244
Robust WAM-based Controller Coordination of FACTS Devices for Power System Transient Stability Enhancement by Sushama Rajaram WAGH Achieving International Excellence This thesis is presented for the degree of Doctor of Philosophy of The University of Western Australia Energy Systems Centre School of Electrical, Electronic and Computer Engineering 2011

Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

  • Upload
    others

  • View
    4

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

Robust WAM-based Controller Coordination of FACTS Devices for

Power System Transient Stability Enhancement

by

Sushama Rajaram WAGH

Achieving International Excellence

This thesis is presented for the degree of Doctor of Philosophy of The University of Western Australia

Energy Systems Centre School of Electrical, Electronic and Computer Engineering

2011

Page 2: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

(i)

ACKNOWLEDGEMENTS

I would like to express my sincere gratitude to my supervisor, Professor Tam Nguyen, for

giving me the opportunity to undertake this research under his guidance, and for the constant

support and encouragement provided by him throughout my PhD candidature at The University

of Western Australia (UWA).

This thesis could not have been completed without the timely support of Prof. Victor Sreeram,

who extended his cooperation at the critical moments of my stay at UWA. I must also thank

Prof. Brett Nener for facilitating my work at UWA. I would like to place on record the support

extended to me by the staff at the Energy Systems Centre and, especial mention is deserved for

the regular technical inputs along with the healthy dose of moral support provided by Dr. Van

Liem Nguyen.

Off course, all of this could not have been possible without the Ad hoc scholarship awarded by

the Energy Systems Centre, and The University of Western Australia.

I am highly obliged to the Board of Directors of Veermata Jijabai Technological Institute

(VJTI) who has supported me in my endeavour for higher learning. I must take this opportunity

to thank, Prof. H. A. Mangalvedekar, who has been my foremost well-wisher and guide in the

academic career. Special thanks go to Prof. N. M. Singh for his support which helped me to

remain focused throughout my research work.

I am incredibly grateful to my family. My mother and sister Vijaya, brother-in-law, Rajesh,

gave me their unconditional support during all these years. I also appreciate the contribution of

Sunitha and Robert Mendonca who have been a pillar of support for my family. It would have

been difficult to continue to work at UWA, without Sudhir Bhil and Asha Sharma taking care of

things back home during my absence.

To my parents, I owe my love of knowledge and desire to excel. I would like to dedicate this

thesis to my mother, Mrs. Yamuna Rajaram Wagh, who took every effort, in providing

education to her children despite the adverse conditions she had to face after the sudden loss of

our father.

Page 3: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

(ii)

ABSTRACT This thesis is devoted to the development of new schemes for transient stability

enhancement using FACTS (flexible alternating current transmission system) devices,

and the control coordination of power systems with FACTS devices in transient state of

post-fault scenario. The key objectives of the research reported in the thesis are, through

online control coordination based on the models of power systems having FACTS

devices, those of maximising and restoring system transient stability following a

disturbance or contingency.

The new schemes are developed in two steps with increasing complexity in modeling

and in terms of considering the size of power system. In the first part of the thesis,

explaining detailed dynamic modeling of power system components with a

comprehensive literature review of controllers used in power system, a new model

predictive control based TCSC controller is designed, developed, and implemented for

SMIB. The key contribution of this new MPC-based TCSC controller is the application

of MPC methodology to power systems which is represented by detailed dynamic

modelling and coordinated with power system primary controllers i.e. exciter and

prime-movers. The simulation study is carried out with a single-machine-infinite-bus

system and the effectiveness of the new controller is validated.

Having established the foundation provided by the comprehensive models developed

for representing power systems with FACTS devices, including the TCSC, the research

in the second part focuses on real-time control coordination of power system

controllers, with the main purpose of restoring power system stability following a

disturbance or contingency.

In the second part of the thesis, a detailed literature review is presented to highlight the

need of real-time controllers in WAN (wide-area network), which is monitored and

operated based on information provided by WAM (wide-area measurement systems).

The practical problems in real-time controller requirements are discussed with review

which gives motivation to the new innovative scheme developed in the second part. The

second part of the thesis develops a novel scheme to predict future system status by a

Page 4: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

(iii)

look-ahead approach and tries to maintain system stability by introducing proper

compensation. The key contribution of this proposed scheme is not only in improving

system stability but it also gives a brief account of various times required for executing

every module of controller in generating required control law.

Although the developed online control coordination scheme is studied and analysed

using only series compensation, the same concept can be further extended for shunt

compensations and improving system performance in different dominations. Therefore,

the model developed in this research can be considered as general.

Another notable contribution with this developed scheme is it is validated using

standard large size power system represented with a detailed dynamic model of every

power system component. The TCSCs used for providing necessary compensation have

been represented by its detailed dynamic models and smooth controller coordination

between power system and TCSCs is achieved.

The key contribution in developing new online control coordination is that of practical

considerations given to various time delays. This thesis develops a general feasible

approach considering the requirements of real-time control while taking into account

various time delays. The communication channel delays and computation delays which

are significant in a large system which is geographically wide-spread are taken into

account for deriving a feasible control coordination scheme. It also considers and

discusses the computation system requirements of a real-time controller, given the large

amount of data to be handled within given time constraints while maintaining stability

in a post-fault region.

Drawing on the constrained optimization based on search technique together with the

new developed control coordination scheme, the method has been validated for a

multimachine power system network of New England. The search technique developed

reduces the computing time significantly as compared to standard optimization methods

such as Newton’s methods.

The third part of the thesis develops the strategy for an estimation of the internal state

variables of a synchronous generator based on available measurements. At present, as

Page 5: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

(iv)

the cost of phasor measurement units (PMUs) and wide-area communication network is

on the decrease, the research proposes and develops a new estimation strategy to

investigate the actual values of internal state variables where direct measurements are

not available and only available limited sets of measurements can be used. The new

estimation procedure eliminates the assumption of systems at steady state and using set

of nonlinear equations, processed through iterative algorithm, finds the exact state

values. The advantage of this new estimation algorithm results in accurate comment on

further power system studies which will be based on accurate values rather than

assumed values.

The research can be further extended to combine the new controller schemes proposed,

with estimation algorithms developed in third part to improve the effectiveness of

controller coordination schemes in enhancing power system stability in given time

bounds.

Page 6: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

(v)

LIST OF PRINCIPAL SYMBOLS SYMBOLS USED IN CHAPTER 2

d-axis direct-axis (machine base)

q-axis quadrature-axis (machine base)

p Derivative operator (d/dt)

is Vector of stator current of synchronous machine

id Direct-axis rotor current (machine base)

iq Quadrature-axis rotor current (machine base)

υr Vector of rotor voltage of synchronous machine

υd Direct-axis rotor voltage (machine base)

υq Quadrature-axis rotor voltage (machine base)

Ψr Vector of rotor flux linkages of synchronous machine

ψfd Main field winding rotor flux

Ψkd Direct-axis damper winding flux

ψfd quadrature-axis damper winding flux

Efd Synchronous machine field voltage

Am, Fm, Km Matrices depending on synchronous machine parameters

ωr Rotor angular frequency of synchronous machine

ωref synchronous speed (reference speed)

δr Rotor angle of synchronous machine

TM mechanical torque input to synchronous machine

Te electromagnetic torque output of synchronous machine

L,G, R, Matrices of synchronous machine parameters

Lss, Lrr, Lsr, Lrs,

Gss, Gsr, Rss, Rrr

Sub-matrices of L,G,R based on synchronous machine

parameters

i vector of stator and rotor currents

υ Vector of stator and rotor voltages of synchronous

machine

Pm, Zm Matrices depending on synchronous machine parameters

and rotor angular velocity

Vref Voltage reference for excitation control system

Page 7: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

(vi)

Vpss Supplementary signal from power system stabiliser (PSS)

XExe State vector of excitation system

VR automatic voltage regulator output

VRmin , VRmax Minimum and maximum voltage limit for AVR

Rf rate feedback

AExc , BExc Matrices depending on gains and time constants of

excitation system controller

XGov State vector for prime-mover controller

AGov , BGov , CGov ,

DGov

Matrices depending on gains and time constants of

prime-mover controller

PC Initial mechanical power

H Synchronous machine inertia constant (sec.kW/KVA)

M Synchronous machine inertia constant (pu)

VD Real part of synchronous machine voltage (system base)

VQ Imaginary part of synchronous machine voltage (system

base)

ID Real part of synchronous machine current (system base)

IQ Imaginary part of synchronous machine current (system

base)

ngen Total number of generators in power system network

nnode Total number of nodes in power system

nload Total number of load nodes in power system

Ii Current injected at node i

IBus Matrix of node currents

VBus Matrix of node voltages

Ig , Vg The current and voltage at the generator node

IL , VL the current and voltages at the loads node

PL , QL Active and reactive power load

Pr active power received

Qr reactive power received

G1 the load conductance

B1 load susptance

Vr Receiving end voltage

XSVC State vector of SVC main controller

Page 8: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

(vii)

ASVC , BSVC , CSVC Matrices depending on gains and time constants of SVC

XC , XL Fixed series capacitor and reactor or TCR

α Thyristor firing angle

Xbypass Bypass reactance of TCSC

Xtcscmin minimum (capacitive) limit of TCSC reactance

Xtcscmax maximum (inductive) limits of TCSC reactance

X1, X2 Intermediate state variables of TCSC model

XSDC Supplementary signal from supplementary damping

controller

Pe Electrical power

Xtcsc State vector for TCSC main controller

Atcsc, Btcsc Matrices depending on gains and time constants of TCSC

controller

SYMBOLS USED IN CHAPTERS 4 AND 5

k The current time instant

N Predicted horizon

NC Control horizon

R, Q Weighting matrices (tuning parameters )

Hp prediction horizon

x Vector of state variables

u Vector of control variables

y Vector of output variables

yref Reference trajectory for output variables

∆t time step

SYMBOLS USED IN CHAPTERS 7, 8 and 9

δi1 Relative rotor angle with respect to reference bus

Tk kth time instant

u Control variable

umax and umin Upper and lower limits of control vector respectively

T Control time window size

Page 9: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

(viii)

T1 First control window

TD1 and TD2 Communication channel delays

TC Computation delay

M Number of control periods

NC Number of polynomial coefficients

NP Number of polynomials

Page 10: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

(ix)

GLOSSARY

AVR Automatic Voltage Regulator

CTW Control Time window

DC Direct Current

EEAC Extended Equal Area Criterion

FACTS Flexible Alternating Current Transmission System

MPC Model Predictive Control

NR Newton-Raphson

PID Proportional-Integral-Differential (controller)

PMU Phasor Measurement Unit

RHC Receding Horizon Control

SCADA Supervisory Control and Data Acquisition

SSSC Static Synchronous Series Compensator

STATCOM Static Synchronous Compensator

SVC Static VAr Compensator

SVR Secondary Voltage Control

TCR Thyristor Controlled Reactor

TCSC Thyristor Controlled Series compensation

TSC Thyristor Switched Capacitor

UPFC Unified Power Flow Controller

WAN Wide-Area Network

WAMS Wide-Area Measurement System

Page 11: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

(x)

LIST OF FIGURES Figure 2.1 Individual synchronous generator connected to external network

Figure 2.2 Angular relationships between external power system network and

individual machine reference axis

Figure 2.3 Transforming individual machine quantities to system frame of

reference for voltage

Figure 2.4 Typical SVC structure with main components

Figure 2.5 SVC dynamic model block diagram

Figure 2.6 STATCOM schematic diagram

Figure 2.7 Phasor diagram of STATCOM operating principle

Figure 2.8 Basic voltage sourced converter structure using GTO thyristors

Figure 2.9 Dynamic model of STATCOM

Figure 2.10 Simplified operating circuit of thyristor controlled series

compensation

Figure 2.11 Typical layout of practical TCSC structure

Figure 2.12 Typical V-I capability characteristics for a single-module TCSC

Figure 2.13 Practical operating range of TCSC for inductive and capacitive

compensation

Figure 2.14 Multi-module TCSC

Figure 2.15 TCSC schematic block diagram with SDC

Figure2.16 Simplified schematic of supplementary damping controller

Figure2.17 Multi-machine power system network having multiple FACTS devices

Figure 3.1 Typical wave specifications expressed in terms of front and tail of

wave

Figure 3.2 Time frame of various transient phenomena

Figure 3. 3 Classification of power system stability studies based on physical

nature of the phenomena

Figure 4.1

Finite horizon and Infinite horizons with no disturbance and perfect

model respectively

Figure 5.1 Proposed strategy for RHC-based TCSC controller

Figure 5.2 Flowchart of RHC implementation algorithm

Page 12: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

(xi)

Figure 5.3 Single-machine-infinite-bus system

Figure 5.4 Generator rotor angle response without RHC controller

Figure 5.5 TCSC reactance reference input using RHC controller

Figure 5.6 Synchronous generator rotor angle response using RHC based TCSC

controller

Figure 6.1 Power flow and information flow network in wide-area network

Figure 7.1 Control scheme timing diagram: Option 1

Figure 7.2 Control scheme timing diagram: Option 2

Figure 7.3 Flowchart of control coordination scheme

Figure 8.1 Control coordination block diagram

Figure 8.2 10-Generator 39-Node New England test system

Figure 8.3 Relative rotor angle transients without online control coordination of

TCSCs

Figure 8.4 TCSC input references for case 1

Figure 8.5 Relative rotor angle transients with online control coordination of

TCSCs for case 1

Figure 8.6 TCSC input references for case 2

Figure 8.7 Relative rotor angle transients with online control coordination of

TCSCs for case 2

Figure 8.8 TCSC input reference responses Option 2

Figure 8.9 Relative rotor angle transients Option 2

Figure 8.10 TCSC input reference responses Approximate control

Figure 8.11 Relative rotor angle transients Approximate control

Figure 9.1 Comparison between actual and estimated rotor angle

Figure 9.2 Comparison between actual and estimated main field flux

Figure 9.3 Comparison between actual and estimated d-axis damper winding flux

Figure 9.4 Comparison between actual and estimated q-axis damper winding flux

Page 13: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

(xii)

TABLE OF CONTENTS

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

1.1 BACKGROUND AND SCOPE OF THE RESEARCH………........ 1

1.2 OBJECTIVES………………………………………………………. 2

1.3 OUTLINE OF THE THESIS……………………………….............. 5

1.4 CONTRIBUTIONS OF THE THESIS……………………………... 9

Chapter 2 Dynamic Modelling of Power System Components……………... 11

2.1 INTRODUCTION…………………………………………………. 11

2.2 SYNCHRONOUS GENERATOR DYNAMIC MODEL…………. 12

2.2.1 Rotor flux dynamics (Generator electrical axis)……………………. 12

2.2.2 Equation of motion (Turbine/Generator mechanical axis)…………. 13

2.2.3 Relation between generator current and voltage……………………. 14

2.2.4 Excitation and automatic voltage regulator (AVR)………………… 14

2.2.5 Prime-mover and governor system…………………………………. 15

2.2.6 Power system stabiliser (PSS)……………………………………… 16

2.3 MULTIMACHINE DYNAMIC MODELLING WITH MACHINE

REFERENCE………………………………………………………..

16

2.4 EFFECT OF MACHINE MODEL USED ON STABILITY

STUDY……………………………………………………………...

19

2.5 MULTI-MACHINE MODELLING WITH SYSTEM

REFERENCE......................................................................................

20

2.5.1 Connection of individual generator to power grid………………….. 20

2.5.2 Transferring quantities from machine reference to system reference

axis …………………………………………………………………

22

2.6 INTRODUCTION TO FACTS DEVICES……………………......... 23

2.7 STATIC VAR COMPENSATOR (SVC)…………………………... 24

Page 14: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

(xiii)

2.8 STATIC SYNCHRONOUS COMPENSATOR (STATCOM)…….. 26

2.8.1 Working principle of STATCOM…………………………………... 27

2.8.2 Dynamic model of STATCOM…………………………………….. 29

2.9 THYRISTOR CONTROLLED SERIES COMPENSATION

(TCSC)………………………………………………………………

29

2.9.1 Structure and operation of TCSC…………………………………… 30

2.9.2 Dynamic model of TCSC with SDC………………………………... 33

2.10 SUPPLEMENTARY DAMPING CONTROLLER (SDC)………… 34

2.11 POWER SYSTEM NETWORK MODELLING…………………… 35

2.12 LOAD MODELLING……………………………………………… 36

2.12.1 Static load modelling……………………………………………….. 38

2.12.2 Dynamic load modelling……………………………………………. 38

2.12.3 Impact of load modelling on power system transient stability……... 39

2.13 CONCLUSIONS……………………………………………………. 40

Chapter 3 Overview of Power System Issues and Solutions………………... 42

3.1 INTRODUCTION TO POWER GRID OPERATION…………….. 42

3.2 POWER SYSTEM ISSUES: CAUSES AND SOLUTIONS………. 43

3.2.1 Power system reliability and security………………………………. 43

3.2.2 Classification of power system transients…………………………... 44

3.2.3 Disturbances/causes of power system problems……………………. 46

3.2.4 Power system stability issues……………………………………….. 47

3.2.5 Solution for power system stability: Need of FACTS devices…....... 51

3.3 TCSC USED FOR POWER SYSTEM PERFORMANCE

ENHANCEMENT…………………………………………………..

52

3.3.1 TCSC implementation history…………………………………........ 53

3.3.2 TCSC placement and its benefits…………………………………… 54

Page 15: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

(xiv)

3.4 REVIEW OF PREDICTIVE CONTROL BASED METHODS…… 55

3.4.1 Voltage control and small-signal stability applications…………….. 56

3.4.2 Predictive control based controllers for transient stability

improvement .……………………………………………………….

57

3.5 CONCLUSION…..…………………………………………………. 59

Chapter 4 Theory and Overview of Predictive Controllers………………. 61

4.1 INTRODUCTION…………………………………………………. 61

4.2 PREDICTIVE CONTROL METHODOLOGY……………………. 62

4.2.1 Model predictive control (MPC)……………………………………. 62

4.2.2 Receding horizon control (RHC)…………………………………… 63

4.2.3 General MPC problem formulation………………………………… 64

4.2.4 Comparison of PID and MPC ……………………………………… 65

4.2.5 Strengths of predictive controlled based methods………………….. 66

4.2.6 Predictive control developments in literature………………………. 67

4.3 MAJOR DRAWBACKS OF PREDICTIVE CONTROL…………. 69

4.3.1 Stability issue……………………………………………………….. 70

4.3.2 Choice of horizon…………………………………………………… 71

4.4 LIMITATIONS OF MPC APPLICATION TO POWER SYSTEM 73

4.4.1 Linear model for nonlinear system issue…………………………… 75

4.4.2 Use of single machine equivalent…………………………………... 76

4.5 CONCLUSIONS…………………………………………………… 77

Chapter 5 MPC-Based TCSC Controller for Power System Transient

Stability Improvement…………………………………………...

78

5.1 INTRODUCTION AND OBJECTIVES………………………….. 78

5.2 AIM OF PROPOSED METHOD…………………………………... 81

5.3 POWER SYSTEM MODELING………………………………… 82

Page 16: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

(xv)

5.4 RHC ALGORITHM……………………………………………….. 83

5.5 LINEARIZATION AND OBJECTIVE FUNCTION……………… 84

5.6 RHC ALGORITHM FLOW CHART……………………………… 86

5.7 SIMULATION RESULTS…………………………………………. 87

5.8 CONCLUSIONS……………………………………………………. 89

Chapter 6 Overview of Real Time Controllers for Power System Stability

Improvement……………………………………………………...

91

6.1 INTRODUCTION………………………………………………...... 91

6.2 WIDE AREA NETWORK OPERATION………………………….. 92

6.3 ROLE OF WAM IN MAINTAINING WAN OPERATIONS……... 95

6.3.1 Advanced technology used in WAMs……………………………… 97

6.3.2 Role of communication network and various time delays………….. 99

6.4 REVIEW OF POWER SYSTEM CONTROLLERS…………......... 102

6.4.1 Controllers for power system performance enhancement………….. 102

6.4.2 Controllers for transient stability improvement…………………….. 103

6.4.3 Review of real time controllers……………………………………... 105

6.4.4 Communication delay consideration in controller design………….. 107

6.4.5 Computation requirements in controller design…………………….. 111

6.5 CONCLUSION…………………………………………………….. 113

Chapter 7 Online Control Coordination of FACTS Devices for Power

System Transient Stability: Control Method Derivation………..

115

7.1 INTRODUCTION………………………………………………….. 115

7.2 BACKGROUND OF PROPOSED SCHEME……………………… 116

7.3 TRANSIENT STABILITY CONTROL PRINCIPLE……………… 118

7.3.1 Power system model used…………………………………………... 118

7.3.2 Time-domain transient stability simulation………………………… 119

Page 17: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

(xvi)

7.3.3 Nonlinear relationship between maximum relative rotor angles and

control variables……………………………………………………..

120

7.3.4 Transient stability control concept………………………………...... 121

7.4 TRANSIENT STABILITY CONTROL SCHEME………………… 123

7.4.1 Option 1…………………………………………………………….. 123

7.4.2 Option 2…………………………………………………………….. 128

7.5 FORMING TRANSIENT STABILITY INDICES………………… 130

7.6 CONTROL COORDINATION FLOWCHART…………………… 132

7.7 CONCLUSIONS……………………………………………………. 134

Chapter 8 Online Control Coordination of FACTS Devices for Power

System Transient Stability: Computing Time Requirement

Analysis and Case-Study…………………………………………..

135

8.1 INTRODUCTION………………………………………………….. 135

8.2 COMPUTING TIME REQUIRMENTS……………………………. 136

8.3 ANALYSIS OF COMPUTING TIME REQUIRMENT…………… 137

8.3.1 Control coordination structure……………………………………… 137

8.3.2 Computing time…………………………………………………….. 138

8.4 CASE-STUDY SIMULATION RESULTS ………………………. 141

8.5 REPRESENTATIVE POWER SYSTEM………………………….. 142

8.6 SYSTEM RESPONSE WITHOUT ONLINE CONTROL

COORDINATION OF TCSCS……………………………...………

143

8.7 OUTLINE OF TCSCS CONTROL COORDINATION STUDY….. 144

8.8 TIME-DOMAIN SIMULATION COMPUTING TIME

REQUIREMENTS…………………………………………………..

145

8.9 COMPUTING TIME REQUIREMENTS FOR NONLINEAR

FUNCTION SYNTHESES…………………………………………

145

8.10 COMPUTING TIME FOR CONSTRAINED OPTIMIZATION….. 146

8.11 CONTROL COORDINATION STUDY RESULTS……………..... 146

Page 18: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

(xvii)

8.11.1 Control Option 1……………………………………………………. 146

8.11.2 Control Option 2…………………………………………………..... 151

8.11.3 Approximate control...……………………………………………… 152

8.12 CONCLUSION……..……………………………………………… 154

Chapter 9 Dynamic Modelling Application for Estimating Internal States

of a Synchronous Generator in Transient Operating Mode

from External Measurements………………..…………………...

155

9.1 INTRODUCTION…………………………………………………. 155

9.2 BACKGROUND THEORY……………………………………….. 156

9.3 DEVELOPMENT OF ESTIMATION PROCEDURE……………... 159

9.3.1 Continuous-time nonlinear dynamical system and estimation

requirement………………………………………………………….

159

9.3.2 Discrete-time domain system model………………………………... 160

9.3.3 Estimation problem formulation…………………………………..... 161

9.3.4 Solution method…………………………………………………….. 162

9.3.5 Estimation process for subsequent time instants…………………… 165

9.4 APPLICATION TO SYNCHRONOUS GENERATOR………...…. 166

9.4.1 Generator dynamic model………………………………………….. 166

9.4.2 Procedure for generator internal state estimation…………………... 167

9.4.3 Measurement requirements…………………………………………. 168

9.4.4 Discussion…………………………………………………………... 169

9.5 REPRESENTATIVE CASE STUDY……………………………… 171

9.5.1 Results……………………………………………………………… 172

9.5.2 Computing time…………………………………………………….. 173

9.6 CONCLUSIONS……………………………………………………. 174

Chapter 10 Conclusions and Future Work…………………………………… 175

Page 19: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

(xviii)

10.1 CONCLUSIONS…………………………………………………… 175

10.2 FUTURE WORK…………………………………………………… 177

10.2.1 MPC based controller for transient stability using multiple FACTS

devices ………………………………………………………………

177

10.2.2 Online control coordination scheme for multiple FACTS devices…. 178

10.2.3 Control coordination for power system stability improvements in

case of loss of communication signal information/ data…………….

178

10.2.4 Estimation of internal state variables for synchronous generator

with incomplete/inaccurate measurements …………………………

179

10.2.5 Estimation of internal state variables for exciter, turbine and

governor system……………………………………………………..

179

Bibliography …………………………………………………….. 181

Appendices……………………………………………………….. 192

Page 20: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

1

Chapter 1 Introduction

1.1 BACKGROUND AND SCOPE OF THE RESEARCH

The economic growth of any country depends on the growth of advanced modern

technological developments, and advanced technology is impossible without reliable

continuous electric power networks. In terms of reliability and continuity, the cost of

‘power that is not supplied’ can be huge when it comes to the manufacturing/production

industry. In this context, the rate of increasing power demand, development of

technology, need of maintaining system reliability and continuity along with quality

will be the decision factors for future trends in power systems. In developed countries

like Australia, to fulfil this ever increasing demand along with increasing concern about

the environment the focus of the power sector has shifted from the creation of additional

capacity to more effective management and efficient utilisation of existing network

capacity. This is equally applicable to developing countries like India [1, 2].

One of the best and most economical options for maintaining continuity and reliability

of power supply is having ties with neighbouring networks by forming an

interconnected grid, which can allow power exchanging/sharing in peak and slack

periods. This is also known as power pooling where the spinning reserve is reduced or

need of having any spare units for planned or unplanned outages is minimised.

Page 21: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

2

Such interconnected ties, however, may sometimes be a source of catastrophic

situations because of the characteristics of electrical power networks. As the power

networks connected in parallel, the system fault level keeps increasing. Unlike other

products, electricity cannot be stored and is generated and consumed at the same time.

With unpredictable load changes and system variations, maintaining a power balance,

which is essential for constant frequency operation, becomes difficult. In addition, any

sudden disturbance like a short circuit in any part of the network can affect other

healthy parts in very short duration of time without giving much margin for power

system operators to take any corrective action. Thus, the interconnected grid formed to

maintain continuity can, during cases of disturbances become the cause of cascade

tripping or even a blackout, because of its complexity and power balance problem.

With the above background, it can be seen that the performance of power systems

decreases with increase in the size, the loading and the complexity of the network.

Because of their geographically dispersed networks, power systems require a

functionally complex monitoring and control system to fulfil the immediate and near

future priorities of higher availability and efficiency maintenance. This is possible only

with good information technology based services in energy management systems.

The major motivation of this research is, therefore, to maintain transient stability of an

interconnected grid via the real-time and optimal adjustment of flexible AC

transmission systems (FACTS) devices controller input references in post-fault

conditions. The control coordination law is derived based on real-time requirements, in

terms of time as well as computation capabilities to have a feasible solution.

1.2 OBJECTIVES

As the transient stability is a critical issue in power system reliability, the objective of

this thesis is to develop a control strategy which will introduce necessary compensation

using effective control of FACTS devices input references to maintain system stability

during post-fault conditions. The total thesis work is divided in four major parts. The

first part discusses the detailed modelling of various power system components,

including synchronous machine, exciter, prime-mover, governor, and various FACTS

Page 22: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

3

devices. After modelling, various power system issues and commonly used methods are

discussed highlighting the need of present research topic of transient stability

improvement using compensating devices.

