

G p12

Ysp 2

E2

Gc 2

U2

G p 21
G p 22

Y2

Figure 4-23 A typical MIMO system, in which the hidden feedback loop is shown in red
lines

144

4.3. CONTROLLER DESIGN METHODOLOGY

Figure 4-23 shows the typical multi-in multi-out (MIMO) system, where Gp11, Gp22 are
two main inner loops, while Gp12, Gp21 denote the transfer functions of two cross-coupling
loops, and Gc1, Gc2 represents the transfer functions of two outer loops. In addition, Ysp1,
Ysp2 are inputs of the outer-loops, while U1(s) and U2(s) are the outputs of the outer-loops
and the inputs to the inner loops and Y1, Y2 are the outputs of the MIMO system.
The closed-loop transfer function of the output Y1(s) to the input Ysp1(s) can be given
by,

Y1 ( s)
G p11
U1 ( s )

(4-11)

For simplicity purpose, all the (s) are ignored in the following analysis. Hence, the
following equations can be given.

Y1 U1 Gp11 U 2 Gp12

(4-12)

U 2 (Ysp 2 Y2 ) Gc 2

(4-13)

Y2 U1 G p 21 U 2 G p 22
Y2 U1 G p 21 (Ysp 2 Y2 ) Gc 2 G p 22

(4-14)

Rearranging equation (4-14), we can obtain

Y2 (1 Gc 2 Gp 22 ) U1 Gp 21 Ysp 2 Gc 2 Gp 22 ,

(4-15)

And assuming Ysp2 =0,

Y2

U1 G p 21

(4-16)

1 Gc 2 G p 22

Substituting (4-16) into equation (4-13) ,

U 2 Y2 Gc 2

Gc 2 G p 21
1 Gc 2 G p 22

U1

(4-17)

Similarly, substituting equation (4-17) into equation (4-12), we can obtain

Y1 U1 G p11
Y1 (G p11

Gc 2 G p 21 G p12
1 Gc 2 G p 22

Gc 2 G p 21 G p12
1 Gc 2 G p 22

U1

) U1

(4-18)

(4-19)

Without considering the feedback loop in red in Figure 4-23, which means leaving the
outer loop appended, then the transfer function becomes,

145

CHAPTER 4: CONTROLLER DESIGN FOR VSC-HVDC CONNECTED TO A WEAK AC GRID

Y1 Gp11 U1

(4-20)

By comparing equation (4-19) and equation (4-20), we can come to the conclusion that
that to involve the power controller in the model can achieve a more precise basement for
AC voltage controller design.
A. AC Voltage Controller Design
Using the methodology proposed in section 4.3.3.3 for power controller design, the
first step is to choose the baselines for controller design. Scenarios 3, 7, 9 and 36 (i.e.
PmaxQmax, P0Qmax, P0Qmin for SCR=2 with angle

and P0Qmin for SCR=7.5 with angle

) are chosen in this study as shown in Figure 4-24.

Magnitude(dB)

Frequency response of Uf_rec(s)/Idref(s)-open loop

40
20
0
-20 0
10

10

10

10

Phase(deg)

-200

-400 0
10

Case3
Case7
Case9
Case36

.
1

10
10
Frequency (rad/sec)

10

Figure 4-24 Bode plots of selected cases in the AC voltage controller design

40

Magnitude(dB)

Magnitude (dB)

0
-10
-20

Filter: 1/(1+0.0075*s)

-30

Phase(deg)

Phase (deg)

-20
1

10

10

10

-45

10

10
10
Frequency (rad/s)

(a)

146

-40 0
10

-40
0

-90
0
10

20

10

-200
-400
-600 0
10

Case3
Case7
Case9
Case36
1

10
10
Frequency (rad/sec)

(b)

10

4.3. CONTROLLER DESIGN METHODOLOGY

(c)

(d)

Figure 4-25 The Bode plots of (a) low-pass filter; (b) Uf(s)/Idref (s) of DB current controlled
VSC with a low-pass filter; (c) the designed PI compensator and (d) open loop transfer
function of Uf(s)/Ufref(s) at the rectifier side.
Similar to the power controller design, a first-order low-pass filter (see Figure 4-25a) is
added to the system to provide an additional attenuation that 20dB/decay which rolls off
when exceeding 133.33rad/s (see Figure 4-25b). Then a set of PI parameters Kp=
0.0314252 and Ki= 1.575 are chosen, which yield a phase margin of better than about 60.9
degree at a gain cross-over frequency of about 41.7 rad/s and a gain margin better than
about 9.6 dB (see Figure 4-25d). The closed-loop responses of Uf(s)/Ufref(s) (including all
36 scenarios) being subject to a 1pu step change in the AC voltage reference are employed
to evaluate the performance of the designed AC voltage controller as shown in Figure 4-26.
1.2

Case 3 Oershoot(%):9.91
At time(seconds):0.0761

Case 36
Settling time(seconds):0.336

Uf(pu)

0.8
0.6
0.4
0.2
0
-0.2
0

0.1

0.2
0.3
0.4
Time (s) (seconds)

0.5

0.6

147

CHAPTER 4: CONTROLLER DESIGN FOR VSC-HVDC CONNECTED TO A WEAK AC GRID

Figure 4-26 Step responses of Uf(s)/Ufref(s)

Note that the simulation results of the step change responses demonstrate a satisfactory
performance, since it settles to the final value within 2% until after 0.336s and with a
maximum overshoot of 9.91% (better than the basic requirement 10%).
B. Small Signal Model for AC Voltage Controller
The block diagram of the AC voltage controller is shown in Figure 4-27. The error of
the reference value Ufref and the measured value Um of the AC voltage go through a low
pass filter 1/(1+TfUf_rec s) , followed by a PI compensator ( K P_Uf +Ki_Uf /s ). Finally, the
current reference of the reactive power loop idref (s) is obtained according to equation
(4-23). The small signal model of the AC voltage controller is given by the equations (4-21)
-(4-23).

U f m
U f ref +

U f_ f
1/(1+TfUf_rec s)

K P_Uf

+
K i_Uf /s

i dref (s)

U fX

Figure 4-27 Control block for AC voltage controller

d U f _ f
dt

U fref U fm U f _ f

d U fX
dt

T fU f _ rec
Ki _ U f U f _ f

idref K P _U f U f _ f U fX

(4-21)

(4-22)
(4-23)

4.3.4 Controller Design for Inverter

After designing the controllers for the rectifier converter, it is time to design the DC
voltage controller and the AC voltage controller for the inverter. However, since the
controllers designed for the rectifier converter in the above sections are also being included
in the complete model, it is important to understand how the operating points of the
rectifier side converter can affect the inverter side controller-design.

148

4.3. CONTROLLER DESIGN METHODOLOGY

(a)

(b)

(d)
(c)
Figure 4-28 The frequency responses of (a) Udc(s)/Iqref(s); (b) Udc(s)/Idref(s); (c) Uf_inv(s)/
Iq_ref_inv(s) and (d) Uf_inv(s)/Id_ref_inv(s) being subject to the operating point variation of the
rectifier side converter (36 scenarios), but with a constant operating point at the inverter
side converter that is PmaxQ0, SCR=2 and with angle equals to 90o.
Note that 36 operating points are considered in the controller design for the rectifier.
However, it is not feasible to consider 36 operating points for both of the inverter and
rectifier operating points together (362) when designing an inverter side controller.
Therefore, to reduce the number of the total operating points considered it is necessary to
examine the sensitivity of the transfer functions for a given frequent operating point of the
inverter first. In such an approach, if the sensitivity is less, any operating point can be
chosen as a base operating point for further controller design, normally the most frequent
operating point is chosen as the base scenario. In such a case, only 36 inverter side
operating points are considered, which significantly reduce the initial total number 362.
Figure 4-28 shows the frequency responses of the transfer functions Udc(s)/Iqref(s),
Udc(s)/Idref(s), Uf_inv(s)/ Iq_ref_inv(s) and Uf_inv(s)/ Id_ref_inv(s) which correspond to the 36

149

CHAPTER 4: CONTROLLER DESIGN FOR VSC-HVDC CONNECTED TO A WEAK AC GRID

operating point variations of the rectifier. In this figure, a constant operating point of the
inverter (PmaxQ0, SCR=2, 90o angle) is assumed. Note that no significant discrepancies are
observed in the results of the two main loops up to 100rad/s (Figure 4-28a, d). However,
results are found to be different in the cross-coupling loops (Figure 4-28b, c). This
indicates that the outer loop controllers for the main loops need a careful design approach
to minimize the cross-coupling effects. Anyhow, it can be concluded that the sensitivity of
the transfer functions of operating points is low, in fact, of which the impact can be ignored
for simplification purpose. Therefore, based on this assumption, only a single frequent
operating point (PmaxQ0, SCR=2, 90o angle) is sufficient for the base scenario of the
rectifier.
It should be emphasized that the VSC losses are not considered in the small signal
model. However, the VSC losses cause inaccuracies in the steady state calculation. This
results in a minor power imbalance further causing minor inaccuracies in the filter bus
voltage. Consequently, the DC voltage also becomes inaccurate, since the calculated DC
voltage is utilized rather than the real DC transmission line feedback data in modeling of
the control system. This approach avoids the potential problems caused by the feedback
dynamics of the transmission line which can complicate the calculation process. Although
the above approach reduced the accuracy of the model, the deviation in the final results is
found small once the loops of the entire system are closed.
4.3.4.1 DC Voltage Controller Design for the Inverter and its Small Signal Model
Figure 4-29a shows the frequency responses of the selected cases in the DC voltage
controller design. By employing the compensator shown in Figure 4-29b, it can be
predicted that a satisfactory performance can be achieved when being subject to a step
change in the DC voltage reference (see Figure 4-29d) based on the analysis of the Bode
plots of the compensated system (see Figure 4-29c).
By testing the frequency response for all 36 scenarios, it is found that the system using
the designed controller yields a phase margin of better than 85.9 degree at a gain crossover frequency of about 9.71 rad/s and a gain margin of better than 24.3 dB (see Figure
4-29c). Furthermore, the results of the step response tests reveal that a fairly low maximum
overshoot 1.89% (at t=0.212s) is obtained in the scenario that PminQmin, SCR=2 and angle
=90o. In addition, a satisfactory maximum settling time at t=0.532 is obtained at the
scenario that PminQmax, SCR=7.5 and angle =90o.

150

4.3. CONTROLLER DESIGN METHODOLOGY

(a)

(b)

(c)
(d)
Figure 4-29 (a) The Bode plots of the selected cases of the DC voltage controller design
Udc(s)/Iqref_inv(s); (b) designed PID compensator (Kp_dc+Ki_dc/s+Kd_dcs/(1+TfUdcs)=-0.0240.088/s+0.0009s/(1+0.06s)); (c) frequency responses of open loop transfer functions of
Udc(s)/ Udc_ref (s) and (d) step responses of the closed-loop Udc(s)/ Udc_ref(s).
The control block diagram of the DC voltage controller is shown in Figure 4-30 and its
small signal model is given by equation(4-24).

U dcm
U dcref
+-

K P _ U dc

K i _ U dc / s
K d _ U dc s

U dcX 1

iqref ( s )

U dcX 2

1 T fU dc s

Figure 4-30 The control block diagram of the DC voltage controller

151

CHAPTER 4: CONTROLLER DESIGN FOR VSC-HVDC CONNECTED TO A WEAK AC GRID

GPID ( s) K p Ki / s K d s / (1 T f s )
GPID ( s)

( K p T f K d ) s 2 ( K p Ki T f ) s Ki

(4-24)

Tf s2 s

For facilitating the derivation of the state space equations, it can be written as a secondorder transfer function as given in equation (4-24).
If we substitute the values A= K p Tf +K d ; B= K p +Ki Tf ; C=Ki; D=Tf; E=1and F=0
into equation (4-24), the general form of the second order form transfer function can be
given by

GPID ( s)

A s2 B s C
D s2 E s F

(4-25)

The state space equation for a general form second order transfer function can be given
by

F
AF

0
C

w
w
d d1
D d1
D

ud
E wd 2 AE
dt wd 2
1

B
D

1 wd 1 A

yd 0

ud
D wd 2 D

(4-26)

Hence, if by i) substituting two states of the general form of the second order transfer
function w d1 and w d2 with UdcX1 and UdcX2 ; ii) relacing u d with Udcref -Udcm and iii)
substituting y d with iqref (s) , the small signal for the DC-PID controller can be obtained as

F
AF
AF

0
C
C

d U dcX 1
D U dcX 1
D
D U dcref

U dcm
E U dcX 2 AE
AE
dt U dcX 2
1

B
B

D
D
1 U dcX 1 A
A U dcref

iqref 0

D U dcX 2 D
D U dcm

(4-27)

4.3.4.2 AC Voltage Controller Design for Inverter and its Small Signal Model
Note that the model used in the AC voltage controller design is a MIMO model by
placing the designed rectifier side power controller, AC voltage controller and the inverter
side DC voltage controller in service. This approach ensures the design is more accurate

152

4.3. CONTROLLER DESIGN METHODOLOGY

since the impacts of the dynamics of the co-operating controllers are also being taken into
account.

(a)

(b)

(c)

(d)

Figure 4-31(a) Bode plots of selected cases for inverter end AC voltage controller design
Uf_inv(s)/Idref_inv(s); (b) designed filter+PI compensator; (c) frequency responses of open
loop transfer functions of Uf_inv(s)/Uf_inv_ref(s); (d) step responses of closed-loop Uf_inv(s)/
Uf_inv_ref(s).
The same methodology is used in this controller design as proposed earlier in section
4.3.3.3 and shown in Figure 4-31a. Using the loop-shaping technique, a set of PID
parameters (Kp_Uf_inv= 0.00842, Ki_Uf_inv=0.752, Kd_Uf_inv=-8.910-5 and Tf_Uf_inv=0.018) are
selected for the PID compensator. The frequency responses of the designed AC voltage

153

CHAPTER 4: CONTROLLER DESIGN FOR VSC-HVDC CONNECTED TO A WEAK AC GRID

controller is shown in Figure 4-31b, and the compensated system is shown in Figure 4-31c.
Note that, the AC voltage loop yields a phase margin of better than 63.2o at a gain crossover frequency of 27.3 rad/s and a gain margin better than 17.8 dB.
The small signal model with respect to the AC voltage is similar to the method
explained earlier in section 4.3.4.1. Figure 4-31d shows the step test responses of the
system operating under 36 different scenarios following a 1pu step change in the reference
value of the filter bus voltage. A satisfactory performance is obtained with a maximum rise
time of 0.404s (reaching to 90% of the final value). In addition, a settling time of better
than 0.73s is obtained in the STATCOM mode in which VSC is embedded in a strong
system which is represented as an inductance in series with a resistance. Moreover, a
maximum overshoot of 7.02% is also obtained within the satisfactory range.

4.3.5 Cross-Coupling Effects Examination

So far, all the initial values of the four controllers of the converters are designed
individually without taking into account the dynamic behaviours of the integrated system.
Therefore, it is important to examine the behaviours of the main loops as well as the crosscoupling loops.
4.3.5.1 Controllers for Rectifier
Two tests are considered to examine the controllers, which are i) for the rectifier: 1%
step change in the filter bus AC voltage reference and ii) 1% step change in the power
reference.
The results given in Figure 4-32 and Figure 4-33 indicate that there is a fairly short
transient change less than 0.4s maximum in the power plots (see Figure 4-32a). Note that
1% AC voltage reference step change causes a comparable change (maximum 0.686%) in
the power plot, which might be a concern. Although the change disappears quickly
(reducing to 1/3 in 0.1s), it should be treated as a not insignificant variation. However,
considering the responses of the power reference step change, the resulted AC voltage
change is found to be very small (see Figure 4-33, b). It should be noted that the levels of
the cross-coupling effects observed in the above mentioned two step test are remarkably
different. Note that a certain cross-coupling effects are also observed between the power
and the DC voltage, but this will not significantly affect the AC grid. It was observed that
10% change in the power reference causes 2.23% increase in the DC voltage response in

154

4.3. CONTROLLER DESIGN METHODOLOGY

which half of the DC voltage step change occurred within a short time 0.5s. It can be
concluded that the cross-coupling effects between the power and the AC voltage induced
by changing the power reference are low. However, the cross-coupling effects between the
rectifier side AC voltage and the power or DC voltage caused by changing the voltage
reference are not insignificant, which is noticeable.

(a)

(b)

(c)

(d)

Figure 4-32 1% step of AC voltage reference at the rectifier end; (a) responses of power at
the rectifier end; (b) responses of the filter bus voltage Uf_rec at the rectifier end; (c)
responses of DC voltage Udc_inv at the inverter end; (d) responses of the filter bus voltage
Uf_inv at the inverter end.

155

CHAPTER 4: CONTROLLER DESIGN FOR VSC-HVDC CONNECTED TO A WEAK AC GRID

(a)

(b)

(c)

(d)

Figure 4-33 1% step of power reference at the rectifier end; (a) responses of the power at
the rectifier end ; (b) responses of the filter bus voltage Uf_rec at the rectifier end; (c)
responses of the DC voltage Udc_inv at the inverter end; (d) responses of the filter bus
voltage Uf_inv at the inverter end.

4.3.5.2 Controllers for Inverter

Two tests are also considered for the inverter side controllers, which are i) 1% step
change in the filter bus AC voltage reference and ii) 1% step change in the DC voltage
reference.
The simulation results shown in Figure 4-34 and Figure 4-35 indicate that only a small
magnitude change is observed in the induced DC step voltage during transients. Although,
it recovers slowly only the DC side is affected, which is not so significant for a point to
point VSC-HVDC transmission link. If there has to have any cross-coupling in the system,
the cross-coupling in terms of DC voltage is desirable since it is the least important

156

4.3. CONTROLLER DESIGN METHODOLOGY

coupling. Although it is not the case in a DC meshed power grid, this is beyond the scope
of this research. The peak of the DC voltage transient is found to be about 5 times less than
the AC voltage change, and the rising time is less than 0.05s. In addition, it settles down to
half of the change in less than 0.5s, which are all satisfactory.

(a)

(c)

(b)

(d)

Figure 4-34 1% Step of the AC voltage reference at the inverter end; (a) response of the
power at the rectifier end; (b) response of filter bus voltage Uf_rec at the rectifier end; (c)
response of the DC voltage Udc_inv at the inverter end and (d) response of the filter bus
voltage Uf_inv at the inverter end.

157

CHAPTER 4: CONTROLLER DESIGN FOR VSC-HVDC CONNECTED TO A WEAK AC GRID

(a)

(b)

(c)

(d)

Figure 4-35 1% step of the DC reference voltage at the rectifier side; (a) the responses of
power at the rectifier; (b) the responses of the filter bus voltage Uf_rec at the rectifier; (c) the
responses of the DC voltage Udc_inv at the inverter end; (d) the responses of the filter bus
voltage Uf_inv at the inverter end.
By observing Figure 4-35d, it is found that the 1% step change in the DC voltage
reference results in about 0.196% change in the AC voltage which is not quite satisfactory
but acceptable. However, the cross-coupling effects can be improved by further slowing
down the DC voltage controller through multiplying the controller coefficient with a gain
less than 1 (e.g. 0.8), but at cost of reducing the response speed of the DC voltage
controller. Moreover, further reducing the speed of the AC voltage controller can also
provide another alternative solution to reduce the cross-coupling effects.
From the above analysis, it can be concluded that the overall performance of the
system controllers behaviour is a trade-off between the primary controller performance

158

4.3. CONTROLLER DESIGN METHODOLOGY

and the cross-coupling effects. In other words, a better cross-coupling effect can be
achieved by slowing down the primary controller, meanwhile the performance of the
primary controllers have to be ensured as well. This requires further fine tuning of the
controller parameters by iterating the controller design procedure to get the final desirable
control system. Also, the root locus analysis and PSCAD simulation verifications can also
assist in tuning such controllers.
4.3.5.3 Analytical Model Verification against Detailed Nonlinear Simulation
It is well known that the small signal models are only accurate around a certain
operating points. Therefore the analytical model verification against detailed nonlinear
simulation (such as PSCAD) is always required.
Since 36 operating scenarios are involved in this research, the bunch file simulations
are required. However, PSCAD is only applicable for a single case simulation. The Multirun block of PSCAD is also not applicable for this particular application. Hence, a selfdefined multi-study component needs to be developed to successively read the input data
from a text file. The procedures for the PSCAD verification is given below,

Compute grid operating points and generate a text file containing these data points
in a form that is suitable as an input to PSCAD.

Create a multi-study component that will read a file from an input file, and read
the operating-point values of the system successively from the above file, and
modify grid impedance, and set control input references accordingly.

Generate a separate Matlab readable output file for each simulation.

Proceed the simulation results in Matlab for comparison of PSCAD results with the
small signal model at various step sizes, operating points and SCRs.

Performing the step tests for 36 operating points and with different step sizes, it is
found that the developed small single model is accurate enough to capture the main
characteristics of the system model developed in PSCAD. It was also observed that the
developed small signal model can be used as a powerful tool for controller design as well
as in the small signal stability analysis of the system. However, due to the space limitations,
for verification purpose, the results are only given for two operating points of the inverter: i)
PmaxQmax, SCR=2 with =90o and ii) P0Qmin, SCR=7.5 with =75o for various step size
changes of 1%, 2%, 4% and 8%.