In the second part, a receding horizon control methodology is developed to control the

TCSC input reference for improving transient stability. The effective control action is

confirmed with MATLAB simulation on a single-machine-infinite-bus system. In the

third part, a real-time controller-coordination based on time-domain simulations and

constrained optimisation is considered to propose optimal adjustment of FACTS

devices controller. The developed control coordination scheme is combined with

practical feasible time constraints to satisfy real-time requirements. The control

algorithm is implemented in two possible options based on how frequently the system

model is updated. Both the control strategies have proven effectiveness in improving

the transient stability of wide-area network validated with a standard New England

system of 10 generators.

The fourth part of the thesis is devoted to the estimation of internal state variables of

synchronous generators using the available measurements, as it is necessary to know the

initial state of the internal state variables which cannot be measured directly. Drawing

on the availability of synchronous generator terminal voltage and current measurements

together with those for rotor speed and/or field voltage, a procedure is derived for

estimating, in transient condition, the generator’s internal operating states, which

include rotor angle and flux linkages associated with field winding and damper

windings. The procedure is based on the fifth-order generator dynamic model formed in

terms of differential and algebraic equations. By applying the numerical integration

formula based on the trapezoidal rule, the generator model is described by a set of

algebraic equations in a recursive form in the discrete time-domain. Combining the

results of external measurements in successive sampling time instants with the set of

recursive equations, leads to a system of nonlinear equations in which the unknown

variables are those representing the internal operating states of the generator. A

minimisation technique based on the sequential quadratic programming method is then

applied to solve the variables in the nonlinear equation system. The estimation

procedure can be applied to any time instants including those in the transient operating

conditions. The effectiveness and accuracy of the procedure developed is tested by

Page 23: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

4

simulation using a representative multimachine power system operating in the transient

mode.

Given the context of the research described above the thesis will have the following

objectives:

(a) Power system modelling for transient stability. The models are to fulfil the

analysis requirements, and at the same time, should be able to give better insight in

internal controller parameters to keep track of the required performance. In this

context, detailed dynamic modelling of synchronous machine, exciter, various

FACTS devices and other power system components is derived and discussed. The

effects of dynamic modelling on transient stability study are discussed in relation to

models considered.

(b) RHC based TCSC controller. A receding horizon based TCSC controller is

developed for power system transient stability improvement for SMIB. The key

feature of the developed controller is that unlike previous literature where classical

models were used for implementation of predictive control methodology, the

detailed power system dynamic model is used for deriving optimal FACTS input

references.

(c) Real-time controller with control coordination schemes for transient stability

enhancement. A real-time control coordination scheme based on time-domain

simulation and constrained optimisation is derived, developed and implemented for

improving multimachine power system transient stability performance using

multiple FACTS devices. To build the most effective control strategy, two different

options are investigated. This is to account for frequent updates in the power system

model and system scenarios due to fault and/or fault clearing occurrences.

(d) Consideration of communication channel delays. Considering the geographically

wide-spread span of wide-area power network, and the necessity for good

information technology, the limitations of communication channel delays are taken

into account while developing control schemes mentioned in (c). Taking the

Page 24: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

5

communication delays into consideration will give a more realistic prediction of

system stability in the future, which is rarely covered in the control designs in

available literature.

(e) Consideration of Computation requirements. Based on the transient stability

control algorithm developed, a comprehensive analysis of the computing time

requirement in implementing the above control coordination algorithm is

performed. The analysis outcome is a set of feasible constraints which are to be

satisfied by computer systems used for online control coordination of FACTS

devices with the aim of enhancing or maintaining power system transient stability

following disturbances. With reference to a representative multimachine power

system having TCSCs, it is demonstrated that even with the available current

technology, the control coordination algorithm can be implemented in real time,

using a cluster of processors operating in parallel. Offline simulation, using realistic

timing parameters for the control scheme, confirms its effectiveness in maintaining

transient stability following a fault disturbance.

(f) Development of internal state estimation algorithm. Drawing on the availability

of synchronous generator terminal voltage and current measurements together with

those for rotor speed and/or field voltage, a procedure is derived for estimating in

transient conditions the generator internal operating states which include rotor angle

and flux linkages associated with field winding and damper windings. The

procedure is based on the fifth-order generator dynamic model formed in terms of

differential and algebraic equations. A minimisation technique based on a

sequential quadratic programming method is then applied for solving the variables

in the nonlinear equation system. The estimation procedure proposed can be applied

to any time instants including those in the transient operating conditions.

1.3 OUTLINE OF THE THESIS

Chapter 1 of the thesis gives the background and scope of research, its objectives and

contribution of the thesis.

Page 25: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

6

Chapter 2 is devoted to dynamic modelling of various power system components,

including the synchronous generator which is the key element to maintain synchronism.

Considering the major drawback of using classical model or third order model in

stability studies, special consideration is given to detailed modelling of synchronous

generator using fifth-order model along with detailed modelling of excitation, prime-

mover and governor system. The single machine modelling is extended to multimachine

dynamic modelling and their connection to the external grid is explained taking into

consideration the individual machine reference and system reference. The chapter

further presents modelling of various shunt and series FACTS devices, load modelling

and supplementary damping controllers. It also gives importance to, and the effects of,

modelling on transient stability study results.

Chapter 3 explains power grid formation, its problems, causes and issues associated in

maintaining reliable, continuity and quality of supply. For improvement in power

system performance, the need for FACTS devices and their applications for improving

stability are discussed and the effectiveness of TCSC is presented in this chapter.

Considering the need of predicting system scenarios in advance, based on a look-ahead

approach, a detailed review of the controllers based on prediction control methodology

is made which gives the foundation for further development of RHC based controller in

Chapter 5.

Chapter 4 gives an overview of receding horizon control methodology, as developed in

the field of control systems and its limitations. It also highlights the issues of concern

for application of predictive control strategies to power system issues. To get familiar

with the development of prediction control methodology, a brief review is presented to

explain the principle of RHC/MPC, its applications in other fields like control system or

chemical industry where it was well developed and well-implemented. Knowing the

practical limitations of MPC applications, a detailed review is presented of MPC

applications to power system specific problems focusing on stability related issues.

Chapter 5 develops a RHC based prediction scheme for investigating the required

TCSC reference inputs for enhancing power system transient stability. The power

system model considered is a single-machine-infinite-bus (SMIB) system, represented

by a detailed dynamic model of synchronous generator including a detailed dynamic

Page 26: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

7

model of an exciter and governor systems. The TCSC is modelled in variable reactance

form and based on the output of a MPC controller. Input is given to TCSC which will

insert the necessary series compensation in post-fault conditions. The method is

validated with case study results of the SMIB system taking into consideration a 3-

phase short circuit where the fault is cleared after critical clearing time. The case study

presented shows that without a controller the system will lose synchronism because of

high acceleration, while with the controller it is proved to regain stable operating

conditions in a short time.

Chapter 6 gives a detailed literature review of operations, maintenance and technologies

used in wide-area network and the role of wide-area measurement systems. It also

highlights the importance of good communication networks for fast and huge

information exchange, giving a brief account of various time delays in wide-area

network and its effect on stability studies. The chapter explains about the structure of

wide-area network and their limitations in terms of communication due to their wide-

spread nature in stability problems. A detailed literature review of methods and schemes

addressing the problems of communication and computation delays for real-time

applications is presented.

A new method is developed in Chapter 7 for real-time transient stability control in a

power system with FACTS devices. Central to the method is the control, in successive

time periods, of a synchronous generator is relative rotor angles to satisfy the nominated

transient stability criterion via the real-time and optimal adjustment of FACTS devices

controllers input references. In each period, the dependencies of maximum relative rotor

angles on input references are expressed as nonlinear functions which are synthesised

from the results of time-domain transient stability simulations, using the prevailing

power system models and conditions. The constrained optimisation problem is then

formed from the synthesised functions, and solved for the optimal input references.

Practical issues related to computing time and communication channel time delays are

considered in the control methodology.

In the context of the online transient stability control method developed in chapter 7,

based on time-domain simulations and constrained optimisation for real-time and

optimal adjustment of FACTS devices controllers’ input references, Chapter 8 has a

Page 27: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

8

focus on the analysis of computing time requirements in implementing the method, and

presents results of study cases. The computing time components of the individual steps

for executing the control algorithm are identified, determined, and then combined to

form overall feasibility constraints to then be satisfied by computer systems used in

control method implementation. Computational tasks which are independent of one

another are identified so that they can be performed using parallel computing systems.

The effectiveness of the control coordination in maintaining transient stability is

verified by a simulation study, with the control scheme having realistic timing

parameters applied to a representative multimachine power system.

Chapter 9 explains and derives new methodologies for internal state variable estimation.

Drawing on the availability of synchronous generator terminal voltage and currents,

rotor speed and field winding voltage measurements, a procedure is derived for

estimating in transient condition the generators internal operating states which are the

rotor angle and flux linkages associated with field winding and damper windings. The

procedure is based on the fifth-order generator dynamic model. By applying the

numerical integration formula based on the trapezoidal rule, the generator model is

described by a set of algebraic equations of a recursive form in the discrete time-

domain. With external measurements, the unknown variables in the equations are those

representing the generator’s internal operating states. The nonlinear equations derived

for successive time instants are solved by applying the Newton-Raphson method.

However, if the number of equations is greater than that of variables, a minimisation

technique based on sequential quadratic programming method is then applied to solve

the nonlinear equation system. The estimation procedure can be applied to any time

instants including those in the transient operating conditions, without the requirement

for specifying the steady-state condition. The effectiveness and accuracy of the

procedure developed are verified by simulation using a representative multimachine

power system operating in the transient mode.

Chapter 10 gives the overall conclusion for the thesis and the future scope of the

ongoing research in relation to further development of the schemes proposed in

Chapters 5, 7, 8 and 9.

Page 28: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

9

Appendices are provided for detailed derivations of major formulation which are used in

the modelling of power system components, developing MPC algorithms and estimation

problem formulation. They also give the complete input data files for the power system

case-studies used in the thesis for validation of results.

1.4 CONTRIBUTIONS OF THE THESIS

The thesis has made five original contributions described below:

(a) Development of an RHC-based TCSC controller for power system transient

stability enhancement considering a detailed dynamic model of a power system.

(b) Development of robust online controller coordination schemes for improvement of

power system transient stability in multi-machine, multiple FACTS devices using

two different strategies.

(c) This controller is further developed to take into consideration the size of power

system networks spread over wide geographical areas causing various time-delays

in WAN communication channels.

(d) Special consideration is given to the computation time and computation burden

forced on computing system requirements in developing this real-time control

coordination scheme for multi-machine, multi-TCSC hybrid network considering

detailed nonlinear dynamic models.

(e) New method is derived to find out values of internal state variables of a

synchronous generator based on practical feasible measurements for real time

application.

Page 29: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

10

The thesis is supported by two published international conference papers and two

journal papers submitted which are under review at present. The list of publications is as

follows:

1) Nguyen, T.T., and Wagh, S.R., “Model Predictive Control of FACTS Devices

for Power System Transient Stability,” Proceedings of the IEEE Transmission

and Distribution Conference, Seoul, Korea, October, 2009.

2) Nguyen, T.T., and Wagh, S.R., “Predictive Control-Based FACTS Devices for

Power System Transient Stability Improvement,” Proceedings of the 8th IET

International Conference on Advances in Power System Control, Operation and

Management, APSCOM 2009, Hong Kong, November, 2009.

3) Nguyen, T.T., and S. R. Wagh, “Application of Dynamic Modelling for

Estimating Internal States of a Synchronous Generator in Transient Operating

Mode from External Measurements”, submitted to IEEE Trans. Power Systems,

2011 (under review).

4) Nguyen, T.T., and S. R. Wagh, “Online Control Coordination of TCSCs for

Power System Transient Stability”, submitted to IET Generation, Transmission

and Distribution, December 2011 (under review).

For reference, copies of the above four publications are given in Appendix H.

Page 30: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

11

Chapter 2 Dynamic Modelling of Power System

Components

2.1 INTRODUCTION

The first step for power system stability study is dynamic modelling of various power

system components. The power system elements include synchronous generators,

transformers, transmission lines, various compensating devices such as that of FACTS

devices which are modelled in forming the complete network model for stability study.

For steady-state power-flow analysis, power system can be well represented using nodal

variables including node voltages and node currents. For dynamic stability studies of

multimachine power system network, first dynamic models of individual items should

be formed and then their interaction with external power system can be considered.

With reference to the transient stability study of power systems after a large disturbance,

the system dynamic performance is decided by rotating machines dynamics and power

system controller including FACTS devices responses which are part of interconnected

network. Hence, the entire power system dynamics can be divided in two major parts,

first, the individual synchronous machine which is working on its own individual

Page 31: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

12

machine axis reference, and second, its connection to the external power system

network which is working on a common system reference axis. To understand the

dynamic behaviour of power systems, first, it is necessary to focus on the detailed

dynamic model of synchronous machines which consist of synchronous machine along

with its exciter and prime mover. In the present chapter the dynamic modelling of

synchronous machine is discussed along with exciter and prime mover which are further

extended and developed for multi-machine system. The latter part of the chapter will

focus on the dynamic modelling of various FACTS devices and loads. The chapter also

discusses the effects of choosing appropriate dynamic models on the power system

performance parameters.

2.2 SYNCHRONOUS GENERATOR DYNAMIC MODEL

Due to the various limitations with classical and reduced order models, in this thesis, the

synchronous machine is represented by the fifth-order model having d-q axis as the

rotor reference frame. To develop basic dynamic equations for balanced , symmetrical,

three-phase synchronous machine, a generator with one field winding on direct axis and

two damper windings, one on d-axis and one on q-axis, is considered [3].

In the two-axis theory of synchronous machines, the three phases of the armature

winding are replaced by fictitious direct-axis and quadrature-axis. Hence the various

self and mutual inductances between stator and rotor circuits can be summarized as

given in [4], with d-q axis terminology in terms of shaft angle θ. From the expression

(A1) of voltage as given in Appendix A, the equations of rotor flux and stator voltages

can be derived in terms of d-q axis terminology which is useful in deriving rotor

dynamic model in the next sections.

2.2.1 Rotor flux dynamics (Generator electrical axis)

As shown in the Appendix A, the advantage can often be taken, in practical stability

studies based on the synchronously rotating reference frame, of reducing the computing

time expanded in analysis by eliminating the terms of stator-voltage transients

corresponding to the rate of change of stator flux linkages with respect to time. These

Page 32: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

13

terms contribute very little dynamic response analyses that are dominated by the inertial

characteristics of rotating machines and they may be readily eliminated. In this research,

the synchronous machine is represented by the fifth-order model of the d-q axis having

a rotor frame of reference. Having one main field winding and two damper windings,

the rotor flux equation can be given by[4, 5] :

rsmrmr viFAp ++= ψψ (2. 1)

which can be expanded for three rotor fluxes separately as:

fdmqmdmkqmkdmfdmfd EKiFiFAAA 111211131211 +++++=•

ψψψψ (2. 2)

qmdmkqmkdmfdmkd iFiFAAA 2221232221 ++++=•

ψψψψ (2. 3)

qmdmkqmkdmfdmkq iFiFAAA 3231333231 ++++=•

ψψψψ (2. 4)

where constants Am and Fm are dependent on machine parameters as derived in

Appendix A. Flux linkages fdψ , kdψ and kqψ constitute the rotor flux linkages vector

while di and qi are stator current components along d- and q-axis respectively.

It should be noted that, physically, exciter output voltage/current and generator field

voltage/current are the same, but distinction is made only in their per unit values to

allow independent selection of the per unit systems for modelling excitation systems

and synchronous machines. Hence, the constant Km11 is used for proper interfacing

between excitation system and synchronous machine field circuits when modelled in per

unit system and is calculated based on machine parameters as explained in [6] .

2.2.2 Equation of motion (Turbine/Generator mechanical axis)

The equation of motion which is also known as the swing equation is expressed as

shown in (2.5) and (2.6). It gives the relationship between the rotor angle and rotor

angular frequency and is represented in rotor mechanical axis.

Page 33: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

14

refr ωωδ −=•

(2. 5)

eM TTM −=•ω (2. 6)

where, rδ is rotor angle, ωref is synchronous speed while ω is rotor angular frequency.

M is calculated as 2H/ωref, where H is the machine inertia constant. TM is the

mechanical torque input to the synchronous generator, while Te is the electromagnetic

torque output of the generator. The electrical torque Te can be expressed in terms of

generator currents as given in (2.7)

BBIIAAIT tss

tse refref ωω += (2.7)

where AA and BB are matrices dependent on machine parameters and are derived in

Appendix A (as A31).

2.2.3 Relation between generator current and voltage

Having discarded stator flux transients which are very fast compared to other dynamics,

the stator voltage vector can be represented in terms of stator current by an algebraic

equation as shown in (2.8) which is derived in Appendix A.

smrms izpv −= ψ (2.8)

where, Pm and Zm matrices are dependent on machine parameters and rotor angular

frequency and can be calculated as shown in Appendix A (given by A13).

2.2.4 Excitation and automatic voltage regulator (AVR)

In transient stability study it is customary to consider excitation circuit model in detail

to obtain an accurate system response. While many excitation models for stability

studies are specified in [7], they generally consist of five major blocks including the

main exciter; regulator; terminal voltage transducer and load compensator; power

system stabiliser; limiters and protective circuits as mentioned in [6] and shown in Fig.

B1 in Appendix B.

In this research, the IEEE Type I excitation model is considered and its schematic block

diagram is shown in Fig. B2. As derived in Appendix B, the exciter model can be given

Page 34: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

15

by a set of equations representing the entire excitation system including an automatic

voltage regulator with its limits. However, the set of equations can be expressed in a

general form as:

)9.2(ExcExcExcExcExc VBXAX +=•

where, XExc is state vector of the excitation system, VExc is the synchronous machine

terminal voltage, VPSS is the supplementary signal from PSS and Vref is the voltage

reference, while AExc and BExc are the constant matrices dependent on the gains and

time constants of the controller. It should be also be noted that (2.9) provides a link

between the exciter controller and the synchronous generator terminal output voltage

through matrix VExc as shown in the detailed model in Appendix B.

2.2.5 Prime-mover and governor system

It is well known that the frequency of the ac voltage at the terminals of a synchronous

generator is determined by its shaft speed and the number of magnetic poles of the

machine. The steady-state speed of a synchronous machine is determined by the speed

of the prime mover that drives the shaft. The prime-movers can be of steam type, gas

turbine or hydro turbines. In [8, 9] various dynamic models for steam and hydro

turbines are explained. A nonreheat type of steam turbine chosen as the prime mover.

Fig. B3 shows the block diagram representation of turbine and governor combined

model. In this present study, steam chest dynamics and the effects of steam valve

positions (PSV) on the synchronous machine torque (TM) are modelled as shown in [10].

Equations (B7)-(B8) give a complete model of turbine and governor systems which can

be represented in compact notations as shown in (2.10) where XGov is the state vector of

the prime-mover controller and AGov, BGov, CGov, and DGov are matrices dependent on

gain and time constants of the controller.

CGovGovGovGovGovGov PDCBXAX +++=•

ωω ref (2.10)

It can be seen from the detailed derivation in Appendix B, that (2.10) provides a link of

interconnection of the prime-mover output to the synchronous generator power.

Page 35: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

16

2.2.6 Power system stabiliser (PSS)

In addition to the above discussed elements, the power system stabiliser (PSS) is

another common stabiliser used to introduce modulating signals through the excitation

system to contribute to rotor oscillation damping.

A block diagram for PSS is shown in Fig. B4, which consists of a gain block, a

washout, a lead-lag block and a limiter. A washout is necessary to guarantee that PSS

responds only for the disturbances and not for any steady-state condition when speed or

power is used as input. The output of PSS is added to the exciter error signal and can be

used as a supplementary signal.

The state equations derived from the given block diagram are explained in detail in

Appendix B and can be summarised in compact form as :

XBXAX PSSPSSPSSPSS += (2.11)

where XPSS is the vector of state variables of PSS as described in Appendix B and, APSS,

BPSS are matrices elements depending on the gain and time constant of the PSS

controller while X is the speed variation.

2.3 MULTIMACHINE DYNAMIC MODELLING WITH MACHINE

REFERENCE

Having modelled the main components of synchronous machines, the overall generator

dynamics for single machine systems can be summarized as a set of differential

equations collected and represented together as follows:

rsmrmr viFA ++=•

ψψ (2.1)

refr ωωδ −=•

(2.5)

eM TTM −=•ω (2.6)

Page 36: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

17

ExcExcExcExcExc VBXAX +=•

(2.9)

CGovGovGovGovGovGov PDCBXAX +++=•

ωω ref (2.10)

For interconnected networks with a number of synchronous machines, a generalised

equivalent machine model can be formed by extending the same modelling concept as

explained in Section 2.2. Using the same terminology for a multimachine system having

N number of machines, the rotor flux dynamics on its own generator electrical axis will

be given by:

rNsNmNrNNmrN viFAp ++= ψψ (2.12)

which can be expanded on the same line as that of a single machine for extracting

individual rotor fluxes for field winding, as well as both damper windings as:

fdNNmqNNmdNNm

kqNNmkdNNmfdNNmfdN

EKiFiF

AAA

111211

131211

++

+++=•

ψψψψ (2.13)

qNNmdNNm

kqNNmkdNNmfdNNmkdN

iFiF

AAA

2221

232221

+

+++=•

ψψψψ

(2.14)

qNNmdNNm

kqNNmkdNNmfdNNmkqN

iFiF

AAA

3231

333231

+

+++=•

ψψψψ (2.15)

where constants AmN and FmN are dependent on machine parameters as explained in

Appendix B. Flux linkages 𝜓𝑓𝑑𝑁, 𝜓𝑘𝑑𝑁 and 𝜓𝑘𝑞𝑁 constitute the rotor flux linkages

vector, while idN and iqN are stator current components along d- and q-axes respectively.

The equation of motion based on the turbine/generator mechanical axis for all given N

synchronous generators can be generalised as:

refNNrN ωωδ −=•

(2.16)

eNMNNN TTM −=•

ω (2.17)

where rNδ is the vector of the rotor angle of all individual machines, ωNref is the

synchronous speed and ωN is the vector of the rotor angular frequency. MN is the

Page 37: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

18

machine inertia constant calculated from the per unit inertia constant H, as 2HN/ωNref,

TMN is the mechanical torque input to the synchronous generator, while TeN is the

electromagnetic torque output of the generator. The mechanical torque in (2.17) will be

as an input to the synchronous generator which is available as an output of turbine and

governor system, while the output expressed as power or electrical torque TeN can be

expressed as:

NtsNsNN

tNseN BBIIAAIT refNrefN ωω += (2.18)

For a transient stability study of multimachine system, the exciter and automatic voltage

regulators can be generally represented in compact notations as:

ExcNExcNExcNExcNExcN VBXAX +=•

(2.19)

The XExc is the vector of excitation system variables and the constants AExc and BExc are

dependent of given excitation system model chosen. The details of the notations used in

(2.19) are as given in the list of symbols.

Similar to the multimachine model formation of synchronous machine or exciter, even

the prime-mover and governor can be generalised for multimachine system. Equation

(2.20) gives a complete model of turbine and governor system which can be represented

in compact notations for multi-machine system having N number of machines, where

XGovN is state vector of prime-mover controller and AGovN, BGovN, CGovN, and DGovN are

matrices dependent on gain and time constants of the controller.

CNGovNNGovNGovNGovNGovNGovN PDCBXAX +++=•

ωω refN (2.20)

The overall generator dynamic model for the transient stability study of multimachine

can be summarized by the following sets of differential equations

rNsNmNrNmNrN viFA ++=•

ψψ (2.12)

refNNrN ωωδ −=•

(2.16)

eNMNNN TTM −=•

ω (2.17)

ExcNExcNExcNExcNExcN VBXAX +=•

(2.19)

CNGovNNGovNGovNGovNGovNGovN PDCBXAX +++=•

ωω refN (2.20)

Page 38: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

19

2.4 EFFECT OF MACHINE MODEL USED ON STABILITY STUDY

Synchronous machines may be modelled in as much detail as possible in the study of

most categories of power system stability. This includes appropriate representation

(subject to the available data) of the dynamics of the field circuit, excitation system, and

rotor damper circuits. With today’s computing tools, it is no longer a necessity to use

the simplified model. Experience has shown that critical problems may be masked by

the use of simplified models often perceived to be acceptable for a particular type of

study.

It is particularly important to represent the dynamics of the field circuit, as it has

significant influence on the effectiveness of excitation system in enhancing large-

disturbance rotor-angle stability. In current literature, generally a classical model for

study is used with only a rotor swing equation. Assuming that the variation in rotor

flux, exciter parameters and governor action is very small the other machine dynamic

equations are ignored and the computational burden and complexity in handling

dynamic equations are reduced. Assumptions for considering use of a classical model

are justified as follows:

(i) As magnetic energy is dissipated in the field winding resistance, flux decrement

effects will cause the rotor emf to decrease with time. If the fault clearing time is very

short then this flux decrement effect can be neglected for transient stability

considerations [11].

(ii) Similarly, the governor being a mechanical system has a large time constant so

compared to the fast action of transients, the effect of variation in governor parameters

is very small on transient performance.

However, if the fault-clearing time is long enough then the decay of this rotor emf will

have a considerable effect [11]. The result of considering flux decrement effects is in

reducing deceleration for given acceleration which will deteriorate the transient

stability. Similarly, most modern generators are equipped with an automatic voltage

regulator (AVR) which increases the synchronising torque component at steady state

[6]. AVR action may reduce the damping of rotor swing following a small disturbance

Page 39: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

20

or large disturbance. When fault occur the generator terminal voltage drops and the

large regulation error forces AVR to increase the generator field current. The effect of

AVR is to increase the field current leading to an increase in the transient emf so a

strong action of AVR may prevent a loss of synchronism after a large disturbance.

Although the fast acting AVR reduces the first rotor swing, it can increase the second

and following swings depending on the system parameters, the dynamic properties of

AVR and the time constant of the field winding. Consequently the use of a classical

model may lead to an optimistic assessment of the critical clearing time although it

reduces complexity and computational burden.

Considering all of the above problems a compromised approach is accepted in most of

the previous literature [12-14] or the sake of simplicity and classical modeling

consisting of rotor swing equations is used for transient stability studies. However, these

devices maintain the terminal voltage of the generator at a specified value and, in the

process, modulate the field voltage, and hence, the field current, thus supplying the

required reactive power to the load.

2.5 MULTI-MACHINE MODELLING WITH SYSTEM REFERENCE

The first part of this chapter has discussed the synchronous machine modelling along

with exciter and governor. However, this modelling was based on its own individual

machine axis and for the study of multimachine networks special care should be taken

to convert all parameters to form a dynamic model on one common base. This section

will describe in detail the connection and interaction of individual machines with the

external system network.

2.5.1 Connection of individual generator to power grid

In general, the dynamics of the post-fault power system after any disturbance can be

described by a set of nonlinear differential-algebraic equations. Typically, the states can

be associated with each machine’s armature and rotor currents, rotor dynamics, AVR,

and turbine-governor dynamics [6, 10, 15]. The overall dynamics can be best

summarized in terms of Fig. 2.1 in the form of block a diagram adopted from [6] and

Page 40: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

21

modified further to explain the structure of differential-algebraic equation model

formation for individual generators (having its own d-q reference frame) connected to

the external power system network (having a D-Q/Re-Im reference frame of

interconnected grid).