159

CHAPTER 4: CONTROLLER DESIGN FOR VSC-HVDC CONNECTED TO A WEAK AC GRID

Figure 4-36 shows the closed-loop responses of the complete control system, under the
step size changes of 1%, 2%, 4% and 8% of the reference value of the inverter side filter
bus voltage (Figure 4-36 a) and the DC voltage (Figure 4-36b). As it can be seen in the
figure the inverter side filter bus voltage results obtained from both of the small signal and
the non-linear model match very well with each other. Although some minor errors are
observed in the cross-coupling loops at the rectifier side, the small signal model does
capture the main characteristic of the system. It can be deduced that the errors may be due
to the inaccuracy introduced during the calculation process of the steady-state value, since
the losses on both side of the converters are not included. However, the level of error is
acceptable and cannot significantly affect the controller design.

Udcinv(%)

0.5

-0.02
0
0.5

I cdrec,I cqrec(%)

0.5

0.5

0
-0.5
0

0.5
Time(Seconds)

1
(a)

160

-0.5
20
Ufinv(%)

Ufrec(%)

-0.2
0
0.02

I cdinv,I cqinv(%)

P(%)

0.2

0.5

0.5

0.5
Time(Seconds)

0
-2
0
2
0
-2
0

4.3. CONTROLLER DESIGN METHODOLOGY

0.5

0
-0.1
0
1

0.5

0
-1
0

0
-2
0
0.5

Ufinv(%)

Ufrec(%)

Udcinv(%)

0
-0.5
0
0.1

I cdrec,I cqrec(%)

I cdinv,I cqinv(%)

P(%)

0.5

0.5
1
Time(Seconds)

0.5

0.5

0.5
Time(Seconds)

0
-0.5
0
5
0
-5
0

(b)
Figure 4-36 The step responses of Ufinv_ref (a) and Udcinv_ref (b) and their induced
performances in their cross-coupling loops as the system working at an operating point
PmaxQmax, SCR=2 with =90o. red: Matlab 1%; cyan: PSCAD 1%; black: PSCAD 2%/2;
magenta: PSCAD 4%/4; green: PSCAD 8%/8.
Figure 4-37 is another step test performed in the STATCOM mode. Firstly, a 1% step
change is applied to the small signal model of the inverter side DC voltage reference,
which is then compared with four results from the detailed PSCAD non-linear model. The
testes set on PSCAD include 1%, 2%, 4% and 8% step changes on both of the AC filter
bus reference voltage and the DC reference voltage. While, the final output results are
manipulated by dividing 1, 2, 4 and 8 respectively. To do in such a way is to examine the
linearity of the PSCAD model. The results especially forcefully confirm the accuracy of
the developed small signal model.

161

CHAPTER 4: CONTROLLER DESIGN FOR VSC-HVDC CONNECTED TO A WEAK AC GRID

Udcinv(%)

P(%)

0.5
0
-0.5
0

0.5

0
-2
0

Ufinv(%)

0
-0.02
0

I cdrec,I cqrec(%)

0.5

0.5
Time(Seconds)

0.1

0.5

0
-1
0

0.5
Time(Seconds)

-0.1
0

I cdinv,I cqinv(%)

Ufrec(%)

0.02

0.5

2
0
-2
0

(a)

Udcinv(%)

x 10

0.5

0.5

0.5

0
-1
0

0.05
0
-0.05
0

-0.05
0

-3

0
-1
0

I cdrec,I cqrec(%)

0.5

Ufinv(%)

Ufrec(%)

-0.01
0

0.05

I cdinv,I cqinv(%)

P(%)

0.01

0.5
1
Time(Seconds)

10
0
-10
0

0.5
1
Time(Seconds)

(b)
Figure 4-37 The step responses of Udcinv_ref (a) and Ufinv_ref (b) and their induced
performances in their cross-coupling loops as the system working at an operating point
P0Qmin, SCR=7.5 with =75o. red: Matlab 1%; cyan: PSCAD 1%; black: PSCAD 2%/2;
magenta: PSCAD 4%/4; green: PSCAD 8%/8.

162

4.4. DISCUSSIONS AND CONCLUSIONS

4.4 Discussions and Conclusions

In summary, the main focus of this chapter is to design three different types of outer
loop controllers i) power controller, ii) AC voltage controller and iii) DC voltage controller
for a VSC-HVDC system. First of all, the causes of the failure operation of a VSC-HVDC
current controller in weak AC grids are examined and it is found that the limited
bandwidth of the system results in the main challenges of such application. Then, the
selected parameter sensitivity (such as SCRs, X/R ratios, loading conditions and power
factors) effects on the main and cross-coupling inner current loops are discussed. In
addtion, the limitations imposed by the RHP zero on the bandwidth of the outer loop
controllers are also investegated and discussed. The study and resolution of these issues
provided an in-depth understanding on the characteristics of the plant to be controlled and
facilitate a more precise controller design.
It is known that one of the drawbacks of the controller design using a simplified model
is the dynamics of the PLL and the filter bus voltage with respect to the entire system
performance are not considered. The secondary drawback is that it does not take into
account the operating points of the system. Therefore, to implicitly understand the
operation of the VSCs working in weak AC grids, it is crucial to carry out a complete
analysis using an accurate representation of the nonlinearities of the PLL, controllers and
the network. Therefore, a robust controller design methodology was firstly proposed in this
chapter by taking into account a set of operating points covering the converter operating
capability (PQ chart) and the various grid conditions in terms of different SCR and X/R
ratios of the grid reactances. In general in this chapter it is stated each part of the control
design is better if it is done with the previously designed controller in place. While, if
unsatisfactory performance were observed of the earlier designed controller when the later
controllers were implemented, it can be re-designed to obtain better performance. By going
through several iteration design process, we can definitely get controllers with improved
performance.
The initial value of the controller is obtained by applying the classical frequency
response technique to a set of open-loop transfer functions of the VSC-HVDC systems
which are obtained by changing the operating points of the developed small signal model
in Chapter 3. This technique is applied to power, AC voltage and DC voltage controller
design respectively. It should also be emphasized that it is more accurate to design a

163

CHAPTER 4: CONTROLLER DESIGN FOR VSC-HVDC CONNECTED TO A WEAK AC GRID

controller with the previous designed controller in service, since the more detailed model
employed will result in a more accurate design for the later controller. In addition, such
approach also allows the cross-coupling effects to be considered.
Finally, a self-defined block is developed in PSCAD to enable the successive PSCAD
simulations for verification purpose. The co-operating performance of the various designed
controllers are examined against the time domain verification. It is found that the designed
power and voltages controllers work very well within all the considered 36 operating
points.

164

Chapter 5: Stability of VSC-HVDC Links

Embedded with the Weak Australian Grid
This chapter evaluates the small signal stability performance of an expanded Australian
grid, in which the DB current controlled VSC-HVDC stations are embedded in parallel
with the existing longitudinal and weak Australian AC grid. In this study, it was found that
the introduction of the new source of geothermal power generation has an adverse impact
on the damping performance of the system. Therefore, this chapter aims to answer how a
VSC-HVDC system can assist in enhancing the system small-signal rotor-angle stability in
a weak multi-machine system.

5.1 Introduction
As is known, the limited availability and environmental concern of fossil fuels, as well
as the continuous growing demand of electricity, have caused renewable energy to become
commercially attractive. Meanwhile, as a result of increasing greenhouse gas emissions,
the Australian government has announced an emissions trading scheme in 2010 termed as
the Carbon Pollution Reduction Scheme [142] and setup new green energy targets to

165

CHAPTER 5: STABILITY OF VSC-HVDC LINKS EMBEDDED WITH THE WEAK AUSTRALIAN

GRID
increase the share of renewable and sustainable electricity generation from 2% in 2004-05
[143] to 20% in 2020 [142].
Among all the potential renewable energy sources, geothermal as a zero-emission, base
load renewable source of electricity, wind and solar energies as environmental friendly
resources offer attractive solutions to long term CO2-e emission reduction scheme.
However, the large-scale integration of renewable energies poses significant challenge on
the existing shared transmission systems, especially in Australia where the grid is one of
worlds

the

longest interconnected

power

systems between Port

Douglas

of

Queensland and Port Lincoln of South Australia with an end-to-end distance of more than
4000 km as shown in Appendix E [144]. In addition, the potential Paralana Geothermal
source is located very remote from the existing shared transmission network. There are
several investigations [145-147] on exploring the potential network extension solution
upon integrating the staged Innamincka generation of 500MW, 2000MW and 5000MW
[145, 146].
This chapter examines the small signal stability performance of this potential extended
Australian grid using an alternative HVDC transmission solution assuming that a 2000MW
capacity is available at Innamincka. It is assumed that the generation 200MW is consumed
locally, and the remaining 1200MW is delivered via a 500kV, 1100km, bi-pole VSCHVDC links to Armadale converter substation located at the half way between Queensland
and New South Wales. It is considered that this arrangement can make full utilization of
the existing transmission corridor, consequently benefiting both States. The remaining 600
MW are transferred to Roxby Down from Innamincka, using the 490km 275kV
transmission line, and then being integrated to the main grid at Port Augusta by a 290km
transmission line. The proposed diagram of the extended South-East Australian power
system with VSC-HVDC links is illustrated in Figure 5-1, which is obtained based on an
IEEE 14 generator 59 bus test system. The test system represents a simplified model of the
Southern and Eastern Australian network which is composed of five areas in which 14
large generators and 5 static var compensators (SVCs) are involved. The data for this
system is available in [148]. Comparing with the other possible solutions, it is considered
that this approach is a realistic option and will be investigated here.

166

5.2. PREPARATION TASKS FOR INTERCONNECTION

406
SVC

SPS_4

GPS_4

N_SVC

AREA4:QSL

407

403

404

405

CPS_4

Grid Extension for geothermal energy

408

402

409
TPS_4
606

601

602

603

410

AREA6:Innamincka

Brisbane

401

VAI Innamincka
604

VAB
411
413

201
414
412
415

ASVC_2

PL34

SVC

SVC BSVC_4

416
VPS_2

BPS_2
205
201
207

203

PL12
206
609

208
EPS_2
202
NPS_3
609 608

209

211

MPS_2

501
TPS_3

204

504

210
Sydney

502

VAS

215

212

505

213
PSVC_5

AREA2:NSW

214

SVC

AREA5:SA

507
Adelaide

216

508 VAA

506

217

503

PPS_3

509

SVC

102

SSVC_5
305

315

311

310

AREA1:Canberra

309

LPS_3

304
308

HPS_1

307

301
303

306

AREA3:VIC
YPS_3

RSVC_3
314

302

SVC
313 VAM Melbourne

312

Figure 5-1 The diagram of the extended Simplified South-East Australian power grid with
VSC-HVDC links.

5.2 Preparation Tasks for Interconnection

This study considers the VSC-HVDC small signal model developed in Chapters 3 and
4. However, for the purpose of obtaining a high accuracy of VSC controllers design, a
dynamic model is utilized to represent the interconnected simplified grid model which is

167

CHAPTER 5: STABILITY OF VSC-HVDC LINKS EMBEDDED WITH THE WEAK AUSTRALIAN

GRID
represented by a source behind impedance or with additional resistance depending on
system short circuit ratio (SCR) and X/R ratio. Nevertheless, the frequency range of
interest in this study is the modal frequency related to electromechanical oscillations which
is more likely encountered in practice i.e. from low frequency inter-area mode of 1.5 rad/s
to higher local frequencies of 12 rad/s [148]. Therefore, it is necessary to test whether it is
appropriate or not to represent the interconnected grid and PCC bus filter as an admittance
matrix within the frequency range of interest.
It should be noted that it is insufficient to employ the constant admittance
representation of grid for torsional oscillations in HVDC interactions studies, since the
frequency of interest in this application is much higher. In [149], a hybrid model was
proposed to offer a feasible solution to this problem. It is realized by including the detailed
dynamics of the transmission network and dynamic devices in adjacent area of HVDC
converters and the remanent parts are modelled as constant admittances in the modeling. It
should be noted that this approach not only ensures sufficient the accuracy, while offers the
simplicity as well. Therefore, the simplification process of the grid and filter dynamic
model will be conducted step by step in the following section to validate the above
mentioned approach.

5.2.1 Admittance Matrix Representation of the Integrated Grid

and Filter
The representation of the grid dynamic model as an admittance is shown in Figure 5-2.
For the explanation of the symbols, refer to Figure 3-1.

u gfI

u gfR

VcRg

g
I cR

REACTANCE
(Dynamic)

VSC

VcIg

g
I gR

FILTER
(Dynamic)

I cIg

I gIg

Grid
(Adimittance)

Figure 5-2 Grid model as an admittance

Hence, the grid impedance dynamic model can be given by the below set of equations,

168

5.2. PREPARATION TASKS FOR INTERCONNECTION

di g g _ R
(1/ Lg ) E g g _ R (1/ Lg )u g f _ R ( Rg / Lg )i g g _ R 0i g g _ I

dt
g
di g _ I
(1/ Lg ) E g g _ I (1/ Lg )u g f _ I ( Rg / Lg )i g g _ I 0i g g _ R

dt

(5-1)

which can be replaced with an grid admittance model and resulting in the following set
of equations,
i gg
i gg

( E g g _ R jE g g _ I ) (u g f _ R ju g f _ I )
Rg jX g
( E g g _ R u g f _ R ) Rg ( E g g _ I u g f _ I ) X g
Rg2 X g2

( E g g _ R u g f _ R ) X g ( E g g _ I u g f _ I ) Rg

(5-2)

Rg2 X g2

In the above equations (5-2), the real part of the grid current i ggR and the imaginary part
of the grid current i ggI can be expressed by the following set of equations,

i
g
gR

i
g
gI

( E g g _ R u g f _ R ) Rg ( E g g _ I u g f _ I ) X g
Rg2 X g2
( E g g _ R u g f _ R ) X g ( E g g _ I u g f _ I ) Rg

(5-3)

Rg2 X g2

After linearizing the above equations (5-3), the small signal model for the admittance
representation of the grid model can be given by

i g gR

Rg
R X

i g gI

2
g

2
g

(E g g _ R u g f _ R )

Xg
Rg2 X g2

Xg
R X

(E g g _ R u g f _ R )

2
g

2
g

(E g g _ I u g f _ I )

Rg
Rg2 X g2

(5-4)

(E g g _ I u g f _ I )

Figure 5-3 shows the simulation results of the dynamic and the admittance
representation of the grid model following a 1% step change on the inverter side input DC
voltage reference. As it can be seen in the figure, the responses of the DC voltage tracking
and the cross-coupled rectifier side power responses are similar. However, the crosscoupling voltage responses in the rectifier and the inverter side have some minor
differences which can be ignored, as highlighted by the circled areas in the graphs.

169

AC vltage-rec(%)

Power-rec(%)

CHAPTER 5: STABILITY OF VSC-HVDC LINKS EMBEDDED WITH THE WEAK AUSTRALIAN

GRID

0.5

Grid Dynamic
Grid Admittance only

0
-0.5
0

0.1

0.2

0.3

0.4

0.5
0.6
Time(Seconds)

0.7

0.05

-0.05
0

0.1

0.2

0.3

0.4

0.5
0.6
Time(Seconds)

0.5
(%)

0.9

Grid Dynamic
Grid Admittance only

Ic

dq

0.8

-0.5
0

0.1

0.2

0.3

0.4

0.5
0.6
Time(Seconds)

0.7

0.8

0.9

Grid Admittance Only Icd

Grid Admittance Only Icq
Grid and filter Admittance Icd
Grid and filter Admittance Icq
0.7
0.8
0.9
1

AC vltage(%)-inv

DC voltage-inv(%)

(a)
2

Grid Dynamic
Grid Admittance only

0
-2
0

0.1

0.2

0.3

0.4

0.5
0.6
Time(Seconds)

0.7

0.2

0.1

0.2

0.3

0.4

0.5
0.6
Time(Seconds)

Ic

dq

(%)

-2
0

0.9

Grid Dynamic
Grid Admittance only

0
-0.2
0

0.8

0.1

0.2

0.3

0.4

0.5
0.6
Time(Seconds)

0.7

0.8

0.9

Grid Admittance Only Icd-inv

Grid Admittance Only Icq-inv
Grid and filter Admittance Icd-inv
Grid and filter Admittance Icq-inv
0.7
0.8
0.9
1

(b)
Figure 5-3 1% DC voltage step responses of the dynamic and grid admittance models (a)
for rectifier side; (b) for inverter side.

170

5.2. PREPARATION TASKS FOR INTERCONNECTION

+
+
g
I gR
g
EgR
EgIg

g
gI

Grid
(Admittance)

u gfR

+
+ g
I fI

g
fR

Filter
(Admittance)

I cRg

I cIg
VSC

VcRg
VcIg

u gfI

Figure 5-4 The grid and filter model as an admittance.

Figure 5-4 illustrates the relationship of the system variables, where the grid and filter
models are represented as admittances, which can be given by the set of equation (5-5).
Firstly, the grid source supplies the filter bus voltages u gfR , u gfI to the converter based on the
first equation of the set of equations (5-5). In accordance with these voltages, the filter
generates the output currents that IgfR and I gfI according to the third equation of the set of
equation (5-5). Further, the converter output currents IgcR and IgcI can be calculated with
g
g
g
g
regards to the voltage drop across the converter reactance (i.e. u fR - VcR
and u fI - VcI ).

Finally, obtain the injected currents to the grid source I ggR and I ggI according to the second
equation of the set of equation (5-5).

u fg Egg Z g I gg
g
g
g
I g I f Ic
I g ug B j
f
f
f

(5-5)

If this set of equations (5-5) is expended and linearized, the small signal model of the
filter and grid admittance representation of the system model can be obtained as,

u g f _ R E g g _ R Rg I g g _ R X g I g g _ I
g
g
g
g
u f _ I E g _ I Rg I g _ I X g I g _ R

(5-6)

g
g
g

I g _ R I f _ R I c _ R
g
g
g
I g _ I I f _ I I c _ I

(5-7)

g
g

I f _ R u f _ I B f
g
g

I f _ I u f _ R B f

(5-8)

171

CHAPTER 5: STABILITY OF VSC-HVDC LINKS EMBEDDED WITH THE WEAK AUSTRALIAN

GRID

Power-rec(%)

0.4

Grid Admittance only

Grid,Filter Admittance

0.2
0
-0.2
-0.4

0.2

0.4
0.6
Time(Seconds)

0.8

AC vltage-rec(%)

0.06
Grid Admittance only
Grid,Filter Admittance

0.04
0.02
0
-0.02

0.2

0.4

0.4
0.6
Time(Seconds)

0.8

Ic (%)

0.2

dq

0
Grid Admittance Only Icd
Grid Admittance Only Icq
Grid and filter Admittance Icd
Grid and filter Admittance Icq

-0.2
-0.4

0.2

0.4
0.6
Time(Seconds)

0.8

DC voltage-inv(%)

(a)

1.5
1
0.5
0
-0.5

Grid Admittance only

Grid,Filter Admittance
0

0.2

0.4
0.6
Time(Seconds)

0.8

AC vltage(%)-inv

0.15
0.1
0.05
0
Grid Admittance only
Grid,Filter Admittance

-0.05
-0.1

0.2

0.5

0.4
0.6
Time(Seconds)

0.8

dq

Ic (%)

0
Grid Admittance Only Icd-inv
Grid Admittance Only Icq-inv
Grid and filter Admittance Icd-inv
Grid and filter Admittance Icq-inv

-0.5
-1
-1.5

0.2

0.4
0.6
Time(Seconds)

0.8

(b)
Figure 5-5 1% DC voltage step response comparison of the admittance representation of
grid model only, and the gird and filter models adopting admittance representation: (a) the
rectifier side; (b) the inverter side.

172

5.2. PREPARATION TASKS FOR INTERCONNECTION

The test results presented in Figure 5-5 reveal that the further simplification of the filter
model does not weaken the accuracy of the results in this application. Hence, the
admittance representation of both the grid and the filter models will be set as the standard
modelling methodology in the following analysis.

5.2.2 Scaling the Existing System

The model developed in chapters 3 and 4 was in SI system. This section will show how
to scale the existing system to make it applicable to any system capacity and rated voltage
level. Figure 5-6 shows the procedure of scaling the system parameters. The main idea of
the scaling is to transfer the original system to per-unit values first, which is then followed
by backing to the new international system of units (SI unit) system by employing equation
(5-9). Therefore, by adopting this methodology, the established VSC-HVDC link model
which is in a particular rating system is able to remain unchanged.

x ( new)

x ( original ) ( new)
xbase
( original )
xbase

Model Parameters
(Original)

(5-9)

New Rating

Parameter Conversion

Model Parameters
(New in SI units)
Model Equations
in SI units

Figure 5-6 System parameter scaling scheme

In the following paragraphs, how to apply the scaling technique to all kinds of
participated controllers are discussed in detail.
5.2.2.1 Inner Current Controller Coefficient
The scaling of the inner current controller coefficients brought one question that
whether they are still correct in comparison with the directly calculated coefficient from
equation (2-48).

173

CHAPTER 5: STABILITY OF VSC-HVDC LINKS EMBEDDED WITH THE WEAK AUSTRALIAN

GRID
A. Method 1-Scaling
Reference to equation (2-48), Figure 5-7 demonstrates how to scale a system to a new
SI system, where N1(p) is the co-efficient of the inner current controller in per-unit system.
It should be noted that all the scaling systems should have exactly the same coefficient N1(p) .
Therefore, using the calculated values of N1(p) based on two different rating systems i)
original system and ii) new system (see equations (5-10) and (5-11)), the co-efficient of the
new rating system can be easily obtained using equation (5-12).