Transmission network equations

Including static loads (AE)

Excitation system (Field Excitation) Efd

Prime mover and governor system (mechanical power input) PM, ω

Generator rotor circuit Rotor flux and emf equations

Main field winding circuitd-axis damper winding circuits q-axis damper winding circuits

(DE)

Acceleration or swing equation Rotor angle

Speed(DE)

Machine rotor base reference frame: d-q

Other generators

Motors

Other dynamic

devices e.g. SVC

Generator voltages Reactive power generated

Generator frequency Active power generated

Stator circuit equations Machine stator base: D-Q

(AE)

Stator

Rotor

Generator connected to outside power system network common system base : Re-Im

Individual machine

Generator outputP, Q, |V|, δ

Fig.2.1 Individual synchronous generator connected to external network [6]

It can be seen from Fig. 2.1 that the transient stability study of power system networks

can be divided into two major parts, the individual synchronous machine and its

connection to the external power system network. While developing the mathematical

models of such an interconnected network, the first stage is the selection of the frame of

reference for the electrical quantities. The equations for each machine are expressed

with reference to pairs of individual axis (d-q) which rotate in synchronism with the

rotors of the machines; while the algebraic power flow equations of the power system

network are with system reference axes (D-Q/Re-Im).

Figure 2.2 shows the reference of frame for two individual machines as d1-q1, d2-q2. It

can be seen that machine 1 reference axes d1-q1 is displaced by δ1 with respect to D-Q

axis while for machine 2 this displacement angle is δ2. In steady state all these axes will

rotate at the same speed, but in transient conditions the angles will vary as the machine

Page 41: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

22

speed varies. Therefore, it is necessary to obtain a relationship between the deviations of

the current and voltage variables of individual machines from their steady-state

equilibrium values after disturbance.

D

Q

d1

d2q1

q2

δ1

δ2

Fig. 2.2 Angular relationships between the external power system network and the

individual machine reference axes

2.5.2 Transferring quantities from machine reference to system reference axis

With reference to the representative voltage vector shown in Fig. 2.3, (2.21) and (2.22)

can be used to transfer quantities from system reference to individual machine reference

while, (2.23) and (2.24) can be used for transforming from individual machine axes to

the main reference. The detailed derivation of these standard transformations can be

found in literature [16] .

δsinVδcosVV QDd += (2.21)

sinδVcosδVV DQq −= (2.22)

sinδVcosδVV qdD −= (2.23)

cosδVsinδVV qdQ += (2.24)

Page 42: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

23

D

Q

d1

Vs

q1

δ

θ

VdVq

VQ

VD

θ- δ

Fig.2.3 Transforming individual machine quantities to system frame of reference for

voltage

In general, while developing a power system model for multimachine system, each

machine quantity will be transformed with their respective dq-DQ displacement using

given transformation.

2.6 INTRODUCTION TO FACTS DEVICES

Having modelled the basic synchronous generator, exciter and prime-mover systems,

the next part of this chapter will focus on dynamic modelling of FACTS devices and

main power system components which plays important role in maintaining power

system performance in normal as well as abnormal conditions.

With the recent developments in power electronics and especially thyristor sizing and

switching technology, the FACTS family has many devices developed for power system

performance improvement depending on targeted issues. FACTS devices consist of

power electronics components and conventional equipment which can be combined in

different configurations. It is therefore relatively easy to develop new devices to meet

extended system requirements. The FACTS family members include static VAr

Page 43: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

24

compensators (SVC), static synchronous compensator (STATCOM), unified power

flow controller (UPFC), and thyristor controlled series compensation (TCSC) which

play different roles in steady-state as well dynamic stability improvement of

interconnected power systems. The SVC, STATCOM have prove their effectiveness in

voltage stability applications while series compensations like TCSC are always useful in

improving power transfer under normal steady state operation as well as after major

disturbances. The SVCs and STATCOM provide fast voltage control, reactive power

control and power oscillation damping features. As reported in an article on power

transmission and distribution published by Siemens, Germany, as an option, SVCs can

control unbalanced system voltages while fixed series compensation is widely used to

improve the stability and transmission capacity in long distance transmissions. In [17-

19] benefits of various FACTS devices in power system performance enhancement are

discussed. This chapter will present a brief dynamic modelling of the most commonly

used FACTS devices followed by dynamic modelling of other power system

components including supplementary damping controller (SDC) and load modelling

which are necessary to consider for system operation. The chapter also highlights on the

effect of modelling these components on performance calculation showing the

importance of using the correct models for study of transient stability.

2.7 STATIC VAR COMPENSATOR (SVC)

Static var compensators (SVCs) are shunt-connected static generators and/or absorbers

whose outputs are varied to control specific parameters of the electric power system.

The term static var system (SVS) is an aggregation of SVCs and mechanically switched

capacitors (MSCs) or reactors (MSRs) whose outputs are coordinated. There are many

basic reactive power control elements as listed in [6] which form all or part of any static

var system, and can be combined to form different types of SVS configurations. In

general, the main components and structure of a typical SVC can be represented as

shown in Fig. 2.4.

Almost always the SVC is connected to the transmission network via a coupling step-up

transformer. At the low-voltage side node of the transformer there are, in general, three

types of elements employed: thyristor controlled reactor (TCR), thyristor switched

capacitors (TSC) and fixed harmonic filters [20].

Page 44: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

25

(i) Thyristor controlled reactor (TCR): A schematic diagram of the single-phase

TCR consisting of the antiparallel thyristor pair and the linear reactor is shown in

typical SVC structure in Fig. 2.4. The controlled switching of the thyristors combined

with the linear reactor response enables the effective supply-frequency reactance of the

TCR, which is a function of the thyristor firing angle. For continuous variation of

reactance the firing angle can be varied smoothly from thyristor fully conducting to

fully non-conducting range (blocking).

High-voltage node

Low-voltage node

Coupling transformer

Filters

TCRTSC

↑QSVC

Fig. 2.4 Typical SVC structure with main components [19]

(ii) Thyristor switched capacitors (TSCs): Thyristors in this model have the function

of switching capacitors on or off as required so as to combine with the TCR to provide a

continuous range covering both inductive and capacitive compensation.

(iii) Fixed harmonic filters: The filters provide low-impedance paths for harmonic

currents generated from the TCR operation. Moreover, the filters also provide

capacitive compensation at the fundamental frequency.

Page 45: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

26

In general, by changing the thyristor firing delay angle, the effective reactance of the

TCR varies which in turn changes the effective reactance of the SVC. By doing so, the

SVC can supply or consume reactive-power from a transmission system. From an

operational point of view, the SVC can be considered as a shunt-connected variable

reactance, which either generates or absorbs reactive-power in order to regulate the

voltage magnitude at the point of connection to the transmission network.

The steady-state control objective of the SVC is that of voltage control function which

is expressed in terms of a V-I characteristic decided by the operating limits of TCR and

TSC under steady-state conditions. However, for a transient stability study, the dynamic

model of SVC is represented as shown in Fig. 2.5 [21, 22] and is simplified as derived

in Appendix C to represent in compact form as:

refSVCtSVCSDCSVCSVCSVCSVC VDVCXBXAX ⋅+⋅+⋅+⋅= (2.25)

In (2.25) SVCX , is a vector of SVC state variables, while ASVC, XSVC, BSVC, CSVC, and

DSVC are matrices whose elements are dependent on the gains and time constants of the

controllers. |Vt| is the magnitude of terminal voltage and Vref is the reference voltage

setting.

Ksvc

Bsvc(min)

svc

svcsTsT

2

111++

svcsT+11 X1svc

Bsvc(max)

Bsvc∑

XSDC

Vref

|Vt|

-

-+

Fig. 2.5 SVC dynamic model block diagram

2.8 STATIC SYNCHRONOUS COMPENSATOR (STATCOM)

The STATCOM performs the same function as that of SVCs. However at voltages

lower than the normal voltage regulation range, the STATCOM can generate more

reactive power than the SVC. This is due to the fact that the maximum capacitive power

Page 46: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

27

generated by a SVC is proportional to the square of the system voltage (constant

susceptance) while the maximum capacitive power generated by a STATCOM

decreases linearly with voltage (constant current). This ability to provide more

capacitive reactive power during a fault is one important advantage of the STATCOM

over the SVC. In addition, the STATCOM will normally exhibit a faster response than

the SVC because with the VSC, the STATCOM has no delay associated with the

thyristor firing (in the order of 4 ms for a SVC) [17].

2.8.1 Working principle of STATCOM

The basic principle of the STATCOM is to use a voltage sourced converter (VSC)

technology based on gate turn-off (GTO) thyristor or insulated gate bi-polar transistor

(IGBT)) that have the capability to interrupt current flow in response to a gating

command. This allows the STATCOM to generate an AC voltage source at the

converter terminal at the desired fundamental frequency with controllable magnitude. A

block diagram of a STATCOM is given in Fig. 2.6 [23].

ø

αjTT eVV =

CqCpC jIII +=

φjdcqCpCC eVkjVVV =+=

XC

Vdc

Fig.2.6 STATCOM schematic diagram

The voltage difference across the coupling transformer reactance produces active-power

and reactive-power exchanges between the network and the STATCOM. As shown in

Fig. 2.7, the phase reference for STATCOM voltage VC is the terminal voltage VT which

in phase with p-axis will decide the direction of active power flow while the exchange

Page 47: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

28

of reactive-power with the network is obtained by controlling the magnitude of the

voltage source. The active-power exchange is only used to control the DC voltage of the

capacitor. In steady-state conditions where the capacitor voltage is constant, the active-

power exchange is, therefore, zero if the VSC losses are discounted.

VCq

D

Qq

p

VC

VT

VCpøα

Fig. 2.7 Phasor diagram of STATCOM operating principle

There are several VSC structures currently used in actual power system operation. Fig.

2.8 shows the basic structure of a three-phase, full-wave converter having six switches

with each consisting of a GTO thyristor connected antiparallel with a diode. With the

aim of producing an output voltage waveform as near to a sinusoidal waveform as

possible, the switching of individual GTO thyristors in the VSC are controlled by the

switching control module designed to minimise the harmonics generated in the VSC

operation and requirement for harmonic filters.

+

-

Va

Vb

Vc

Output

Fig.2.8. Basic voltage sourced converter structure using GTO thyristors

Page 48: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

29

2.8.2 Dynamic model of STATCOM

The dc voltage represented in Fig. 2.6 can be combined with the transfer functions

shown in Fig. 2.9 to form a complete dynamic model for a STATCOM main controller

as shown in (2.26) - (2.30).

φstadc VAV = (2.26)

SDCstaCqstaTstaTrefsta XEIDVCVBV +++= (2.27)

φstaSDCstaTstaTrefstadcstastaC VLXKVJVHVGVFX +++++= (2.28)

φNXMφ staCsta += (2.29)

where, sinφVV Tφ = (2.30)

and Asta, Bsta, Csta, Dsta, Esta, Fsta, Gsta, Hsta, Jsta, Ksta, Lsta, Msta, and Nsta are matrices

elements dependent on STATCOM and its controller parameters are as derived in

Appendix C.

∑+VTref

-+

XSDC

|VT|

-

droop ICq

sK sta1 ÷

Ksta

|V|+

Vdc

stasT+11XC ø

max

min

∑( )

sta

stastasT

TK

2

22 1++

Fig.2.9 Dynamic model of STATCOM

2.9 THYRISTOR CONTROLLED SERIES COMPENSATION (TCSC)

The series counterpart of a shunt-connected SVC is a TCSC, which is connected in

series with a transmission line to provide improved stability of interconnected power

systems, increasing power transfer and directing power flows in desired transmission

Page 49: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

30

paths [24, 25]. A typical TCSC module in one phase is given in Fig. 2.10 [24]

consisting of a series capacitor in parallel with a TCR.

XFC

XP

Xc

XL Xtcsc

Fig.2.10 Simplified operating circuit of thyristor controlled series compensation

The operating constraints of a TCSC are associated with its reactance and can be

expressed in terms of the following inequalities:

tcscmaxtcsctcscmin XXX ≤≤ (2.31)

In (2.31), Xtcscmin and Xtcscmax are the minimum (capacitive) and maximum (inductive)

limits of TCSC reactance. They are dynamic limits which depend on the transmission

line current and can be determined using the capability characteristics. If the TCSC

reactance hits a limit, the specified control function will no longer be applicable, and the

TCSC will behave like a fixed reactance corresponding to the respective limit value.

2.9.1 Structure and operation of TCSC

Based on the above simplified operation circuit, it can be seen that a TCSC module can

operate in two extreme modes, i.e. thyristor path totally blocked, using it like a

conventional capacitor XC or continuously gated where it appears as a small inductance

with net reactance of Xbypass. In other words as mentioned in [26, 27], TCSC is often

treated as a variable reactance of a transmission line.

However, in practice, the practical layout of TCSC structure is as shown in Fig. 2.11

[28] in which an indispensable component is a highly nonlinear metal oxidized varistor

(MOV) which operates in both steady state and transient process of the power system.

Page 50: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

31

breaker

Varistor

TCSC Reactor

breaker

Capacitor bank

Fig.2.11 Typical layout of practical TCSC structure

During transient operation, the MOV bypasses excessive currents and limits the voltage

across the capacitor banks. This produces distortion of the TCSC voltage waveform and

consequently, sharply changes the fundamental frequency reactance of the TCSC.

Effective design and accurate evaluation of the TCSC control strategy depend on the

simulation accuracy of this process.

The operation of a TCSC module is constrained by an overvoltage protection provided

by the MOV which imposes the voltage limit on the TCSC operation in the capacitive

zone, and by harmonic and thyristor current ratings which constrain the TCSC operation

in the inductive zone. Fig. 2.12 [23] shows a typical TCSC capability characteristic for a

single module in terms of voltage versus the line current. Considerations of the thyristor

delay angle limits, voltage limits for the safe operation of the series capacitor and

thyristor current limits lead to the operating boundaries of the form in the voltage-

current plane in Fig. 2.12 within which the TCSC operation is allowed. The response

time periods are considered in the construction of the operating boundaries which are

applicable for short-term transient, long-term transient and continuous (steady-state)

operations, respectively.

Page 51: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

32

With reference to the above discussion of practical operating limits for a single module,

there is a gap as shown in Fig. 2.13 in the control range between blocked reactance, XC,

and bypassed reactance, Xbypass, for which no thyristor firing angle exists. This restricts

the application of TCSC in the transmission system where a smooth variation in the

combined reactance of the TCSC and transmission line is often required.

In order to eliminate this gap, the TCSC is split into multiple modules as shown in Fig.

2.14 which operate independently in the inductive and capacitive modes. By doing so, a

continuous transition from the capacitive to the inductive domain becomes feasible.

The larger the number of the modules into which the TCSC is divided, the narrower is

the gap between the capacitive and inductive regions. With a sufficient number of

modules, the TCSC reactance can vary continuously from capacitive value to inductive

value, and the reactance limits approach a closed locus within which TCSC operation is

feasible.

Vol

tage

(pu)

Indu

ctiv

eC

apac

itive

Maximum Thyristor Current

-2

0

-2

1 2

MOV protection Level

Max

imum

Firing

Adv

ance

Maximum Firing Delay

Harmonic Heating Limit

No Thyristor Current (blocked, slope= XC)

Full Thyristor Conduction(slope=Xbypass)

Line current (pu)

Fig.2.12 Typical V-I capability characteristics for a single-module TCSC

Page 52: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

33

δ (d

eg)

Xbypass XC

900

0 2-2 -1 1 300

1800

Maximum Firing Delay

Maximum Firing advance

Xnet(pu)

Unavailable

Fig. 2.13 Practical operating range of TCSC for inductive and capacitive compensation

MOV MOV MOV

Conventional Series Capacitor

VC

ILine

Multi-module TCSC

Fig. 2.14 Multi-module TCSC

2.9.2 Dynamic model of TCSC with SDC

For a transient stability study, the schematic block diagram of TCSC is shown in Fig.

2.15 in which TCSC dynamics are represented by one first-order block along with the

supplementary damping controller (SDC) block consisting of washout and two lead-lag

Page 53: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

34

compensators. The time lag used with TCSC is associated with the firing controls and

natural response of the TCSC and is represented by single time constant Ttcsc.

∑csc

csc1 t

tsT

K+

Pe

Xref

-

+

XSDC

Xtcsc

Xtcsc(max)

Xtcsc(min)XSDC(max)

XSDC(min)

SDC4

SDC311

sTsT

+

+ X2SDC

SD C2

SD C111

sTsT

+

+ X1SDC

SDC

SDC1 sT

sT+ SDC1K

Fig. 2.15 TCSC schematic block diagram with SDC

The reactance limits of the TCSC must be considered for static modelling as well as

dynamic. These limits are relatively complex and time dependent. The TCSC dynamic

model is derived from the above block diagram and represented in compact notations

as:

ettttT PBXAX••

+= cscTcscTcscTcsc (2.32)

In (2.32), ATtcsc and BTtcsc matrix constants are dependent on controller design (i.e. gains

and time constants) and are derived as shown in Appendix C.

Although TCSC control parameters can be tuned with changing system status and

controller output as discussed in [29], for the purpose of this study the tuning

parameters are fixed throughout to avoid complexity in focusing on transient stability

improvement for a multimachine system, considering its nonlinear dynamics.

2.10 SUPPLEMENTARY DAMPING CONTROLLER (SDC)

While using TCSCs for transient stability improvement focusing on first-swing stability,

there are chances of system oscillations in the next successive cycles because of slow

action of AVR. Hence, after taking care of the first-swing the next successive

oscillations can be taken care by using supplementary damping controllers (SDC) which

work almost the same as power system stabilisers.

Page 54: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

35

It has a washout filter and dynamic compensator as two main components followed by a

limiter. The simplified structure of SDC is represented as shown in Fig. 2.16. As the

SDC is expected to respond only for transient variations and avoid responding to any dc

offset of input signal, this washout filter acts like a high pass filter allowing only

frequency of interest. This can be achieved by selecting proper time constants to control

local modes or inter area modes. A dynamic compensator is a two-stage lead-lag

network and a limiter is used to keep the controller response within a specified tolerance

limit restricting it from large deviations.

PeSDC

SDC1 sT

sT+ ( )sT

( )minSDCX

( )maxSDCX

Washout Dynamic compensator

Fig.2.16 Simplified schematic of supplementary damping controller

Though there are many parameters which can be used as input signals to this SDC,

including line current, active power, reactive power and bus voltage magnitude in

present research active power input is used as an input signal to SDC. The two main

reasons for using active power as an input signal are, firstly, it is available with local

measurements and secondly, this will provide correct control action in maintaining

power transfer over lines when severe fault occurs. In [29], a qualitative analysis is

presented with different pros and cons in selecting a particular parameter as input signal.

2.11 POWER SYSTEM NETWORK MODELLING

Having formed the dynamic models of key compensating devices in power system, to

estimate the power flow over given transmission lines or voltages and phase angles at

every node, it is necessary to conduct power flow study. Hence, a proper representation

of power system network is necessary. The general power system network can be

represented as shown in Fig.2.17 which constitutes generator nodes, load nodes

including multiple FACTS devices inserted at some nodes for compensation.

Page 55: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

36

Representing total generator nodes by ngen and nsvc as the number of SVC nodes, nsta as

the total of STATCOM nodes and ntcsc as the nodes at which TCSCs are inserted, the

system has in all nnode nodes including load nodes.

Following the notations of Vk as the nodal voltage vector and Ybus as the general

admittance matrix based on a power system structure as shown in Fig. 2.17, the vector

of the current injected at every node including generator and load nodes, can be

represented in general form as:

bus

n

1kbusbus YVI

node

∑=

= (2.33)

where,

Ibus=

noden

i

1

I

I

I

; Vbus=

noden

i

1

V

V

V

; Ybus =

nodenodenode

node

nn1n

ii

2221

1n1k1211

YYY

YYYYYY

(2.34)

The general active and reactive power flow equations at any given node are represented

as given in (2.35):

= ∑

*

Rek

kikii VYVP and

= ∑

*

Imk

kikii VYVQ

(2.35)

where, i=2,3,. . ., nnode and Pi and Qi, are active and reactive powers injected at ith node

respectively.

2.12 LOAD MODELLING

As explained in [30], a load model is a mathematical representation of the relationship

between a bus voltage (magnitude and frequency) and the power (active and reactive) or

current flowing into the bus load. The term ‘load model’ may refer to the equations

themselves or the equations plus specific values for the parameters (e.g. coefficients,

exponents) of the equations. Depending on the computational implementation of these

equations in a specific program, the load power or current may not be included

explicitly, but it is useful to think of the model in these terms.

Page 56: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

37

It is difficult to quantify the benefits of improved load representation. However, several

studies, reported in literature, have demonstrated the impact that different load models

can have on the different types of study results. In some cases the impact can be

significant. A common philosophy, in the absence of accurate data on load

characteristics, is to assume what is believed to be a pessimistic representation, in order

to provide some safety margin in the system design and operating limits.

1 (slack)

2

ngen

gen

SVC1 ngen+1

SVCn_svc ngen+nsvc

STATCOM1 ngen+nsvc+1

STATCOMn_sta ngen+nsvc+nsta

TCSC1

TCSCn_tcsc

ngen+nsvc+nsta+1

ngen+nsvc+nsta+ntcsc

ngen+nsvc+nsta+2*ntcsc+1

n_svc1

n_svcn_svc

n_sta1

n_stan_sta

ngen+nsvc+nsta+ntcsc+1

ngen+nsvc+nsta+2*ntcsc

n_node

svc

sta

TCSC

nodeload

Fig.2.17 Multi-machine power system network having multiple FACTS devices

Page 57: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

38

2.12.1 Static load modelling

Static load modeling can be defined as in [30] as model that expresses the active and

reactive power at any instant of time as a function of the bus voltage magnitude and

frequency at the same time. Static load models are used both for essentially static load

components, e.g. resistive and lighting load and as an approximation for dynamic load

components e.g. motor-driven loads.

In this research, the static loads are modelled as equivalent admittances. The required

data for calculations of these admittances is obtained from load flow studies. Thus for

an active load of PL and reactive load of QL, at any load bus having voltage as VL, the

equivalent load admittance at a given bus can be calculated as:

2L

L2

L

LL

V

QjV

PY −= (2.36)

Although the static load model of the constant admittance form is most popular one,

other static models such as those based on constant current, constant power and

exponential functions have also been proposed and reported in [30-32] .

2.12.2 Dynamic load modelling

The dynamic load model as defined by [30], is a model that expresses the active and

reactive powers at any instant of time as a function of voltage magnitude and frequency

at any past instant of time and, usually including the present instant. Difference or

differential equations can be used to represent such models.

Even though power system load has gain more attention in literature, it is still

considered as one of the most uncertain and difficult components to model due to the

large number of diverse load components, its high distribution, variable compositions

and also because of lack of precise information on the composition of the load [33]. The

induction motor type of loads connected to the power system, maintain their stability,

when unbalances due to voltage changes arise by shifting their driving points slightly.

Page 58: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

39

The induction motor is a typical load under constant torque operation. It compensates

for a shortage of power by increasing the slip and for a surplus of power by decreasing

its slip, resulting in balanced operation. Such a mechanism is called self-controllability.

As mentioned in [33], with various developments in power system, it’s not only

induction motors which should be considered as dynamic loads, but tap-changers or

spontaneous load variations should also be considered for load modelling. Navarro in

[33], has discussed various types of load and their static as well as dynamic load

models such as, frequency load models, induction load models and exponential dynamic

load models.

On the time scale of importance for transient stability properties, generators and motor

loads are the primary source of angular dynamics following a network fault. However,

in common with [34, 35] and [13] systems are considered with no motor loads. This

assumption is proposed for three reasons: an accurate representation of load behaviour

is difficult to obtain; measurement of motor load dynamics is difficult; and the constant

impedance load assumption simplifies the control design process. Though the effect of

dynamic loading in stability study is discussed in previous literature, in present research

only static load models are considered. However, the control schemes developed in this

thesis for RHC based TCSC controllers or online control coordination schemes can be

equally applicable for dynamic load consideration by modifying existing power system

models with additional equations of dynamic load models.

2.12.3 Impact of load modelling on power system transient stability study

The load modelling is important from a first-swing stability study point of view or

transient stability point of view. A first-swing problem exhibits large and rapid voltage

excursions during the initiating fault and slower voltage excursions during the first

power-angle swing, which lasts one second or less. Load response to these voltages is

important. System voltages are normally depressed during the first angular swing

following the fault. The power consumed by the loads during this period will affect the

generation-load power imbalance and thereby affect the magnitude of the angular

excursion and the first swing stability of the system. As mentioned in [30], in case of a

constant impedance load model, the power consumption would vary with the square of

Page 59: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

40

the voltage and, therefore, would be lower than the actual load during the depressed

voltage period. For loads near accelerating machines, this will give pessimistic results,

since the generation-load imbalance will be increased. However, for loads remote from

accelerating machines (near to decelerating machines), there will be an optimistic

impact on the results. On the other hand, a constant MVA load model would have the

opposite impact on the results since it would hold the load power at higher value during

the depressed voltage. It is therefore difficult to select a model that is guaranteed to be

conservative for all parts of the system and for various disturbances.

There is also a brief frequency excursion during the power-angle swing, so frequency

characteristics of loads close to accelerating or decelerating generators can also be

important. Chandrashekhar and authors in [36] have discussed the roles played by

structure and load models in direct stability assessment. One of the major drawbacks of

using classical modelling with constant-impedance loads is difficulty in accounting

rigorously for transfer conductance between the machine voltage nodes and the

dependency of stability properties on the structural features of the network getting

masked. These disadvantages are essentially due to the process of assuming impedance

models for loads and then using a reduced network model. The resistive part of the

loads leads to transfer conductances and the network structure is lost in the reduction.

However, with present advanced computer technology the network reduction can be

avoided and system structure can be preserved for better accuracy.

2.13 CONCLUSIONS

The state-of-the-art models for the synchronous generator in dynamic condition have

been explained. The dynamic modelling of exciter, prime-mover and PSS are explained

for transient stability study. The importance and effect of using accurate models of

synchronous machine, exciter and governor are discussed. The interaction of exciter

and prime-mover with synchronous generator is explained and how to incorporate these

individual machine models with external power system networks for transient stability

study is shown by proper transformation of reference axes.

The chapter has discussed the dynamic modelling of various key FACTS devices

including SVCs, STATCOMs and TCSCs. Using simplified schematic representative

Page 60: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

41

dynamic models, the effect of modelling these equipments on transient stability is

discussed in brief. The power system network equations are considered using multiple

FACTS devices. Based on the various aspects of dynamic modelling of key power

system elements for transient stability, a comprehensive model will be formed and used

for deriving various control schemes which are proposed in the next part of this thesis.