I ( SI )

I (P)

V (P)

N1( p )

I base

Vbase

V ( SI )

Figure 5-7 Scaling of inner current controllers

I base1
N1( sys1) / Zbase1
Vbase1

(5-10)

Ibase 2
N1( sys 2) / Zbase 2
Vbase 2

(5-11)

N1( P ) N1( sys1)

N1( P ) N1( sys 2)

N1( sys 2) N1( sys1)

Zbase 2
Zbase1

(5-12)

B. Method 2-Directly Calculating from Equation (2-48)

This method can be demonstrated using the following steps, where the superscript (1)
represents the original system, and the superscript (2) represents the new system, and the
superscript (P) indicates per units system.
Firstly, calculate the key coefficient N1 and the time constant T of N2 for both of the
original system and new updated system,

(1)
R (1)
N1 1 N (1) ;

(2)
N (2) R
;
1
1 N 2(2)

Set T

(1)

Set T

(2)

R (1)
( (1) );
L
R (2)
( (2) );
L

(5-13)

Then, calculate the per unit value of coefficients R(p), L(p) and T (p) of the systems.

174

5.2. PREPARATION TASKS FOR INTERCONNECTION

( P ) R (1)
L(1)
; L( P )
R
Rb1
Lb1

(1)
(1)
T ( P ) R / Rb1 R Lb1

L(1) / Lb1 L(1) Rb1

(5-14)

Then calculate the updated coefficients R(2) and L(2) of the new system
(2)
( P)
(1) Rb 2
R R Rb 2 R R ;

b1

L(2) L( P ) L L(1) Lb 2 ;
b2

Lb1

(5-15)

Finally, It can be obtained that the coefficient N2 for both systems are the same by
calculating the time constant T of N2 and further to calculate N2,
Rb 2
Rb1
R
R (1)
( (2) )
( (1) ) T (1)
L
L
L
L(1) b 2
Lb1
R (1)

(2)

T (2)

(5-16)

(1)
Therefore, it can be derived that N(2)
2 =N 2 . Then using the calculated value of N2, we

can obtain.

N1(2)

(2)

R
1 N 2(2)

N1(2) N1(1)

Rb 2
)
Rb1

;
1 N 2(2)
R (1) (

Rb 2
Z
N1( sys1) base 2
Rb1
Zbase1

(5-17)
(5-18)

From the above analysis, it can be concluded that the DB controller coefficient directly
calculated using the equation (2-48) should be the same as the scaled value using the
original system.
5.2.2.2 PLL Controller Coefficient
The PLL controller coefficient can be modified by multiplying by Vb, as given in the
below equation (5-19).

Y /U
Y /U

SI

pu

/ VSI

/ b
VSI / Vb

Vb / VSI

(5-19)

where, the base value of the angle b is equal to 1.

175

CHAPTER 5: STABILITY OF VSC-HVDC LINKS EMBEDDED WITH THE WEAK AUSTRALIAN

GRID
5.2.2.3 Coefficients of the Power and the AC voltage Controller
For the type of PI with an additional filter controller as shown in Figure 5-8, there is no
need for any change in the coefficient Tf since it is simply a time constant. However, the
unit of Kp in the power controller is kA/MW, and kA/MW/s in Ki. Therefore, the new
system PI coefficients can be given by equations (5-20) and (5-21) using equation (5-12).
Similarly, by replacing the Pbase with Vbase, the PI coefficients for the AC voltage controller
loop can be easily obtained.

K p(2) K p(1) (

Pbase1 I base 2
)(
)
Pbase 2
I base1

(5-20)

Ki(2) Ki(1) (

Pbase1 I base 2
)(
)
Pbase 2
I base1

(5-21)

Kp

Pe( MW ) / U (f kV )

1
1 Tf s

+
+

Ki
s

( kA )
( kA )
I qref
/ I dref

Figure 5-8 Scaling of PI controller with additional filter

5.2.2.4 Coefficient of the DC Voltage Controller

As proposed in Chapter 4, a PID controller is designed for the DC voltage controller,
which is composed of 3 parts (proportional, integral and derivative) as shown in Figure 5-9.

Kp

U dc( kV )

Ki
s
Kd s
1 Tf s

( kA )
I qref

Figure 5-9 Scaling of the PID controller

In order to obtain the new PID coefficients for the DC voltage loop, it is necessary to
derive the units for each individual part as shown in Table 5-1.

176

5.2. PREPARATION TASKS FOR INTERCONNECTION

Table 5-1 The units derivation for the PID coefficients

Proportional part:
Udc Kp=Iqref Kp= Iqref/Udc=kA/kV

Integral part

Udc Ki/s=Iqref

Udc Ki dt=d Iqref

Udc Ki t=Iqref
Ki=kA/(kV )

Udc Kd/(1+Tf s)=Iqref

Derivative part :

Udc Kd= Iqref + Tf

Iqref/ dt)

dt Udc Kd=dt Iqref + Tf

Iqref

Kd=(Iqref/ Udc) ((t+ Tf)/t)

Kd=kA/kV

Therefore, according to Table 5-1, the new PID coefficients for the DC voltage loop
can be given by

K p(2) K p(1) (

Vdc _ base1
Vdc _ base 2

Ki(2) Ki(1) (
K d(2) K d(1) (

Vdc _ base1
Vdc _ base 2
Vdc _ base1
Vdc _ base 2

)(

I base 2
)
I base1

(5-22)

)(

I base 2
)
I base1

(5-23)

)(

I base 2
)
I base1

(5-24)

5.2.2.5 Description of the Methodology to Verify the Accuracy of the Model

Figure 5-10 shows the procedure to verify the accuracy of the system scaling technique,
and Figure 5-11 shows the corresponding results. Note that the results confirm that all of
four systems have the same eigenvalues, which can also be used to explain the results
given earlier in section 5.2.1that Figure 5-3 and Figure 5-5.
Original admittance model-VS-Original dynamic model

Original admittance model-VS-Scaled admittance model

Scaled admittance model-VS-Scaled dynamic model

All Model has to have consistent eigen-values

Figure 5-10 Flow chart of the verification methodology of modeling

177

CHAPTER 5: STABILITY OF VSC-HVDC LINKS EMBEDDED WITH THE WEAK AUSTRALIAN

GRID

250

Original admittance model

Original dynamic model
Scaled admittance model
Scaled dynamic model

200

Im

150

100

50

0
-300

-250

-200

-150
Re

-100

-50

Figure 5-11 Eigenvalue map of the original and scaled systems

It should be noted that the established VSC-HVDC system is still in SI unit, but the
interconnected grid model is in per-unit. Therefore, there is a need for a unit transformation
block at the integrating interface. This means that the input bus voltages to the VSC links
have to be multiplied by the rated voltage base value Vbase. For the same reason, the output
grid currents from the VSC link have to be divided by the rated current base value Ibase to
transform the original SI unit system to the new per-unit system. In addition, a special
attention has to be paid to identify the standard direction of the systems to be
interconnected, which have to be consistent.

5.2.3 DC Link Parameter Sensitivity

As demonstrated in the previous section, the established model can be scaled to any
rating system as required. However, it is important to note that the DC link parameters are
not be able to be scaled since it is physically present, which create a question about the
robustness and sensitivity of the designed controllers in chapter 4.
The DC link parameters utilized in this chapter are obtained from the project titled The
GezhoubaNan Qiao HVDC Project which is a 500kV 1046km long Line Commutate
Converter (LCC)-HVDC transmission system with a capacity of 1200MW [150]. The
system data is available in [151], where R Link =26, LLink =0.46H, Clink =13.73F for a
single line. Although the voltage rating 500kV is for conventional LCC-HVDC scheme,
but the voltage rating for VSC-HVDC system is gradually progressing presently. As
reported

178

in

[16],

the

highest

pole

voltage

in

recorded

operation

was

5.3. INTEGRATING WITH THE SIMPLIFIED AUSTRALIAN GRID

350 kV (Caprivi Link, Namibia)

and the highest transmission voltage was 500kV

(Skagerrak 4, Denmark Norway). Therefore, the selection of the DC link parameters in this
research is considered realistic.
Figure 5-12 shows the eigenvalue map of the system with a new series of updated DC
link parameters. It can be seen from Figure 5-12 that there is significant changes in a pair
of modes: 41.83 rad/s moves to 84.91rad/s and 184 rad/s moves towards 569.2 rad/s.
Therefore, it is important to determine whether this change will affect the efficiency of the
designed controllers.
To investigate this issue, a review on performance evaluation in terms of all the power
and voltage controllers with a new series of DC link parameters are also undertaken, and
the simulation results are given in Appendix F, which are then being compared with the
results in Chapter 4. It was concluded that no significant effects are observed on the
controllers within the frequency range considered. As it was also observed in the power
and the DC voltage control loops, the pattern of the frequency responses is much more
consistent due to the changes in the eigenvalues. It can be noted that this change decreases
the difficulties of controller design for these VSC-HVDC links embedded in a weak AC
system. Hence it can be concluded that the robustness and sensitivity to the DC link
parameters of the designed controllers are acceptable.
600

New Updated System

Original System

400

Im

200

-200

-400

-600
-100

-80

-60

-40

-20

Re

Figure 5-12 The comparison of the system eigenvalues between the new updated DC link
system and the original system

5.3 Integrating with the Simplified Australian Grid

In this sub-section, the small signal stability performance of the extended simplified

179

CHAPTER 5: STABILITY OF VSC-HVDC LINKS EMBEDDED WITH THE WEAK AUSTRALIAN

GRID
Australian grid which is operating in parallel with the DB controlled VSC-HVDC
links is examined. The integration of VSC-HVDC links with the simplified Australian grid
is done by employing the block interconnection technique developed in section 3.4.2
according to the block diagram presented in Figure 5-4 of which the inputs is provided by
the IEEE 14 generator 59 bus test system as mentioned before in section 5.1.

5.3.1 Accuracy Evaluation of Designing VSC Controllers Based

on Simplified Grid Admittance Model
Prior to the investigation of the small signal stability of the system, the effectiveness of
designing the VSC controllers based on the simplified grid admittance model is evaluated
by comparing the performance of the Australian model and the simplified grid admittance
model. Figure 5-13a,b illustrate the frequency responses of

the transfer functions of

Vr(s)/Ir(s), Vi(s)/Ir(s), Vr(s)/Ii(s) and Vi(s)/Ii(s) for both of the rectifier side at Innamincka
and the inverter side at Armadale. Note that if the source is represented by a constant
voltage behind source impedance (Zs=Rs+jXs), then the above transfer functions will have
the following four forms: Vr(s)/Ir(s) =Rs, Vi(s)/Ir(s) =Xs, Vr(s)/Ii(s) =-Xs and Vi(s)/Ii(s) =Rs.
This confirms the results given in Figure 5-13a (Rs) to be the high frequency responses of
the system analyzed. Further analysis of Figure 5-13a also demonstrates the resistance and
inductance values at the inverter side,

Rs 10( 42.5/20) 0.0075 pu M

X s 10( 28/20) 0.0398 pu M

(5-25)

base 100 MW

(5-26)

base 100 MW

Similarly, at the rectifier side, these values are

Rs 10( 63/20) 0.0007 pu M

(5-27)

base 100 MW

X s 10( 32.5/20) 0.0237 pu M

(5-28)

base 100 MW

However, at very low and at intermediate frequencies, the source cannot be represented
by a voltage behind impedance. For example, at very low frequencies at the inverter side,
the results of the transfer functions are,

Vr (s) / I r (s) 10( 12.5/20) 0.2371 pu M

Vi (s) / I r (s) 10( 13/20) 0.2239 pu M

180

base 100 MW

base 100 MW

(5-29)
(5-30)

5.3. INTEGRATING WITH THE SIMPLIFIED AUSTRALIAN GRID

Vr (s) / Ii (s) 10(2.5/20) 1.3335 pu M

Vi (s) / Ii (s) 10(0/20) 1 pu M

base 100 MW

base 100 MW

(5-31)
(5-32)

As shown above Vr (s)/Ir (s) is not equal to Vi (s)/Ii (s) , and Vi (s)/Ir (s) is not equal to
-Vr (s)/Ii (s) . This indicates that the VSC controllers adopted in chapter 4 are in fact to

represent the source by a voltage behind equivalent impedance at high frequencies.

Therefore, it is necessary to examine whether this representation is adequate in the
controller design.

(a)

(b)
Figure 5-13 The frequency responses for the rectifier side (a) and for inverter side (b) for
the higher order grid impedance models (I to V).

181

CHAPTER 5: STABILITY OF VSC-HVDC LINKS EMBEDDED WITH THE WEAK AUSTRALIAN

GRID
The following eight figures from Figure 5-14 to Figure 5-21 present the frequency
responses of the open loop transfer functions of all four controllers using the equivalent
high frequency grid admittance model, and their corresponding step responses. The
following two primary conclusions can be drawn from these results:
1) In the active power and DC voltage loops (Figure 5-14 and Figure 5-18), no change
was observed in the major loop using different grid models. In the cross-coupling AC
voltage loops, however discrepancies are observed on the transient magnitude change
and the oscillation frequencies, but can be ignored in this application.
2) In the AC voltage loop results at both ends (Figure 5-16 and Figure 5-20), the step
response time of the voltage of the higher order grid admittance model appears not as
fast as the equivalent simplified grid admittance model. This might be caused by the
interaction between the VSC-HVDC voltage controllers and the nearby SVCs.
Although there are adverse impacts on the cross-coupling control loops as in the above
conclusion, however, this is assumed to be acceptable since the level of the observed
distinctions is small and can be ignored reasonably.
In summary, using the frequency responses of the open loop transfer function of the
target system with all the controllers in service, it can be concluded that the frequency
responses differ slightly at very low frequency range. This confirms to represent the grid as
a source behind impedance is appropriate.

Magnitude(dB)

200
100
0
-100
-200
-300 -4
10

-2

10

10

10

10

10

Phase(deg)

-100
-200
-300
-400
-500 -4
10

Grid simplified
Grid detailed
-2

10

10
10
Frequency (rad/sec)

10

10

Figure 5-14 The open loop frequency responses, P_rec/P_ref

182

5.3. INTEGRATING WITH THE SIMPLIFIED AUSTRALIAN GRID

0.15
Inverter side DC voltage(%)

Simplified
Detailed

0.5

-0.5
0
0.01

0
-0.01
-0.02
-0.03
0

2
3
Time(Seconds)

Simplified
Detailed
4
5

Inverter side AC voltage(%)

Rectifier side AC voltage (%)

Rectifier side Power(%)

1.5

Simplified
Detailed

0.1

0.05

-0.05
0
0.01

0
-0.01
-0.02
-0.03
-0.04
0

2
3
Time(Seconds)

Simplified
Detailed
4
5

Figure 5-15 The step responses test of the power reference at the rectifier side

Magnitude(dB)

200
0
-200
-400 -4
10

-2

10

10

10

10

10

Phase(deg)

0
-200
-400
-600 -4
10

Grid simplified
Grid detailed
-2

10

10
10
Frequency (rad/sec)

10

10

Figure 5-16 The open loop frequency responses, Uf_ref_rec/Uf_rec

183

Simplified
Detailed

0.6
0.4
0.2
0
-0.2
0
1.5

5
Inverter side AC voltage(%)

Rectifier side AC voltage (%)

Rectifier side Power(%)

0.8

Inverter side DC voltage(%)

CHAPTER 5: STABILITY OF VSC-HVDC LINKS EMBEDDED WITH THE WEAK AUSTRALIAN

GRID

Simplified
Detailed

1
0.5
0
-0.5
0

2
3
Time(Seconds)

0.15

Simplified
Detailed

0.1
0.05
0
-0.05
0
0.04

Simplified
Detailed

0.02
0
-0.02
-0.04
0

2
3
Time(Seconds)

Figure 5-17 The step response test of the AC voltage reference at the rectifier side

Magnitude(dB)

200
100
0
-100
-200 -4
10

-2

10

10

10

10

10

Phase(deg)

0
-100
-200
-300
-400 -4
10

Grid simplified
Grid detailed
-2

10

10
10
Frequency (rad/sec)

10

10

Figure 5-18 The open loop frequency responses, U_dc_ref_rec/U_dc _rec

184

Simplified
Detailed

0.1
0.05
0

Rectifier side AC voltage (%)

-0.05
0

0.5

0.01

Simplified
Detailed

0.005
0
-0.005
-0.01
0

0.5
Time(Seconds)

0.05

Simplified
Detailed

0
-0.05
-0.1
-0.15
0

Inverter side AC voltage(%)

Rectifier side Power(%)

0.15

Inverter side DC voltage(%)

5.3. INTEGRATING WITH THE SIMPLIFIED AUSTRALIAN GRID

0.5

1.5

Simplified
Detailed

1
0.5
0
-0.5
0

0.5
Time(Seconds)

Figure 5-19 The step response test of the DC voltage reference at the inverter side

Magnitude(dB)

200
100
0
-100
-200
-300 -4
10

-2

10

10

10

10

10

Phase(deg)

-100
-200
-300
-400
-500 -4
10

Grid simplified
Grid detailed
-2

10

10
10
Frequency (rad/sec)

10

10

Figure 5-20 The open loop frequency responses, Uf_ref_inv/Uf_inv

185

CHAPTER 5: STABILITY OF VSC-HVDC LINKS EMBEDDED WITH THE WEAK AUSTRALIAN

GRID

Inverter side DC voltage(%)

0.1

0.05

-0.05
0

0.01
Rectifier side AC voltage (%)

0.05

Simplified
Detailed

0.005

-0.005

2
3
Time(Seconds)

-0.05
-0.1

-0.15
0

1.5

Simplified
Detailed

-0.01
0

Simplified
Detailed

Inverter side AC voltage(%)

Rectifier side Power(%)

0.15

Simplified
Detailed

0.5

-0.5
0

2
3
Time(Seconds)

Figure 5-21 The step response of the AC voltage reference at the inverter side

5.3.2 Stability Analysis

Table 5-2 summarizes the inter-area modes of the interconnected model, in which
fourteen generators are equipped with the power system stabilizers (PSSs) with an
equivalent damping gain of 10 per unit, while only the three Innamincka machines are
assumed to operate without PSSs.
Table 5-2 Inter-area modes of the integrating system
Case: Three Generators (Innamincka) with PSSs out of service
Mode

Real

Imag

Damping Ratio

Mode Description

I 40

-0.286

3.907

0.073

Innamincka & VIC & Canberra VS SA

I 35

-0.290

3.417

0.085

Innamincka VS VIC& SA

I 25

-0.242

2.273

0.106

I 15

-0.154

1.703

0.090

Innamincka & SA& QLD VS

VIC &NSW& Canberra
QLD VS Innamincka & SA& VIC

The eigenvalue map of the main electro mechanical modes of the interconnected
system is given in Figure 5-22, which is obtained by using Matlab m-file. The main inter-

186

5.3. INTEGRATING WITH THE SIMPLIFIED AUSTRALIAN GRID

area modes of I40, I35, I25, and I15 are found to be lightly damped in the figure. The
damping criterion used here is chosen according to the Australian National Electricity
Rules which require the halving time of the least damped electro-mechanical mode of
oscillation is not more than five seconds (1/T ln(2)/5) [144], which corresponds to a
damping constant with a magnitude greater than -0.14 Nepers/Second (Np/s).
11

Damping Criterion
10

-0.14Np/s

9
8

Im

7
6
5
4

I40
I35

3
I25

2
1
-1.5

I15
-1

-0.5

Re

Figure 5-22 Eigenvalue map of the interconnected system

A. Mode I40 Eigen Vector and Participation Factor Analysis
In the following of this paragraph, mode I40 associated figure (right eigenvector) is
presented to indicate the energy exchange of the power system considered (see Figure
5-23). In the figure, the circle represents the phasor diagram of associated generators (such
as ING1, ING2, ING3, YPS_3, LPS_3, HPS_1, NPS_5, TPS_5 and PPS_5). For the
representations of the symbols, refer to Figure 5-1. There are nine participating generators
which clearly form two groups whose phasor differ by 180o. The consequence is if
oscillations were observed in one group, similar oscillations would be observed in the antiphase group as well. For this particular mode (I40), it is implied that the Innamincka
machines, the Victoria machines and the Canberra machine together oscillate against the
South Australia machines.

187

CHAPTER 5: STABILITY OF VSC-HVDC LINKS EMBEDDED WITH THE WEAK AUSTRALIAN

GRID
Extended Right Eigenvector Prototype
15 Jul 2013 - 18:38
LF_CASE01_R3_S 14-GENERATOR, SIMPLIFIED SYSTEM MODEL.
AREA4->AREA2->AREA3->AREA5 500-1000-500 MW.
Speed eigenvector components for I40 mode
Mag
Ph()
0.990 164.58
0.990 164.58
0.990 164.58

VAR
W
.ING3
W
.ING1
W
.ING2

0.472 -164.51 W
0.432 -162.87 W
0.128 -147.06 W

.YPS_3
.LPS_3
.HPS_1

CASE
(A)
(A)
(A)

CASE
(A)
(A)
(A)

VAR
W
.NPS_5
W
.TPS_5
W
.PPS_5

Mag
Ph()
0.835
9.34
0.815
4.18
1.000
0.00

(A)
(A)
(A)

(A)-0.286 +3.907 i

Figure 5-23 Right eigenvector prototype of mode I40

It can be easily seen from the participation factor plot (see Figure 5-24) that the main
participants are the rotor speed angles and the rotor speed frequencies of generators. The
most significant participants are the generators located at SA and Victoria with a value of
ranging from 15.5% to 8.68%, which are followed by the generators of Innamincka group
where the participation factors are equal to 0.0553.
The expected left-shift by fitting a PSS to a particular generator in a multi-machine
power system at a selected mode can be predicted by the following formula [99],
DPSS (

p
2H

(5-33)

where, DPSS represents damping gain of the PSS on the machine rating, p is the
magnitude of the participation of the machine rotor speed in the selected electromechanical mode and H is the inertia constant of the machine on the machine rating.

188

5.3. INTEGRATING WITH THE SIMPLIFIED AUSTRALIAN GRID

Therefore, by assuming DPSS to be equal to 1, the approximated left-shift damping ratio

by introducing additional PSS at Innamincka machines can be calculated as
3 (

0.0553
) 0.0259
2 3.2

(5-34)

State Variable

Participation factor for Mode:-0.28616 +/- 3.9073 j

DEL .PPS_5
W
.PPS_5
DEL .LPS_3
W
.LPS_3
DEL .TPS_5
W
.TPS_5
W
.ING2
DEL .ING2
DEL .ING1
W
.ING1
DEL .ING3
W
.ING3
DEL .NPS_5
W
.NPS_5
DEL .YPS_3
W
.YPS_3
x081.LPS_3
Eq' .LPS_3
Ed' .PPS_5
Eq' .NPS_5
Eq' .TPS_5
Eq' .YPS_3
EF .TPS_5
Eq' .PPS_5
x031.TPS_5
x091.TPS_5
x111.NPS_5
x112.NPS_5
x041.NPS_5
0

0.02

0.04

0.06 0.08
0.1
0.12
Participation factor

0.14

0.16

Figure 5-24 Participation factor for mode I40

B. Mode I35 Eigen-Vector and Participation Factor Analysis
The right eigenvector prototype of Mode I35 (see Figure 5-25) shows that the machines
at Innamincka oscillate against with the machines in the region of Victoria and South
Australia.
As suggested in the participation factor analysis plot of the mode I35 (see Figure 5-26),
the main participants are the two machines locating at Victoria, LPS and YPS, which
account for a factor of 23.6% and 7.17% respectively. The participation factors of the
Innamincka machines are 5.6% each, followed by the machines BPS at New South Wales
and PPS at the region of South Australia.
The approximate left-shift damping ratio by introducing additional PSS at Innamincka
machines is
3 (

0.056
) -0.026
2 3.2

(5-35)

189

CHAPTER 5: STABILITY OF VSC-HVDC LINKS EMBEDDED WITH THE WEAK AUSTRALIAN

GRID
Extended Right Eigenvector Prototype
15 Jul 2013 - 18:28
LF_CASE01_R3_S 14-GENERATOR, SIMPLIFIED SYSTEM MODEL.
AREA4->AREA2->AREA3->AREA5 500-1000-500 MW.
Speed eigenvector components for I35 mode
Mag
Ph()
0.632 146.49
0.648 146.62
0.173 175.10

VAR
W
.LPS_3
W
.YPS_3
W
.HPS_1

0.467
0.404
0.468
0.122
0.172
0.166

W
W
W
W
W
W

-138.14
-136.55
-123.08
-110.61
-100.43
-95.23

.PPS_5
.TPS_5
.NPS_5
.MPS_2
.BPS_2
.VPS_2

CASE
(A)
(A)
(A)

CASE
(A)
(A)
(A)

(A)
(A)
(A)
(A)
(A)
(A)

VAR
W
.ING2
W
.ING1
W
.ING3

(A) W

.EPS_2

Mag
Ph()
1.000
0.00
1.000
0.00
1.000
0.00

0.169

-89.07

(A)-0.290 +3.417 i

Figure 5-25 Right eigenvector prototype of mode I35

State Variable

Participation factor for Mode:-0.28986 +/- 3.4173 j

DEL .LPS_3
W
.LPS_3
DEL .YPS_3
W
.YPS_3
Eq' .LPS_3
x081.LPS_3
W
.ING3
DEL .ING3
DEL .ING1
W
.ING1
DEL .ING2
W
.ING2
DEL .BPS_2
DEL .PPS_5
W
.BPS_2
W
.PPS_5
Eq' .YPS_3
DEL .MPS_2
Eq' .HPS_1
DEL .EPS_2
DEL .TPS_5
W
.MPS_2
W
.TPS_5
W
.EPS_2
x121.YPS_3
EF .YPS_3
DEL .VPS_2
x082.LPS_3
x122.YPS_3
Eq' .MPS_2
W
.VPS_2
DEL .NPS_5
W
.NPS_5
x021.LPS_3
DEL .HPS_1
SV1 .RSVC_3
Eq' .NPS_5
0

0.05

0.1
0.15
0.2
0.25
Participation factor

0.3

Figure 5-26 Participation factor for mode I35

190

0.35

5.3. INTEGRATING WITH THE SIMPLIFIED AUSTRALIAN GRID

C. Mode I25 Eigenvector and Participation Factor Analysis

It can be seen from Figure 5-27 that the machines at Innamincka, South Australia and
Queensland oscillate against the machines sitting in Victoria, New South Wales and
Canberra.
Extended Right Eigenvector Prototype
15 Jul 2013 - 18:20
LF_CASE01_R3_S 14-GENERATOR, SIMPLIFIED SYSTEM MODEL.
AREA4->AREA2->AREA3->AREA5 500-1000-500 MW.
Speed eigenvector components for I25 mode
Mag
Ph()
VAR
0.310 177.64 W
.BPS_2
0.312 178.21 W
.EPS_2

0.334
0.300
0.339
0.370
0.341

-179.40
-178.95
-174.12
-166.95
-164.17

W
W
W
W
W

.MPS_2
.VPS_2
.HPS_1
.YPS_3
.LPS_3

CASE
(A)
(A)

CASE VAR
(A) W
.ING2
(A) W
.ING3

(A)
(A)
(A)
(A)
(A)
(A)
(A)
(A)

(A)
(A)
(A)
(A)
(A)

W
W
W
W
W
W
W
W

.ING1
.NPS_5
.CPS_4
.TPS_5
.PPS_5
.TPS_4
.GPS_4
.SPS_4

Mag
Ph()
1.000
0.00
1.000
0.00

1.000
0.617
0.628
0.515
0.569
0.438
0.658
0.662

-0.00
-17.57
-19.39
-20.59
-20.96
-22.98
-23.52
-24.44

(A)-0.242 +2.273 i

Figure 5-27 Right eigenvector prototype of mode I25

By conducting participation factor analysis (see Figure 5-28), it is suggested that the
important affecting factor can be classified into three categories. Firstly, the participant
GPS and SPS machines at Queensland gain the most priority which account for 13.4% and
7.59% respectively. Secondly, the machines CPS and TPS located at Queensland and the
machines MPS, BPS and EPS located in New South Wales take over weights ranging from
4.47% to 2.4%. Finally, the machines at Innamincka count 2.05% each.
The approximate left-shift damping ratio by introducing additional PSS to the
Innamincka machines is
3 (

0.0205
) -0.0096
2 3.2

(5-36)

191

CHAPTER 5: STABILITY OF VSC-HVDC LINKS EMBEDDED WITH THE WEAK AUSTRALIAN

GRID

State Variable

Participation factor for Mode:-0.24248 +/- 2.2725 j

DEL .GPS_4
W
.GPS_4
DEL .SPS_4
W
.SPS_4
DEL .CPS_4
DEL .MPS_2
W
.CPS_4
W
.MPS_2
DEL .BPS_2
DEL .TPS_4
W
.BPS_2
DEL .LPS_3
W
.TPS_4
DEL .EPS_2
W
.LPS_3
W
.EPS_2
DEL .HPS_1
DEL .ING1
W
.ING1
W
.ING2
W
.ING3
DEL .ING3
DEL .ING2
DEL .VPS_2
DEL .PPS_5
W
.PPS_5
W
.VPS_2
Eq' .EPS_2
Eq' .HPS_1
DEL .TPS_5
x081.HPS_1
W
.TPS_5
W
.HPS_1
0

0.02

0.04

0.06 0.08
0.1
0.12
Participation factor

0.14

0.16

Figure 5-28 Participation factor for mode I25

D. Mode I15 Eigenvector and Participation Factor Analysis
It is important to note that the Mode I15 is the least damped inter-area mode in this
study. Figure 5-29 indicates that the machines in Queensland oscillate against with that in
Innamincka, SA and Victoria. In addition, the participation factor analysis results as shown
in Figure 5-30 also suggest that the important affecting factor can be separated into three
categories. First priority gives to the machines at SA approximately 7.58%-10.7%,
followed by the Innamincka machines about 6.71% each. The final group involves the NPS
machine in SA, and the GPS machine and the SPS machine in Queensland and the LPS
machine in Victoria ranging from 5.05% to 2.29%.
The potential left-shift damping ratio by introducing PSSs to the Innamincka machines
can be calculated as

3 (

192

0.0671
) -0.031453
2 3.2

(5-37)

5.3. INTEGRATING WITH THE SIMPLIFIED AUSTRALIAN GRID

Extended Right Eigenvector Prototype

15 Jul 2013 - 18:15
LF_CASE01_R3_S 14-GENERATOR, SIMPLIFIED SYSTEM MODEL.
AREA4->AREA2->AREA3->AREA5 500-1000-500 MW.
Speed eigenvector components for I15 mode
Mag
Ph()
0.186
95.43
0.185
96.12
0.149
97.83
0.180
98.78

VAR
W
W
W
W

.SPS_4
.GPS_4
.TPS_4
.CPS_4

CASE
(A)
(A)
(A)
(A)

CASE VAR
(A) W
.ING3

(A)
(A)
(A)
(A)
(A)
(A)
(A)

W
W
W
W
W
W
W

.ING1
.ING2
.NPS_5
.PPS_5
.TPS_5
.LPS_3
.YPS_3

Mag
Ph()
1.000
0.00

1.000
1.000
0.777
0.776
0.729
0.224
0.221

-0.00
-0.00
-6.24
-8.26
-8.37
-27.44
-31.87

(A)-0.154 +1.703 i

Figure 5-29 Right eigenvector prototype of mode I15

State Variable

Participation factor for Mode:-0.15417 +/- 1.7033 j

DEL .PPS_5
W
.PPS_5
DEL .TPS_5
W
.TPS_5
DEL .ING3
W
.ING3
W
.ING2
DEL .ING1
W
.ING1
DEL .ING2
DEL .NPS_5
W
.NPS_5
DEL .GPS_4
DEL .LPS_3
W
.GPS_4
W
.LPS_3
DEL .SPS_4
W
.SPS_4
Eq' .YPS_3
DEL .TPS_4
Eq' .LPS_3
DEL .CPS_4
W
.CPS_4
Eq' .TPS_5
W
.TPS_4
DEL .YPS_3
W
.YPS_3
x091.TPS_5
0

0.02

0.04
0.06
0.08
Participation factor

0.1

0.12

Figure 5-30 Participation factor for mode I15

193

CHAPTER 5: STABILITY OF VSC-HVDC LINKS EMBEDDED WITH THE WEAK AUSTRALIAN

GRID

5.3.3 Damping Controllers Design

The equivalent damping gain of the PSSs of the three Innamincka machines are set to
be increased from zero (no PSSs in service) to 30 pu in 5 pu steps (see Figure 5-31). It is
obvious that the supplementary PSSs equivalence fitted to the Innamincka machines can
enhance the system damping performance, since all of the inter-area modes are left-shifted.
11

0p.u. A
5p.u. B
10p.u. C
15p.u. D
20p.u. E
25p.u. F
30p.u. G

10
9
8

Im

7
6
5
4

I40
I35

I25
I15

2
1
-1.5

-1

-0.5

Re

Figure 5-31 Eigenvalue evolution for the PSS damping gain on each generator is increased
from zero (no PSS in service) to 30 pu on machine base (750MVA) in 5 pu steps.
Table 5-3 Approximate improvements on system damping: comparison of the results
obtained from adding with equivalent damping torque (Figure 5-31) and the analysis for
participation factor.
Participation factor Analysis
Mode
Figure 5-31
(Approximation based on 30pu)
I40

-0.3699-(-0.2862)= -0.0837

-0.0259 30 100/750= -0.1036

I35

-0.3771-(-0.2899)= -0.0872

-0.026 30 100/750= -0.104

I25

-0.2683-(-0.2425)= -0.0158

-0.0096 30 100/750= -0.0384

I15

-0.2638-(-0.1542)= -0.1096

-0.031453 30 100/750= -0.12581

Table 5-3 gives the comparison results of the improvements on the system damping. In
the table, the third column data is calculated by the participation factor, and the data in the
second column is computed by using the system model by adding an equivalent damping

194

5.3. INTEGRATING WITH THE SIMPLIFIED AUSTRALIAN GRID

gain to the torque. In this study, the base capacity of the system (Sbase) is 100MVA, while
the machine base (Mbase) is set to 750MVA. It is shown that both of these cases have
similar results despite an error of 0.02 occurred in each mode. This also implies that the
approximation based on the participation factor is an effective way to evaluate the damping
performance of PSS.
5.3.3.1 PSS Design and Assessment
The approach for PSS design employed in this study is explained in detail in [99, 152].
As stated in [99],
The objective of PSS design is to induce on the shaft of the generator a torque of
electromagnetic origin which is in-phase with rotor-speed perturbations of the generator
Here, the procedure to design the PSS is given below:
Step 1: Determine the torque coefficients induced by PSS. In other words, specify the
DAMP(s) coefficient, add wash out filter and low pass filter to the coefficient making the
transfer function proper to obtain the transfer function of the stabilizer DAMP(s),
Step 2: Obtain the gain and phase characteristic of the excitation system, the generator
and the power system (i.e. the PVr characteristic),
Step 3: Apply curve fitting to the selected PVr characteristic,
Step 4: Derive the PSS transfer function,
Step5: Assess the performance of PSS by using eigenvalue analysis and by comparing
the torque coefficients induced by the PSS specified with the specified value.
A. Specifying the DAMP(s)
It was also stated in [99] that The wash-out and low-pass filter time constant are
chosen so that, over the frequency range of interest, DAMP(s) has near constant gain
(

) and phase between 0 and -30 degrees (depend on the damping requirements of

local- and inter-area modes).

Typically, DAMP(s) has the following form.

where

DAMP(s) De B(s)

(5-38)

sTw
1
B( s ) (
)(
)N
1 sTw 1 sTp

(5-39)

195

CHAPTER 5: STABILITY OF VSC-HVDC LINKS EMBEDDED WITH THE WEAK AUSTRALIAN

GRID
The frequency response of the DAMP(s) specified in this case example is shown in
Figure 5-32, where De equals to 20pu on the basis of machine rating and the time constants
for the wash out filter and low-pass filter are set to be 3s and 0.05s respectively.

Bode Diagram

Magnitude (dB)

30
20
10
0

Phase (deg)

-10
90
45
0
-45
-90
-2
10

-1

10

10
10
Frequency (rad/s)

10

10

Figure 5-32 Bode plot for DAMP(s) with DPSS=20pu on Mbase 750 MVA, Tw=3s and
Tp=0.05s.

B. PVr characteristic and curve fitting for Innamincka Generator 1

To disable the shaft dynamics of all the machines, the transfer function from the
voltage reference input (Vr) of Automatic Voltage Regulator (AVR) on Innamincka
generator #1 to the torque of the electrical power output (P) that PVr(s) can be easily
calculated as shown in Figure 5-33 (solid blue line). For design purpose, the PVr(s) should
be fitted to a lower order plant as given in equation (5-40) below. Its corresponding Bode
plot is given in Figure 5-33 (dotted red line).

PVr (s)

18.6s 3268
0.0056916s 1
54.805
s 15.25s 59.63
0.01677s 2 0.25574s 1
2

(5-40)

Then converting PVr(s) from pu on Sbase (100 MVA) to pu on Mbase (750 MVA) and
specify the DAMP(s) function,
PVr _ M ( s) K sm G( s) K sm

1 0.0056916s
0.01677s 2 0.25574s 1

here, Ksm Ks Sbase / M base 54.805 100 / 750 7.3073

196

(5-41)

5.3. INTEGRATING WITH THE SIMPLIFIED AUSTRALIAN GRID

P/Vr Characteristic ING1

Magnitude(dB)

40
20
0
-20 -1
10

10

10

10

Phase(deg)

0
-100
-200
-300 -1
10

Full load case

Fitting curve
0

10

10

10

Frequency (rad/sec)

Figure 5-33 PVr characteristic and curve fitting for Innamincka generator #1

C. Derivation of the PSS Transfer Function

The transfer function of PSS that PSS(s) and the system PVr characteristics PVr(s) are
related in terms of the following equation,
PSS (s) PVr (s) DAMP(s)

(5-42)

Then the PSS transfer function can be easily derived as,

PSS ( s) DAMP( s) / PVr _ M ( s)

3s
1
0.01677s 2 0.25574s 1
( DPSS / K sm ) (
)(
)(
)
(5-43)
1 3s 1 0.05s
0.0056916s 1
3s
1
0.01677s 2 0.25574s 1
(0.13685 DPSS ) (
)(
)(
)
1 3s 1 0.05s
0.0056916s 1
D. Assessing the PSS Performance
a) Eigenvalue Analysis
Figure 5-34a shows that the damping performance of the system under high loading
condition has been enhanced by the designed PSS. However, it can be noted that the
compensation added by PSS is not so significant, and it can be further verified by
conducting a small step test (0.01 pu) on the voltage reference of Innamincka generator #1
(see Figure 5-35). Figure 5-35 shows that the response of the generator output power is
damped slightly by comparing the case with (dashed black curve) and without PSS (solid
red line). In addition, the damping performance of the system under the light loading

197

CHAPTER 5: STABILITY OF VSC-HVDC LINKS EMBEDDED WITH THE WEAK AUSTRALIAN

GRID
condition has not been changed significantly, which can be verified by the eigenvalue map
shown in Figure 5-34b. As it can be seen in the figure, the eigenvalues of this operating
mode have not been significantly left-shifted. However, it is worth noting that it does not
deteriorate the damping performance of the system at least (see Figure 5-35). To further
improve the damping performance of the system, the additional damping toque added by
introducing supplementary POD controllers will be shown in the next section.

(b) Light loading condition

(a) High loading condition

Figure 5-34 The comparison of the rotor modes of the system with the equivalent damping
torque (blue star) and the system fitted with PSS (red dot) under the high loading condition
and light loading condition. The damping gain in Innamincka generators are increased
from zero (no PSS in service) to 30 pu in 5 pu steps.
0.3

without PSS(high)
with PSS(high)
without PSS(Light)
with PSS(Light)

P.ING1(p.u.)

0.2
0.1
0
-0.1
-0.2
0

10
Time(Seconds)

15

20

Figure 5-35 Comparison on the power output of generator #1 from Innamincka under the
condition with and without PSS in service by applying a small disturbance 0.01pu on the
reference voltage of Innamincka generator #1

198

5.3. INTEGRATING WITH THE SIMPLIFIED AUSTRALIAN GRID

b) Comparing the Torque Coefficients

As indicated in [99] To obtain the torque coefficient induced by PSS on the shaft of the
machine, the shaft dynamics is disabled and a speed test signal was injected into the PSS.
Figure 5-36a,b show the induced damping torque and synchronous by PSS torque in
Innamincka generator #1 (red curve) compared with the specified damping torque
coefficient (blue curve). Results in these plots confirm the successful implementation of
the PSS.

(a)

(b)

Figure 5-36 (a) Damping torque coefficients introduced by the PSS in the Innamincka
generator #1 (red curve) compared with the specified damping torque coefficient (blue
curve) (b) induced synchronizing torque (De equals to be 20 pu on Mbase)
5.3.3.2 POD Design and Assessment
The auxiliary device fitted to the FACTS devices for enhancing power system stability
are called the power oscillation damping controllers, which is designed to enhance the
damping performance of certain inter-area modes. Based on the source of the feedback
signals, two types of POD, local POD and wide area POD (WAPOD) are designed
systematically based on the residue method. Note that the residue analysis gives an
indication on the selection of the best suitable feedback signal which is the one with the
largest residue, since such signal is defined to be capable of improving the targeting
oscillation mode with less efforts [153].
Figure 5-37 shows the general structure of the AC/DC hybrid network with POD
damping controllers. The POD controllers are composed of an amplification block, a wash-

199

CHAPTER 5: STABILITY OF VSC-HVDC LINKS EMBEDDED WITH THE WEAK AUSTRALIAN

GRID
out and a low-pass filter and

stages lead-lag blocks. The transfer function G(s) of the

POD controllers can be given by,

G( s) k

sTw 1 sTlead mc
1

[
]
1 sTLP 1 sTw 1 sTlag

(5-44)

where, TLP denotes the time constant of the low-pass filter; similarly, Tw stands for the
time constant of wash-out filter; Tlead and Tlag imply time constants for the lead-lag block;
mc is the number if compensation stages (usually mc = 2).