Page 61: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

42

Chapter 3 Overview of Power System Issues and

Solutions

3.1 INTRODUCTION TO POWER GRID OPERATION

In recent years, as a consequence of the deregulation of the electric power industry,

power sources and consumers geographically dispersed, resulting in bulk power

exchanges over long distances. Though the supply should be increased to cope with

increasing demand, the development of generation is much slower than the rate of load

increment. As mentioned in [37] , with reference to report [38] from the Electric Power

Research Institute, the generation capacity margin has consistently decreased in the past

20 years. For example, the generation capacity margin in 2000 was only one third of a

half of the increase of electricity demand. In short, the demand is ever increasing and

the new transmission infrastructure to cope with this increasing demand is restricted

because of several factors like economy and environment. This forces the existing

power system network to work at its maximum possible permissible limits, making it

vulnerable to becoming unstable in case of any sudden disturbance. In such

circumstances to maintain, continuity and reliability, interconnection of various areas

becomes essential. Forming interconnected grid helps in many ways including less

Page 62: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

43

installed capacity, less spinning reserve required, efficient loading which results in

better transmission efficiency, maintaining continuity, giving breathing time for some

outages and maintenance work. However, interconnected systems operating close to

their transient stability limit are vulnerable to any major sudden disturbances, where

cascade tripping problem and total or partial blackouts can occur if necessary corrective

action is not taken in time.

3.2 POWER SYSTEM ISSUES: CAUSES AND SOLUTIONS

3.2.1 Power system reliability and security

As explained in [39] an electrical power system consists of numerous components

connected together to form a large, complex system which is generating, transmitting

and distributing electrical power. Electric power systems and additional preventive

control schemes are designed in such a way, that the system should be able to withstand

any single contingency, that is, outage of any single component without loss of stability

and all system variables kept within predefined ranges [6]. Not all possible

disturbances, however, can be foreseen at the planning stage and these may result in

instability leading eventually to collapse or islanding of the system. Furthermore,

because of environmental constraints on the extension of the transmission capacity,

increased electricity consumption and new economic constraints imposed by the

liberalised power market, power systems are operated closer and closer to their stability

limits. During normal operation, the focus is on economic optimisation of system

operation, while during more challenging network conditions (alert or emergency

situations) the focus shifts to stability consideration rather than economy [40]. The

ultimate objective is keeping as much as possible of the network and generators

connected to the grid intact, as breakdown will normally result in more severe problems.

Hence, the main concern in the emergency state is the system security. Security is an

online operational characteristic which describes the ability of power systems to

withstand different contingencies without service interruptions. Security is closely

related to reliability: an unreliable system cannot be secure. The security level of the

power system desired is to be high enough to enable robust operation changes

dynamically as the power system operation changes and depends on the factors outside

the control of power system operators such as weather [41] . With this background, it is

Page 63: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

44

vital to give top priority to dynamic security and stability before pursuing other targets,

like economical operation, optimal load flow and fair deregulation in power market

[42]. To understand the dynamic behaviour of system, it is necessary to know the

classification of power system transients and its time frames of response.

3.2.2 Classification of power system transients

Power systems comprise of many components including generators, loads, transmission

lines, which interact with each other in various ways and on various time scales (i.e

dynamic speed ranging from milliseconds up to years). Dynamics present in power

systems may be very nonlinear and have a hybrid nature having both continuous and

discrete state variables [43] . At every instant power system networks experience some

sort of disturbances that may be small or large in amplitude and varying over different

time duration. The severity of transients is classified based on the rate at which the wave

rises from 10% of its value to 90%, i.e. value of front of waveform and the time for

which it stays on system i.e. fall time measured from peak to 50% of its value, which is

also known as the tail of wave as shown in Fig.3.1. The sharp front can stress the

system pushing suddenly away from normal operating conditions, while the long tail

may keep the system oscillating for a longer time.

0.90

0.10

TailFront

Time

0.50

Fig.3.1 Typical wave specifications expressed in terms of front and tail of wave

Page 64: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

45

The time frame for some important disturbances is shown in Fig. 3.2 [44, 45]

classifying the transients based on their behaviour. The class A type of transients are

known as ultrafast transients and are also categorized as surge phenomena, having a

very steep wave front. The class B type of transients are medium-fast and are due to

short-circuit phenomena, while the class C represents slow transients which comes

under the category of transient stability varying in time range of few seconds to few

minutes.

10-7 10-5 10-3 10-1 101 103 10-5

Lightning

Switching

Subsynchronous resonance

Transient stability

Long term dynamics

Tie-line regulation

Daily load following

SVC,TCSC etc

Generator control

Protection

Prime mover control

LFC

Operator actions

1 cycle 1 second 1 minute 1 hour 1 day

Pow

er sy

stem

phe

nom

ena

Pow

er sy

stem

con

trol

s

Timescale (seconds)Class AClass B

Class C

Fig: 3.2 Time frame of various transient phenomena

The modelling of power system equipment and their responses are also compared in

Fig.3.2 for these same time frame disturbances and it can be seen that no power system

control can really respond to class A types of transients. However, with fast switching

thyristor controls, FACTS devices can be used to enhance power system performance

Page 65: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

46

for medium-fast or class B types of transients. It can be observed that prime mover and

generator controls are active to respond during class C transients which are transient

stability responses indicating that assumption or simplified modelling of these elements

may lead to pessimistic results. With reference to Fig.3.2 and as explained in Chapter 2

it can be seen that any accuracies in dynamic modelling of power system elements will

result in inaccurate prediction of system stability in transient stability conditions.

3.2.3 Disturbances/causes of power system problems

Even knowing the classification of transients on power systems, the response of various

power system elements to these transients and having most accurate designed protection

schemes, power systems are always exposed to various serious disturbances which can

lead to the interruption of power supply to consumers. Even the best planned system

cannot predict all possible contingencies, and any unpredictable events can stress the

system beyond planned limits. Some of the more well-known reasons as well as those

mentioned in [41], why completely reliable system operation is not achievable [46]:

(a) Globalisation/liberalisation: Deregulation and privatisation [18] has opened the

market for independent suppliers and transmission companies giving rise to different

load patterns than those for which it was previously designed.

(b) Transmission congestion and stressed conditions: problems of uncontrolled loop

flows, over loading and excessive short-circuit levels gives rise to instabilities and

outages which are bottlenecks in transmission. This gives rise to an infinite number of

possible operating contingencies in modern interconnected power system networks. The

evolving nature of power systems generating unpredictable changes, giving power

system operation a totally different scenario from the expectations of the system

designers, particularly during emergency. In addition, a combination of unusual and

undesired events such as human error combined with heavy weather and scheduled or

unscheduled maintenance outages of the important system element. For example,

deregulation provided financial motivation to transfer power from generation e.g.

independent power producers to remote loads. As existing power systems were not

designed for those transfers, the system has to bear additional stress.

Page 66: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

47

(c) Weak connections: Extension of interconnected systems is one of the solutions for

an economical, efficient and reliable power system. However, the increased power

exchange among the long interconnected links can give rise to weak link connections

which can put the system in a condition of cascade tripping.

(d) Unexpected events, stability threats or hidden failures in protection systems

The development of transmission systems follows closely the increasing demand on

electrical energy. However, the performance of the power systems decreases with

increasing size and complexity of the networks. This is related to problems with load

flow, power oscillations and voltage quality. Such problems have been highlighted by

the deregulation of the electrical power markets, worldwide. Contractual transactions

now result in power flows that are much different from those of the original network

design criteria, and the operational constraints of the existing network, especially where

the connecting AC links are weak.

The interconnection of power systems offers numerous benefits for power transmission,

including pooling of various energy resources, reduction of reserve capacity in the

systems and increasing the transmission efficiency. However, if the size of the system is

too large, dynamic problems can occur which can jeopardize the reliability and

availability of the synchronous operation of the interconnected grids. By using the

advanced solutions, based on modern power electronics, the performance of

transmission systems can be improved.

3.2.4 Power system stability issues

The severity and response time for power system stability will vary depending on the

type of transients arising in power systems as explained in Section 3.2.2. Deregulation

has resulted in more competition in the power market so that it is becoming necessary to

allow the operating points of the generators to vary quickly to meet the needs according

to the power purchase agreement between the seller and buyer. Conventional control

strategies based on approximate linearised models have proven to be very valuable to

ensure satisfactory transient performance due to the inherent nonlinear nature of the

Page 67: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

48

power system. In fact, the nature of a power system is not only highly nonlinear but it

exhibits uncertainty somewhat as their operating condition keeps changing frequently.

Particularly after a large disturbance, the power system topology can change drastically

due to fault clearing which may be by removal of any of the transmission lines [47]. In

such circumstances, it is necessary to know whether the system will regain any stable

post-fault operation or collapse. With reference to this, various stability definitions are

given in literature.

As mentioned in [48, 49] , stability can be defined as, the ability of power system for a

given initial operating condition, to regain a state of operating equilibrium after being

subjected to a physical disturbance, with most system variables bounded so that

practically the entire system remains intact. During normal operating conditions an

interconnected grid will help for power exchange over tie-lines, showing its usefulness.

However, during any sudden disturbances it also affects the entire network and

“stiffness” of the system decides the system stability in such conditions. In such

circumstances, the stability of the power system is defined by its ability to restore to

normal operation after any disturbance.

The transient stability is defined as the ability of the power system to come back to

stable operating conditions, subjected to any sudden disturbance. Following a

disturbance, protective relays will sense the type of disturbance and will give the

tripping command to the corresponding circuit breaker, to isolate the faulty part from

healthy network. However, system stability cannot be guaranteed even after fault

clearance, as total fault clearing time will decide the system behaviour in its post-fault

region. For guaranteed system recovery after such disturbances it is necessary to

understand the behaviour and response of various transients which arise in power

systems.

Though power system stability is a single problem, because of its high dimensionality

and complexity, to analyse specific types of problems using an appropriate degree of

details of system representation, simplifying assumptions can be used. Analysis of

stability, including key factors that contribute to instability and devising methods of

improving stable operation, is greatly facilitated by the classification of stability into

appropriate categories. Classification, therefore, is essential for meaningful practical

Page 68: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

49

analysis and resolution of power system problems which is shown in Fig.3.3.

Phenomena that create wide-area power system disturbances are divided into the

following major categories: angular stability, voltage stability, overloads, power system

cascading, etc. using variety of protective relaying and emergency control measures are

employed to combat them [41].

The angular instability or loss of synchronism condition occurs when generators in one

part of the network accelerate while other generators in some other area decelerate

thereby creating a situation where the system is likely to get separated into two parts.

The most common predictive scheme to combat loss of synchronism is the equal area

criteria and its variations. This method assumes that a power system behaves like a two-

machine model where one area oscillates against the rest of the system. Whenever the

underlying assumption holds true, the method has potential for fast detection.

Often the causes of angular instability can be overloading or short circuiting. Overloads

frequently occur during wide-area disturbances due to the increasingly high utilisation

of equipment capability. These overloads may result in faults or equipment damage if

overload protection is not provided. Outage of one or more power system elements due

to overload may result in overload of other elements in the system. If the overload is not

alleviated in time, the process of power system cascading may start, leading to power

system separation. Uncontrolled separation often occurs as a result of a transmission

line short-circuits protection system interpreting power swing as a short-circuit. When a

power system separates, islands with an imbalance between generation and load are

formed with a consequence of frequency deviation from nominal value. If the imbalance

cannot be handled by generators, load or generation, shedding is necessary. A quick,

simple and reliable way to re-establish active power balance is to shed load by

underfrequency relays.

While the system frequency is a final result of the power deficiency, the rate of change

of frequency is an instantaneous indicator of power deficiency and can enable incipient

recognition of the power imbalance. However, change of the machine speeds is

oscillatory by nature, due to the interaction among generators. The main cause of

system oscillations is smaller system inertia which gives rise to larger peak-to-peak

value for rate of change of frequency oscillations [41].

Page 69: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

50

Power system Stability

Rotor-angle stability

Frequency stability

Large-disturbance angle stability /

transient stability

Voltage stability

Small-disturbance angle stability /

small-signal stability

Large-disturbance voltage stability

Small-disturbance voltage stability

Slow and small amplitude

disturbance

Different modes of oscillations

Sudden large amplitude

disturbance

First swing and oscillations

problem

3-5s following disturbance is

period of interest

Few seconds to several minutes

following disturbance is the period of

interest

Fig.3.3 Classification of power system stability studies based on physical nature of the

phenomena

The rotor-angle stability also known as simply angle stability is concerned with the

ability of interconnected synchronous machines of a power system to remain in

synchronism under normal operating conditions and after being subjected to a

disturbance [50]. A fundamental factor in this aspect of stability is the manner in which

torque or power outputs of the synchronous machines vary as their rotors oscillate. The

mechanism by which the synchronous machine maintains synchronism with one another

is through the development of restoring torques whenever there are forces tending to

accelerate or decelerate the machine with respect to each other. When system and

synchronous generators to remain in synchronism and to give desired performance in

post-disturbance condition, use of some compensating devices becomes essential to

maintain system performance.

Page 70: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

51

3.2.5 Solution for power system stability: Need of FACTS devices

With the formation of wide-area network (WAN), if the power should be transmitted

through an interconnected system over longer distances, transmission needs to be

supported, which is possible with the help of recent developments in FACTS devices.

One of the major key consequences of deregulation as mentioned in [18] is that the

power transfer across the system is nowadays much more wide spread and fluctuating

than initially designed by the system planners. The system elements are going to be

loaded up to their limits, with the risk of losing that (n-1) safety criterion. The major

blackouts as reported in [18] show that, once the cascading sequence is started, it is

difficult or even impossible to stop it, unless the direct causes are eliminated.

This motivates researchers to find some solution for improving transient stability by

some controller, compensations, fast enough to help in the recovery of system in a post-

fault region. With recent developments in power electronics and especially thyristor

technology, the FACTS family is taking a leading role in power system performance

improvement. These FACTS devices are used not only in transient stability

improvement but also used in normal healthy operations to improve power transfer

capability, voltage stability etc.

In order to study the effects of FACTS devices installed in power systems, proper

models of power systems with FACTS elements need to be established which is already

discussed and derived in Chapter 2. However, Table 3.1 shows selected FACTS devices

with their basic scheme and comparative performance in terms of their impact on

various power system studies (load flow and transient stability) based on the report

[51].

For most applications in AC transmission systems and for network interactions, SVC,

FSC and TCSC are fully capable to match the essential requirements of the grid. Out of

all FACTS devices, the only device which is most commonly inserted in series, for

power system performance is TCSC. Thyristor-controlled series compensation can

provide improved stability for interconnected power systems allowing higher power

transfer levels and directing flows on desired transmission paths. Knowing the necessity

of FACTS controllers for power system performance, the next section of this chapter

Page 71: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

52

will give a brief review of various power system controllers used for stability

enhancement.

Table: 3.1 Overview and functions of most popular FACTS devices

Principle Devices Scheme Impact on system performance Load

flow Stability Voltage

quality Variation of line impedance : Series compensation

FSC TPSC TCSC

Small Small Medium

Strong Strong Strong

Small Small Small

Voltage control: Shunt compensation

SVC STATCOM

No/ Low No/ Low

Medium Medium

Strong Strong

FSC: Fixed series compensation

TPSC: Thyristor protected series compensation

TCSC: Thyristor controlled series compensation

SVC: Static VAr compensation

STATCOM: Static synchronous compensator

3.3 TCSC USED FOR POWER SYSTEM PERFORMANCE ENHANCEMENT

The power system transient structure forms firstly, with continuous dynamics which is

to be controlled and secondly with the discrete-event controller. To trigger the action of

the controller and link these two, some trajectory reference is set with suitable

parameters. The sensitive parameter is used to activate the controller which can be an

error in the given state variables. For example the reference load angle delta or speed

can be specified. In an occurrence of any disturbance the system dynamics start

evolving and deviate from steady conditions introducing errors in actual system

parameters and reference values set, which give rise to triggered action of the controller.

Page 72: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

53

In [52] , broad classification of various emergency schemes, such as event-based and

response-based is given. The event-based controller works based on predefined

disturbances. For example, look-up tables while response-based will react based on

getting triggered for a particular situation. As a power system is nonlinear and non-

stationary, the response-based controller will be adaptive and flexible to provide a best

control action to assure system stability, compared to event-based type control. Event-

based control schemes require no intervention of system operator and have a special

protection system as a defence scheme which responds automatically when the

contingency is detected [53].

Though there are many controllers developed for various power system applications

including voltage control, power system oscillations damping or transient stability

improvement using various FACTS devices, the focus of this research is using TCSC

for improvement of transient stability. The next section will provide the implementation

history of TCSC to discuss its practical performance and explain its benefits.

3.3.1 TCSC implementation history

Authors in [24], have reported the world’s first 3-phase, 2x165 MVAr TCSC

installations in 1992, in a 230kV transmission line at the Kayenta Substation in Arizona

and the Slatt Substation in Oregon. The aim of these installations was to increase power

transfer up to the thermal limits of line and to evaluate the effect of TCSC in controlling

power flow to damp electromechanical power oscillation. However, soon it was realized

that in addition to increase in power transfer by 30%, TCSC provides effective means

for damping electromechanical power oscillations. The on-site observations also

showed that it can provide series compensation without causing risk of sub-synchronous

resonance (SSR) as in case of a fixed series capacitor. With this experience, ABB

installed, the world’s first TCSC for sub-synchronous resonance in Stode, Sweden in

1998. As mentioned in [54], a thyristor-switched series capacitor (TSSC) controller

was experimented for compensating the 345kV transmission line at the Kanawa River

Substation in West Virginia. A TCSC controller is also installed on a 500kV

transmission line in Fengtun, China. Kirschner et.al. in [18] report that the rating of

shunt connected FACTS controllers is up to 800 MVAr while, series FACTS devices

are implemented on high voltage levels up to 765 kV, this can increase the line

Page 73: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

54

transmission capacity up to several GW. Also [18] found that the interconnection of the

1000 km long, North-South Brazilian system operating at 500kV AC is unstable without

the damping function of TCSC. If only one TCSC is in operation, the interconnection

becomes stable, and with both devices working, the inter-area oscillations are well

damped. It is also reported that, under increased load conditions, the TCSC damping

function is activated up to several hundred times per day, thus saving power

transmission and keeping the return investments constantly ‘running’. To find out the

optimised placement of TCSC and the effect of load increase and generation

rescheduling, a method based on trajectory sensitivity has been given by [55-57].

3.3.2 TCSC placement and its benefits

The authors in [24] have given some fundamental and concise study results of TCSC

behaviour. There are two strategies in the use of TCSC, firstly, the ‘constant line power

strategy’ where the power flow over a particular line is maintained constant with TCSC

series compensation and secondly by the ‘constant angle strategy’. While, in general,

the system stability condition improves with the placement of a TCSC in a multi-

machine system, the placement of the TCSC in some of the lines may have a

detrimental effect on system stability. Therefore, as mentioned in [57], it is extremely

important to identify the lines that give maximum benefit, and the lines that adversely

affect the system stability.

Zhou and Liang [58] give a detailed survey of various control schemes for enhancing

power system stability performance using TCSC. TCSC is not only useful in large

disturbances but it also improves the performance of systems under normal conditions.

Further benefits include scheduling of power flow over line under normal operating

condition, small-signal stability, damping oscillations, limiting short-circuit currents,

and many more the explained in [59] and [17]. However, for transient stability

problems, TCSC is considered to be most effective solution as compared to other shunt

types of compensators and FACTS devices. The authors in [47] have presented a

nonlinear control scheme for the TCSC to improve transient stability and to dampen

power system oscillations. However, an established nonlinear mathematical model

which has proven to be an affine nonlinear system is based on a single-machine-infinite-

bus (SMIB) system. On the use of the SMIB approach is given in Chapter 4. In a study

Page 74: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

55

of TCSC controller design for power system stability improvement, [29] provides a

detailed analysis of some of the fundamental aspects of proper TCSC controller design.

The limitations of using linear control techniques for controller design detail with large

disturbance in a realistic power system network are discussed in. A detailed analysis of

TCSC control performance for improving system stability with different input signals is

discussed and the need for proper input signal selection and coordination of the different

control levels are highlighted. Author, Sun in [60], has discussed the dynamic response

of TCSC through time domain digital simulation of an example TCSC circuit and then

proposed an improved closed loop reactance control method for the application of

TCSC.

Following a disturbance, protective relays will act to clear the fault by tripping the

faulted component. However, the power system is not guaranteed to be stable after a

clearing fault. In order to improve the transient stability performance, early detection of

fault, and fast fault clearance is most important. Along with fast controllers, the

prediction algorithms can help to predict the system conditions in advance so that

necessary corrective action can be implemented accurately well before the system

moves in critical stages. Knowing the potential of various prediction algorithms, the

next section of the chapter is devoted to reviewing of predictive control based methods

which are used for power system performance enhancement.

3.4 REVIEW OF PREDICTIVE CONTROL BASED METHODS USED FOR

POWER SYSTEM APPLICATIONS

Predictive controllers are well-developed in control systems and have proved their

effectiveness in chemical industry applications [61, 62]. However, Power systems

exhibit several features of complex systems, such as hybrid nature (mixed and

continuous dynamics), nonlinear dynamics and very large size. Such complex features

of power systems, offer challenges to predictive controls which require reasonable time

in optimisation computation. Several authors have addressed difficulties in applying

predictive methodology for power systems because of its complexity, large size, and

hybrid nature. The power system can be modelled as a hybrid system incorporating

nonlinear dynamics, discrete events and discrete manipulated variables. The nonlinear

behaviour of the system calls for methods based on a dynamic model in order to account

Page 75: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

56

for changes of operating point and the network state. This is complicated by the fact that

most control moves are inherently discrete-valued, for example switching on of

capacitor banks. However with recent advances in computation, communication and

power system instrumentation technology, and more specifically phasor measurement

units and wide-area measurements, the coordination and model based approach have

become more tractable. To deal with these challenges, Marek Zima [43] tried to use

MPC for power system applications using a trajectory sensitivity concept which claimed

that trajectory sensitivities allow an accurate reproduction of the nonlinear system

behaviour using a considerably reduced computational burden as compared to full

nonlinear integration of the system trajectories. Unless employed control changes

significantly, trajectory sensitivity reproduces the system behaviour quite accurately,

even considering nonlinear dynamics. However, it is very difficult to bind a region, in

which the changes can be considered ‘reasonably small’. Another possible source of

errors may be certain types of discrete events. A discrepancy between the model and the

actual system response can be corrected in the receding horizon manner. With this

background, the next section of this chapter will discuss various predictive control

applications for power system stability enhancement.

3.4.1 Voltage control and small-signal stability applications

There are a number of attempts of applying predictive control methodology for voltage

stability and small-signal stability. In [63, 64], Zima has used trajectory sensitivities

based model predictive control (MPC) for emergency control of voltage stability

problems. It has used simplified MPC concepts in which only one control input set is

computed and tracking of certain reference values are disregarded. Using the same

trajectory sensitivities in order to reduce the modelling complexity and computational

burden [43] has presented closed loop MPC used in both emergency and normal

operation conditions in power system. Geyer and Larsson [39] used mixed logic

dynamics framework in connection with MPC for dealing with the hybrid nature of

power systems to predict and prevent voltage collapse in a power system.

In [65], for predictive frequency stability control based on wide-area phasor

measurements, a method is proposed in terms for a predictive control strategy. First a

single-machine equivalent model of the system is formed based on the collected

Page 76: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

57

measurements, which is then used to estimate the active power imbalance in the system

and subsequently a predicted steady-state frequency. The same model is then used to

find out the amount of load shedding required to keep the frequency above some target

value. Finally, the calculated amount is allocated to different feeders using a simple

iterative method considering the actual load on feeders. To maintain frequency stability,

load shedding is an easy and quick solution; however, it is not a preferred or

recommended solution when it comes to quality and continuity of supply.

3.4.2 Predictive control based controllers for transient stability improvement

The interconnection of power systems offers numerous benefits for power transmission,

such as pooling of various energy resources, reduction of reserve capacity in the

systems and increasing the transmission efficiency. However, if the size of the system is

too large, dynamic problems can occur which could and can jeopardize the reliability

and availability of the synchronous operation of the interconnected grids [51] .

Following a disturbance, protective relays will act to clear the fault by tripping the

faulted component. However, the power system is not guaranteed to be stable after a

clearing fault. In order to improve the transient stability performance, early detection of

fault, and fast fault clearance is most important so a neural network-based system was

proposed in [52]. For making accurate predictions of transient stability status of power

systems, in [52], training examples were added continuously to reflect the most recent

operating conditions.

In the emergency control given in [65], a prediction method is used for estimating

future frequency decline after disturbance and necessary corrective action to maintain

wide-area frequency stability problem. However, the prediction is based on

approximation. Authors in [66] have proposed a response-based emergency control

scheme to prevent transient instability with the help of a hybrid model developed with

continuous dynamics and discrete events interacted. The problem of deciding

emergency control action is embedded into the MPC framework. For response-based

emergency schemes, an essential but challenging step is to make real-time predictions

of the transient stability status of a power system within a short time interval, say about

15ms after clearing the fault. In such critical conditions, a compromise has to be made

Page 77: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

58

between prediction accuracy and speed. Though many event-based and response-based

emergency control schemes are reported in literature, the response-based emergency

control leads over event-based because of its adaptability and flexibility to changing

operating conditions and disturbance.

It has been found in [67] that, when the transmission line impedance is controlled by

FACTS equipments like TCSC, the control action is multiplicative. This necessitates

accurate nonlinear models to make use of the most effective, economical use of

available control resources. In this context, variable structure control [68] and nonlinear

model-based self-tuning control [69] have been proposed for a class of transient

stability problems. In [67], a nonlinear time series, model-based, generalised predictive

controller for a simplified power system, using rotor angles as the measured output and

a TCSC controller is designed and developed. Generalised predictive controllers offer

advantages of being easy to implement in real-time and allowing systematic methods of

handling input constraints.

Ford et al. [13] have tried to develop an efficient and robust model predictive control for

first swing transient stability using FACTS devices. However, it is based on simplified

low order model and unrealistic fault duration times and is more focused on the power

transfer capacity of transmission lines and several other assumptions.

To study the effectiveness of various control methods applied to power systems, [67]

compared generalised predictive control (GPC), feedback-linearisation and LQR

methods. The authors in [67] observed that the nonlinear GPC provides the best

damping out of the three within the shortest possible time and using the least amount of

control. Rajkumar et al [69] presented generalised predictive control schemes for TCSC

to raise the transient stability limits and provide rapid damping to the power system

oscillation. However, for nonlinear systems, the online optimisation numerical

computation burden is huge and the demand of real-time control may not be satisfied.

To address this issue, [70] designed a TCSC controller with closed-form analytical

solution control law based on nonlinear optimal predictive control theory. The authors

claim that this controller does not require online optimisation and, hence, will satisfy the

demand of real-time control. A major advantage of the nonlinear GPC over the other

two controllers is that it allows a systematic way of handling control constraints albeit

Page 78: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

59

numerically intensive, which may restrict the look-ahead horizon for real-time

implementation.

Feedback-linearising controllers demand excessive control efforts, compared with

LQR’s due to the extended control efforts of feedback-linearising the nonlinear

dynamics. When the control saturates, the nonlinear dynamics are not cancelled entirely,

and the system is left with residual nonlinear dynamics with properties that are different

from the original nonlinear power system. Feedback-linearising controllers need a

perfect reference model and measurements to provide the exact cancellation.

Uncertainties in the reference model can lead to deteriorated robustness of the

controller. There exists the danger of destabilisation, by imperfect cancellation power

system dynamics which possess an unstable equilibrium, e.g. dynamically unstable

situations. The presence of dynamic uncertainties, such as time-varying infinite-bus

voltage, can lead to non-robust performance of the feedback linearising controller. LQ

regulators perform worst than the nonlinear GPC, but better than the feedback

linearisers, in identical conditions. Although the performance of LQR on nonlinear

systems is sub-optimal, it is seen to have good robustness properties.