VSC-HVDC POD Damping Controller

1 sTlead
1 sTlag

1
1 sTLP

1 sTlead
1 sTlag

sTw
1 sTw

PL 34 , BI

Pmax

Pmin

POWER
Rdc _ rec

SYSTEM
Ldc _ rec

Ldc _ inv

Rdc _ inv
I dc _ inv

I dc _ rec

Cdc _ rec

U dc _ rec

Rdc _ rec

Pref _ REC PREC

Pref _ REC+

+-

Ki
s

Ldc _ rec

Ldc _ inv

Rdc _ inv

Cdc _ inv

Power Plant

I cq

Kp

1
1 Tf s

U dc _ inv

U cdc

+
+

+-

I cqref

DB

Vcqref

VSC-HVDC Outer Power and Inner Current Controller

Figure 5-37 The schematic diagram of the VSC-HVDC damping system

The theoretical analysis and the systematic design approach are well documented in [98,
100, 153]. Assuming that the excitation mode of the system is the Target mode h only,
the simplified approximate model of the target system can be given by

, where Rh is

the open-loop residue from the selected input to the output of the MIMO system for the
particular hth complex mode. Furthermore, assuming that the POD controller G(s) is in
place, the closed-loop transfer function for the POD loop and the balanced equation for h
via the eigenvalue calculation can be obtained. Furthermore, by applying the small signal
technique and first order Taylor series expansion, it can be obtained the left-shift of a
target mode caused by the POD controllers satisfies the below equation,

200

5.3. INTEGRATING WITH THE SIMPLIFIED AUSTRALIAN GRID

h k Rh G(h )

(5-45)

where k is chosen such that

1
G ( s)
Rh (
)
s s h

(5-46)

In order to left-shift the target mode horizontally by using a POD controller as in

equation (5-44), the parameter selections of Tlead and Tlag must satisfy the following
relationship [100],

comp arg(G(h )) 1800 arg( Rh )

(5-47)

where for G(s),

T
c lead
Tlag

Tlead

1 sin(
1 sin(

i c

comp
mc

comp
mc

(5-48)
)

; Tlag c Tlag

(5-49)

A. Residue Analysis
Table 5-4 presented in the following shows the residue analysis for the interconnected
system. For the local POD, the power flow in adjacent AC lines PL34 is selected as the
input signal since it has relatively higher residue [98]. For the same reason, the bus voltage
angle difference between Brisbane and Innamincka BI is chosen as the input signal for
the WAPOD.
Note that in Table 5-4, 17 different modes are considered. In this table, two types of
stabilizing input signals are considered: (i) local signals such as power flow in adjacent AC
lines (PL12, PL34) and (ii) wide-area signals such as bus voltage angles at key nodes
(VAB,VAS,VAM,VAA,VAI) in the various regions of the system. For explanations of the
symbols, refer to Figure 5-1. In this table, and denote the bus voltage angle and
frequency respectively, while the subscript B, S, M, A, I denote the 5 areas Brisbane,
Sydney, Melbourne, Adelaide and Innamincka respectively.

201

10

11

12

13

14

15

16

17

202
57.841

70.108

-71.047 62.521

57.854

-67.306 30.107

-0.7921 46.431

58.184

7.3503

53.306

53.037

-43.571 -88.744 33.621

33.911

8.6735

1.06E-05 6.53E-05

-61.049 -87.127

3.12E-07 2.02E-06

8.6169

49.935

61.9

-0.6005 -32.847

-43.598 88.446

70.661

53.367

-83.896 -16.451 -39.647 44.162

89.679

6.28E-10 1.02E-08 3.68E-08 2.68E-08 2.93E-12 2.73E-10 2.02E-09 1.34E-08 9.96E-09 2.55E-12 3.23E-09 1.06E-10 7.83E-06

-48.311 -32.77

2.9391

0.005093

-54.016 84.466

-50.364 67.417

-13.875 -24.096 28.82

69.314

-14.135

-87.795 -12.588 60.114

6.00E-12 7.60E-11 4.03E-10 3.39E-10 3.52E-13 3.28E-11 5.14E-11 3.43E-10 2.54E-10 6.43E-14 8.78E-11 2.93E-12 2.33E-07

-61.132 -87.197

2.76E-05

-3.3829 -29.084

33.923

-39.292 44.134

3.12E-07 2.03E-06

8.5585

4.4388

-88.415 0.1195

-43.805 -71.096 -7.9853

62.615

-23.45

54.818

1.07E-05 6.53E-05

-30.191 -29.199 31.532

47.711

0.0036817 6.01E-10 1.24E-09 1.32E-08 1.71E-08 9.24E-23 1.10E-20 1.93E-09 1.36E-08 4.12E-08 2.55E-12 3.24E-09 1.08E-10 1.93E-05

78.327

-64.707 10.01

-62.937 5.6807

83.933

-24.17

63.324

5.96E-12 9.05E-12 1.45E-10 2.14E-10 3.08E-24 3.69E-22 4.84E-11 3.50E-10 1.02E-09 6.45E-14 8.79E-11 3.00E-12 5.54E-07

-86.186 72.057

2.01E-05

32.683

34.932

-60.123 -87.256

43.344

81.031

3.05E-07 2.02E-06

7.2523

30.911

17.075

49.151

1.04E-05 6.53E-05

43.926

-38.049 44.862

0.0014634 2.75E-10 1.21E-08 2.04E-08 1.01E-08 1.09E-22 9.96E-21 1.79E-09 2.65E-08 3.69E-09 2.36E-12 3.23E-09 1.56E-10 1.36E-07

-81.344 -41.848 -37.84

-67.376 41.226

-18.249 -35.075 -14.098 24.426

-66.381 52.442

-24.031 -53.842 4.7301

4.4266

-55.874

18.144

2.72E-12 8.86E-11 2.24E-10 1.26E-10 3.65E-24 3.34E-22 4.49E-11 6.83E-10 9.13E-11 5.97E-14 8.76E-11 4.34E-12 3.91E-09

26.285

7.99E-06

-88.562

-88.66

-3.715

3.76E-07 2.16E-06

21.912

76.37

-46.966 -41.833

67.228

-2.3052

-30.137 49.683

8.3962

1.28E-05 6.96E-05

-50.128 19.393

-62.319 28.663

0.0006068 7.50E-11 1.19E-08 6.81E-09 2.14E-09 1.47E-22 1.39E-20 2.57E-09 2.35E-08 5.26E-10 2.52E-12 3.40E-09 2.51E-10 1.62E-07

-45.802 21.757

-23.801 -11.947 -78.912 83.337

-60.697

2.31E-06

26.162

7.96E-08

-59.069

2.16E-06

25.2

7.31E-08

-35.928

1.43E-06

48.34

4.82E-08

-23.082

1.00E-04

61.186

3.40E-06

Local Signal

-60.239 -70.941 -85.879 15.437

-54.671 40.21

8.9947

75.545

-89.091

17.186

7.43E-13 8.67E-11 7.46E-11 2.67E-11 4.90E-24 4.66E-22 6.45E-11 6.05E-10 1.30E-11 6.37E-14 9.23E-11 7.00E-12 4.64E-09

-46.361 -73.421 56.038

3.31E-06

-86.268 57.894

63.213

-18.959 -77.818 32.284

-75.286 52.865

73.425

89.684

7.977
7.11E-07

18.854

0.0066914 2.54E-09 4.01E-08 2.04E-07 1.92E-07 1.45E-22 5.76E-20 1.94E-09 6.46E-09 2.80E-07 9.22E-14 4.28E-12 4.51E-11 0.0002839 1.59E-07 2.73E-07

25.881

Mode

78.691

9.8428

-3.4094

0.00049 0.0009516 5.01E-04

0.0045244 5.79E-10 9.15E-08 5.56E-08 1.83E-08 8.53E-21 1.20E-18 2.20E-08 1.97E-07 1.03E-08 2.46E-11 3.18E-08 2.71E-09 8.74E-06

CHAPTER 5: STABILITY OF VSC-HVDC LINKS EMBEDDED WITH THE WEAK AUSTRALIAN

GRID
Table 5-4 Residue analysis of the system
RECTIFIER SIDE POWER REFERENCE INPUT
Wide area signal

Mag Ang Mag Ang Mag Ang Mag Ang Mag Ang Mag Ang Mag Ang Mag Ang Mag Ang Mag Ang

5.3. INTEGRATING WITH THE SIMPLIFIED AUSTRALIAN GRID

B. Local POD controller design

It can be seen from Table 5-4 that the residue angle from PREF_REC to PL34 under the
least damped mode 17 (at frequency of 1.7033rad/s) is 89.684o. Using the equations (5-47)
to (5-49), Tlead and Tlag can be calculated as 1.4229s and 0.24224s respectively. Then the
transfer function of the local POD can be given by
10s
1
1+1.4229s 2
POD( s) KSS Q( s) KSS (
)(
)(
)
1 10s 1 0.01s 1+0.24224s

(5-50)

Figure 5-38 shows the effectiveness of the designed POD control. The result meets the
compensation requirement, since sum of the compensated angle of the POD (92.3o) and the
residue angle of the target mode (89.684o) is approximate 180o, which ensures the
horizontally left-shift of the mode.
Bode Diagram

Magnitude (dB)

40
20
0
-20
-40

Frequency(rad/s:1.7)
Phase(deg):92.3

Phase (deg)

-60
135
90
45
0
-45
-90

-2

10

10
10
Frequency (rad/s)

10

Figure 5-38 Frequency response of lead compensator Q(s) for the local POD

a) Root Locus Analysis

The root locus analysis has been conducted and the eigenvalue map of the system for
different values of the amplification block KSS is shown in Figure 5-39. Note that the
target inter-area mode I15 is left-shifted from -0.1549 Np/s to -0.3862 Np/s under high
loading condition (see Figure 5-39a), which also resulted in shift of the inter-area modes of
I40, I35, I25. Although the Mode I20 is shifted to the right but which is still with enough
damping ratio. This is also the cause of the first swing appeared in the with local POD

203

CHAPTER 5: STABILITY OF VSC-HVDC LINKS EMBEDDED WITH THE WEAK AUSTRALIAN

GRID
only plot of the Figure 5-41. In addition, for the system under light loading condition (see
Figure 5-39b), the modes I50, I30 and I25 remain unchanged, except the mode I40 but still
within the acceptable limit. Similar to the high loading condition, mode I20 is also shifted
to the right but with an adequate damping performance.

Damping Criterion

Damping Criterion

-0.14Np/s
I50

-0.14Np/s
I40

I40

I35
I20

I30
I20

I25
I15

(a)High loading condition

I25

(b) Light loading condition

Figure 5-39 The eigenvalue map of system fitted with POD, where the letters represent the
gains of the POD and KSS is increased from 3.642 10-3 to 7.03642 10-2 with a step size of
0.01.

b) Eigenvalue Analysis of POD in cooperation with PSS

By comparing the eigenvalue maps of the system with the coefficient KSS varying
from 3.642 10-3 to 7.03642 10-2 at a step size of 0.01, and with PSS on and off, it can be
concluded that the inclusion of PSS has significantly improved the inter-area mode I40.
However, the other three inter-area modes display small improvements including the high
loading conditions (see Figure 5-40a). For the system under the light loading condition, the
superposed effects of PSS and POD can be observed in Figure 5-40b. Note that no
significant impact has been observed by placing these damping controllers in service.
Although the mode I40 has been slightly deteriorated, it is still within the acceptable limit.

204

5.3. INTEGRATING WITH THE SIMPLIFIED AUSTRALIAN GRID

12
Damping Criterion

10

-0.14Np/s

Pentagram: PSS De= 0pu

Star
: PSS De=30pu

Damping Criterion

Im

I40

I35

I25
I15

I20

0
-4

-3

-2
Re

-1

-0.14Np/s

I50

I40

I30
I25

2
1
0

0 -1.5

-1

-0.5

Re

(a)High loading condition

(b) Light loading condition

Figure 5-40 The eigenvalue map of system fitted with POD and PSS. Here the letters are
the gains of the POD and KSS is increased from 3.642 10-3 to 7.03642 10-2 with a step size
of 0.01.
By applying the small-signal disturbance test (0.01 pu on the reference voltage
command) to Innamincka generator #1, it can be easily seen in Figure 5-41 that the
damping performance of the system is improved by placing the PSS and POD controller.
The plot labelled with POD (Local) only provides a better damping performance than
the PSS only controller. However, all these supplementary control schemes can provide
satisfactory improvements on damping the system inter-area modes.
0.4

without PSS & POD(Local)

with PSS only
with POD (Local) only
with PSS & POD(Local)

0.3
0.2

P.ING1(p.u.)

Im

Pentagram: PSS De=0pu

Star : PSS De=30pu

0.1
0
-0.1
-0.2
-0.3
0

4
6
Time(Seconds)

10

Figure 5-41 Power output of Innamincka generator #1 following a 0.01 pu step change on
the voltage reference of Innamincka generators.

205

CHAPTER 5: STABILITY OF VSC-HVDC LINKS EMBEDDED WITH THE WEAK AUSTRALIAN

GRID
Figure 5-42 provides the proof of the adequate performance of the VSC controllers cooperating with the supplementary PSS and local POD controllers. A sudden 0.01pu step
change of the power reference command at the rectifier side are applied to the system, the
top-left rectifier power plot in the figure displays some oscillations at the first 10s as a
consequence of the POD controllers taking effects. It can be concluded in this result that all
the controllers perform well as expected.
-5

DC-INV (p.u.)

P-REC (p.u.)

0.015
0.01

0.005

10

15

20

10

x 10

-5

Time(Seconds)

-1

-2

15

20

-6

x 10

AC-INV (p.u.)

AC-REC (p.u.)

-5

10

Time(Seconds)

10

15

20

Time(Seconds)

x 10

0
-5
-10
-15

10

15

20

Time(Seconds)

Figure 5-42 The performance evaluation of the VSC controllers with the supplementary
controllers in service following a step change of 0.01 pu on power reference of the rectifier
side converter.

C. WAPOD Controller Design

As displayed in the residue analysis, Table 5-4, the residue angle from

to

of the lightly damped mode 17 is 70.661o. Similarly, for the WAPOD, the results of

the calculated Tlead and Tlag are 1.8435s and 0.18697s respectively. The transfer function
for the WAPOD can be obtained by
10s
1
1+1.8435s 2
POD( s) KSS Q( s) KSS (
)(
)(
)
1 10s 1 0.01s 1+0.18697s

(5-51)

Figure 5-43 shows the frequency response of the designed lead compensator Q(s) for
WAPOD, which meet the design requirements.

206

5.3. INTEGRATING WITH THE SIMPLIFIED AUSTRALIAN GRID

Bode Diagram

Magnitude (dB)

40
20
0
-20
-40

Frequency (rad/s): 1.7

Phase (deg): 111

Phase (deg)

-60
135
90
45
0
-45
-90

-2

10

10
10
Frequency (rad/s)

10

Figure 5-43 Frequency response of lead compensator Q(s) of the WAPOD

a) Root Locus Analysis
By increasing the amplification block coefficients KSS from 3.642 10-3 to 0.12 with a
step size of 0.02, the eigenvalue evolution map was obtained by conducting the root-locus
analysis as given in Figure 5-44 and 5-45. The results in Figure 5-44 indicate a significant
improvement of the damping performances of the all 4 inter-area modes under high
loading condition. To examine the system sensitivity to the employment of WAPOD
(designed to improve the damping performance of high loading condition), the damping
performance of light loading condition is also tested.

Damping Criterion

Damping Criterion

-0.14Np/s

-0.14Np/s

F*

(a) High loading condition

(b) Zoomed version of left figure (circled area)

Figure 5-44 The eigenvalue map of system fitted with WAPOD under high loading
condition, where the letters represent the gains of the WAPOD, as KSS is increased from
3.64210-4 to 0.12 with a step size of 0.02.

207

CHAPTER 5: STABILITY OF VSC-HVDC LINKS EMBEDDED WITH THE WEAK AUSTRALIAN

GRID

Damping Criterion
-0.14Np/s
Damping Criterion
-0.14Np/s

(a)Light loading condition

(b) Zoomed version of left figure (circled area)

Figure 5-45 The eigenvalue map of system fitted with WAPOD under light loading
condition, where the letters represent the gains of the WAPOD, while KSS is increased
from 3.64210-4 to 0.12 with a step size of 0.02.
The simulation results suggest that the involvement of the WAPOD does not
deteriorate the light loading scenario, but not significantly improve it as well (see Figure
5-45).
b) Eigenvalue Analysis on WAPOD in cooperation with PSS
Figure 5-46 shows the eigenvalue analysis of the system operating under high loading
condition by employing the WAPOD damping controller and the PSS controller. It can be
observed from the figure that the damping performance of the system has been improved
significantly with employing the PSS only. This is further confirmed by the time response
plot shown in Figure 5-47, which demonstrate that the adequate damping performance can
be solely obtained by the WAPOD controller in service. In addition, it is shown that the
performance of WAPOD together with PSS is better than that of local POD with PSS
controller (red curve and magenta curve in Figure 5-47). Figure 5-48 illustrates the
performance of the designed VSC-controllers is still satisfactory as they operate well even
with the additional WAPOD controller.

208

5.3. INTEGRATING WITH THE SIMPLIFIED AUSTRALIAN GRID

Damping Criterion
-0.14Np/s
Damping Criterion
-0.14Np/s

(a)High loading condition

(b) Zoomed version of left figure (circled area)

Figure 5-46 Eigenvalue analysis of WAPOD with PSS under high loading condition,
where KSS is increased from 3.642 10-4 to 0.12 with a step size of 0.02

0.4

without PSS & WAPOD

with WAPOD only
with PSS & WAPOD
with PSS only
with PSS & POD(Local)

0.3

P.ING1(p.u.)

0.2
0.1
0
-0.1
-0.2
-0.3
0

4
6
Time(Seconds)

10

Figure 5-47 Power outputs of Innamincka generator #1 following a step change of 0.01 pu
on voltage reference of Innamincka generators.

209

CHAPTER 5: STABILITY OF VSC-HVDC LINKS EMBEDDED WITH THE WEAK AUSTRALIAN

GRID

-5

DC-INV (p.u.)

0.01

0.005

AC-REC (p.u.)

10

15

20

-5

10

15

Time(Seconds)
-6

x 10

-1

Time(Seconds)

-2

10

-5

AC-INV (p.u.)

P-REC (p.u.)

0.015

x 10

10

15

20

Time(Seconds)

20

x 10

0
-5
-10
-15

10

15

20

Time(Seconds)

Figure 5-48 The performance evaluation of the VSC controllers with PSS and WAPOD in
service following a step change of 0.01 pu on power reference of the rectifier side
converter.

5.4 Conclusion and Discussion

This chapter presented a small-signal rotor-angle stability analysis of a model of the
Australian power system with embedded VSC based HVDC links. We examined the
adequacy of using a constant admittance matrix to model the AC transmission system for
the analysis of the electromechanical modes. In addition, to facilitate this analysis for
various rating systems, a flexible scaling technique was introduced. The parameter
sensitivity in terms of the length of the DC link was also examined in the chapter, which
does not affect the performance of the designed controller, but on the contrary, enhanced
the system stability to some extent. Through the back examination of the basement for the
controller design against a higher order detailed Australian grid, it is confirmed that the

210

5.4. CONCLUSION AND DISCUSSION

assumption that designing the control system of VSC-HVDC based on the first order grid
model is sufficiently accurate.
Furthermore, for analytical purposes, a simplified model of the Australian power
system is used to connect the high capacity, but as yet undeveloped, geothermal resource in
the region of Innamincka in northern South Australia via a 1,100 km HVDC link to
Armadale in northern New South Wales. By diagnosing the lightly damped inter-area
modes of the hybrid AC/DC grid, it is found that the introduction of the new source of
geothermal power generation has an adverse impact on the damping performance of the
system. Hence, the supplementary damping controllers (i) generator power system
stabilizers (PSS) fitted to the synchronous machines which are used to convert geothermal
energy to electrical power and (ii) power oscillation damping controllers fitted to the VSCHVDC link are well designed to provide adequate damping in rotor modes of oscillation.
In the design of POD, two types of stabilizing input signals are considered: (i) local signals
such as power flow in adjacent AC lines and (ii) wide-area signals such as bus voltage
angles at key nodes in the various regions of the system. It was concluded that the smallsignal rotor-angle stability of the interconnected AC/DC system was greatly enhanced by
employing the designed damping controllers. Lastly, it should be emphasized that the
analysis tool utilized in this chapter is MATLAB m-file.

211

CHAPTER 6: CONCLUSION

Chapter 6: Conclusion
6.1 Summary
Due to the increasing demand on producing energy from renewable resources, VSCHVDC has become a common and effective solution for the grid integration of such energy
resources, which also promoted the ongoing developments in the field. As it is a widely
employed and proven VSC-HVDC technique, it can not only increase the transmission
capability but also enhance the damping ability and transient ability of the system by
employing an additional damping controller.
Significant levels of renewable energy resources (such as wind, solar and geothermal)
are available worldwide and economically harvested locally. However, the locations of
such resources are usually far from the main load centres and cities. Therefore, utilization
of these energies is very challenging. For example, the power grid in Australia is one of the
worlds longest interconnected power systems starting from Port Douglas in Queensland
and ending in Port Lincoln in South Australia with a total distance more than 4000
kilometres.

212

6.1. SUMMARY

The motivation of this research study has been to design and develop a detailed
mathematical understanding of the control system of a VSC-HVDC that can be integrated
into such a long and weak AC system.
As presented in the thesis, the outer power loop, the inner current loop, synchronization
method, and the input-output impedance of a weak AC interconnected VSC-HVDC system
have been partially studied in the literature. However, most of these earlier studies
typically employed a single-machine infinite-bus test system and did not study a wide
range of operating conditions that is likely to be encountered in practice. Therefore, one of
the primary aims of this thesis was to design a robust control system with a simple
structure and study a wide range of loading conditions that may occur in practice.
To achieve this main aim, this research can be divided into two main sections. The first
section (Chapter 2-4) of the research has investigated three potential linear inner current
control schemes to determine the most suitable type. Then an analytical model of the DB
controlled VSC was developed as the suitable controller. The discussions in the thesis
include a set of linearized representation of system components which prepared the ground
work. This resulted in the development of a new method for eliminating the
modulator/demodulator blocks. Using this new model, the classical frequency response
technique was also applied to a set of linear models using a number of operating points to
develop a robust outer loop controller.
In the second section of the research (Chapter 5), the impact of an admittance
representation of the grid to the controller design in the VSC-HVDC system and a scaling
approach were investigated. The investigations of this section also include the integration
of the designed VSC-HVDC links into the Extended Simplified South-East Australian
power grid, which aimed to form a large AC/DC integrated power system. In addition,
extensive eigen-sensitivity studies were performed to determine the characteristics of the
interaction and to identify the critical parameters. Finally, the power damping controllers
as PSS, local POD and WAPOD were designed to enhance the power system stability.

213

CHAPTER 6: CONCLUSION

6.2 Original Contributions and associated Key Results

6.2.1 Original Contributions
In this thesis, the applicability of three types of linear controllers, namely PI, PR and
DB, for VSC-HVDC transformation systems was investigated in depth to identify the most
suitable current control method.
In addition, a novel small-signal model for the digital DB controlled voltagecontrolled oscillator was also developed and systematically verified. To achieve this, the
study developed a new methodology to linearize the modulator/demodulator blocks which
are used to develop the small signal models for several key components.
The most significant contribution of the thesis is in designing a controller of the VSCHVDC system capable of working with a weak AC system. I proposed a robust controller
design methodology which takes into account a set of operating points covering the
converter operating capability (PQ capacity chart) and various grid parameters including
X/R ratios. I realized this method by applying the frequency response technique to a set of
open loop transfer functions of the target system, and bring up with practical
recommendations.

6.2.2 Associated Key Results of the Thesis

1. Three types of linear inner current controllers, PI, PR, and DB, were discussed for a
VSC-HVDC transmission system (Chapter 2).

A new method for the selection and optimization of the parameters of the PI
compensators in the various control loops using a decoupled control strategy was
proposed. In this study, the initial value of the PI compensator parameters for
input to the optimization algorithm were obtained using the classical frequency
response design approach to simplified linear models of the open-loop transfer
functions of VSC-HVDC control system. An optimization algorithm based on the
simplex method was adopted. The objective was to simultaneously minimize the
weighted sum of the integral of the time absolute-error products (ITAE) of the
active power, the reactive-power, the DC voltage and the inner current controllers
of the respective VSCs. It was concluded that if necessary the weightings of the
error signals may need to be modified in this stage.

214

6.2. ORIGINAL CONTRIBUTIONS AND ASSOCIATED KEY RESULTS

Four different DB controllers were also studied in the thesis by including one
sample delay with the reduced gain and Smith Predictor, two sample delays
based on IMC control design and solving feedback transfer function. It was
concluded from the simulation results that the solving feedback transfer function
method proves to be more superior to the other methods specifically during the
large step response with a fast settling time and with an acceptable reactive
current overshoot

2. A new small signal model for the digital DB controlled VSC was developed
(Chapter 3).

Pad Approximation was employed to transfer the discrete DB controller to

continuous domain.

Three approximation methods were investigated and compared: the first order
Pad Approximation, the third order Pad Approximation and first order lag
approximation since it was essential to model the ZOH, the DB control block and
the delay block. In this investigation the first-order Pad Approximation was
found to be accurate enough to approximate the pure one sample delay.

PLL was also investigated in detail. The principle of PLL, and the parameter
design were studied and verified using the small signal and the large signal
system disturbances.

In this part of the study, a new approach to eliminate the modulator and
demodulator block is proposed and realized the freely transformation among abc
natural, and dq synchronous reference frames. In addition, the limitation of
this methodology was also examined in this thesis, based on which it is suggested
that the best strategy for controller design is to design in dq synchronous
reference frame but implemented in abc natural reference frame.

The small signal model for VSC itself adopted here was the average value model
(AVM) which was verified in this project. Although it was found that to
represent it as an ideal converter is not so accurate due to its non-linearity nature,
the deviation of final results contributed by these differences was small once the
loops of the whole system are closed.

215

CHAPTER 6: CONCLUSION

Every component of the control system was characterized and verified step by
step by comparing the small signal and the large signal models. This provided a
solid foundation to verify the accuracy of the complete model. Furthermore, the
state-space models of the system components were also developed to obtain the
model of a large system.

In the development and analysis stages of this research, a block connection

method was proposed which was based on modular concept.

3. Controller design of the VSC-HVDC system which is capable of working in a large

set of grid impedance specifically in a weak AC system was proposed in this
project by considering a wide range of operating scenarios (Chapter 4).

The causes of failures in a weak AC grid interconnected VSC-HVDC system

were studied from the viewpoint of low frequency stability, in which the system
tends to operate at a lower bandwidth as the SCR of the interconnected system
decreasing.

Selected parameter sensitivity effects on the main and cross-coupling inner

current loops were discussed which involved the following factors SCRs, X/R
ratios of grid reactances, loading conditions and different power factors.

The limitation imposed by the RHP zero on the bandwidth of the outer loop
controllers was also discussed.

A robust controller design methodology was proposed by taking into account a

set of operating points covering the converter operating capability (PQ chart) and
the various grid conditions in terms of different SCR and X/R ratios of grid
reactance. The method was realized by applying the classical frequency response
technique to a set of open loop transfer functions of the target systems to be
controlled. It was concluded that it is more accurate to design a controller with
the previous designed controller in service, since the more detailed the system
will result in a more accurate design for the later controller. In addition, such
approach also allows the cross-coupling effects to be included.

A self-defined block was developed in PSCAD for facilitating the successive

PSCAD simulations for verification purpose.

216

6.3. SUGGESTIONS FOR FURTHER RESEARCH

Time domain verification for the co-operating performance of the various

designed controllers was also provided.

Finally, the small signal model development for each controller such as PID, PI
with low pass filter was presented.

4. To avoid the adverse impacts of the introduction of the new source of geothermal
power generation where the power are delivered via VSC-HVDC links fed into a
weak AC grid (Australian), three types of damping controllers (PSS, POD and
WAPOD) were designed to enhance the small-signal rotor-angle stability of the
power system. To achieve this, the following tasks were completed (Chapter 5).

A constant admittance matrix to model the AC transmission system was studied,

and it was concluded that using such representation is adequate for the analysis of
the electro-mechanical modes.

A scaling technique was introduced which made the model applicable to the
systems with different ratings.

The parameter sensitivity, in terms of the length of the DC link, was also
examined in the thesis, which did not weaken the performance of the designed
controller, but enhanced the system stability to some extent.

Back examination of the basement for the controller design against a higher order
detailed Australian grid confirmed the accuracy of the assumption that design the
control system of VSC-HVDC based on first order grid model.

It was observed that the introduction of the new source of geothermal power
generation has an adverse impact on the damping the inter-area modes by using
the small signal stability analysis.

Three types of damping controllers (PSS, POD and WAPOD) were designed and
verified systematically to enhance the power system stability.

6.3 Suggestions for Further Research

Although, a number of research topics were indentified as potential continuing works,
they are not included in this thesis to avoid diversion from the primary aims. Some of these
works are listed below to guide the future researchers:

217

CHAPTER 6: CONCLUSION

PSCAD verification of the grid integrated damping controllers can be done to be

able to provide more proofs for the credibility of the designed damping controllers.

The dynamics of the grid frequency can be introduced for the small signal
model of DB controlled VSC, which may offer a better precision for the model.

The control strategy employed in this thesis can be improved to allow a much
weaker system to be studied to increase the operating limits of the conventional DB
controlled VSC.

The control system design is able to be improved by employing a more advanced

DB current controller which can be designed by using the augmented model (a
higher third-order plant model) proposed in [40]. Such an advanced DB controller
can offer better control performance, which is able to facilitate the outer loop
controller design.

Coordination of PSS, POD/WAPOD and control system of VSC.

The coordinating controllers parameters can be tuned according to the input/output

impedance criterion as given in [65].

Finally, the DB current controller can be applied to a grid integrated energy storage
system, and the system stability can be studied. Such study can provide in-depth
understanding on how the energy storage system can enhance the power system
security.

218

Appendix A: VSC-HVDC System Parameters

219

APPENDICES

Rated Power
rec
inv

Rated ac Voltage
Rated Frequency
Switching Frequency
Sampling
ratio
commuting transformer

equivalent
inductance
L(pu)
R(pu)

Equivalent Filter
L(pu)

rec
inv
rec

0.15 pu

inv
rec
inv
rec
inv

Rated dc Voltage
DC-link Capacitor

0.0015 pu
0.15 pu
130 kV
C=500F

rec
inv

Per Unit system

Pn=75MW
Vn,LL(RMS)=13.8kV
Vn,LL(RMS)=115kV
fn=50Hz
fsw=1350Hz
fsamp=1350Hz
Tsamp=7.410-4
13.8 kV / 62.5 kV
62.5 kV / 115 kV

Ub ( rec ) 2 / 3 U n, LL ( rec )
Ub (inv ) 2 / 3 U n, LL (inv )

rec

2 S
Ib ( rec ) b
3 U b ( rec )

inv

2 S
I b (inv ) b
3 U b (inv )
Ub ( dc ) 130kV

Ib( dc ) Sb / Ub ( dc )
C pu CUb2( dc ) / Sb [116]

220

Appendix B: Method (II) for Elimination of Modulator/De-modulator System

Appendix B: Method (II) for Elimination of

Modulator/De-modulator System
This method is derived from the rotary factor perspective based on state-space
equations.
A. Block Transformation Method Derivation for General Modulator/De-modulator
System
The general Modulator/De-modulator system in reference frame is able to be
modelled using a set of differential equations and a set of algebraic equations as given
below, where u , Z , y denote the inputs, state variables and outputs of the system
quantities in reference frame.

Z A0 Z B0u

y C0 Z D0u

(B.1)

u
y
u ; y
u
y

G( s)

y
u

C0 ( sI A0 )1 B0 D0

(B.2)

221

APPENDICES

Zt At Zt Bt udq

ydq Ct Zt Dt udq

(B.3)

ud
yd
udq ; ydq
yq
uq

u e j (0t )udq

j (0t )
ydq
y e

(B.4)

Z AZ Be j (0t )udq

j (0t )
udq
y CZ De

(B.5)

cos(0t ) sin(0t )
Z AZ BRudq
Set R

sin(0t ) cos(0t )
y CZ DRudq

(B.6)

In which,
A
A 0
0

0
B
;B 0

A0
0

0
C
;C 0

B0
0

0
D
;D 0

C0
0

0
D0

This assumption is valid because system is symmetrical,

Z AZ BRudq

(B.7)

Multiply R 1 both sides,

R1 (Z AZ ) R1BRxdq

(B.8)

cos(0t ) sin(0t ) B0 0 cos(0t ) sin(0t )

R 1BR

sin(0t ) cos(0t ) 0 B0 sin(0t ) cos(0t )

cos(0t ) B0 cos(0t ) sin(0t ) B0 sin(0t ) cos(0t ) B0 sin(0t ) sin(0t ) B0 cos(0t ) (B.9)

sin(0t ) B0 cos(0t ) cos(0t ) B0 sin(0t ) sin(0t ) B0 sin(0t ) cos(0t ) B0 cos(0t )

1 0
B0

0 1
B0

R 1 ( Z AZ ) B0udq
R 1Z R 1 AZ B0udq

222

(B.10)

Appendix B: Method (II) for Elimination of Modulator/De-modulator System

R 1 Z ( R 1 ) Z - ( R 1 ) Z R 1 AZ B0udq

( R 1 Z ) ( R 1 ) Z R 1 AZ B0udq

R 1 e j0t ;( R 1 ) j0 e j0t j0 R 1

( R 1 Z ) j0 R 1 Z R 1 AZ B0udq
cos(0t ) sin(0t ) A0 0 cos(0t ) A0 sin(0t ) A0
R 1 A

sin(0t ) cos(0t ) 0 A0 sin(0t ) A0 cos(0t ) A0

0 cos(0t ) sin(0t ) A0 cos(0t ) A0 sin(0t )
A
AR 1 0

0 A0 sin(0t ) cos(0t ) A0 sin(0t ) A0 cos(0t )

R 1 A AR 1

( R 1 Z ) j0 ( R 1 Z ) A( R 1Z ) B0udq
s ( R 1 Z ) j0 ( R 1 Z ) A( R 1Z ) B0udq

R B0 udq
B0
R 1 Z

Z
udq
[( s j ) A]
[( s j ) A]

(B.11)

Substituting (B.4) into(B.5), it can be easily obtained

Rydq CZ DRudq ydq R1CZ R1DRudq

(B.12)

Then substituting (B.11) into (B.12), we can obtain,

G( s)

ydq
udq

R 1CR[( s j ) A]1 B0 R 1DR

Since, R 1CR C0 ; R 1DR =D0

(B.13)

G ( s) C0 [( s j ) A]1 B0 D0
Therefore, the transfer function transferred from

to dq reference frame can be

easily done by replacing Laplace factor s with s+j item.

B. Compute State Space Equations and the Frequency Response Matrix
The original reference frame state space equation is as follows,
Z AZ Bu
sZ AZ Bu

y CZ Du
y CZ Du

(B.14)

Transfer to dq reference frame is done by replace s with s+j

( s j ) Z dq Adq Z dq Bdqudq
sZ dq ( Adq j ) Z dq Bdqudq

ydq Cdq Z dq Dudq

ydq Cdq Z dq Dudq

(B.15)

223

APPENDICES

Z d Adq Z d Bdq 0 ud
s


Z q Adq Z q 0 Bdq uq

0 ud
yd Cdq 0 Z d Ddq

y
0 Cdq Z q 0 Ddq uq
q

(B.16)

As the original reference frame transfer function is

G ( s) C ( sI A) 1 B D
G ( s) dq C[( sI j) A)]1 B D
1
BD
( sI j) A
1
G ( s ) dq C
BD
( sI A) j
( sI A) j
G ( s ) dq C
BD
( sI A) 2 2
sI A

G ( s) dq (C
B D) jC
B
2
2
( sI A)
( sI A) 2 2
sI A

Yd jYq [(C
B D ) jC
B](U d jU q )
2
2
( sI A)
( sI A) 2 2
sI A

Yd (C
B D)U d C
BU q
2
2
( sI A)
( sI A) 2 2

sI A
Yq C
BU d (C
B D)U q
2
2
( sI A)
( sI A) 2 2
Yq
Yq
Yd
Y
sI A

sI A
C
B D; d C
B ; C
B; C
BD
2
2
2
2
2
2
Ud
( sI A)
Uq
( sI A) U d
( sI A) U q
( sI A) 2 2
G ( s ) dq C

224

Appendix C: Method (I) for Elimination of Modulator/Demodulator System

Appendix C: Method (I) for Elimination of

Modulator/Demodulator System
This method is derived from the transfer matrix state space equations point of view
avoiding imaginary component during derivation
A.

Continued to Equation (3-83)

T

Substituting [ X , X ] in equation (3-80) into equation (3-83), substituting [ X , X ]

in equation (3-78) into equation (3-83),
cos In sin In d Z d A
sin In cos In dt Z 0

q
B
0

0 cos In sin In Z d sin In cos In d

A sin In cos In Z q cos In sin In dt
0 cos
B sin

(C.1)

sin ud

cos uq

cos In sin In
sin In cos In

Aside: Compute

Zd
Z
q

(C.2)

According to the Woodbury matrix identity theory, this matrix has the form that
1

A1 A1BPCA1 A1BP E F
A B

C D

PCA1
P G H

(C.3)

225

APPENDICES

where, P ( D CA1B)1
A cos In
B sin In

C sin In
D cos In

(C.4)

1
In
cos

(C.5)

A1
P [cos In sin In(

1
) In( sin In)]1
cos

cos 2 sin 2 1
P [
] In cos In
cos
E ( A1 A1 BPCA1 )

(C.6)

from(C.3)

1
1
1
In (
In)( sin In) (cos In)(sin In) (
In)
cos
cos
cos
1
sin 2

In
In
cos
cos
1
(
In)(1 sin 2 ) In
cos
cos In

F A1 BP
1
In)( sin In)(cos In)
cos
F sin In from(C.3)
F (

G PCA
1
) In
cos
from(C.3)

G ( cos In)(sin In)(

G sin In
H P
H cos In

Thus, equation (C.2) can be written as equation (C.7) according to equation (C.3).
1

cos In sin In
E
sin In cos In G

F cos In sin In

H sin In cos In

From equation (C.1), the equation (C.7) can be written as,

226

(C.7)

Appendix C: Method (I) for Elimination of Modulator/Demodulator System

Z d cos In sin In A 0 cos In sin In Z d

Z q sin In cos In 0 A sin In cos In Z q
cos In sin In sin In cos In d

sin In cos In cos In sin In dt

Zd
Z
q
cos In sin In B 0 cos sin ud

sin In cos In 0 B sin cos uq
cos A sin A cos In sin In Z d

sin A cos A sin In cos In Z q
(cos sin sin cos ) In
d
(cos 2 sin 2 ) In

2
2
(sin cos ) In
( sin cos cos sin ) In dt

cos B sin B cos In sin In ud

sin B cos B sin In cos In uq

Zd
Z
q

Zd
(cos 2 sin 2 ) A
( cos sin sin cos ) A Z d 0 In d Z d


Z

(sin 2 cos 2 ) A
q In 0 dt Z q
Z q ( sin cos cos sin ) A

(cos 2 sin 2 ) B
( cos sin sin cos ) B ud

u
(sin 2 cos 2 ) B
( sin cos cos sin ) B
q
d
Now assume that, which is significantly important,
0
dt

Zd A

Z q In0

In0 Z d B 0 ud


A Z q 0 B uq

(C.8)

Substituting [ X , X ]T in equation (3-78) into equation (3-84), substituting [u , u ]T

in equation (3-81) (a) into equation(3-85),
cos In sin In yd C 0 cos In sin In Z d D 0 cos
sin In cos In y 0 C sin In cos In Z 0 D sin

yd cos In sin In C
y

q sin In cos In 0
cos C

sin C

0 cos In sin In Z d cos In sin In D

C sin In cos In Z q sin In cos In 0

sin C cos In sin In Z d cos D sin D cos

cos C sin In cos In Z q sin D cos D sin

yd C 0 Z d D
y

q 0 C Zq 0

0 cos
D sin

sin ud

cos uq
sin ud

cos uq

sin ud

cos uq

0 ud

D uq

(C.9)

Finally, rewrite equation(C.8) and(C.9) below; they can be combined in equation

227

APPENDICES

In0 Z d B 0 ud


A Z q 0 B uq

Zd A

Z q In0

(C.10)

yd C 0 Z d D 0 u d

Z 0 D u
y
0
C

q
q
q
These two equations can be combined as equations (3-86).
B. Compute the Frequency Response Matrix
sZ dq Adq Z dq Bdqudq
( sI Adq ) Z dq Bdqudq
Z dq ( sI Adq ) 1 Bdqudq

ydq (Cdq (sI Adq )1 Bdq Ddq )udq

Gdq ( s)
1
Consider the term (sI Adq ) = sIn

ydq
udq

Cdq ( sI Adq ) 1 Bdq Ddq

1

0 Adq 0 In (sIn Adq ) 0 In

(sIn Adq )
sIn 0 In Adq 0 In

(C.11)
1

(C.12)

Now use equation (C.3) to compute equation(C.12), in the following A, B, C, D in

equation (C.3) are replaced by a,b,c,d to avoid confusion with the state-matrix.

a sIn Adq ;
b 0 In;

(C.13)

c 0 In;
d sIn Adq ;

P (d ca 1b) 1
a ca 1b

a 0 Ina 1 (0 In)
(a 02 a 1 ) 1

228

(C.14)

Appendix C: Method (I) for Elimination of Modulator/Demodulator System

E a 1 a 1bpca 1
a 1 a 1b(a 02 a 1 ) 1 ca 1
a 1 02 a 1 (a 02 a 1 ) 1 a 1
a 1 ( In 02 (a 02 a 1 ) 1 a 1 )

(C.15)

a 1 ( In 02 (a (a 02 a 1 )) 1 )
a 1 ( In 02 (a 2 02 In) 1 )
a(a 2 02 In) 1
F a 1bp
a 10 (a 02 a 1 ) 1
0 a 1 (a 02 a 1 ) 1

(C.16)

0 ((a 02 a 1 )a) 1
0 (a 2 02 In) 1
G pca 1
(a 02 a 1 ) 10 Ina 1
0 ((a 02 a 1 ) 1 a 1 )

(C.17)

0 (a(a 02 a 1 )) 1
0 (a 2 02 In) 1

Hp
(a 02 a 1 )1
The Woodbury identity is

( A CBCT )1 A1 A1C ( B1 CT A1C )1CT A1

Define A a, C 0 In, CT 0 In, B a 1
H (a 02 a 1 ) 1 a 1 a 102 (a 02 a 1 )1 a 1
a 1 ( In 02 (a 02 a 1 ) 1 a 1 )

Then:

a 1 ( In 02 (a(a 02 a 1 )) 1 )
a 1 ( In 02 (a 2 02 In) 1 )
a(a 2 02 In) 1

H=E

(C.18)

Thus
E ( s) F ( s) a(a 2 02 In)1 0 (a 2 02 In)1
( sI Adq )1

2
2
1
a(a 2 02 In)1 (C.19)
F ( s) E ( s) 0 (a 0 In)

229

APPENDICES

Substitute a sIn Adq into equation(C.19),

( sI A)[( sI A)2 02 ]1
0 [( sI A)2 02 ]1
( sI Adq )

2
2 1
( sI A)[( sI A)2 02 ]
0 [( sI A) 0 ]
1

E(s) (sI A)[(sI A)2 02 ]1; F (s) 0[(sI A)2 02 ]1

C 0 E ( s ) F ( s ) B 0 D
Gdq ( s )

0 C F (s) E (s) 0 B 0
CE ( s ) CF ( s) B 0 D 0

CF ( s ) CE ( s) 0 B 0 D
CF ( s ) B
CE ( s ) B D

CF ( s ) B CE ( s ) B D
Gdd ( s ) Gdq ( s )

Gqd ( s ) Gqq ( s )

(C.20)
(C.21)

0
D

(C.22)

Gdd ( s ) CE ( s) B D
C ( sI A)[( sI A) 2 02 ]1 B D
Gdq ( s) CF ( s) B
C 0 [( sI A) 2 02 ]1 B

(C.23)

Gqd ( s ) CF ( s ) B
C 0 [( sI A) 2 02 ]1 B
Gdq ( s )

Gqq (s) Gdd (s)

where, Gdq ( s) and Gqd ( s) are cross-coupling items between dq axes due to modulation d
e-modulation.