In view of the many attempts of predictive control algorithms to power system

applications, the next chapter will develop the theory and scheme for transient stability

enhancement using MPC based TCSC controllers.

3.5 CONCLUSION

The chapter has reviewed the problems, causes and power system issues. Knowing the

limitations of limited infrastructure and maximum utilisation of transmission networks,

the chapter has highlighted the role of FACTS devices in enhancing power system

performance and has given emphasis on the effectiveness of series compensation in

maintaining transient stability. With a brief review of power system controller

requirements the chapter has discussed various prediction based methods its limitations

and applications related to power systems. In view of the many attempts of applying

predictive control algorithms to power system applications, the first part of the research,

developing a receding horizon control based TCSC controller for implementation and

validation single machine infinite bus sample system. With reference to this the next

Page 79: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

60

chapter will discuss the problems associated with application of existing predictive

control schemes to power systems and will prepare a background theory which is used

for developing the RHC-based controller in subsequent chapters.

Page 80: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

61

Chapter 4 Theory and Overview of Predictive

Controllers

4.1 INTRODUCTION

Having reviewed various predictive control-based controllers used for power system

stability improvement in Chapter 3, this chapter will develop the necessary background

theory for a new proposed predictive control-based TCSC controller for transient

stability enhancement. The chapter will first explain predictive control theories

developed basically in control and chemical fields in general, along with their pros and

cons. While the second part of the chapter will cover necessary background theory for

the application of predictive control strategy in power systems. While there are many

predictive control schemes developed and reported in literature [71-73], in this research

a FACTS controller is developed based on the receding horizon principle (RHC).

The three control strategies which had been investigated independently, RHC is one

type of predictive control while the other two well-known predictive controls are

generalised predictive control (GPC) and model predictive control (MPC). GPC is

based on single-input and single-output models such as the auto regressive moving

Page 81: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

62

average, or the controlled auto-regressive integrated moving average models, which

have been used for most adaptive controls. The basic idea of GPC is to calculate a

sequence of future control signals in such a way that it minimises a multistage cost

function defined over a prediction horizon. The index to be optimised is the expectation

of a quadratic function measuring the distance between the predicted system output and

some predicted reference sequence over the horizon plus a quadratic function measuring

the control efforts [73] .

The term ‘predictive’ is used in GPC since the minimum variance is given in predicted

values on the finite future time. The term ‘predictive’ is used in MPC since the

performance is given in predicted values on the finite future time which can be

computed by using the model. The performance for RHC is the same as the one for

MPC. Thus, the term ‘predictive’ can be incorporated in RHC as the receding horizon

predictive control (RHPC) or simply RHC.

4.2 PREDICTIVE CONTROL METHODOLOGY

4.2.1 Model predictive control (MPC)

MPC has been developed on a model basis process industry area as an alternative

algorithm to the conventional proportional integrate derivative (PID) control that does

not utilise the model. The purpose of MPC is to achieve online accurate tracking of the

trajectory delivered by the dynamic real-time optimiser. The MPC solves a constrained

optimisation problem online and determines an optimal control input over a fixed future

time horizon, based on the predictive future behaviour of the process. Although more

than one control move is generally calculated, only the first one is implemented. At the

next sampling time, the optimisation problem is reformulated and solved with new

measurements obtained from the system. The optimal reference trajectories for the

manipulated and controlled variables are produced by the dynamic real-time trajectory

optimiser and passed to the MPC.

The status of system variables (including control variables), updates of the model

parameters and estimates of disturbance signals are assumed to be available online and

able to be imputed to the MPC. Given factors such as the initial system status,

Page 82: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

63

information about the type of disturbances, and the reference trajectories, the optimiser

produces the manipulated variables such that the input and output trajectories follow the

reference trajectories as closely as possible subject to the constraints imposed in the

optimisation.

The commercially available, model predictive controllers vary in many details but they

all are based on finite time horizon optimisation problems, based on one linear model at

a time. The characteristic feature of MPC is that the control strategy is determined by

the optimisation of a performance function on a finite time interval. This interval

stretches from the current time to a time instant, which is a fixed time slot ahead. The

optimal control is calculated and implemented only until new measurements are

available. Based on new measurements, an update of the control strategy is determined

by repeating the optimisation of the performance function at the next time step. In this

way, the control strategy depends on the measurements and could therefore be known as

feedback type.

MPC can be described in short as a control methodology, which allows explicit

integration/inclusion of the constraints (imposed on the controller system and/or

employed controls) and explicit expression of the control quality criteria in the control

objective [71, 74]. The main features of MPC are:

(a) It can be used to control a great variety of systems, including those with non-

minimum phase, long time delay or open-loop unstable characteristics;

(b) It can deal with multivariable, multi-input-multi-output as well as single-input-

single-output systems; and

(c) Its system constraints can be readily treated within the optimisation process

4.2.2 Receding horizon control (RHC)

RHC, which is based on state-space framework, has been developed in academia as an

alternative control to the LQ controls. The basic concept of RHC is that: at the current

time, the optimal control is obtained,(either closed-loop type, or open-loop type), on a

finite fixed horizon, from the current time k, say [k, k+N]. Among the optimal control

Page 83: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

64

on the entire fixed horizon [k, k+N], only the first one is adopted as the current control

law. The procedure is then repeated the next time, say [k+1, k+1+N]. The term ‘receding

horizon’ is introduced, since the horizon recedes as time proceeds.

4.2.3 General MPC problem formulation

One of the reasons for the fruitful achievements of MPC algorithms is the intuitive way

it addresses the control problem. In comparison with conventional control, which often

uses a pre-computed state or output feedback control law, predictive control uses a

discrete-time model of the system to obtain an estimate (prediction) of its future

behaviour. This is done by applying a set of input sequences to a model, with the

measured state/output as the initial condition, while taking into account constraints. An

optimisation problem built around a performance oriented cost function is then solved

to choose an optimal sequence of controls from all feasible sequences. The feedback

control law is then obtained in a receding horizon manner by applying to the system

only the first element of the computed sequence of optimal controls, and repeating the

whole procedure at the next discrete-time step.

In short MPC is built around the following key principles:

(a) The explicit use of a process model for circulating predicting of future plant

behaviour.

(b) Optimisation of an objective function, subject to constraints which yields an optimal

sequence of control.

(c) A receding horizon strategy, so that at each instant the horizon is moved towards the

future, which involves the application of the first control signal of the sequence

calculated at each step.

MPC intends to force the controlled system, which are expressed by the system state, to

follow the desired trajectory. The trajectory applies an optimal sequence of manipulated

control inputs in the time instant/samples within the specified time horizon. Both the

system state values, as well as control inputs, can be subjected to the inequality and/or

equality constraints. To capture the system dynamics and predict the system response to

the control inputs, the model of the system is needed; which is accomplished here by

introducing the equality constraints. The various MPC algorithms propose different cost

Page 84: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

65

functions for obtaining control law as given in [43, 63, 75] and shown in Appendix D.

However, the general aim is that the future output (y) on the considered horizon should

follow a determined reference signal yref and at the same time, the control efforts (∆u)

necessary for doing so should be penalised. The general expression for such an

objective function will be [71]:

2

1

2

1

221 ])1([)]()([),,( ∑∑

==−+∆++−+=

uN

j

N

Nju jtjttjtNNNJ uRyyQ ref (4.1)

The parameters N1 and N2 are the minimum and maximum prediction horizons and Nu is

the control horizon, which does not necessarily have to coincide with the maximum

horizon. The N1 and N2 mark the limits of the instants in which it is desirable for the

output to follow the reference. The term yref, which is also known as a reference

trajectory provides an advantage in predictive control. If the future evolution of the

reference is known a priori, the system can react before the change has effectively been

made, thus avoiding the effects of delay in the system response. However, in

minimisation the majority of methods usually use a reference trajectory which does not

necessarily have to coincide with the real reference as is the case in most of the power

system scenarios.

The weight matrices/vectors Q and R express the importance of the close tracking of the

reference for various states. The weight matrix R can be used to define the control

efforts. The overall controller performance is tuned using Q and R (e.g. accuracy,

aggressiveness).

4.2.4 Comparison of PID and MPC

The PID controllers by far are the most dominating form of feedback in use for more

than 90% industry applications used for wide range of problems, such as process

control, motor drives etc. However, the performance of conventional PID controllers

can be severely degraded if a process has a relatively large time delay compared to the

dominant time constant, and PID controllers cab only be detuned to retain closed loop

stability resulting in sluggish performance [76]. The widespread use and success of

Page 85: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

66

MPC applications attests to the improved performance of MPC compared to PID for

control of difficult process dynamics for more advanced controls. In study carried out in

[77], it has been reported that, the MPC controller was capable to maintain the variation

of the controlled variables much closer to the set points than the classical PID

controllers. In addition, it was claimed that using MPC, it is not only possible to save

equipment and energy cost, but the plant can also be exploited at its maximum capacity.

As MPC finds an edge over traditional PID controllers, next section will describe in

detail about various strengths of the predictive controlled based methods.

4.2.5 Strengths of predictive controlled based methods

As mentioned in [14], an issue with power systems is the control of large complex

nonlinear systems. MPC has been shown to be successful in addressing many large

scale nonlinear control problems and therefore is worth considering for stabilisation of

power systems. While MPC is suitable for almost any kind of problem, it displays its

main strength when applied to problems with;

(a) a large number of manipulated and controlled variables;

(b) constraints imposed on both the manipulated and controlled variables;

(c) changing control objectives and /or equipment failure (sensor/actuator); and

(d) time delays.

The strengths of MPC that are relevant to the task of power system stabilisation are the

explicit handling of constraints. Predictive controls based on the state space model can

be dealt with in terms of RHC instead of MPC although MPC based on the state-space

model is same as RHC. The advantages of RHC control as given in [73]:

(a) Applicability to a broad class of systems. The optimisation problem over the

finite horizon, on which RHC is based, can be applied to a broad class of systems,

including nonlinear systems and time-delayed systems. Analytical or numerical

solutions often exist for such systems.

Page 86: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

67

(b) Systematic approach to obtain closed-loop control. While optimal controls for

linear systems with input and output constraints or nonlinear systems are usually open-

loop controls, RHCs always provide closed-loop controls due to the repeated

computation and the implementation of only first control.

(c) Constraint handling capability. For linear systems with the input and state

constraints that are common in industrial problems, RHC can be easily and efficiently

computed by using mathematical programming, such as quadratic programming (QP)

and semidefinite programming (SDP). Even for nonlinear systems, RHC can handle

input and state constraints numerically in many cases due to optimisation over finite

horizon.

(d) Good tracking performance. RHC presents good tracking performance by

utilising the future reference signal for a finite horizon that can be known in many cases.

In infinite horizon tracking control, all future reference signals are needed for the

tracking performance. However, they are not always available in real applications and

the computation over the infinite horizon is almost impossible. In PID control, which

has been most widely used in the industrial applications, only the current reference

signal is used even when the future reference signals are available on a finite horizon.

This PID control might be too short-sighted for the tracking performance and thus has a

lower performance than RHC, which makes the best of all future reference signals.

There are many such advantages listed in [73] such as adaption to changing parameters

which is very important from a power system application point of view where power

system topology may change after disturbance.

4.2.6 Predictive control developments in literature

MPC was popularised in the 1970s for control of petroleum refinery operations, which

often operate at constraints on manipulated variables or controlled variables. Since then,

MPC has become the benchmark for complex constrained multivariable control

problem in the process industries. A good literature review of past, present and future of

model predictive control is presented in [78-82] . MPC and RHC are forms of control in

which current control action is obtained by online solving at each sampling instant using

Page 87: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

68

the current state of the plant as the initial state. The sequence and the first control in

this sequence is applied to the system [83]. Linear MPC refers to a family of MPC

schemes in which linear models are used to predict the system dynamics even though

the dynamics of the closed-loop system is nonlinear due to the presence of various

constraints. The linear MPC approach has found successful applications, especially in

the process industries as reported in [84] .

The authors in [85] proposed a predictive control approach for the optimal control of

nonlinear systems, in which it has been claimed that the main features of an explicitly

analytical form of the optimal predictive control are - online optimisation is not

required; stability of the control-loop system is guaranteed, the whole design procedure

is transparent to designers; and the resultant controller is easy to implement. By

establishing the relationship between the design parameters and time-domain transients

it is shown that the design of an optimal generalised predictive controller to achieve

desired time-domain specifications for nonlinear systems can be performed by looking

up tables.

In a survey paper of theory and practice in MPC [82], authors have addressed the

important issues for any control system and then reviewed a number of design

techniques emanating from MPC, namely Dynamic Matrix Control, Model Algorithmic

Control, Inferential Control and Internal Model Control. These are put in perspective

with respect to each other and the relation to more traditional methods like Linear

Quadratic Control is examined.

It has been observed that nonlinear model predictive control is an attractive strategy for

controlling complex systems as it offers good dynamic performance while ensuring

operation within certain physical limits. This feature enables the system operator to run

the system near constraint boundaries, which can increase productivity without

sacrificing quality [86]. In general, MPC computes an optimal sequence of manipulated

inputs, which minimises a tracking error, i.e. the difference between the desired

reference output and its real value, subject to constraints on inputs and outputs. As

explained in [61, 62], this can be formulated in the continuous time domain for a

general case, applicable to nonlinear systems.

Page 88: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

69

Nonlinear generalised predictive control has been observed to provide a powerful means

of stabilising power systems following large faults. The controller is obtained by

optimising a quadratic criterion over a fixed horizon, using nonlinear reference models

of the power system. Normally, the feedback loop is closed with the first control, and

the computations are repeated at the next sample. This scheme has the disadvantage of

requiring large horizons to assure stability. In [87], it is mentioned that appropriate

selection of the reference trajectory permits the use of short prediction horizons for the

controller to conduct the power system states to a neighbourhood of the post-fault

equilibrium. In this region, local asymptotic stabilisation is provided by the linear

controller.

Though the predictive control method is well established in chemical and control

engineering it has some major issues which will be discussed in next the two sections of

this chapter. The first section will discuss the control strategy related issues while the

second section will highlight the issues focused on application of predictive control to a

power system point of view.

4.3 MAJOR DRAWBACKS OF PREDICTIVE CONTROL

The two major drawbacks of predictive control are: firstly by, its open control loop

nature, and secondly by the computational burden associated with the solution of the

optimisation problem. The first obstacle can be overcome by introducing an implicit

feedback in the form of repetitive computation of control laws in a receding horizon

manner which has been proven for infinite horizon control. The second obstacle has

restricted the application of MPC, mostly to control slower processes with the

dynamics, in an order of minutes (such as in the chemical industry). This is also

probably the factor limiting a wider spreading of MPC in power systems up to now

[43]. Getting an appropriate model of the system is also one of the significant limitation

as mentioned in [74].

In order to solve the predictive control problem, there must be a way of computing the

predicted values of the control variables. This must cover from the best estimation of

the current state and the assumed future inputs as well as the latest input and the

assumed future changes. The way in which the predictions are made has great effect on

Page 89: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

70

the performance of the closed-loop system running under predictive control. So the

choice of prediction strategy is another ‘tuning parameter’ for predictive control, just as

are choice of horizon and cost functions. Furthermore, the prediction strategy follows in

a rather systematic way from assumptions made about disturbances acting on the system

measurements errors such as noise. Hence, it can be said that rather than choosing a

prediction strategy, a model of the environment is specified.

4.3.1 Stability issue

Predictive control, using the receding horizon idea, is a feedback control policy. There

is therefore a risk that the resulting closed loop might be unstable. Even though the

performance of the plant is being optimised over the prediction horizon, and even

though the optimisation keeps being repeated, each optimisation ‘does not care’ about

what happens beyond the prediction horizon, and so could be putting the system into

such a state that it will eventually be impossible to stabilise. This is particularly likely to

occur when there are constraints on the possible control input signals. The problem

arises because the prediction horizon is too short-sighted, and it turns out that stability

can usually be ensured by making the prediction horizon long enough, or even infinite.

Another way of ensuring stability is to have any length of horizon, but to add a terminal

constraint, which forces the state to take a particular value at the end of the prediction

horizon.

Stability has been one of the main problems in MPC, ever since early MPCs for linear

systems were criticised for their loss of stability [84, 85]. This problem has been solved

for linear systems in various ways such as infinite horizon predictive control, terminal

constraints and the fake algebraic Riccati equations (FARE). Though there are some

good results shown in literature for nonlinear system for addressing stability issues [88],

from a computational point of view solving a nonlinear dynamic optimisation problem

with equality constraints is highly computationally intensive, and in many cases is

impossible to perform within a limited time. In [88], Chen et al, developed a practical

MPC with guaranteed stability for general nonlinear systems. It tried to address several

issues in the implementation of nonlinear MPC, including computational delay, loss of

optimality in the optimisation procedure and stability, but computational burden was not

considered.

Page 90: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

71

The key feature of all model based predictive control methods is open loop optimal

control rather than closed-loop control in a moving horizon. Application of the MPC

concept to nonlinear systems like power systems leads, in general, to involved nonlinear

programming problems. In general, the optimisation problem is nonconvex and leads to

many difficulties impacting on implementation of MPC. These difficulties are related to

feasibility and optimality, computation and stability aspects. The important distinction

in nonlinear is not linear versus nonlinear, but rather convex versus nonconvex. If the

resulting nonlinear optimisation problem is convex, there exist methods which ensure

convergence to a global minimum, which is unique if the performance criterion is

strictly convex.

On the other hand, if the system to be controlled is nonlinear, even if the cost function

and constraint sets are convex, the control problem will be, in general, a nonconvex

nonlinear optimisation problem. Therefore, finding a global optimum can be a difficult

and computationally very demanding task. In other words, non-convexity makes the

solution of the nonlinear programming uncertain [89].

To overcome the limitations of nonlinear and linear controllers, a combination of both

linear as well as nonlinear controllers is suggested in [87]. When the faults concerned

are large, the proposed nonlinear predictive controller based on TCSC returns the power

system state to a small neighbourhood of the post-fault equilibrium. In this region,

linear controllers are designed to provide effective damping to the origin. The

coordination of linear and nonlinear controller offers powerful means of extending the

stability limit of power systems.

4.3.2 Choice of horizon

One of the main issues in optimal control is whether or not the closed-loop system

under derived optimal control law is stable. As discussed in [72] the first obvious option

can be an infinite horizon technique to minimise the performance objective determined

by cost. However, the open-loop optimal control problem, that must be solved online, is

often formulated in finite horizon. The input is parameterised finitely in order to allow a

real-time numerical solution of the nonlinear open-loop control problem. It is obvious

Page 91: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

72

that the shorter the horizon the less costly the online optimisation solution. Thus, from a

computational point of view it is desirable to implement a predictive control with a

short horizon. However, when a finite prediction horizon is used, the actual closed-loop

input and trajectories will differ from the predicted open-loop trajectories, even if no

model mismatch and no disturbances are present.

Fig.4.1 (a) and (b) shows the basic difference in finite and infinite receding horizon

formulation of predictive control. With reference to Fig.4.1, at time k, a particular

trajectory is optimal over the prediction horizon Hp at time (k+1) with no disturbance

and no model mismatch. The system will be in same state as it was predicted at the

previous time step. With this initial observation, it is expected that the optimal trajectory

over prediction horizon from time (k + 1) to (k +1+Hp) should coincide with previously

computed optimal trajectory. But in previous optimisation the interval between (k + Hp)

to (k +1+ Hp) was not considered. As a result when a new time interval enters it may

give a totally different optimal trajectory than what was computed in earlier steps.

(a)

(b)

Fig.4.1. (a) and (b) Finite horizon and Infinite horizons with no disturbance and perfect

model respectively

On the other hand, with second case, at time k, an optimal trajectory is determined for

the entire infinite horizon implicitly. As a result, at time (k+1), no new information

enters in the optimisation problem and optimal trajectory continues as the tail of the

previously computed trajectory. The Bellman’s principle of optimality states that the tail

of any optimal trajectory is itself the optimal trajectory from its starting point which is

thus applicable to the infinite horizon problem, while, the finite horizon it does not

apply because at every step new optimisation problems arise.

Page 92: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

73

With reference to above discussion, it can be concluded that short prediction horizon

gives rise to too short-sighted control and may also give rise to stability issues while

stability can be ensured using longer horizons. However, as the horizon length increases

the computational burden increases making solution costlier. In addition, too long a

prediction horizon can degrade the performance as errors in prediction are also large

[90]. In such a scenario, the stability can be ensured by three ways-first by imposing

terminal constraints, second using infinite horizons as explained above and third, the

FARE approach.

4.4 LIMITATIONS OF MPC APPLICATION TO POWER SYSTEM

There are many difficulties that limit the use of this kind of model such as:

(a) computational load in applying MPC to large systems with fast time constants;

(b) lack of identification techniques for nonlinear processes;

(c) requirement of an appropriate system model.

The general tools for nonlinear MPC are not necessarily well developed for the specific

nonlinearities of the power system. The major disadvantage associated with RHC, is

longer computation time compared with conventional nonoptimal controls. In [75],

Zima and Anderson have discussed various techniques in relation to minimisation of

computation efforts. The first option given is, the use of full nonlinear mode and

computations which is most accurate but most time consuming also. This type of

approach can be applied for small sized systems with slow dynamics. The second

frequently used method is to linearise the system equations around the present operating

(equilibrium) point and apply linear MPC. Discrepancy between the linearised model

and the actual system behaviour is then compensated in the next controller step.

However, this involves a risk that the system may undergo large excursions from the

optimal trajectory and even violate imposed constraints. The third approach is based on

approximation of the expected trajectory (Euler prediction) of the system if control

inputs would remain unchanged and the numerical computation of sensitivities of

control inputs impact on this trajectory.

Page 93: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

74

Camponogara, in [90] has discussed distributed model predictive control from a power

system point of view. Typically, MPC is implemented in a centralised fashion. The

complete system is modelled, and all the control inputs are computed in one

optimisation problem. This is possible for small size plants such as in the chemical

industry where it is possible to get all measurements. In large-scale applications, such as

power systems, it is useful (sometimes necessary) to have distributed or decentralised

control schemes where local control inputs are computed using local measurements and

the reduced-order models of the local dynamics. With this background, [90] has tried to

achieve some degree of coordination among agents that are solving MPC problems with

locally relevant variables, costs and constraints, but without solving centralised MPC

problem. Such coordination schemes are considered to be useful when local

optimisation problems are much smaller than a centralised problem, such as new

deregulated power markets. A power system with a two area case is studied in [90] for

load-frequency problems using the distributed MPC technique considering

communication network and coordination problems in power systems. However, the

power system model used is very much simplified.

As concluded in [14] that centralised or global MPC may face some difficulties such as

lack of knowledge from system and computation costs because of complex and highly

nonlinear structure of power systems, the authors have developed a new approach to the

control angle difference of multi-machine systems using a combination of MPC and

energy function. However, the power system model used is again a simplified low order

dynamic model. Centralised MPC is not well suited for control of large-scale,

geographically expansive systems such as power systems. However, the performance

benefits obtained with centralised MPC can be realized through distributed MPC

strategies. Such strategies rely on decomposition of the overall system into

interconnected subsystems, and iterative exchange of information between these

subsystems [91]. As TCSC devices are mainly installed in long distance, high voltage

AC transmission system, [70] has considered simplified models of interconnected

power systems for developing nonlinear optimal predictive controller for TCSC and

then extended it for multi-machine power systems with some approximations and

assumptions.

Page 94: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

75

4.4.1 Linear model for nonlinear system issue

For computational reasons, MPC applications largely have been limited to linear models

i.e. those in which the dynamics of the system models are linear. Such models often do

not capture the dynamics of the system adequately, especially in regions that are not

close to the target state. In these cases, nonlinear models are necessary to describe

accurately the behaviour of physical systems. From an algorithmic point of view,

nonlinear model predictive control (NMPC) requires the repeated solution of a

nonlinear optimal control problem.

A fundamental difficulty with the NMPC approach is that the implementation platform

must be capable of solving a constrained optimisation problem within a specified time

limit. This time decreases as the speed of the dynamics to be controlled increases. As a

result, the implementation of NMPC has, to date, been generally limited to plants with

slow or otherwise very simple dynamics so that the time constraints in computing a

solution are relaxed. Surmounting this difficulty of computational overhead to achieve

the benefits of MPC for linear systems has attracted research attention [86].

The relationship between the linear model given in Appendix D by (D1-D3) and the real

system need careful consideration of predictive control. In most control methodologies

the linear model is used offline, as an aid to analysis and design. In predictive control it

is used as part of the control algorithm, and the resulting signals are applied directly to

the system. Therefore careful attention must be paid to appropriate treatment of

measurements before using them in the control computation algorithm, and the

computed control signal.

The two major problems of this computational burden are first, large computational

delay and second, achieving global and sometimes even local minimum in a given time

limit of each optimisation cycle. To tackle the online computational issue, one solution

is proper choice of prediction horizon. However, it may be difficult to predict system

output for a longer horizon and a smaller horizon may not guarantee required close-loop

stability [72]. Though there have been many efforts to extend MPC from linear systems

to nonlinear systems the two major obstacles to the extension of MPC from linear to

Page 95: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

76

nonlinear systems like power systems are: stability and the online computational burden

[88].

In [87], Rajkumar has proposed a high performance nonlinear predictive controller

using TCSC for the stabilisation and damping of multi-machine power systems. The

controller is designed using the classical model of power systems and effective means

of extending the stability limit is achieved by such linear and nonlinear controller

coordination. After a large disturbance, the nonlinear predictive controller based on

TCSC, brings the system to a small neighbourhood of the post-fault equilibrium. Linear

controllers will then provide effective damping to the origin.

4.4.2 Use of single machine equivalent

It has been observed that, to overcome computational requirements and avoid

complexity of power systems, the control strategies are developed based on a single

machine equivalent model. The methods relying on a one-machine infinite bus (OMIB)

equivalent are based on the observation that the loss of synchronism of a multimachine

power system originates from the irrevocable separation of its machines into two groups

that they successively replace by a two-machine equivalent and further by a one-

machine infinite bus equivalent. Thus an OMIB may be viewed as a transformation of

the multidimensional multimachine dynamic equations into a single dynamic equation.

Depending on power system modelling and the assumed behaviour of the machines

within each group, OIMB can be distinguished in three types as explained in [92, 93] as

time-invariant, time varying and generalised. The time-invariant OMIB is based on

assumption of simplified power system model and coherency of the machine within

each one of the two groups, while time-invariant is based on simplified model but no

coherency in the group. Using detailed power system model instead of simplified one is

categorised as generalised OMIB.

With reference to various advantages and disadvantages [92], generalised OMIB is

found much better for quantifying the severity of instability and also deciding required

compensation to make such a system stable. To restore transient stability [94] has

proposed a well-behaved optimal power flow model with embedded transient stability

constraints which can be used for both dispatching and re-dispatching. The transient

Page 96: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

77

stability constraints are formulated by reducing the initial multi-machine model to a

one-machine infinite-bus equivalent FACTS controller input signal derivation.

4.5 CONCLUSIONS

Starting from a basic introduction, the chapter has discussed various strengths and

limitations of predictive control strategies to develop its application. After critically

reviewing the problems associated with predictive control based methods to power

systems, first an attempt will be made to develop an RHC-based TCSC controller for a

single-machine-infinite-bus system for improving transient stability improvement.

It has been seen that insertion of thyristor controlled series compensation (TCSC) can

lead to increased power transfer and in turn transient stability. Following the

disturbance, the entire system dynamics change and many parameters undergo variation

which need to be controlled, but the relative rotor angle of synchronous machine is the

key parameter which will decided the overall dynamics of the perturbed system. As per

the first swing stability criteria, the relative rotor angle swing will be an ideal indicator

of system stability. The issues related to the prediction scheme will be selected to

maximise the performance and assure system stability in post-fault conditions.

Page 97: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

78

Chapter 5 MPC-based TCSC Controller for

Power System Transient Stability

Improvement

5.1 INTRODUCTION AND OBJECTIVES

The previous chapter has given a basic introduction to predictive control strategies and

discussed their strengths and limitations with application to power systems. This chapter

focuses on developing an RHC-based TCSC controller, to address the power system

stability related problem, using MPC controller.

It has been seen that insertion of TCSC can lead to increased power transfer and in turn

transient stability. Following a disturbance, the entire system dynamics change and

many parameters undergo variations which need to be controlled. The relative rotor

angle of synchronous machine is the key parameter which will decided the overall

dynamics of a perturbed system. As per the first swing stability criteria, the relative

rotor angle swing will be an ideal indicator of system stability. The issues related to the

prediction scheme will be selected to maximise the performance and assure system

Page 98: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

79

stability in post-fault conditions. In current literature where RHC controllers are

developed for power system, in most cases, the model used is classical. This has

limitations which does not give insight in machine dynamics which plays important role

in system stability. Knowing the various pros and cons of RHC related methods and

their applications in power system, an attempt has been made here to apply RHC

controller to power system transient stability in order to maintain its detailed fifth order

rotor dynamic model, having 10 variables including exciter and prime-mover systems.

In order to improve the transient stability performance, early detection of faults, and fast

fault clearance is most important. In an initial attempts reported in [95], a bang-bang

type control of switched series capacitors was proposed but the technique for

determining the required switching instant was not given. In addition, the use of fixed

series capacitors can give rise to the risk of sub-synchronous resonance (SSR) [24].

However, with recent advanced developments in power electronics, the FACTS devices

can be used for power system stability enhancement in healthy or post-fault conditions,

as explained in [47, 96].

FACTS devices of both the shunt and series form can contribute to stability

enhancement. However, it is accepted in general that series compensation is more

effective in improving or maintaining system transient stability. Among the FACTS

devices-based series compensators, the TCSC is the most popular, and is extensively

used in power systems, particularly where long-distance transmission interconnections

are required. A TCSC is directly connected in series with the transmission line for

which it provides the compensation, without the use of a coupling transformer.

At present, the reference input to a TCSC, either active-power reference or reactance

reference, is determined, based mainly on a steady-state operating condition and offline

calculation. Small-disturbance stability enhancement is achieved by a supplementary

damping controller (SDC) in conjunction with the TCSC main controller. There has

been extensive research on the design of SDCs, including the online and adaptive tuning

of their parameters [97]. However, research on the real-time control of TCSC for

enhancing or maintaining power system transient stability following a large disturbance

has been very limited, where a simplified power system is adopted for forming the

control law [13]. It is acknowledged that the real-time and optimal control of the

Page 99: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

80

TCSC reactance reference is the most effective means of transient stability enhancement

[12-14] .

In this chapter an attempt has been made to derive the control law based on the power

system detailed dynamic model in the prevailing operating condition. Based on the

concept of receding horizon control (RHC) [12, 13] the overall control is subdivided

into a series of time horizons, and the power system dynamic model is used in a

predictive mode in each time horizon. At the start of each time horizon, for which

current system dynamic response is available via wide-area measurements (WAMs),

subsequent system responses within the time horizon are predicted using the system

dynamic model, and are optimised, subject to FACTS devices operating limits, to obtain

the optimal reference inputs to FACTS devices controllers. The objective function in the

optimisation represents the system dynamic performance index (DPI) expressed in

terms of relative rotor angles. The relationship, required in forming the objective

function, between the DPI and the FACTS controllers input references, is derived

through the linearisation of the power system model around the current operating point.

The variables in the relationship derived are the FACTS controllers input references,

and their optimal values obtained from the optimisation are used for setting the input

references to the FACTS devices controllers. The RHC is applied repeatedly for

successive time horizons in each of which an optimal set of FACTS devices input

references is derived and implemented.

The control algorithm proposed is implemented in software, and then tested by

simulation with a single-machine-infinite-bus (SMIB) power system having a TCSC.

The results presented in the chapter indicate the effectiveness of the control

methodology proposed. There is definitely a need for predicting in advance where the

system goes before the control actions are deployed [46] i.e. a look-ahead approach. A

simulation environment is proposed using fresh real-time information from the EMS, in

addition to historical data recordings that can be fed into the online dynamic simulation.

The accuracy of the system state prediction simulation environment depends on how

reliable our schedules and limits are in the short-term time frame. Under catastrophic

conditions a window of approximately few minutes ahead may help to reduce the risk of

system operation.

Page 100: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

81

The fast state prediction simulation and, a change of simulation scope and region of

interest may help. The state prediction could be triggered automatically (or on demand)

to assess control movements and reconfiguration plans under disturbances. A lot of

potential problems can be caught after control deployment. For emergency conditions,

models including only the measurements that can be delivered in the quickest time to

the control centre are required. These measurements may include items such as:

important generator outputs, frequency and voltage measurements, important flow

measurements, most important high transmission backbone lines internal to a specific

control area; and important new measurements such as angle measurements. The goal is

to estimate where the system is heading and what the security level is going to be.

The system recovery depends very much on new changed system topology. The online

dynamic simulation will recognise and characterise situations, it will predict collapse

and unstable behaviour and it will recommend re-adjustment to prevent, or corrections

to recover, from failures. The need for high frequent solutions of a quadratic program

and repeated linearisation of the nonlinear model determines the main computational

load for MPC.

5.2 AIM OF PROPOSED METHOD

The SMIB system is represented by two mechanical axes or swing equations, three rotor

flux equations, three differential equations describing excitation dynamics and two

differential equations for prime-mover and governor dynamics. The controller

coordination between FACTS devices, power system dynamics and proposed MPC

controller is shown in Fig.5.1.

The output of the controller will decide the input reference for FACTS device for which

this predictive controller needs some reference trajectory to follow which is represented

by Yref in Fig.5.1 above. The reference output Yref, may be pre-selected on the basis of

contingency studies. Keeping the reference output fixed at the post-fault equilibrium

may impose large computation burdens on the controller, since a large horizon may

have to be selected to assure a solution to the minimisation problem. Different options

for the selection of the reference output, such as e.g. keeping it fixed at some value

close to the post-fault equilibrium, can be helpful in realising the control objectives with

Page 101: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

82

shorter horizon. Though, in the present study, reference output is kept the same as that

of pre-fault system status, it can be changed as per the changed power system network

or any other value based on selected criteria. The system performance is observed in

every stage of pre-fault, during-fault and post-fault. In post-fault the system stability is

observed when fault clearing time is greater than critical clearing time.

Model Predictive ControllerOriginal Nonlinear system model

Linearized Model for optimization

Quadratic programming QP problem

Cost function and Constraints

𝛥x

Y

YrefY

FACTS

Power System

Fig.5.1 Proposed strategy for RHC-based TCSC controller

5.3 POWER SYSTEM MODELLING

The power system which incorporates both continuous dynamic and discrete events can

be divided into the two parts. Firstly, the continuous dynamical system which is

modelled as a differential algebraic equation (DAE) system which is formed from two

ordinary differential equations (ODE) of the load and the algebraic equations (equality

constraints). The ODE as well as the model of the algebraic equations is nonlinear.

Additionally, the saturation of the internal Automatic Voltage Regulator (AVR) of

generators can be included as part of the continuous dynamics.

As proved in various studies [98] that rotor angle trajectory is the dominating factor in

determining stability, in the present proposed scheme, rotor angle is considered as

stability performance index. Similarly, for the choice of FACTS devices which can

provide effective control in power system stability improvement, TCSC is preferred

over others. One of the reasons why TCSC is preferred is because of its distinctive

quality of extremely simple main circuit topology. The capacitor inserted directly in

series with the transmission line and thyristor controlled inductor mounted directly in

parallel with this capacitor, thus requires no interfacing equipments such as high voltage

Page 102: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

83

transformers. This makes TCSC more economical than other competing FACTS

devices. There are various advantages of TCSC as compared to other FACTS devices

which are addressed well in literature [18, 24, 99].

In the first part of the thesis, as the focus is to use a nonlinear dynamic model and

coordinate it with the MPC controller with FACTS devices for first swing control, a

simplified single machine system is considered. To keep the complexity within the

limited range, TCSC is represented as a simple variable reactance. In the second part of

the thesis, multimachine systems and detailed dynamic models of TCSC will be

developed and considered.

5.4 RHC ALGORITHM

The RHC module solves online a constrained optimisation problem and determines an

optimal control input over a fixed future time-horizon, based on the predicted future

behaviour of the system and on the desired reference trajectory. As by now, linear MPC

theory is quiet mature and important issues such as online computation, the interplay

between modelling/identification and control and issues like stability are well addressed

[84]. In this proposed method, the predicted future system behaviour is represented as

the sum of a nonlinear prediction component and a component based on linear time-

varying models defined along the reference trajectory, which needs to be tracked. The

first component constitutes a future output prediction using non-linear simulation

models, given initial system inputs and disturbance history. The second component uses

linearised models for prediction of future process output as required for calculation of

optimum future system manipulation.

The constrained optimisation problem leads to a quadratic programming problem which

is a convex optimisation problem. The status of real-time optimisation is transferred to

the RHC to ensure feasibility and proper functionality of all system components. A need

for high frequent solutions of a quadratic program and the repeated linearisation of the

nonlinear model determine the main computational load for RHC. As explained in [71,

73] the RHC concept is well suited for finding control laws in an optimal way for

hybrid systems like power networks.

Page 103: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

84

The concept of RHC [72, 100] in the discrete time-domain is that of subdividing the

control into a number of durations each of which is referred to as a time horizon with a

nominated number of time steps. The aim of RHC is to derive a set of control variables

for individual time steps within a time horizon. The control variables derivation is based

on the optimisation of the system responses in a given time horizon. If a particular time

horizon to be considered at the present time, tp, is [ ]tNttttt ppp ∆+∆+∆+ ,,2, where

tt p ∆+ is the start of the horizon, t∆ the time step, and N the number of time steps in

the horizon, then the optimal set of values for the control variables, following the

optimisation, are represented as ( ) ( ) ( ){ }tNttttt ppp ∆+∆+∆+ uuu ,,2, . One option is

to implement these control variable values at time instant

[ ]tNttttt ppp ∆+∆+∆+ ,,2, respectively and then to move on to the next time

horizon starting at time ( )( )tNt p ∆++ 1 . However, this option might present a problem

if within the horizon ( ) ( ) ( ){ }tNttttt ppp ∆+∆+∆+ ,,2, , there are events occurring after

pt , which have not been represented in the system model at time pt , used for response

evaluations and optimisation. The second option is to implement the control value

( )tt p ∆+u at time tt p ∆+ , and move on to the next horizon which starts at time

tt p ∆+ 2 , i.e. the time horizon to be considered is ( )[ ]tNttt pp ∆++∆+ 1,,2 . The

system model is then updated for time tt p ∆+ , and the RHC procedure is then repeated

for the horizon starting from tt p ∆+ 2 , which will lead to the optimal value for the

control vector at time tt p ∆+ 2 to be implemented.

5.5 LINEARISATION AND OBJECTIVE FUNCTION

In principle, the nonlinear power system model as developed in Chapter 2 can be

represented by the following sets of differential equations and algebraic equations as:

), uyh(x,x =•

(5.1)

( ) 0uy,x,g = (5.2)

Page 104: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

85

In (5.1) and (5.2), h and g are, in general, nonlinear vector functions; x is the vector of

state variables arising from modelling synchronous generators together with their

controllers, FACTS devices and dynamic loads; y is the vector of network nodal voltage

variables, and u is the vector of control variables to be optimised in real time for

maximising transient stability margins. In the present work, the control variables

represent the input references to the main controllers of FACTS devices which have

been selected for participating in the control.

Equation (5.1) and (5.2) can be applied, through a numerical integration, for predicting

the system responses required in individual time horizons. However, it is, particularly in

terms of computing time, difficult, if not impossible, to carry out the optimisation of the

system responses which are implicit nonlinear functions of the control variables. To

avoid this difficulty, the MPC method is based on the linearisation of the nonlinear

model in (5.1) and (5.2) by which the state variables describing power system dynamic

responses are expressed explicitly in terms of a linear vector function of control variable

vector u. This allows the prediction of the system responses within each time horizon in

terms of the variation of the control variables, which provides the basis for forming the

objective function to be minimised. A representative objective function often quoted in

control theory literature is

( )( ) ( )( )∑∑∑∑

====∆+∆+−∆+

N

jpiij

L

i

N

jrefkpkkj

M

ktjtuWZtjtZW

1

2

11

2

1 (5.3)

In (5.3), kZ ’s (for k= 1, 2, . . . , M) denotes the power system responses selected for

optimisation which are formed approximately using the responses at time tp and the

deviation predicted by the linearised system model; refkZ represents the values of the

system responses in the pre-disturbance condition, and the second summation in (3) has

the purpose of minimising the deviation of the control vector from that at time tp. The

objective function in (5.3) is a quadratic function in ( )tjtu pi ∆+∆ (for i= 1, 2, . . . , L

and j= 1,2, . . . , N). Constrained optimisation based on the quasi-Newton method is

directly applicable for minimising the objective function subject to bounds imposed on

Page 105: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

86

the control variables, to identify the optimal values for ( )tjtu pi ∆+∆ from which the

total values of the control variables are formed and implemented. If the second option

referred to in Section 5.4 is adopted, the control vector to be implemented at time

tt p ∆+ is ( ) ( )ttt pp ∆+∆+ uu , and the RHC procedure will be repeated for the next

horizon, starting at time tt p ∆+ 2 as described in the algorithm of Section 5.4.

5.6 RHC ALGORITHM FLOW CHART

Fig.5.2 shows the complete block diagram for implementation of a RHC controller

applied to the SMIB system for transient stability improvement. At the start of any RHC

time horizon, the power system configuration together with its operating state is

available via WAMs and current statuses of the circuit breakers and isolators. This in

conjunction with the system database allows the construction of the nonlinear power

system model relevant to transient stability analysis and simulation. With the RHC

algorithm adopted, linearisation is then required to form the objective function for

minimisation. The outcome of the minimisation is the updated value of the control

variable, which in this case is the TCSC reactance reference. The updated control

variable is sent to the TCSC main controller, for adjusting its input reference as shown

in Fig.5.2. The RHC sequence is repeatedly applied for successive time horizons as

indicated in the loop shown in Fig.5.2.

The solution for output of the optimisation problem is a required TCSC reactance

setting value which will be used to set the input reference for given FACTS devices, i.e.

in this case TCSC. The TCSC input reference is adjusted as per the output of the

controller which will be inserted in series with the transmission line to modify its

reactance to improve system performance in post-fault conditions. The updated system

variables are again used to form a power system model for next time instant and entire

process will be repeated for every time step to update control law, FACTS devices input

references as per changed system conditions by using principle of receding horizon.

Page 106: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

87

Power systemWAMs

Circuit breakers/isolators statuses Database

Nonlinear power system model

Linearisation and prediction

Objective function formulation and

Constrained minimization

Updating the FACTS-devices control variables

FACTS Devices

New Xtcsc optimized value

Fig.5.2 Flowchart of RHC implementation algorithm

5.7 SIMULATION RESULTS

The general RHC algorithm developed in Section 5.2 is applied to the single-machine-

infinite-bus power system with a TCSC as shown in Fig.5.3. The system data is given in

[101]. For dynamic simulation of the power system and the RHC, the fifth-order

generator model [10, 101] is used. The excitation controller is based on IEEE Type-

ST1, the prime-mover and governor is as per [9] and the TCSC, represented in terms of

variable reactance associated with reactance control mode, is adopted from [55].

Fig.5.4 shows the variation of TCSC reactance reference derived from the output of the

RHC. The effectiveness of the control applied to the TCSC input reference is confirmed

in the rotor angle response of Fig.5.5 which indicates that the system transient stability

is maintained throughout the transient operating period, and a new steady-state

condition is reached after about three seconds subsequent to the fault onset.

Page 107: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

88

Fig.5.3 Single-machine-infinite-bus system

The TCSC operates in the reactance control mode. Based on this, the control variable in

the study is that which represents the TCSC reactance reference input. For transient

stability control, the system response selected for forming the objective function in (5.3)

is the generator rotor angle, with the infinite bus as the reference. The system details

and data are given in Appendix E.

The simulation is carried and analysed with two cases. A three-phase-to-earth fault is

applied at bus 3, with the fault clearing time of 200 ms. With reference to the time

origin of Fig.5.4, the fault starts at time 0.5s. The time step length adopted in the study

is 10 ms. The first case study is that when there is no transient stability control by RHC

where the TCSC reference input remains fixed at the pre-fault value. The rotor angle

response in Fig.5.4 indicates that the system transient stability is lost after fault and fault

clearance. The second case study represents the RHC for optimally adjusting the TCSC

reactance reference input signal, based on the control sequence in Fig.5.5.

Fig.5.4 Generator rotor angle response without RHC controller

0 1 2 3 4 50

500

1000

1500

2000

Time (s)

Del

ta (D

eg)

Page 108: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

89

Fig.5.5 TCSC reactance reference input using RHC controller

Fig.5.6 Synchronous generator rotor angle response using RHC based TCSC controller

5.8 CONCLUSIONS

The chapter has developed a comprehensive, flexible, and systematic method for the

formulation of RHC-based TCSC control law, together with its software

implementation which has been validated with many case-studies on SMIB of which

representative results are presented in this chapter.

0 1 2 3 4 5

-0.25

-0.2

-0.15

-0.1

-0.05

0

0.05

Time (s)

TCSC

Rea

ctan

ce (p

u)

0 1 2 3 4 50

50

100

150

200

Time (s)

Del

ta (D

eg)

Page 109: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

90

The simulation is carried out in two parts, first without a RHC controller, which shows

continuous increase of rotor angle with rapid rate, leading systems to a ‘run away’

situation because of acceleration. The second part of the simulation is carried out with a

RHC controller switched on after a clearing fault, where the fault-clearing time is again

the same as that of case one, which is greater than critical clearing time. It has been

observed that as the RHC-based TCSC controller provides the required series

compensation the system regains its stability even if the fault is cleared after the critical

clearing time.

The research and results presented in the chapter have confirmed that it is possible to

form control law which is adaptive to power system operating condition, and effective

in improving or maintaining its transient stability. This is achieved by directly deriving

the relationship between the relative rotor angle and the control variable through

linearisation in individual time horizons, which leads to the objective function to be

minimised for forming successive optimal values for the control variable.

The further development in this research is to extend this strategy for multi-machine

systems with large numbers of generators. The major concerns in this development are

first, due to computation burden because of scale of operations and requirements, the

optimisation problem to be solved online are large. Secondly, the solution must be

obtained in a limited amount of time because it is implemented in a receding horizon

fashion. The third aim is to attempt to develop this RHC algorithm with capabilities for

control of different system dynamics and for the various constraints handling along with

good computational efficiency in optimisation to allow online RHC application.

Page 110: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

91

Chapter 6 Overview of Real Time Controllers

for Power System Stability

Improvement

6.1 INTRODUCTION

The first part of this thesis (Chapter 2 to Chapter 5) has covered detailed dynamic

modelling of various power system components along with FACTS devices as

compensating devices. A new predictive control based FACTS controller is developed

for enhancement of transient stability performance of power systems. Although the

proposed predictive scheme for FACTS controller has shown its effectiveness by

applying linear models for a nonlinear power system and validated it for a single-

machine infinite bus, in reality, power systems are inherently nonlinear. Together with

higher quality specifications increasing demand, tighter environment regulations and

demanding economic considerations are pushing power systems closer to the

boundaries of the admissible operating regions. In such circumstances, linear models are

often inadequate to describe the system dynamics, and nonlinear models must be used.

In addition, the simplified dynamic models of components which are used to overcome

Page 111: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

92

the problem of computation requirement may not reflect correct system status which

might lead to wrong prediction and in turn system stability. This has motivated the

development of the second part of the thesis in which an online control coordination

strategy is developed considering nonlinear models for a real multimachine system with

real time requirements.

To address these limitations, the second part of thesis will consider the requirements for

real multimachine systems, including their detailed dynamic models for accuracy. It will

also consider computation time requirements and computation burden that computing

systems should have for real-time application of controllers. With reference to this,

given that recent advances have been made in computer technology and wide-area

measurement systems, the second part of the thesis is devoted to real-time controllers

and their requirements. Starting with the review of real-time controllers this chapter will

discuss the requirements of real-time power system controllers, including those of

FACTS devices, for achieving power system stability. With the background of reviews

presented in this chapter, the next chapter will develop new online control coordination

of FACTS devices for transient stability improvement for a realistic multimachine

power system network.

6.2 WIDE AREA NETWORK OPERATION

The growing trend towards restructuring the power industry and the ever increasing

demand for power exchange calls for the employment of WAMS for near to real-time

measurements to maintain or improve the stability of the system. To achieve technical

and economical advantages, power systems have been extended by interconnections to

the neighbouring systems. Regional systems have been built-up towards national grids

and later to interconnected systems with the neighbouring countries. Such large systems

came into existence, covering parts of or even whole continents, to gain the following

well known advantages:

(a) Reduction of reserve capacity in the systems (i.e. less spinning reserve);

(b) Utilisation of the most efficient energy resources;

Page 112: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

93

(c) Parallel operation to share power in peak and off peak demand periods making

power generation economical as well as efficient;

(d) Reliability and continuity along with the quality of supply

With the above advantages of interconnection, generators and loads that are over

thousands of miles apart are connected to form a large system. As a result, the general

configuration of a modern power system consisting of generation, transmission and

distribution is often geographically dispersed.

The general trend in power system planning utilises tight operating margins, with less

redundancy because of new constraints placed by economical and environmental

factors. At the same time, factors such as - addition of non-utility generators and

independent power producers, an interchange increase; an increasingly competitive

environmental; and the introduction of FACTS devices make the power system more

and more complex to operate and to control, and thus more vulnerable to a disturbance.

The interesting example of an actual blackout that occurred in Italy can be reviewed

here to explain power system problems, their cause and the complexity in controlling

the severity of these problems. As mentioned in [18], the Italian blackout was initiated

by a line trip in Switzerland. Reconnection of the line after the fault was unsuccessful

because of too large phase angle difference which was about 60 degrees, leading to

blockage of the synchro-check devices. Twenty minutes later a second line tripped,

followed by a fast trip sequence of all interconnecting lines to Italy due to overload.

As such, this example shows that interconnected system have high fault levels and are

prone to becoming unstable, leading to total blackout due to its complexities and power

balance problems when faced with large and severe disturbances. Hence, imminent if

not immediate priorities must be the higher availability and maintaining of efficiency

which is possible only with good information technology based services in power

system management. Conventional SCADA and Energy management system (EMS)

stability control systems currently do not provide efficient solutions in occurrences of

cascaded outages, through any coordinated or optimised stabilising actions. [41]

Page 113: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

94

In modern interconnected networks, a fast developing emergency may involve a wide

area. Since operator response may be too slow and inconsistent, fast automatic actions

are implemented to minimise the impact of the disturbance. These automatic actions

may use local or centralised intelligence or a combination of both. In [40], the author

has discussed the trends in wide area protection of power system. The normal automatic

control can provide either preventive or corrective action. During normal operation, the

focus is on economic optimisation of system operation, while during more challenging

network conditions, such as alert state, or emergency situations, the focus of control

shifts towards stability considerations. The ultimate objective is keeping the maximum

of possible networks intact and the generators connected to the grid. The breakdown

normally results in one or more severe problems in the power system. The main concern

in the emergency state is the system security. System protection schemes form a last

line of defence in the case of severe disturbances.

The authors in [102, 103] have discussed the basic design and special applications of

wide-area monitoring and control systems which complement classical protection

systems and SCADA, also known as Energy Management Systems application.

Currently, the local automatic actions are conservative, act independently from central

control, and the prevailing state of the whole affected area is not considered. Actions

incorporating centralised intelligence are limited to the information anticipated to be

relevant during unforeseen contingencies. There are few schemes that are adaptive to

intelligence gathered from a wide area that respond to unforeseen disturbances or

scenarios.

Historically, only centralised control was able to apply sophisticated analysis because

only at this higher level could computers and communication support be technically and

economically justified. However, with the increased availability of sophisticated

computers communication and measurement technologies, more intelligence can now

be used at a local level. The possibility to close the gap between central and local

decisions and actions will depend on the degree of intelligence put in the local

subsystems. Decentralised subsystems that can make local decisions based on local

measurements and remote information (system wide and emergency control policies)

Page 114: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

95

and/or send pre-processed information to higher hierarchical levels are an economical

solution to the problem.

Measurements provided by the sensors are usually collected by a host computer in

substations over a local-area communication network. This substation based sensor

network is, in turn, a node of a wide-area network which collects and collates data from

various substations. It then performs various application tasks in order to arrive at

protection and control decisions [104]. These decisions are then communicated to the

substation computers and through them to various actuators in the substations. The

power, communication and computer infrastructures are thus closely interlinked. The

Fig.6.1 shows typical flow of information and communication for various levels of

primary and secondary protections for given disturbances (sensing, computing and

communication infrastructure). Such geographically dispersed system require

functionally complex monitoring and control systems, as the performance of the power

system decreases with the increasing size, loading and complexity of the network.

6.3 ROLE OF WAM IN MAINTAINING WAN OPERATIONS

From Fig.6.1 the power system can be viewed as a large-scale, multi-input, multi-

output, nonlinear system distributed over large geographic areas and needing fast

communication as well as accurate control to maintain reliability. When a major power

system disturbance occurs, protection and control systems have to limit impact, stop the

degradation and restore the system to a normal state by appropriate remedial actions.

WAM and protection systems limit severity of disturbances by early recognition as well

as proposition and execution of coordinated stabilising actions.

There are three approaches to control power system dynamics using wide-area

measurements. First is a control-room operator response to information derived from

WAMS; second is a discontinuous control, such as switching control modes or

protection schemes in response to specific observed dynamic conditions; and third is a

continuous control using wide-area signals as controller input signals.

Page 115: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

96

System-wide controller and coordinator

Data processing

Data Delivery

PMU 8

PMU 1

PMU 10

PMU 2 PMU

4

PMU 5

PMU 3

PMU 7

PMU 9

PMU 6

Local area control Local area control Local area control

SS 1 SS 2 SS 3Data Acquisition Data Acquisition

Data Acquisition

Power flow direction within local area

Information signal flow (communication links for data flow)

Fig.6.1 Power flow and information flow network in wide-area network

The scheme of WAM alarming during heightened risk of instability with automated

diagnostics and operator responses can help to manage potential dynamics problems.

This is because of their prior knowledge and mitigation procedures, identified in real-

time, for responding unanticipated events. The comparison table of various control

schemes with their advantages and disadvantages is given in [105]. It is argued that the

tools for control-room operators to observe and take action on dynamics problems are a

necessary step before automated systems for security against dynamic issues can be

widely deployed. This stage enables the operator to use discretion in balancing the

increased risk of instability against the cost associated with a dispatch action to reduce

the risk. On the other hand, the advanced measurement and communication technology

in wide-area monitoring and control, FACTS devices can prove better tools to control

the disturbance and a better way to detect and control an emergency.