230

Appendix D: Lead-Lag Block

Appendix D: Lead-Lag Block

State space equation of lead-lag block

x
k

1 Ta s
1 Tb s

Figure D.1 Block diagram of lead-lag-gain

x k (1 Ta s ) y (1 Tb s )
kx kxTa s y yTb s
kx y s (Tb y KTa x)

(D.1)

T
kx y
s( y k a x)
Tb Tb
Tb
set y k
sz

Ta
T
x z y zk a x
Tb
Tb

T
k
1
x ( z k a x)
Tb
Tb
Tb

(D.2)

(D.3)

231

APPENDICES

1
k k Ta

sz T z ( T T T ) x

b
b
b b

y zk a x

Tb
T
1
k k Ta
Set A ; B
; C 1; D k a
Tb
Tb Tb Tb
Tb

232

(D.4)

Appendix E: Regional Boundaries for the National Electricity Market & Committed
Developments

Appendix E: Regional Boundaries for the

National Electricity Market & Committed
Developments

233

APPENDICES

234

Appendix F: Performance Evaluation of all the Power and Voltage Controllers with a New
set of DC Link Parameters

Appendix F: Performance Evaluation of all

the Power and Voltage Controllers with a New
set of DC Link Parameters
A.

Rectifier Side Power Controller with Updated DC Link

50

Magnitude(dB)

Magnitude(dB)

40
20
0
-20 -1
10

10

10

10

10

10

10

10

10

Phase(deg)

Phase(deg)

-500 -1
10

-50
-100 -1
10

500

Phalf-Qmax
Phalf-Qmin
P -Qmin
0

10

10
Frequency (rad/sec)

(a)

10

10

-500

-1000 -1
10

Phalf-Qmax
Phalf-Qmin
P -Qmin
0

10

10
Frequency (rad/sec)

10

10

(b)

Figure F-1 Open loop frequency response (a) without and (b) with power controller under
the updated DC link condition

235

APPENDICES

1.4
System: untitled9
Peak amplitude: 1.04
Overshoot (%): 4.35
At time (seconds): 0.219

1.2
1

System: untitled12
Settling time (seconds): 0.506

pu

0.8
0.6
0.4
0.2
0
0

0.2

0.4
0.6
0.8
Time (s) (seconds)

1.2

Figure F-2 Step response on power reference with updated DC link parameters
B.

Rectifier Side AC Controller with Updated DC Link

50

Magnitude(dB)

Magnitude(dB)

40
20
0
-20 0
10

10

10

10

10

10

10

Phase(deg)

Phase(deg)

-400 0
10

-50
-100 0
10

-200

case 3
case 7
case 9
case 36
1

10
10
Frequency (rad/sec)

10

(a)

-200
-400
-600 0
10

case 3
case 7
case 9
case 36
1

10
10
Frequency (rad/sec)

10

(b)

Figure F-3 Open loop frequency response (a) without and (b) with rectifier side AC voltage
controller under the updated DC link condition

236

Appendix F: Performance Evaluation of all the Power and Voltage Controllers with a New
set of DC Link Parameters

1.2
System: untitled12
Peak amplitude: 1.09
System: untitled36
Overshoot (%): 9.02
At time (seconds): 0.123 Settling time (seconds): 0.337

0.8

pu

0.6
0.4
0.2
0
-0.2
0

0.1

0.2

0.3
0.4
Time (s) (seconds)

0.5

0.6

Figure F-4 Step response on rectifier side AC voltage reference with updated DC link
parameters
C.

Inverter Side DC Voltage Controller with Updated DC Link

100

Magnitude(dB)

Magnitude(dB)

50
0
-50
-100
10

10

10

-100
-200 0
10

200

10

10

10

10

Phase(deg)

Phase(deg)

100
0
-100
-200
10

10
Frequency (rad/sec)

(a)

10

-500

-1000 0
10

10

10
Frequency (rad/sec)

10

10

(b)

Figure F-5 Open loop frequency response (a) without and (b) with inverter side DC voltage
controller under the updated DC link condition

237

APPENDICES

1.2

System: untitled1
Peak amplitude: 1.02
Overshoot (%): 1.97
At time (seconds): 0.426

1
0.8

System: untitled18
Settling time (seconds): 0.483

pu

0.6
0.4
0.2
0
-0.2
0

0.5

1
Time (s) (seconds)

1.5

Figure F-6 Step response on inverter side DC voltage reference with updated DC link
parameters
D.

Inverter Side AC Voltage Controller with Updated DC Link

A.

(a)

(b)

Figure F-7 Open loop frequency response (a) without and (b) with inverter side DC voltage
controller under the updated DC link condition

238

Appendix F: Performance Evaluation of all the Power and Voltage Controllers with a New
set of DC Link Parameters

1.2
1

System: untitled3
Peak amplitude: 1.06
Overshoot (%): 5.82
At time (seconds): 0.135

0.8

System: untitled36
Settling time (seconds): 0.452

pu

0.6
0.4
0.2
0
-0.2
0

0.1

0.2

0.3
0.4
Time (s) (seconds)

0.5

0.6

0.7

Figure F-8 Step response on inverter side AC voltage reference with updated DC link
parameters
E.

Step Tests for System with All Controllers in service under Updated DC
Link Condition

a) Step test on rectifier side power controller

(a) Rectifier side Power response

(b) Rectifier side AC voltage response

239

APPENDICES

(c) Inverter side DC voltage response

(d) Inverter side AC voltage response

b) Step test on rectifier side AC voltage controller

(c)

240

(a) Rectifier side Power response

(b) Rectifier side AC voltage response

Inverter side DC voltage response

(d) Inverter side AC voltage response

Appendix F: Performance Evaluation of all the Power and Voltage Controllers with a New
set of DC Link Parameters
c) Step test on inverter side DC voltage controller

(c)

(a) Rectifier side Power response

(b) Rectifier side AC voltage response

Inverter side DC voltage response

(d) Inverter side AC voltage response

d) Step test on inverter side AC voltage controller

(a) Rectifier side Power response

(b) Rectifier side AC voltage response

241

APPENDICES

(c) Inverter side DC voltage response

242

(d) Inverter side AC voltage response

Appendix G: VSC-HVDC System (II) Parameters

Appendix G: VSC-HVDC System (II)

Parameters
Basement for AC system
Nominal three-phase power of the ac
grid

75MW

Nominal peak phase voltage at the ac

side

=51.03kV

=0.9798kA

Nominal peak phase current

Nominal impedance
Converter resistance of reactance

0.0015pu=0.0015*

Converter inductance of reactance

0.15pu=0.15*

Base frequency

2 50Hz

0.0781
=0.0249H

*The base for per unit transformation is chosen as to achieve a power invariant transformation,
so that the ac and dc side power is the same.
Basement for DC system

243

APPENDICES

Nominal DC voltage

130kV

Nominal DC current

0.5769kA

Nominal DC impedance

225.33

Time constant

244

Reference

Reference
[1]

E. S. I. P. Council, "Annual Planning Report," Adelaide, Australia, 2009.

[2]

A. Yazdani and R. Iravani, Voltage-sourced converters in power systems :

modeling, control, and applications. Hoboken, N.J.: IEEE Press/John Wiley, 2010.

[3]

D. J. Vowles, C. Samarasinghe, M. J. Gibbard, and G. Ancell, "Effect of wind

generation on small-signal stability A New Zealand Example " in Power and
Energy Society General Meeting - Conversion and Delivery of Electrical Energy in
the 21st Century, 2008 IEEE, 2008, pp. 1-8.

[4]

J. Pan, R. Nuqui, K. Srivastava, T. Jonsson, P. Holmberg, and Y.-J. Hafner, "AC

grid with embedded VSC-HVDC for secure and efficient power delivery," in
Energy 2030 Conference, 2008. ENERGY 2008. IEEE, 2008, pp. 1-6.

[5]

V. K. Sood, HVDC and FACTS Controllers: Applications of Static Converters in

Power Systems: Springer, 2004.

[6]

F. Schettler, H. Huang, and N. Christl, "HVDC transmission systems using voltage

sourced converters design and applications," in Power Engineering Society Summer
Meeting, 2000. IEEE, 2000, pp. 715-720.

[7]

W. Pan, Y. Chang, and H. Chen, "Hybrid Multi-terminal HVDC system for large
scale wind power," in Power Systems Conference and Exposition, 2006. PSCE'06.
2006 IEEE PES, 2006, pp. 755-759.

245

APPENDICES

[8]

N. Flourentzou, V. G. Agelidis, and G. D. Demetriades, "VSC-Based HVDC Power

Transmission Systems: An Overview," Power Electronics, IEEE Transactions on,
vol. 24, pp. 592-602, 2009.

[9]

J. Liang, O. Gomis-Bellmunt, J. Ekanayake, and N. Jenkins, "Control of multiterminal VSC-HVDC transmission for offshore wind power," in Power Electronics
and Applications, 2009. EPE'09. 13th European Conference on, 2009, pp. 1-10.

[10]

G. Asplund, K. Eriksson, and K. Svensson, "DC transmission based on voltage

source converters," in CIGRE SC14 Colloquium, South Africa, 1997.

[11]

Y. Liu, R. Zhang, J. Arrillaga, and N. Watson, "An overview of self-commutating

converters and their application in transmission and distribution," in Transmission
and Distribution Conference and Exhibition: Asia and Pacific, 2005 IEEE/PES,
2005, pp. 1-7.

[12]

H. Jiang and A. Ekstrom, "Multiterminal HVDC systems in urban areas of large

cities," Power Delivery, IEEE Transactions on, vol. 13, pp. 1278-1284, 1998.

[13]

W. Lu and B.-T. Ooi, "Premium quality power park based on multi-terminal

HVDC," Power Delivery, IEEE Transactions on, vol. 20, pp. 978-983, 2005.

[14]

(12June).
List
of
HVDC
http://en.wikipedia.org/wiki/List_of_HVDC_projects

[15]

T. Westerweller, K. Friedrich, U. Armonies, A. Orini, D. Parquet, and S. Wehn,

"Trans bay cable world's first HVDC system using multilevel voltage-sourced
converter," presented at the CIGR session, Paris, 2010.

[16]

B. W. B.JACOBSON, M.P.BAHRMAN, "500 kV VSC Transmission System for

lines and cables " presented at the 2012 San Francisco Colloquium, San Francisco,
2012.

[17]

B. T. Ooi and X. Wang, "Voltage angle lock loop control of the boost type PWM
converter for HVDC application," Power Electronics, IEEE Transactions on, vol.
5, pp. 229-235, 1990.

[18]

X. Wang and B. T. Ooi, "High voltage direct current transmission system based on
voltage source converters," in Power Electronics Specialists Conference, 1990.
PESC '90 Record., 21st Annual IEEE, 1990, pp. 325-332.

[19]

B. T. Ooi and X. Wang, "Boost-type PWM HVDC transmission system," Power

Delivery, IEEE Transactions on, vol. 6, pp. 1557-1563, 1991.

[20]

B. T. Ooi, F. D. Galiana, D. McGillis, H. C. Lee, X. Wang, Y. Guo, J. W. Dixon,

H. L. Nakra, and J. Belanger, "Research in pulse width modulated HVDC
transmission," in AC and DC Power Transmission, 1991., International Conference
on, 1991, pp. 188-193.

246

projects.

Available:

Reference

[21]

M. Lindgren, Modeling and control of voltage source converters connected to the

grid: Chalmers University of Technology, 1998.

[22]

J. Svensson, "Grid-connected voltage source converter-Control Principles and wind

energy applications ", Chalmers University of Technology, 1998.

[23]

J. Svensson, "Voltage angle control of a voltage source inverter, application to a

grid-connected wind turbine," presented at the 6th European Conference on Power
Electronics and Applications, Sevilla, 1995.

[24]

J. Svensson, "Inclusion of Dead-Time and Parameter Variations in VSC Modelling

for Predicting Responses of Grid Voltage Harmonics," in 7th European Conference
on Power Electronics and Applications (EPE'97), Trondheim ,Norway, 8-10
september 1997, pp. 216-221.

[25]

C. Du, "VSC-HVDC for Industrial Power Systems," PhD, Division of Electric

Power Engineering, Chalmers University of Technology Goteborg, Sweden, 2007.

[26]

J. Yong, R. Zhen, O. Kaijian, and Y. Jun, "Parameter estimation of regulators in

Tian-Guang HVDC transmission system based on PSCAD/EMTDC," in Power
System Technology, 2002. Proceedings. PowerCon 2002. International Conference
on, 2002, pp. 538-541 vol.1.

[27]

F. y. YANG, Z. XU, and J. ZHANG, "Study on Parameter Optimization of HVDC

PI Controllers " Power System Technology, vol. 11, p. 007, 2006.

[28]

A. Gole, S. Filizadeh, R. Menzies, and P. Wilson, "Electromagnetic transients

simulation as an objective function evaluator for optimization of power system
performance," in Proceedings, IPST'03-International Conference of Power Systems
Transients, New Orleans, 2003.

[29]

Z. Chengyong, L. Xiangdong, and L. Guangkai, "Parameters Optimization of VSCHVDC Control System Based on Simplex Algorithm," in Power Engineering
Society General Meeting, 2007. IEEE, 2007, pp. 1-7.

[30]

M. Zakaria Moustafa and S. Filizadeh, "Simulation of a VSC transmission scheme

supplying a passive load," in Industrial Electronics, 2008. IECON 2008. 34th
Annual Conference of IEEE, 2008, pp. 942-946.

[31]

T. M. Undeland, J. A. Suul, M. Molinas, and C. Bajracharya, Understanding of

Tuning Techniques Of Converter Controllers for VSC-HVDC, 2008.

[32]

J. A. Suul, M. Molinas, L. Norum, and T. Undeland, "Tuning of control loops for

grid connected voltage source converters," in Power and Energy Conference, 2008.
PECon 2008. IEEE 2nd International, 2008, pp. 797-802.

[33]

"IEEE Guide for Planning DC Links Terminating at AC Locations Having Low

Short-Circuit Capacities," IEEE Std 1204-1997, p. i, 1997.

247

APPENDICES

[34]

W. Liying and N. Ertugrul, "Selection of PI compensator parameters for VSCHVDC system using decoupled control strategy," in Universities Power
Engineering Conference (AUPEC), 2010 20th Australasian, 2009, pp. 1-7.

[35]

G. Ledwich and H. Sharma, "Connection of inverters to a weak grid," in Power

Electronics Specialists Conference, 2000. PESC 00. 2000 IEEE 31st Annual, 2000,
pp. 1018-1022 vol.2.

[36]

A. Farag, M. Durrant, H. Werner, and K. Abbott, "Robust control of a VSC HVDC

terminal attached to a weak AC system," in Control Applications, 2003. CCA 2003.
Proceedings of 2003 IEEE Conference on, 2003, pp. 173-177 vol.1.

[37]

M. Durrant, H. Werner, and K. Abbott, "Synthesis of multi-objective controllers for

a VSC HVDC terminal using LMIs," in Decision and Control, 2004. CDC. 43rd
IEEE Conference on, 2004, pp. 4473-4478 Vol.4.

[38]

M. Durrant, H. Werner, and K. Abbott, "A Comparison of LMI and GA Based

Robust Controller Designs for VSC HVDC," in Decision and Control, 2006 45th
IEEE Conference on, 2006, pp. 3990-3995.

[39]

Y. A. R. I. Mohamed and E. F. El-Saadany, "A Robust Natural-Frame-Based

Interfacing Scheme for Grid-Connected Distributed Generation Inverters," Energy
Conversion, IEEE Transactions on, vol. 26, pp. 728-736, 2011.

[40]

Y. A. R. I. Mohamed, M. A-Rahman, and R. Seethapathy, "Robust Line-Voltage

Sensorless Control and Synchronization of LCL -Filtered Distributed Generation
Inverters for High Power Quality Grid Connection," Power Electronics, IEEE
Transactions on, vol. 27, pp. 87-98, 2012.

[41]

Y. A. R. I. Mohamed, "Suppression of Low- and High-Frequency Instabilities and

Grid-Induced Disturbances in Distributed Generation Inverters," Power
Electronics, IEEE Transactions on, vol. 26, pp. 3790-3803, 2011.

[42]

Y. A. R. I. Mohamed, "Mitigation of Dynamic, Unbalanced, and Harmonic Voltage

Disturbances Using Grid-Connected Inverters With Filter," Industrial Electronics,
IEEE Transactions on, vol. 58, pp. 3914-3924, 2011.

[43]

M. Liserre, R. Teodorescu, and F. Blaabjerg, "Stability of photovoltaic and wind

turbine grid-connected inverters for a large set of grid impedance values," Power
Electronics, IEEE Transactions on, vol. 21, pp. 263-272, 2006.

[44]

M. Liserre, A. Dell'Aquila, and F. Blaabjerg, "Genetic algorithm-based design of

the active damping for an LCL-filter three-phase active rectifier," Power
Electronics, IEEE Transactions on, vol. 19, pp. 76-86, 2004.

[45]

S. Cobreces, E. Bueno, F. J. Rodriguez, F. Huerta, and P. Rodriguez, "Influence

analysis of the effects of an inductive-resistive weak grid over L and LCL filter

248

Reference

current hysteresis controllers," in Power Electronics and Applications, 2007

European Conference on, 2007, pp. 1-10.
[46]

S. Guoqiao, Z. Xuancai, C. Min, and X. Dehong, "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.

[47]

S. Guoqiao, X. Dehong, C. Luping, and Z. Xuancai, "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.

[48]

J. Dannehl, F. W. Fuchs, Th, x00F, and P. B. gersen, "PI State Space Current
Control of Grid-Connected PWM Converters With LCL Filters," Power
Electronics, IEEE Transactions on, vol. 25, pp. 2320-2330, 2010.

[49]

J. Dannehl, F. W. Fuchs, S. Hansen, Th, x00F, and P. B. gersen, "Investigation of

Active Damping Approaches for PI-Based Current Control of Grid-Connected
Pulse Width Modulation Converters With LCL Filters," Industry Applications,
IEEE Transactions on, vol. 46, pp. 1509-1517, 2010.

[50]

T. Erika and D. G. Holmes, "Grid current regulation of a three-phase voltage source

inverter with an LCL input filter," Power Electronics, IEEE Transactions on, vol.
18, pp. 888-895, 2003.

[51]

R. Ottersten, "On Control of Back-to-Back Converters and Sensorless Induction

Machine Drives," PhD Doctoral thesis, Deparment of Electric Power Engineering,
Chalmers Gothenburg, Sweden, 2003.

[52]

J.Hu, "Deatbeat Controlled PWM converter," Master, Dept. Elect. Eng., McGill
Univ., Montreal,QC, Canada, 1999.

[53]

T. Lianxiang, "Control and protection of multi-terminal DC transmission systems

based on voltage-source converters," Ph.D, Elect. Eng., McGill Univ.,
Montreal,QC, Canada, 2003.

[54]

A. Timbus, M. Liserre, R. Teodorescu, P. Rodriguez, and F. Blaabjerg, "Evaluation

of Current Controllers for Distributed Power Generation Systems," Power
Electronics, IEEE Transactions on, vol. 24, pp. 654-664, 2009.

[55]

W. Liying, N. Ertugrul, and M. Kolhe, "Evaluation of dead beat current controllers

for grid connected converters," in Innovative Smart Grid Technologies - Asia (ISGT
Asia), 2012 IEEE, 2012, pp. 1-7.

[56]

D. Jovcic, N. Pahalawaththa, and M. Zavahir, "Analytical modelling of HVDCHVAC systems," Power Delivery, IEEE Transactions on, vol. 14, pp. 506-511,
1999.

249

APPENDICES

[57]

V. Khatri, V. K. Sood, and H. Jin, "Analysis and EMTP simulation of a

conventional gate firing unit for HVDC converters operating with weak AC
systems," in Electrical and Computer Engineering, 1994. Conference Proceedings.
1994 Canadian Conference on, 1994, pp. 173-177 vol.1.