Page 116: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

97

In [106], the commercial availability of a wide area monitoring system and its

experiences are described. With the emphasis on higher utilisation of power systems,

monitoring of its dynamics is becoming increasingly important. This requires

information with higher accuracy and update rates that are faster than those usually

provided by traditional protection systems. The introduction of phasor measurement

units as well as advances in communication and computational equipments has made it

technically feasible to monitor the stability of the power systems online, using a wide

area perspective. Power utilities have already deployed PMUs in their grids, mainly for

manual data acquisition and processing [106]. Wide area monitoring systems provide

central data acquisition from already installed and planned PMUs enabling utilities to

utilise phasor information wherever it is needed. The three main goals of WAMS are:

monitoring of the dynamic system behaviour (i.e. stability assessment); monitoring of

transmission corridors (i.e. congestion management); and finally, disturbance analysis

and system extension planning (i.e post-mortem analysis). WAMS includes all types of

measurements that can be useful for system analysis over the wide-area of an

interconnected system. Real-time performance is not required for this type of

application, but is no disadvantage. The main elements are time tags with enough

precision to unambiguously correlate data from multiple sources and the ability to

convert all data to a common format. Accuracy and timely access to data is important as

well. Certainly with its system-wide scope and precise time tags, phasor measurements

are a prime candidate for WAMS.

6.3.1 Advanced technology used in WAMS

As mentioned in the previous section, monitoring, operation and control are the three

major parts of energy management systems and wide-area measurements are the

integral part of power networks which will help in ensuring operation of transmission

networks within their operational limits in the present climate of deregulation.

Emerging techniques in computer technology, communication technology and PMU

technology are being used in WAMS and form the basis for real-time dynamic

monitoring, online security assessment and wide-area stability control of power

systems. In all, WAMS is playing a vital role in interconnected power systems [107] by

providing a wide area system view and increased stability [108].

Page 117: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

98

In recent years, many attempts have been made to design an integrated controller and

with the wide application of synchronised phasor measurement units in power systems,

WAMS has enabled the use of a combination of measured signals from remote locations

for global control purpose. It is found that if remote signals from one or more distant

locations of the power system are applied to local controller design, system dynamic

performance can be enhanced [109].

Phasor measurement unit (PMU) technology is attractive since it can provide

synchronised, real-time measurements of voltage, incident current phasors, rotor angle

and electrical power at the system buses and lines. Once these real-time signals of the

whole system are available in the form of synchronised phasors, the operators can give

online monitoring of power system operating conditions. Moreover, PMU is also

introduced for the use of stability assessment and wide-area control of the power

system. Due to the large scale and intricate structure of modern power systems, the

demand for structuring WAMS has been increasing. WAMS provides a dynamic

coverage of the wide-area power network and is also able to handle cascaded outage

through coordinated and optimised stabilising actions.

Authors in [110] have proposed a two-level hierarchical structure to optimise the

control of transient swings in multimachine power systems. The control technique

involves a number of independent local controllers communicating with a central

coordinating controller which accounts for nonlinearities and yield global optimal

transient performance. On a similar line, to address voltage regulation and rotor

oscillation problems simultaneously, [111] has proposed a two-level hierarchical

controller based on wide-area measurement for multimachine power systems. The

solution given consists of a local controller for each generator at first level helped by a

multivariable central one at secondary level. The secondary-level controller uses remote

signals from all generators and improves the local controller performance.

Information is now obtained fast and fresh from the synchronised measurement [46].

The power utilities are now placing these devices in selected locations to measure the

voltage and current phasors at the same time. They are transmitted to a central place

where they are compared, analysed and processed. The technology of synchronised

Page 118: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

99

phasor measurements is well established. It provides an ideal measurement system to

monitor and control power systems, in particular during conditions of stress. The

essential feature of the technique is that it measures positive sequence (and negative

sequence and zero sequence quantities, if needed), voltages and currents of a power

system in real time with precise time synchronisation. This allows accurate comparison

of measurements over widely separated locations as well as potential real-time

measurements based control actions. The synchronisation is achieved through a global

positioning satellite (GPS) system as shown in Fig.6.1.

Starting from WAMS and moving to wide area control systems i.e. WACS is the

challenge of the new century. Authors in [112] have explained about how advances in

digital and optical communication and computation can be exploited to gain the specific

advantage of WACS.

Communication systems are a vital component of a wide area protection system. These

systems distribute and manage the information needed for operation of the wide-area

relay and control system [41]. To meet these difficult requirements, the communications

network will need to be designed for fast, robust and reliable operation.

The introduction of PMU technology can significantly improve the observability of the

power system dynamics, and it can enhance different kinds of wide-area protection and

control [46]. The control actions can either be preventive, or corrective. During normal

operating condition the focus is always on the economics of the system but during

cascading condition the focus is on control shifts towards ensuring power system

security. The objective in that case is to keep, as much as possible, an intact electrical

network with all generators connected to the grid. The energy management systems

EMS applications can improve the security margin using optimisation techniques or

sensitivity routines that, along with time domain simulation, could predict the control

actions to return the system to normal.

6.3.2 Role of communication network and various time delays

Though the above scenario looks very appealing in terms of fast communication, in

reality, there exist various time delays (lags) in power system measurement. A major

Page 119: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

100

component of system-wide disturbance protection is the ability to receive system-wide

information and commands via the data communication system and to send selected

local information to the SCADA centre. This information should reflect the prevailing

state of the power system [41].

Communication delay plays a crucial role in WAMS by determining the time-lag before

control action is initiated to dampen power system oscillations. As mentioned in [108],

the PMUs measure voltage, current, and frequency phasors using discrete Fourier

transform (DFT) and can detect transients or surges within milliseconds of their

occurrence. In addition to the propagation delay of the particular link, the message

format of the PMU and data rate of the link determine the communication delay in the

system. Furthermore, there is also a processing delay due primarily to the window size

of the DFT. The standard delays calculation associated with various communication

links is given in [108]. It indicates the delays of various communication links when

using PMUs in a WAMS environment and could provide useful delay statics that can be

integrated into simulation and performance analysis of WAMS.

Normally it is assumed that the time frame of the disturbances is shorter than the

response time of the human operators of the power system (less than several minutes).

Thus the scenarios address systems for automatic protection or control as opposed to

manual control. A complete wide-area protection and control system would have the

capability to not only detect incipient disturbance, but also to respond in real time with

effective control action.

The major options for communication used in WAMS are both wireless (micro-wave,

satellites) as well as wired (telephone lines, fibre-optics, power lines) network options

[107]. The basic flow of data and information starts from measurements with the help of

various sensors. These measurements are then sent to signal distributors to send on to

the equipment which use these measurements. This signal is then digitised by signal

converting equipment and then communicated. Traditional short circuit protection

systems measure local signals and respond in four to 40 miliseconds to disturbances in

the local area [104]. Wide-area protection and control systems would gather information

from multiple locations on the system and issue wide area controls as deemed necessary

to respond to disturbances in a somewhat longer time frame.

Page 120: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

101

The main causes for delay in communication are as mentioned in [108], transducer

delays; delays because of window size of the DFT; processing times; data size of the

PMU output; multiplexing and transition; communication links involved; and data

concentrators which are primarily data collecting centres located at the central

processing unit and are responsible for collecting all the PMU data that is transmitted

over the communication link.

The propagation delay is dependent on the medium characters and the physical route

distance in WAMS. For a local controller, the time delay of feedback signals is very

small (less than 10ms), so the small time delay is often ignored in the controller design.

However, for a wide-area controller, the time delay in an interconnected power system

can vary from ten to several hundred milliseconds or more. An experimental research in

[108] has shown that the time delay caused by different communication links are

different, but all of the delays are more than 100ms. In the case of a satellite link, the

propagation delay could be as high as above 700ms. There could be larger delays when

a large number of signals are to be routed and signals from different areas are waiting

for synchronising. Such large time delays can invalidate many controllers that work

well in with no signal delayed input and even cause disastrous accidents [107]. The

impact of time delay on controller performance has been ignored for a long time in

power systems, but it has significant effect in wide-area control.

Today’s wide area communication topologies such as the synchronous optical network

SONET) are capable of delivering messages from one area of a power system to

multiple nodes on the system in as just 6ms [104]. Assuming decision time of 50ms, a

disturbance on a system could be detected and a corrective response delivered in less

than 200ms. In all, communication systems are a key component of wide area protection

systems. These systems distribute and manage the remote information needed for

operation of the wide area protection and control systems. With rapid advancements in

WAMS technology, the transmission of measured signal to a remote control centre has

become relatively simple.

Page 121: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

102

6.4 REVIEW OF POWER SYSTEM CONTROLLERS

6.4.1 Controllers for power system performance enhancement

For effective compensation, it is necessary to find the optimal locations for using

FACTS devices depending on the targeted parameter such as voltage stability or

damping oscillations. The given location may be very effective for one type of stability

problem while it may deteriorate the other performance parameters. The authors in

[113] have given a brief literature review of various methods used for finding the best

locations for effective shunt and series compensations for various issues, and have

proposed a method to identify effective locations based on sensitivity analysis for

voltage stability enhancement using series compensation. The authors in [114] have

reported an eigenvalue sensitivity approach for location and controller design of series

compensation for damping power system oscillations.

To tackle with the problem of power system oscillations, Korba in [115] has developed

a model-based approach for monitoring of dominant electromechanical oscillations in

real time. With a similar concept, the authors in [116] have addressed online estimation

of electromechanical oscillatory modes in power systems using dynamic data like

currents, voltages and angle differences measured online across transmission lines.

Authors in [117] have discussed and compared control techniques for damping

undesired inter-area oscillations in power systems, by means of PSS, SVC and

STATCOM. The study on different controllers, their locations and use of various

control signals for effective damping of these oscillations is explained. When multiple

controllers are used in large system, it is necessary to coordinate them properly to

enhance performance. Poor coordination may result in degrading the system

performance. Based on coordination or interactions of multiple FACTS devices, the

authors in [118], have shown interactions between various stabilisers (PSS and/or

FACTS) in multimachine power systems. The issue of coordinated design and

performance of multiple FACTS devices is reported in [119, 120] for power system

oscillation damping, while [121, 122] have used TCSC for damping these inter-area

oscillations. The oscillation problem is analysed from the point of view of Hopf

bifurcations, an extended eigenvalue analysis to study different controllers, their

Page 122: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

103

locations and the various control signals for the effective damping of these oscillations.

An application of a normalised H-infinity loop shaping technique for design and

simplification of damping controllers in the linear matrix inequalities framework is

illustrated in [123]. Authors in [124] have investigated the use of multiple input signals

for the design of PSS and TCSC controllers for damping power system oscillations. In

[34], a robust damping control is designed for multiple swing modes damping in a

typical power system model using global stabilising signals. A multiple-input-single-

output (MISO) controller is designed for a TCSC to improve the damping of inter-area

nodes.

Using the linearised model of power system, [125] has investigated the enhancement of

damping the power system oscillations via coordinated design of PSS and STATCOM

controllers using an SMIB system. The FACTS devices’ actions may produce additional

synchronisation and/or damping torque improving system capability to absorb

disturbance impacts in the sense of enhancing the capacity of restoring the system

equilibrium and/or making faster the attenuation of the oscillation. To verify this fact, in

[126] the problem of power system stability including effects of TCSC and SVC on

synchronising torque and damping torque are analysed by means of the properties of the

potential part of the transient energy as well as with Lyapunov function time

derivatives. A Lyapunov function was derived for SMIB by a model and it was claimed

that, though some approximations were made by linearisation, the more important

nonlinearities were preserved.

6.4.2 Controllers for transient stability improvement

FACTS controllers like TCSC can be placed in one of the lines of a power system with

a suitable control scheme to improve the transient stability condition of the system

[127]. In [128], it has been shown that the energy of the controlled line can be used to

devise a discrete control scheme for TCSC, with minimum measurements.

In [96, 129] evaluation of transient stability margin is discussed with the help of

multiple FACTS devices, using trajectory sensitivity analysis. To get the information

about each generator, trajectory sensitivity and its corresponding indices are used. It was

shown that the best possible location of the FACTS devices is totally dependent on the

Page 123: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

104

location of fault. Also STATCOM proved to provide great assistance in the restoration

of the post-fault voltages at different load buses in comparison with the TCSC.

However, this study was carried out using a classical simplified model which may not

be able to reflect true system dynamics.

Athay and authors in [130] have described the development and evaluation of an

analytical method for direct determination of transient stability. It has been claimed that

the study has developed a practical approach, sufficiently accurate and applicable to

realistic problems in power system operation and planning. [131] has given a practical

concern to both transient stability requirements as well as voltage regulation. With these

being the need for two different model requirements, a global controller is proposed to

coordinate the transient stabiliser and voltage regulator. However, the power system is

modelled with a simplified low order model and used for single-machine-infinite-bus

system.

A new fast method for assessing transient stability is proposed in [132] based on the

relationship of transient stability power limits, post-fault static stability power limits and

power impact caused by accelerating power of the failure process. Furthermore, the

method claims that transient stability margins can be assessed efficiently but the power

system model used in the method is a simplified classical model. The authors in [133]

have implemented the transient stability assessment in real-time and investigated the

effectiveness of various available methods as an important tool for energy management.

Various methods have been developed for system assessment considering the

importance of transient stability issues and problems thereafter. However, while, these

tools are capable of capturing the system responses quite accurately, they do not

inherently provide information regarding the degree of stability [134]. In the paper

‘keeping an eye on power system dynamics’ [135], the authors have discussed present

power system scenarios, given some measurements and analysis of events and have

highlighted post-disturbance monitor issues. It gives importance to online

measurements not only in maintaining system stability under disturbance conditions, but

also for post-mortem analysis to prevent (or minimise severity of) the same problems in

future.

Page 124: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

105

Actually, each of the FACTS devices controls one or more of the network variables

with criteria based on specific control schemes as mentioned in [17, 23, 113]. However,

compared to other FACTS devices such as shunt compensations by SVC or

STATCOM, series compensation by TCSC has proved to be more effective for transient

stability enhancement. Knowing the advantages of TCSC compensation for

performance enhancement in post-fault condition, the focus of the present research is

basically on improvement of transient stability using TCSC. The construction,

modelling and working of TCSC and other FACTS devices is already discussed in the

Chapter 3, so next the section of this chapter will discuss the performance of TCSC and

its applications, highlighting its features.

6.4.3 Review of real time controllers

The introduction of time delay in a feedback loop has a destabilising effect and reduces

the effectiveness of control system damping. In some cases, the system synchronism

may be lost [136] so in order to satisfy the performance specifications for wide-area

control systems, the design of a controller should take into account delay. Moreover, the

controller should tolerate not only the range of operating conditions desired but also the

uncertainty in delay. The impact of time delay on robust controller designs has been

ignored in power systems, but becomes a pertinent topic in recent years with the

proposal of wide-area power system controls.

The advent of real-time phasor measurements and improved communication makes it

attractive to consider new solutions to transient stability. The communication delays

including both propagation and processing delays seem to be approaching values that

are acceptable when considering a transient swing with a period of approximately one

second.

The modern energy system management is supported by SCADA, by numerous power

system analysis tools such as state estimation, power flow, optimal power flow, security

analysis, or transient stability analysis etc in addition to linear and nonlinear

optimisation programs. The available time for running these application programs is the

limiting factor in applying these tools in real time during emergency, and a tradeoff with

accuracy is required. The real time optimisation software and security assessment and

Page 125: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

106

enhancement software do not include dynamics. Further propagation of a major

disturbance is difficult to incorporate into suitable numerical algorithms, and a heuristic

procedure may be required.

With the increased availability of sophisticated computer, communication and

measurement technologies, more ‘intelligent’ equipment can be used at the local level to

improve the overall emergency response. There seems to be a great potential for wide

area protection and control systems, based on powerful, flexible and reliable system

protection terminals, high speed communication, and GPS synchronisation in

conjunction with careful and skilled engineering by power system analysts and

protection engineers in co-operation [40].

The real-time measurement equipments and associated communication system (i.e.

PMUs and WAMS) can be exploited for developing advanced control techniques and

centralises response-based control architectures. The main idea is that power system

trajectories, acquired in real-time, allow the identification of threats to system security

and degraded dynamic states. If necessary, the control centre can evaluate suitable

corrective control actions and successively transmit the corrective signals to the

actuators. Remedial actions can then be applied through any fast actuator devices such

as FACTS devices [53].

As mentioned in [52] the essential but challenging step in a response-based type

controller is to know the real-time system status using prediction. Online simulations

should be executed continuously to reflect the most current operating condition. A

further, important task is using this system information, deriving the optimal control

action to make the system stable in a short time interval. Counter measures and new

approaches to system security can be based on the adoption of FACTS and HVDC

technologies, dynamic security assessment methodologies, PMU technologies, real-time

measurements and control systems with wide-area measurement systems and wide-area

control systems, automation and control methodology [53].

Power system security can be quantitatively enhanced by developing an online

environment where all control features are implemented on the transient time-scale.

Though Chu and Liu have reported online learning applied to power system transients

Page 126: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

107

in [52], the attempt was to bring the system back to pre-fault conditions even after

disturbance in post-fault regions. The degree of stability measured by this, the

performance index was hence defined by the deviation of system performance in post-

fault, from its pre-fault values. In reality, there are chances that after disturbance the

system topology is changed due to the loss of one or more transmission networks. In

such cases, giving importance in maintaining system synchronism for keeping

frequency constant, the system operating conditions can be changed with the changed

transmission network. In short, it may not be feasible to bring a system back to the same

pre-fault operating conditions after a disturbance in all given circumstances.

6.4.4 Communication delay consideration in controller design

Traditional stability controlling strategies in power systems only used the local

measuring data. Time delays of the local measuring data were usually very small (<10

ms) [37], and were generally ignored in the past stability study and controller design.

With the rapid development of PMU/WAMS, coordinated stability control strategy has

been paid more and more attention. It uses the remote measuring information from

PMU/WAMS. Since time delay in wide area measurements is usually obvious, it is

important to properly consider its impact on the stability analysis and controller design

in power systems.

Employing phase-measurement units (PMUs) it is possible to deliver the signal at a

speed of as high as 30Hz sampling rate [109]. It is possible to deploy the PMUs at

strategic locations of the grid and obtain a coherent picture of the entire network in real-

time. As mentioned in [137], wide-area can be 10-12 times more effective than local

decentralised control of wide-area oscillations. However, the cost and associated

complexities restrict the use of such sophisticated signal-transmission hardware on a

large scale commercial scale. As a more viable alternative, the existing communication

channels are often used to transmit signals from remote locations. The major problem is

the delay involved between the instant of measurements and that of the signal being

available to the controller. The delay can typically be in the range of a few hundred

milliseconds depending on the distance, protocol of transmission and several other

factors. To consider this effect of time-delay on system stability, [137] has adopted a

predictor approach and discussed its implementation and experimental verification

Page 127: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

108

using FACTS devices. However, the focus of this time-delay was to investigate its

impact on small-signal stability of power system.

In distributed systems such as protective relay systems, the time delay or latency is

usually less than 10ms [138]. Unlike the small time delay in local control, in wide-area

power systems the time delay can vary from tens to several hundred milliseconds or

more. Fibre optic digital cables are reported to have approximately 38ms for one way,

while considering delays it is over 80ms. Communication systems that entail satellites

may have an even longer delay. The delay of a signal feedback in a wide-area power

system is usually considered to be in the order of 100ms. If routing delay is included,

and if a large number of signals are to be routed, there is a potential of experiencing

long delays and variability.

In [107] the authors have designed a TCSC controller including feedback signal delay,

using a theoretical approach based on time-delay dynamic systems combined with LMI

techniques. The design is discussed and implemented for showing effect of time-delay

and its performance. The analysis of the time delay impact on wide-area system control

is addressed in [138]. In order to eliminate the effect of a power system model’s

nonlinearity and wide-area information’s uncertainty including time delays and

incompleteness, [139] has proposed a nonlinear robust integrated controller.

Optimally tuned conventional controllers, using linear time-invariant (LTI) lead-lag, are

simpler and often provide effective solutions to improve the damping of selected

oscillatory modes. However, they only work within a limited operating range and, in the

case of a changed system configuration with new operating points, can still cause poorly

damped or unstable oscillations due to parameters being tuned to a previous different

setting. To make these types of controllers adaptive for real-time implementation, an

adaptive controller for FACTS has been proposed in [97]. However, while making its

operating range wider than any LTI controllers, it has resulted in high computational

efforts on hardware and large computational time.

The primitive power system controllers such as excitation controllers have been well-

developed and applied to power plants widely. With the use of TCSC equipments, the

applicability of the primitive controllers and their original control strategies under the

Page 128: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

109

least cost performance index is a considerable problem. The coordination among the

TCSC controller and other controllers is also important. As mentioned in [140], [42]

have given some results about the interactions between TCSC and other controllers.

However, those conclusions are mainly obtained when all the controllers are equipped

with conventional control approaches, such as PID or linear control. When nonlinear

control approaches are employed, the coordination problem should be considered in

more depth.

In fact, multi-machine power systems equipped with TCSCs are characterised by high

nonlinearity and strong coupling, and influenced by exogenous disturbances such as

change of operating points, power system faults, etc. These disturbances have great

impacts on the control design. Although many control design approaches have been

developed for FACTS devices to enhance the power system transient stability, most of

the existing controls are based on the approximately linearised model of the power

system and conventional control principles that are not suitable for the cases with large

disturbances. Though there are many references in literature where TCSC nonlinear

control law is derived, it is based on fixed structure and parameters without considering

the system uncertainty such as loss of transmission line which changes network

topology.

In [140] the authors have developed coordinated nonlinear robust control of TCSC and

excitation for multi-machine systems. Although all the above problems of nonlinear

models, changed system structures, etc. are taken into consideration, the power system

model used to represent multi-machine network is a simplified one.

In the past, studies on time delays in power systems mainly focused on evaluating the

time delay impact on the controller design [37, 107, 109, 138, 141]. In [142] the

influence of time delays on small signal stability regions is investigated and shows that

time delays in power systems can bring both negative and positive influence to the

system’s small signal stability. However, little work has been done to directly analyse

the impact of the time delay on power system stability, especially on transient stability.

In [110] an optimal two level structure for the transient stability problem in

multimachine power systems is proposed. The solution involves a number of

Page 129: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

110

independent local controllers communicating with a central coordinating controller,

which optimises a global cost function.

At the most basic level of response measurement is the problem of predicting whether

an on-going transient swing is stable or unstable. This is essentially a problem of out-of-

step relaying where there is no control perse but rather a decision to block relay tripping

if the swing is stable and to effect a pre-planned separation if the swing is unstable. This

is a difficult task only with local measurements. In some special situations new

approaches are possible with real-time phasor measurements. If the instability is

modelled as a two-machine system, then the parameters of the two-machine model can

be inferred from the number of measurements in the real world. The equal area-criteria

can be then used to predict whether the swing is stable or not. Unfortunately, when

more than two machine models are required this simple criterion is not useful. Most of

the familiar techniques for investigating transient stability were developed for offline

planning studies and have no direct applicability to real-time problems. The various

energy function techniques can be viewed as a means of avoiding the solution of a large

number of differential equations. In real time, however, nature provides a solution to the

differential equations and the decision about stability or instability must be made on the

basis of measured system variables [104].

Othman et al. [143], proposed a model in d-q axis domain assuming that the line current

is a forcing function to TCSC and therefore an independent quantity. This was a major

drawback of this model, as a change in the TCSC firing angle will cause a change in

line current and hence, it is unreliable in terms of accuracy. The conventional

controllers used for TCSC control utilises a TCSC reactance to firing angle (i.e. Xtcsc

α conversion table) which is generated offline for a given TCSC. This table depends

on the characteristics of devices, so to address all these limitations, in [144], a linearised

discrete-time model of a TCSC-compensated transmission line is presented. The model

derived considers the proper characteristics of TCSC, i.e. the variation in its impedance,

with the firing angler of the TCR. Through the digital computer simulation it is shown

that the eigenvalues of a TCSC-compensated transmission line has two complex

conjugate pairs of poles whose real parts depend only on the line resistance and

reactance. The model thus derived in the d-q axis frames is then linearised around the

nominal operating point and is shown to predict any disturbance very accurately. Three

Page 130: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

111

different controllers were designed based on this linearised model, which requires the

measurements of local variables only. Two of these controllers were used for regulation,

and were able to reject any major or minor disturbance, whereas the third one was used

for tracking any changes on the power set point.

A lot has been learnt from the offline applications experience, as the first step towards

online applications [46]. Online dynamic analysis could be conducted based on the

most recent wide-area system information PMU data, analogs, statuses and topology

structures. The security of the system could anticipate failure of more than one critical

component and simulations will be available to prevent actions or correct situations.

The interface with short-term simulations will provide a list of devices whose behaviour

could drive the system to instability given the current conditions. The secure region is

constrained by specific limits such as frequency stability and transient stability limits,

whose assessment can be done by full time-domain simulations and approximate

methods [46]. Hiskens, in [145], has established a deterministic nonlinear time-delay

model and incorporated it into systematic hybrid (continuous/discrete) system

representation. It has been shown that time-delay affects the differential-algebraic

model.

A major component of system-wide disturbance protection is the ability to receive

system-wide information and commands via the data communication system and to

send selected local information to the SCADA centre. This information should reflect

the prevailing state of the power system [41].

6.4.5 Computation requirements in controller design

For wide spread geographical power system networks for system monitoring and

operation, factors such as communication, collection of data, interaction and exchange

of information in time bound limits, put lot of pressure on available communication

networks and computation facilities. In addition to communication delays, computation

delays are also playing an important role. Actually, it is not only computational delay,

but even computation burden is one of the major concerns, as the size of power system

network increases. Based on whatever simulation-based methods are used in literature,

their limitations and problems in terms of computation complexity handling are

critically reviewed in [146].

Page 131: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

112

With the deregulation and constant expansion in power systems, the demand of high

performance computing for power system adequacy and security analysis has been

increased rapidly. High performance plays an important role in ensuring efficient and

reliable communication for power systems operation and control. As explained in [147],

grid computing technology is an infrastructure, which can provide high performance

computation (HPC) and communication mechanism for providing services in these

areas of power systems. In [147], the authors have presented a review on the current

research that has been done in the adoptability of grid computing technology in power

system analysis, operation and trading. It has explained the problems with parallel

processing where the task is divided into a number of subtasks of equal size and allotted

to different workers. For this purpose, it has been pointed out that all machines need to

be dedicated with the same configuration and processing speed, otherwise thus can face

the problem of time difference in giving output results from each one of them. Based on

the various experiments carried out in Western Australian Supercomputer program

(WASP) at The University of Western Australia (UWA), it was observed that, the

proportion of computing speed to size of system do not vary linearly. For very small

systems, parallel processing may not be efficient, while as the size of the system and

amount of computation required increases, the speed up factor of computation time also

increases. However, after a certain threshold, again this may not be efficient as there

may be lot of time consumed in distributing work to different processors and also in

collecting results and compiling it for final output. The overall computing time

requirement is affected by the actual processing speed and time for data transfer

depending on the amount of data transfer and channels available for this transfer. The

parallel processing techniques involve tight coupling of machines and using a

supercomputer is possible only for justifiable size of power system network and

depends on its importance. To address problems of computation, [147] has suggested

grid computing technology by listing various interesting features of grid computing in

terms of sharing resources, for example, taking advantage of time difference of zones

operating on a grid.