[58]

J. A. Suul, A. Luna, P. Rodriguez, and T. Undeland, "Frequency-adaptive Virtual

Flux estimation for grid synchronization under unbalanced conditions," in IECON
2010 - 36th Annual Conference on IEEE Industrial Electronics Society, pp. 486492.

[59]

J. A. Suul and T. Undeland, "Impact of Virtual Flux reference frame orientation on

voltage source inverters in weak grids," in Power Electronics Conference (IPEC),
2010 International, 2010, pp. 368-375.

[60]

J. A. Suul, A. Luna, P. Rodriguez, and T. Undeland, "Voltage-Sensor-less

Synchronization to Unbalanced Grids by Frequency-Adaptive Virtual Flux
Estimation," Industrial Electronics, IEEE Transactions on, vol. PP, pp. 1-1.

[61]

C. Zhe. Control of Power Electronic Converters.

[62]

Z. Lidong, L. Harnefors, and H. P. Nee, "Power-Synchronization Control of GridConnected Voltage-Source Converters," Power Systems, IEEE Transactions on,
vol. 25, pp. 809-820, 2010.

[63]

Z. Lidong, "Modeling and Control of VSC-HVDC Links Connected to Weak AC

Systems," PhD, SCHOOL OF ELECTRICAL ENGINEERING, ROYAL
INSTITUTE OF TECHNOLOGY, Stockholm 2010.

[64]

Z. Lidong and H. P. Nee, "Multivariable feedback design of VSC-HVDC

connected to weak ac systems," in PowerTech, 2009 IEEE Bucharest, 2009, pp. 18.

[65]

S. Jian, "Impedance-Based Stability Criterion for Grid-Connected Inverters,"

Power Electronics, IEEE Transactions on, vol. 26, pp. 3075-3078, 2011.

[66]

L. Harnefors, M. Bongiorno, and S. Lundberg, "Input-admittance calculation and

shaping for controlled voltage-source converters," Industrial Electronics, IEEE
Transactions on, vol. 54, pp. 3323-3334, 2007.

[67]

H. Jinwei and L. Yun Wei, "Generalized Closed-Loop Control Schemes with

Embedded Virtual Impedances for Voltage Source Converters with LC or LCL
Filters," Power Electronics, IEEE Transactions on, vol. 27, pp. 1850-1861, 2012.

[68]

T. Midtsund, J. A. Suul, and T. Undeland, "Evaluation of current controller

performance and stability for voltage source converters connected to a weak grid,"
in Power Electronics for Distributed Generation Systems (PEDG), 2010 2nd IEEE
International Symposium on, 2010, pp. 382-388.

250

Reference

[69]

D. Jovcic, "Control of High Voltage DC and Flexible AC Transmission Systems,"

Electrical and Electronic Engineering)--University of Auckland, 1999.

[70]

P. Fischer de Toledo, "Modelling and control of a line-commutated HVDC

transmission system interacting with a VSC STATCOM," KTH, 2007.

[71]

D. Jovcic, L. A. Lamont, and L. Xu, "VSC transmission model for analytical

studies," in Power Engineering Society General Meeting, 2003, IEEE, 2003, p.
1742 Vol. 3.

[72]

A. Tabesh and R. Iravani, "Small-signal model and dynamic analysis of variable

speed induction machine wind farms," Renewable Power Generation, IET, vol. 2,
pp. 215-227, 2008.

[73]

N. P. W. Strachan and D. Jovcic, "Stability of a Variable-Speed Permanent Magnet

Wind Generator With Weak AC Grids," Power Delivery, IEEE Transactions on,
vol. 25, pp. 2779-2788, 2010.

[74]

N. R. Chaudhuri, R. Majumder, B. Chaudhuri, and P. Jiuping, "Stability Analysis

of VSC MTDC Grids Connected to Multimachine AC Systems," Power Delivery,
IEEE Transactions on, vol. 26, pp. 2774-2784, 2011.

[75]

G. O. Kalcon, G. P. Adam, O. Anaya-Lara, S. Lo, and K. Uhlen, "Small-Signal

Stability Analysis of Multi-Terminal VSC-Based DC Transmission Systems,"
Power Systems, IEEE Transactions on, vol. PP, pp. 1-13, 2012.

[76]

Y. Pipelzadeh, N. Chaudhuri, B. Chaudhuri, and T. Green, "System stability

improvement through optimal control allocation in voltage source converter-based
high-voltage direct current links," Generation, Transmission & Distribution, IET,
vol. 6, pp. 811-821, 2012.

[77]

A. M. Alsseid, D. Jovcic, and A. Starkey, "Small signal modelling and stability

analysis of multiterminal VSC-HVDC," in Power Electronics and Applications
(EPE 2011), Proceedings of the 2011-14th European Conference on, 2011, pp. 110.

[78]

C. Sao and P. W. Lehn, "A Block Diagram Approach to Reference Frame

Transformation of Converter Dynamic Models," in Electrical and Computer
Engineering, 2006. CCECE '06. Canadian Conference on, 2006, pp. 2270-2274.

[79]

P. W. Lehn and S. Podrucky, "Small Signal modeling of Power Electronics

Converters with Resonant Controllers," in International Conference on Power
System Transients IPST 2009, pp. 3-6.

[80]

S. Podrucky, "Small Signal Modeling of Resonant Controlled VSC Systems,"

University of Toronto, 2009.

251

APPENDICES

[81]

D. Povh, D. Retzmann, E. Teltsch, U. Kerin, and R. Mihalic, "Advantages of large

ac/dc system interconnections," Report B4-304, CIGRE Session, 2006.

[82]

H. F. Latorre, M. Ghandhari, and L. Soder, "Use of local and remote information in

POD control of a VSC-HVdc," in PowerTech, 2009 IEEE Bucharest, 2009, pp. 1-6.

[83]

L. Weixing and B. T. Ooi, "Simultaneous inter-area decoupling and local area

damping by voltage source HVDC," in Power Engineering Society Winter Meeting,
2001. IEEE, 2001, pp. 1079-1084 vol.3.

[84]

R. J. Piwko and E. V. Larsen, "HVDC System Control for Damping of

Subsynchronous Oscillations," Power Apparatus and Systems, IEEE Transactions
on, vol. PAS-101, pp. 2203-2211, 1982.

[85]

P. Kundur, Power System Stability And Control: McGraw-Hill Education (India)

Pvt Limited, 1994.

[86]

M. J. Gibbard, D. J. Vowles, and P. Pourbeik, "Interactions between, and

effectiveness of power system stabilizers and FACTS device stabilizers in
multimachine systems," in Power Engineering Society Winter Meeting, 2000.
IEEE, 2000, p. 1532 vol.2.

[87]

Y. Chang, "Design of HVDC and SVC Coordinate Damping Controller Based on

Wide Area Signal," International Journal of Emerging Electric Power Systems,
vol. 7, p. 6, 2006.

[88]

C. Yong, X. Zheng, C. Gaihong, and X. Jiani, "A novel SVC supplementary

controller based on wide area signals," in Power Engineering Society General
Meeting, 2006. IEEE, 2006, p. 7 pp.

[89]

K. Uhlen, L. Vanfretti, M. de Oliveira, A. Leirbukt, V. H. Aarstrand, and J. O.

Gjerde, "Wide-Area Power Oscillation Damper implementation and testing in the
Norwegian transmission network," in Power and Energy Society General Meeting,
2012 IEEE, 2012, pp. 1-7.

[90]

J.-H. Ying, H. Duchen, K. Linden, M. Hyttinen, P. F. de Toledo, T. Tulkiewicz, A.

K. Skytt, and H. Bjorklund, "Improvement of subsynchronous torsional damping
using VSC HVDC," in Power System Technology, 2002. Proceedings. PowerCon
2002. International Conference on, 2002, pp. 998-1003 vol.2.

[91]

Z. Chao, X. Zhou, and L. Ruomei, "Dynamic Modeling and Transient Simulation

for VSC based HVDC in Multi-Machine System," in Power System Technology,
2006. PowerCon 2006. International Conference on, 2006, pp. 1-7.

[92]

H. F. Latorre, M. Ghandhari, and L. Soder, "Control of a VSC-HVDC Operating in

Parallel with AC Transmission Lines," in Transmission & Distribution Conference
and Exposition: Latin America, 2006. TDC '06. IEEE/PES, 2006, pp. 1-5.

252

Reference

[93]

H. F. Latorre and M. Ghandhari, "Improvement of voltage stability by using VSCHVdc," in Transmission & Distribution Conference & Exposition: Asia and
Pacific, 2009, 2009, pp. 1-4.

[94]

R. Preece, J. V. Milanovic, A. M. Almutairi, and O. Marjanovic, "Damping of

Inter-Area Oscillations in Mixed AC/DC Networks Using WAMS Based
Supplementary Controller," Power Systems, IEEE Transactions on, vol. 28, pp.
1160-1169, 2013.

[95]

J. L. Domnguez-Garca, O. Gomis-Bellmunt, F. D. Bianchi, and A. Sumper,

"Power oscillation damping supported by wind power: A review," Renewable and
Sustainable Energy Reviews, vol. 16, pp. 4994-5006, 2012.

[96]

A. Heniche and I. Kamwa, "Assessment of Two Methods to Select Wide-Area

Signals for Power System Damping Control," in Power Systems, IEEE
Transactions on vol. 23, ed, 2008, pp. 572-581.

[97]

N. Modi, M. Lloyd, and T. K. Saha, "Wide-area signal selection for power system
damping controller," in Universities Power Engineering Conference (AUPEC),
2011 21st Australasian, 2011, pp. 1-6.

[98]

P. Pourbeik and M. J. Gibbard, "Tuning of SVC stabilisers for the damping of

inter-area modes of rotor oscillation," presented at the Proc. Aust. Universities
Power Engineering Conf, Perth, 1995.

[99]

M. Gibbard, "Tuning of Power System Stabilizaers(PSS)," Department of Electrical

and Electronic Engineering The university of Adelaide Australia1998.

[100] D. J. Vowles, "Tuning of Stabilizers for FACTS Devices," Department of Electrical

and Electronic Engineering The university of Adelaide Australia1998.
[101] M. Gibbard, "Co-ordinated design of multimachine power system stabilisers based
on damping torque concepts," in IEE Proceedings C (Generation, Transmission
and Distribution), 1988, pp. 276-284.
[102] M. J. Gibbard, "Robust design of fixed-parameter power system stabilisers over a
wide range of operating conditions," Power Systems, IEEE Transactions on, vol. 6,
pp. 794-800, 1991.
[103] M. J. Gibbard, N. Martins, J. J. Sanchez-Gasca, N. Uchida, V. Vittal, and L. Wang,
"Recent applications of linear analysis techniques," Power Systems, IEEE
Transactions on, vol. 16, pp. 154-162, 2001.
[104] C. Spallarossa, Y. Pipelzadeh, B. Chaudhuri, and T. Green, "Assessment of
disturbance propagation between AC grids through VSC HVDC links using
reduced Great Britain model," 2012.

253

APPENDICES

[105] J. Wang, C. Fu, and Y. Zhang, "Design of WAMS-Based Multiple HVDC

Damping Control System," Smart Grid, IEEE Transactions on, vol. 2, pp. 363-374,
2011.
[106] W. Grainger, C. Parker, and P. Wright. (April 2001) "Commissioning a new
interconnection between two regions in Australia". Cigre Electra magazine. pp.1017.
[107] S. Hiley, "Redesign of the Blackwall Static Var Compensator Power Oscillation
Damper Controller," School of Information Technology and Electrical Engineering,
The University of Queensland, 24 Oct. 2007.
[108] N. Mai Huong, T. K. Saha, and M. Eghbal, "Investigation on the impact of hybrid
multi-terminal HVDC system combining LCC and VSC technologies using system
identification," in Universities Power Engineering Conference (AUPEC), 2012
22nd Australasian, 2012, pp. 1-6.
[109] M. H. Nguyen, M. Eghbal, T. K. Saha, and N. Modi, "Investigation of Oscillation
Damping for connecting remote generators to a large power system," in Innovative
Smart Grid Technologies Asia (ISGT), 2011 IEEE PES, 2011, pp. 1-8.
[110] F. Magueed, Converter Interfaced Distributed Generation-Grid Interconnection
Issues: Chalmers University of Technology, 2007.
[111] M. Bongiorno, "On Control of Grid-connected Voltage Source ConvertersMitigation of Voltage Dips and Subsynchronous Resonances," Chalmers University
of Technology, 2007.
[112] M. M. T.M. Haileselassie, T. Undeland. , "Multi-Terminal VSC-HVDC System for
Integration of Offshore Wind Farms and Green Electrication of Platforms in the
North Sea," presented at the NORPIE/2008, Nordic Workshop on Power and
Industrial Electronics, June 9-11, 2008.
[113] S. Ruihua, Z. Chao, L. Ruomei, and X. Zhou, "VSCs based HVDC and its control
strategy," in Transmission and Distribution Conference and Exhibition: Asia and
Pacific, 2005 IEEE/PES, 2005, pp. 1-6.
[114] C. Schauder and H. Mehta, "Vector analysis and control of advanced static VAr
compensators," Generation, Transmission and Distribution, IEE Proceedings C,
vol. 140, pp. 299-306, 1993.
[115] A. Lindberg, PWM and Control of Two and Three Level High Power Voltage
Source Converters: Royal Institute of Technology, 1995.
[116] A. M. Gole, S. Filizadeh, R. W. Menzies, and P. L. Wilson, "Optimization-enabled
electromagnetic transient simulation," Power Delivery, IEEE Transactions on, vol.
20, pp. 512-518, 2005.

254

Reference

[117] D. Graham and R. C. Lathrop, "The synthesis of "optimum" transient response:

criteria and standard forms," American Institute of Electrical Engineers, Part II:
Applications and Industry, Transactions of the, vol. 72, pp. 273-288, 1953.
[118] D. N. Zmood, D. G. Holmes, and G. H. Bode, "Frequency-domain analysis of
three-phase linear current regulators," Industry Applications, IEEE Transactions
on, vol. 37, pp. 601-610, 2001.
[119] J. G. Hwang, P. W. Lehn, and M. Winkelnkemper, "A Generalized Class of
Stationary Frame-Current Controllers for Grid-Connected AC/DC Converters,"
Power Delivery, IEEE Transactions on, vol. 25, pp. 2742-2751, 2010.
[120] F. Blaabjerg, R. Teodorescu, M. Liserre, and A. V. Timbus, "Overview of Control
and Grid Synchronization for Distributed Power Generation Systems," Industrial
Electronics, IEEE Transactions on, vol. 53, pp. 1398-1409, 2006.
[121] M. Liserre, R. Teodorescu, and F. Blaabjerg, "Stability of grid-connected PV
inverters with large grid impedance variation," in Power Electronics Specialists
Conference, 2004. PESC 04. 2004 IEEE 35th Annual, 2004, pp. 4773-4779 Vol.6.
[122] R. Teodorescu and F. Blaabjerg, "Proportional-resonant controllers. a new breed of
controllers suitable for grid-connected voltage-source converters," 2004.
[123] D. N. Zmood and D. G. Holmes, "Stationary frame current regulation of PWM
inverters with zero steady-state error," Power Electronics, IEEE Transactions on,
vol. 18, pp. 814-822, 2003.
[124] G. K. Hung, C. C. Chang, and C. L. Chen, "Analysis and implementation of a
delay-compensated deadbeat current controller for solar inverters," Circuits,
Devices and Systems, IEE Proceedings -, vol. 148, pp. 279-286, 2001.
[125] M. Bojrup, "Advanced control of active lters in a battery charger application,"
Licentiate Thesis, Lund Institute of Technology, Lund,Sweden, 1999.
[126] T. Lianxiang and O. Boon-Teck, "Elimination of Harmonic Transfer Through
Converters in VSC-Based Multiterminal DC Systems by AC/DC Decoupling,"
Power Delivery, IEEE Transactions on, vol. 23, pp. 402-409, 2008.
[127] P. Mattavelli, G. Spiazzi, and P. Tenti, "Predictive digital control of power factor
preregulators with input voltage estimation using disturbance observers," Power
Electronics, IEEE Transactions on, vol. 20, pp. 140-147, 2005.
[128] V. Blasko and V. Kaura, "A new mathematical model and control of a three-phase
AC-DC voltage source converter," Power Electronics, IEEE Transactions on, vol.
12, pp. 116-123, 1997.
[129] S. Chiniforoosh, J. Jatskevich, A. Yazdani, V. Sood, V. Dinavahi, J. A. Martinez,
and A. Ramirez, "Definitions and Applications of Dynamic Average Models for

255

APPENDICES

Analysis of Power Systems," Power Delivery, IEEE Transactions on, vol. 25, pp.
2655-2669, 2010.
[130] J. Jatskevich, S. D. Pekarek, and A. Davoudi, "Parametric average-value model of
synchronous machine-rectifier systems," Energy Conversion, IEEE Transactions
on, vol. 21, pp. 9-18, 2006.
[131] A. Yazdani, "Electromagnetic transients of grid-tied photovoltaic systems based on
detailed and averaged models of the voltage-sourced converter," in Power and
Energy Society General Meeting, 2011 IEEE, 2011, pp. 1-8.
[132] C. Karawita and U. D. Annakkage, "Multi-Infeed HVDC Interaction Studies Using
Small-Signal Stability Assessment," Power Delivery, IEEE Transactions on, vol.
24, pp. 910-918, 2009.
[133] S. Cole, J. Beerten, and R. Belmans, "Generalized Dynamic VSC MTDC Model for
Power System Stability Studies," Power Systems, IEEE Transactions on, vol. 25,
pp. 1655-1662, 2010.
[134] D. J. Vowles and M. J. Gibbard, "Mudpack Software Package for the SmallSignal
Analysis of Power Systems (Version 10R-19)," The University of Adelaide April
2008.
[135] P. Rault, F. Colas, X. Guillaud, and S. Nguefeu, "Method for small signal stability
analysis of VSC-MTDC grids," in Power and Energy Society General Meeting,
2012 IEEE, 2012, pp. 1-7.
[136] R. Teodorescu, F. Blaabjerg, M. Liserre, and A. Dell'Aquila, "A stable three-phase
LCL-filter based active rectifier without damping," in Industry Applications
Conference, 2003. 38th IAS Annual Meeting. Conference Record of the, 2003, pp.
1552-1557 vol.3.
[137] M. Liserre, F. Blaabjerg, and S. Hansen, "Design and control of an LCL-filterbased three-phase active rectifier," Industry Applications, IEEE Transactions on,
vol. 41, pp. 1281-1291, 2005.
[138] (24-May-13). National Electricity Rules: Current Rules (55 ed.). Available:
http://www.aemc.gov.au/electricity/national-electricity-rules/current-rules.html
[139] Wikipedia. (24 May ). Available: https://en.wikipedia.org/wiki/Newton's_method
[140] J. Z. A. M. Gole, "VSC transmission limitations imposed by AC system strength
and AC impedance characteristics," presented at the The 10th International
Conference on AC and DC Power Transmission ACDC 2012 4 - 6 Birmingham,
UK, December,2012.
[141] S. Skogestad and I. Postlethwaite, Multivariable Feedback Control: Analysis and
Design, 2nd Edition. West Sussex, England: John Wiley & Sons Ltd., 2005.

256

Reference

[142] Wikipedia, the free encyclopedia. (15 October 2009, 30th Oct). Mandatory
renewable
energy
targets.
Available:
http://en.wikipedia.org/wiki/Mandatory_renewable_energy_targets
[143] Electricity Supply Industry Planning Council "ANNUAL PLANNING REPORT ",
ed, June 2009.
[144] AEMO. "AEMO
www.aemo.com.au

Australian

Energy

Market

Operator".

Available:

[145] R. d. S. a. A. Robbie, "Network Extensions to Remote Areas Part 2 Innamincka

Case Study " 26 November 2009.
[146] R. d. S. a. A. Robbie, "Network Extensions to Remote Areas Part 1 Planning
Considerations " 26 November 2009.
[147] M. H. Nguyen, T. K. Saha, and M. Eghbal, "Hybrid multi-terminal LCC HVDC
with a VSC Converter: A case study of Simplified South East Australian system,"
in Power and Energy Society General Meeting, 2012 IEEE, 2012, pp. 1-8.
[148] D. V. M. Gibbard, "Simplified 14-Generator Model of the SE Australian Power
System," ed. The University of Adelaide, South Australia, 2008.
[149] C. Karawita and U. D. Annakkage, "A Hybrid Network Model for Small Signal
Stability Analysis of Power Systems," Power Systems, IEEE Transactions on, vol.
25, pp. 443-451, 2010.
[150] The
Gezhouba
Shanghai
HVDC
project.
Available:
http://www.abb.com/industries/ap/db0003db004333/3940678d3a84b3dbc12577490
03bcd23.aspx
[151] F. YANG, "Study on Parameter Optimization of HDVC PI Controllers," Power
System Technology, vol. 30, pp. 15-20, June 2006.
[152] M. J. Gibbard and D. J. Vowles, "Reconciliation of methods of compensation for
PSSs in multimachine systems," Power Systems, IEEE Transactions on, vol. 19, pp.
463-472, 2004.
[153] R. Sadikovic, G. Andersson, and P. Korba, "Damping controller design for power
system oscillations," Intelligent Automation & Soft Computing, vol. 12, pp. 51-62,
2006.

257