The ideal scenario for control of a power system would be to have the capability to

instantly compute optimum operating conditions and keep the system at those operating

conditions using the available controls. This will require knowledge of system topology

Page 132: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

113

information and the system’s real and reactive load. In this context, instantly or real-

time means fast enough to provide a required bandwidth. In this scenario, considering a

model-based control algorithm, upon occurrence of a system contingency, the change in

system topology will be instantly detected or estimated at a central location. Then as

repeatedly as necessary, the trajectory to the final operating point will be planned

instantly starting from the initial system state, taking into account all constraints , and

the controller will be driven to achieve tracking of the planned trajectory [1].

In order to satisfy specifications for wide-area control systems, the design of a controller

must take into account this delay, to provide a controller robust enough not only for the

range of operating conditions desired, but also for the uncertainty in delay. The impact

of time delay on robust controller designs has been ignored in power systems but has

become a pertinent topic in recent years with the proposal of wide area power system

control [37].

It is found that if a controller is designed for a delay-free system but applied to the

delayed-input system, the close-loop system may lose stability resulting in unacceptable

performance. The aim of a generalised predictive algorithm developed in this research is

to judge the transient stability status of a power system after fault occurrence. The time

interval for prediction should be very short so that it is made in advance before losing

synchronism, leaving enough time for remedial action to be effective. With reference to

above background, the prediction algorithm developed considers the computation time

required for prediction and optimisation to find necessary TCSC reactance. In addition,

it also considers the further communication delay in implementing the corrective

actions, in deciding the effectiveness of controller.

6.5 CONCLUSION

The chapter has traced through the evolution of, and advances made in the field of wide-

area measurements and technology use for maintaining wide area network stability. The

comprehensive review has identified two key issues which require further research and

development. The first is that, most of the time, communication delays are ignored in

control law derivations, or if it is included at all at any time, the system used is SMIB

with a classical model. Such approximate or simplified models or control laws without

Page 133: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

114

consideration of practical issue of communication time delay may not give a correct

picture of system stability. The second key issue which is ignored in previous research

is requirements of computation systems and computation time delays. For development

of real-time controls it is customary to consider these communication and computation

delays and also check for their feasibility as per real-time requirements.

The evolving power system with dispersed generation, power electronics devices

connecting low and medium voltage levels, which are geographically at a far distance,

will cause widening of the interconnected system making it more complex. The new

research should be focused on making this interconnected system safe enough for any

contingency which is possible if propagation of the loss of stability is taken in to

account. Keeping the central theme of tracking in advance the generators which are

expected to fall out of synchronism because of sever disturbances, and system

performance for given emergency situation and control action, a new online controller

coordination scheme is proposed in the next chapter to enhance the transient stability of

an interconnected power system using series compensation.

The total online control coordinated scheme is given in the next two chapters. Chapter 7

will develop the necessary theory for this control coordination scheme, while Chapter 8

will verify the various time requirements for feasibility of a real time controller.

Page 134: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

115

Chapter 7 Online Control Coordination of

FACTS Devices for Power System

Transient Stability: Control Method

Derivation

7.1 INTRODUCTION

As discussed in the previous chapter with its detailed review of real-time controller

requirements, strong robust communication architecture is essential to meet the features

of accurate detection, decision and reaction times for any system. In the context of real-

time controllers, a new method is developed for real-time transient stability control in a

power system which has FACTS (flexible AC transmission systems) devices. Central to

the method is the control in successive time periods of synchronous generator relative

rotor angles to satisfy the nominated transient stability criterion via the real-time and

optimal adjustment of the input references of the controller FACTS devices. In each

period, the dependencies of maximum relative rotor angles on input references are

expressed as nonlinear functions which are synthesised from the results of time-domain

Page 135: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

116

transient stability simulations, using the prevailing power system model and condition.

The constrained optimisation problem is then formed from the synthesised functions,

and solved for the optimal input references. Practical issues related to computing time

and communication channel time delays are considered in the control methodology. The

next chapter will examine the computing time requirement, and present the results of the

control coordination study to verify the effectiveness of the proposed control method.

7.2 BACKGROUND OF PROPOSED SCHEME

Transient stability is one of the principal considerations in power systems planning,

design and operation. Following the restructuring and deregulation of the power supply

industry, it is generally acknowledged that the power system transient stability margin is

being reduced. This is a consequence of transmission companies increasing their

competitiveness in market environments. Much research has arisen in addressing this

critical issue [13, 97, 148, 149], drawing on the availability of FACTS (flexible AC

transmission systems) device controllers, and advancements in communications,

measurements and computing. In the area of small-disturbance stability, recent research

[97,137, 149] has applied WAMs based on phasor-measurements units (PMUs), and

self-tuning/adaptive controllers, to coordinate and enhance the dynamic performance of

FACTS device controllers.

There has been very limited development in the area of real-time control for transient

stability enhancement and for large disturbance follow-up maintenance . This is despite

extensive research [94, 98, 150-152] in small-disturbance stability enhancement and

damping of inter-area modes of oscillation. Further where these measures are

formulated and implemented with postulated fault disturbances at time of load dispatch

have not resulted in a workable solution. The key difficulties encountered in the

development of real-time control schemes are those of nonlinearity and high dimension

inherent in power systems, and computing time required for executing the control

algorithms. Notwithstanding the difficulties, there have been some proposals based on

model predictive control (MPC) and extended equal area criterion (EEAC) techniques

[13, 153] for real-time transient stability control. However, the use of simplified

classical generator models only gives approximations and these are limitations in the

application EEAC in MPC as identified in [13]. A number of assumptions have been

Page 136: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

117

adopted for developing an MPC scheme with cutset EEAC stability measures, which

can lead to inaccuracies near equilibrium conditions and possible compromise in the

control robustness [13].

Against the above background, the objective of this research is to develop a new

control scheme for optimally coordinating the input references to FACTS device main

controllers. This will be implemented in real time, in order to maximise the system

transient stability margin, and maintain transient stability subsequent to a large

disturbance. A key feature of this control scheme is online synthesis with individual

control periods for a set of nonlinear functions that express a generator’s maximum

relative rotor angles in terms of FACTS device controllers input references. These are

referred to as the control variables. As it is difficult, if not impossible, to determine

analytically, in a closed form, these nonlinear functions, a series of time-domain

transient stability simulations are performed to provide the results or data for the

function syntheses. The synthesised functions are then used as system transient stability

indices for deriving the control coordination algorithm. The overall control coordination

is subdivided into a number of control periods. Optimal values of FACTS device

controllers are determined in advance for each control period. These are then

implemented using a constrained optimisation method which minimises the input

reference of the rotor angle that is relatively largest. This is subject to the constraint of

individual rotor angles that are less than the nominated threshold and that control

variables must stay within their bounds. The constrained optimisation problem is

formulated in terms of the synthesised nonlinear functions.

In the present work, nonlinear functions in the form of polynomials are adopted to

represent the nonlinear relationships between relative rotor angles and control variables.

Nonlinear regression analysis is performed, following transient stability simulations

with various values of the control variables (i.e. perturbations of the control variables)

within their ranges, to determine the coefficients in the polynomials. As time-domain

simulations are used, it is straightforward to include detailed dynamic models for

generators, controllers and dynamic loads. The starting values of all of the power

system variables for the transient stability simulations related to each control period are

obtained directly from wide-area measurements, and/or derived from them based on

models for individual items of the generation plant.

Page 137: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

118

The power system model for simulation is directly constructed from the circuit-breaker

and isolator status data together with generator and controller dynamic models, and the

parameters for individual items of plants are held in the database. There are no

requirements for system model identification. Any control actions are included in the

system model in a straightforward manner, and their effects are represented directly in

the time-domain response evaluations from which transient stability indices, expressed

in terms of maximum relative rotor angles, are formed.

As the control coordination method is based on nonlinear time-domain simulation and

prevailing power system configuration and operating conditions, the following

advantages are achieved:

(a) Inclusion of nonlinear and detailed dynamic models for items of the plant

(b) Being adaptive to variation in power system conditions

(c) High accuracy and robustness in the control

The control coordination process is initiated by circuit-breaker opening operations.

These events include those of faults, and subsequent fault clearance by protection

systems.

The development of the algorithm for the control coordination takes account of practical

issues related to computing time and communications channel time delay. The

feasibility in terms of computing time requirements for implementing the control

coordination proposed using a cluster of high-performance and low-cost processors, will

be investigated in the next chapter.

7.3 TRANSIENT STABILITY CONTROL PRINCIPLE

7.3.1 Power system model used

The starting point of the proposed transient stability control is to construct the model to

represent the power system in the prevailing operating conditions. The control sequence

is initiated by circuit-breaker opening operations. If there have been short-circuit faults

Page 138: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

119

in the power system, these circuit-breaker operations would be those arising from fault

clearance, subsequent to which, the faults are isolated from the system. On this basis,

the prevailing power system configuration is established directly from the status data of

circuit breakers and isolators received at the control centre. The power system model

relevant to the transient stability phenomena is then constructed from the system

configuration, detailed models adopted for individual items of plant and their

parameters held in the database. The details of forming dynamic models of various

power system components is explained in Chapter 2, which is expressed in symbolic

notations by the following sets of differential equations and algebraic equations:

), uyh(x,x =•

(7.1)

( ) 0uy,x,g = (7.2)

In (7.1) and (7.22), h and g are, in general, nonlinear vector functions; x is the vector of

state variables arising from modelling synchronous generators together with their

controllers, FACTS devices and dynamic loads; y is the vector of network nodal voltage

variables, and u is the vector of control variables to be optimised in real time for

maximising transient stability margins. In the present work, the control variables

represent the input references to the main controllers of FACTS devices which have

been selected for participating in the control.

In terms of representing system loads, composite load models based on admittances

and/or aggregate dynamic load components are used. Their parameters are determined

on the basis of the data obtained from PMUs, pre-determined load compositions at

individual load buses and generic dynamic load models (i.e. individual motor

equivalents) and inertia data [6] .

7.3.2 Time-domain transient stability simulation

Starting from a given set of values for the elements of the state variable vector x at any

instant of time, the sets of equations in (7.1) and (7.2) can be solved simultaneously,

using numerical integration. This gives system responses for any specified future time

periods for control coordination subsequent to the starting time-instant. In principle, this

time-domain simulation process is a straightforward one if there are direct

Page 139: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

120

measurements of all of the state variables in vector x to be used as the initial condition.

In practice, these direct measurements might be difficult, to achieve. For example,

generator rotor angles, rotor flux variables, and state variables associated with

controllers’ models can be difficult, if not impossible, to measure. However, it would be

possible to obtain the values of the state variables through the available measurements

by synchronised PMUs of network variables such as nodal voltages, nodal currents,

nodal power, branch currents and power flows together with generator rotor speed

and/or field winding voltage measurements. These available measurements combined

with the individual dynamic models for synchronous generators and controllers would

give, through dynamic simulation in the time-domain, solutions to the state variables

associated with the models up until the current time instant. In this way, state variable

vectors would be available for initialising the simulation with the system model

described in (7.1) and (7.2) from any specified time instant (at present or in the past) to

any future time instant as required for control coordination purpose.

7.3.3 Nonlinear relationship between maximum relative rotor angles and control

variables

The key property on which the proposed control coordination draws is that, for any

given period of time subsequent to a disturbance, the relative rotor angle response of

any synchronous generator (except the reference generator) is a nonlinear function of

the control variables (i.e. the input references to the FACTS devices controllers), with a

given power system configuration and parameters. As the focus of the present work is

on transient stability control, individual maximum relative rotor angles, and their

dependence on the control variables are of interest and relevance to the development of

the control algorithm. In any specified time period, the dependence on the control

variables of the maximum (either positive-going or negative-going) relative rotor angle

of each generator is expressed in a compact form as follows, using a functional notation:

( )( ) ( )kiki TfT ,max1 u=δ Ni ,,3,2 = (7.3)

In (7.3), the reference generator is identified by 1; i denotes the ith generator; N is the

total number of generators; ( )( )ki Tmax1δ is the maximum relative rotor angle of the ith

Page 140: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

121

generator with respect to the reference generator in the time period Tk, and ( )ki Tf ,u is

the nonlinear function of the control vector u, for time period Tk. The kth control period,

Tk, is defined as that from a given time instant tk to the time instant tk+T where T is a

chosen control time window (CTW).

Except for simple or trivial cases, it is not possible to derive analytically in a closed

form the nonlinear function ( )ki Tf ,u referred to in (7.3), for a transient stability control

purpose. The present research proposes a scheme based on a number of time-domain

simulations related to individual control periods Tk to form functions ( )ki Tf ,u . In

Section 7.5 the scheme is developed, based on control vector perturbations and time-

domain simulation, by which fi is synthesised, and expressed in terms of a polynomial

in control variables in vector u.

7.3.4 Transient stability control concept

Subsequent to a disturbance (for example, fault and fault clearance), the control is

subdivided into a number of control periods to be considered sequentially. An optimal

control vector is to be determined in advance for subsequent implementation for each

control period with the objective that all of the relative rotor angle maximum values in

that period are to be within the nominated transient stability threshold, and, for

maximising the transient stability margin, the greatest relative rotor angle is to be

minimised. With this objective, and using the functional notation in (7.3), the

constrained optimisation problem is formulated as follows, for each control period:

Minimise ( )[ ]2

, kj Tf u Mk ,,2,1 = (7.4)

subject to:

( ) thkith Tf δδ ≤≤− ,u Ni ,,3,2 = (7.5)

and

( ) ( )maxmin mmm uuu ≤≤ Lm ,,2,1 = (7.6)

In (7.4), j is the identifier for the synchronous generator having the greatest relative

rotor angle among (N-1) relative rotor angles within the kth control period; M is the

number of control periods; δth is the nominated transient stability threshold (for

Page 141: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

122

example, 1800) in terms of relative rotor angle; um is the mth element of the control

vector u having upper and lower limits um(max) and um(min) respectively, and L is the

number of control variables. The objective function in (7.4) is formed as the square of

the greatest relative rotor angle, ( )kj Tf ,u , to include the case of its being negative.

The solution of the constrained optimisation problem in (7.4)–(7.6) for each control

period is the optimal values of the control variables (i.e. elements of vector u) which are

used for setting the input references to the participating controllers. In relation to the

frequency at which the controllers’ input references are updated, there are two options:

(a) Option 1: The controller input references are updated, using their optimal values, at

the start (referred to as tk) of the control period, Tk , and then kept constant until the end

of the control period, Tk, which is tk+T. The control sequence will then be repeated for

the next control period Tk+1.

(b) Option 2: Similar to option 1, the controller input references are updated at time

instant tk, using their optimal values. However, the input references, once updated, will

be kept constant only until time instant tk+Tx with Tx < T. The control sequence will

then be repeated for the next control period Tk+1. This implies that, although the

constrained optimisation problem described in (7.4)–(7.6) is solved with respect to the

control period Tk from tk to tk+T, the outcome of the optimisation is used for the period

from tk to tk+Tx (with Tx < T ) only. In this option, the frequency at which the controller

input references are updated is higher, in comparison with that in option1, and there are

overlaps among control periods.

With option 2, it is possible to take account of the change in power system conditions

more closely in the control. However, this has implications in terms of computing time

requirements. In the next section 7.4 are presented the detailed control schemes for both

options1 and 2, taking into account the computing time and communication channel

delay.

Page 142: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

123

7.4 TRANSIENT STABILITY CONTROL SCHEME

7.4.1 Option 1

Fig. 7.1 shows the control scheme timing diagrams for two successive control cycles

identified by 1 and 2. The timing diagram in Fig.7.1(a) is related to the first control

cycle, following the reception at the control centre of the information that circuit

breakers opening operations were completed at time tc and with reference to the time

origin at which t = 0 in Fig.7.1(a).

There are synchronised measurements of the power system operating states by PMUs at

a given sampling frequency. Time instant tx1 following tc as shown in Fig. 7.1(a) is that

which is closest to tc, and coincides with a time instant when synchronised

measurements are completed. Their results are sent to the control centre which will

receive them at time-instant ty1, after the communication channel time delay of TD1. As

it is not necessary that various measured results would arrive and be received

simultaneously at the control centre, TD1 is set to be the greatest time delay encountered

among the channels, including those for circuit breakers and isolators’ status data. The

calculation related to optimal control coordination for the first control cycle (i.e. k = 1)

is to be carried out, starting at time instant ty1, and completed within the specified time

interval, TC, as indicated in Fig.7.1(a). There are five key calculation steps for the first

control cycle:

(a) Updating the power system model;

(b) Time-domain transient stability simulation;

(c) Nonlinear function synthesis;

(d) Determining optimal values for the control variables; and

(e) Updating the FACTS device controller input references.

In detail, the steps are described as follows:

a) Updating the power system model. Based on the most recently available circuit-

breaker and isolator status data after the circuit-breaker opening operations which is

received at time ty1 the power system model is updated. Parameters and models of

Page 143: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

124

individual items of plant which are held in the database are used for power system

model construction as described in Section 7.3.1. Also included in this calculation step

is the assembling of the system nodal admittance matrix and its LU factorisation.

b) Time-domain transient stability simulations. This is for the purpose of

generating data for synthesis of nonlinear functions referred to in (7.3) of Section 7.3.3.

At each time step in the simulation, the sets of equations in (7.1) and (7.2) described in

Section 7.3.1 are solved. The set of differential equations in (7.1) is transformed into an

algebraic equation system via a numerical integration formula such as the trapezoidal

rule. This equation system is combined with the network equation set in (7.2) defined

by the nodal admittance matrix formed in step (a).

A series of simulations are performed in this step, using the model from step (a), and

measured initial condition at time tx1 which is received at time ty1. The solution time for

the first simulation is tf1 – tx1 where tf1 is the end of the first control cycle, as indicated in

Fig. 7.1(a).

The first simulation uses the existing values of the control variables (i.e. the input

references to the FACTS device controllers) for the whole time duration from tx1 to tf1.

ty2 tz2 t2 tf2

0T

T

tx2

Timet

(b)

TCTD1 TD2

(a) tx1 ty1 tz1 t1 tf1

0TD1 TD2TC T

Timet tc

Fig.7.1 Control scheme timing diagram: Option 1 (a) The first control cycle: 1=k (b) The second control cycle: 2=k

ty2 = ty1 + T; t2 = tf1

Page 144: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

125

Each subsequent simulation is carried out for the period from t1 to tf1 which are the start

and end of the first control period respectively, drawing on the initial condition at t1

obtained from the first simulation. For these individual simulations, the FACTS device

controllers input references are perturbed, within their limits, from the existing values.

There is no requirement for the perturbations to be small ones as in linear sensitivity

analysis. From the results of each and every simulation, including the first one, the

maximum value (either positive-going or negative-going) of each relative rotor angle

within t1 and tf1 (i.e. within the control period) is determined, and recorded for

subsequent use in nonlinear function synthesis. In the case of the relative rotor angle

having the maximum value at the start (i.e. t1) of the control period from t1 to tf1, the

value at the end of the period (i.e. tf1) instead of the maximum value is used. This makes

provision for response time delays inherent in the FACTS device controllers and power

systems. It is due to this that the relative rotor angles at the start of each control period

(for example, at t1 for the first control period) would not be influenced by any updated

or revised settings of the controllers input references implemented at the control period

starting time.

The number of simulations depends, in general, on the form of the nonlinear functions

adopted for representing the maximum relative rotor angles. This is specifically in terms

of FACTS device controllers input references, and the number of controllers

participating in the control. Once all of the required simulations are completed, the data

set of maximum relative rotor angles and FACTS device controllers input reference

values which have been used in the simulations from t1 to tf1, is available for the

subsequent syntheses of nonlinear functions expressing the relationships between

maximum relative rotor angles and control variables (i.e. the FACTS device controllers

input references).

c) Nonlinear function synthesis. Each maximum relative rotor angle within the first

control period T1 (period T1 starts at t1 and ends at tf1 = t1+T ) is to be represented by a

nonlinear function, ( )1,Tfi u , (for i = 2, 3, ... , N) as discussed in Section 7.3.3.

In this step, the data set obtained in step (b) is used for the function syntheses. This

involves the postulation of the functional form and determining the parameters or

coefficients in each function, using the data set. The outcome of this step is a set of

nonlinear functions, ( )1,Tfi u ’s, expressed explicitly in terms of the control variables in

Page 145: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

126

vector u. In Section 7.5 a numerical procedure for forming the nonlinear functions is

presented based on polynomials of control variables and nonlinear regression.

d) Determining optimal values for the control variables. With the nonlinear

functions ( )1,Tfi u , (for i = 2, 3, ... , N) formed in step (c), the constrained optimisation

problem described in (7.4)–(7.6) is formed completely for the first control period with k

= 1. The jth nonlinear function, ( )1,Tf j u , used in forming the objective function in (7.4)

is for the generator with the largest relative rotor angle in period T1, identified from the

results of the simulations performed in step (b).

Solution of the constrained optimisation problem gives the optimal values of the control

variables in vector u. Established methods such as the quasi-Newton method or

sequential quadratic programming [154, 155] can be applied for solving the problem.

As the typical number of FACTS device controllers in a power system is not large, the

dimension, as determined by the number of control variables, of the constrained

optimisation problem encountered in the optimal control coordination is relatively

small. The optimisation is a static one for each control cycle, within which there are no

time-dependent quantities or variables. Time-domain simulations are excluded from the

optimisation loop.

With a relatively small number of FACTS devices controllers, it is possible to solve the

optimisation problem described in (7.4)-(7.6) in Section 7.3.4 by calculating the

objective function and constraint functions for a finite set of specified values of control

vector u with high resolution, and based on the results of the calculation, optimal vector

u is determined. This search method is very suitable for implementation by parallel

computing systems as the function calculations for individual values of control vector

can be performed in parallel and independently of one another. This is an effective

method for exploiting the inherent parallelism in the calculations for solving the

constrained optimisation problem with low computing time, using a computer system

with parallel processing capability.

e) Updating the FACTS device controller input references. The calculations

described in steps (a)–(d) are completed by tz1 (i.e. within the allowed computational

time duration TC). At time instant tz1, the optimal values of the input references,

Page 146: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

127

determined in step (d), are sent from the control centre to individual FACTS device

controllers. After a time delay TD2 of the communication channels, taking into account

the longest delay encountered in them, the controller input references are updated at

time t1, the start of the first control period. In option 1, the controller input references

are kept constant, following the updating until tf1, the end of the first control period.

Following the completion of the calculations at tz1 related to the first control cycle, the

calculations for the second control cycle (i.e. k=2) are started at time instant ty2 as

shown in the timing diagram of Fig. 7.1(b). With the constraint that the CTW is not less

than the time allowed for calculations, TC, the timing diagram in Fig. 7.1(b) ensures

that, for ty2= ty1+T:

• The calculations for the second control cycle start after the completion of those

for the first control cycle.

• The start, t2, of the second control period coincides with the end, tf1, of the first

control period.

The steps of calculations are similar to those for the first control cycle. If there are no

changes in circuit-breaker and isolator status data, then the power system configuration

to be used would be the same as that for the first control cycle. Load models might have

to be updated, depending on the measured load demands at time tx2. The information on

the measured power system state at tx2 is received, after communication channel time

delay of TD1, at the control centre at time ty2, which provides the initial condition for

time-domain transient stability simulations to be carried out and completed within the

time interval TC between ty2 and tz2 as shown in Fig. 7.1(b). The calculation sequence as

described in steps (b)–(d) for the first control cycle is then repeated. For the first

transient stability simulation in the second control cycle, known values of the control

variables are used throughout the duration from tx2 to tf2 (the end of the second control

period). For the subsequent simulations for the duration from t2 to tf2, perturbed values

of the control variables as with those in the first control cycle are used.

At tz2, (i.e. the end of the calculation period), new optimal values of the control

variables for the second control cycle are sent to the FACTS device controllers for

Page 147: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

128

updating their input references. Following the updating, they are held constant at the

optimal values until tf2 (the end of the second control period).

The control coordination steps (a)-(e) are then repeated for the subsequent control

cycles (i.e. for k = 3, 4, … , M). The control coordination stopping criterion is as

follows:

• All of the relative rotor angles being within the nominated transient stability

threshold, and

• The amplitudes of relative rotor angle oscillations being within a specified upper

limit.

From t1 (the start of the first control period) to the end of the last control period, the

frequency of updating the FACTS device controllers input references, using option 1, is

1/T. The time delays TD1 and TD2 in the timing diagrams in Fig. 7.1 are settled by the

communication channels themselves. The control coordination designers need to

determine time intervals TC and T, taking into account the computer system processing

capability and the dependence of power system dynamic performance on CTW.

7.4.2 Option 2

Except the possibility that the time intervals TC and T can be different from those of

option 1, the timing diagram for the first control cycle (i.e. k = 1) in option 2 as shown

in Fig. 7.2(a) is the same as that in Fig. 7.1(a) for option 1. However, the control

coordination calculations related to the second control cycle (i.e. k = 2) in option 2 start

at ty2 = ty1+ Tx instead of ty1+T, as indicated in Fig. 7.2(b), with the following constraint:

TTT xC <≤ (7.7)

The constraint in (7.7) that Cx TT ≥ ensures that the calculations for the second control

cycle will start after the completion of those for the first control cycle. With Tx < T, as

indicated in Figs. 7.2(a) and (b), the new optimal values for the control variables

determined for the second control period will be implemented at time t2 which is prior

to tf1, the end of the first control period. The optimal values for the control variables

Page 148: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

129

determined for the first control period are implemented at time t1, and then kept constant

up until t1+Tx instead of t1+T as in the case of option 1. Similarly, from the control

coordination calculations for the second control period, the FACTS device controller

input references are updated at t2 and kept constant up until t2+Tx which is prior to tf2,

the end of the second control period. The sequence is then repeated for the subsequent

control cycles, with the control coordination stopping criterion being identical with that

of option 1. The calculation steps in Section 7.4.1 (a)–(d) in individual control cycles

are the same for both option 1 and option 2.

In option 2, the optimisation for each control period is carried out for the period T but

the control variable optimal values determined by that optimisation are implemented

only for the period Tx < T before they are updated with the new optimal values

determined for the next control period. With option 2, there are overlaps between

control periods, and the frequency of updating the FACTS device controller input

references is 1/Tx. With the same CTW, the updating frequency for option 2 is higher

than that (i.e. 1/T) for option 1.

The possible benefits offered by option 2 include a closer representation than with

option 1 in the control coordination of any changes in power system operating condition

as detected by measurements. However, the constraint that Tx is to be greater than the

calculation time TC would, in general, lead to the requirement of computer systems with

higher processing capability in comparison with that for option 1.

ty2 tz2t2 tf20

TD1 TD2

tx2

Timet

TTx

Tx(b)

TC

tx1 ty1 tz1 t1 tf10

TD1 TD2TC T

Timet

Tx(a)tc

Fig.7.2 Control scheme timing diagram: Option 2

(a) The first control cycle: 1=k (b) The second control cycle: 2=k

Page 149: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

130

7.5 FORMING TRANSIENT STABILITY INDICES

As referred to in Section 7.3.3, a key aspect in the control coordination proposed is to

form nonlinear relationships between the maximum relative rotor angles used as

transient stability indices and the control variables, through a series of time-domain

simulations for each control period described in Section 7.4.1 (b) and (c). In principle,

various forms of nonlinear functions can be postulated for expressing the required

relationships. In the present work, polynomial-type functions of the control variables are

adopted in forming the transient stability indices. For the ith generator, the nonlinear

function, fi (u,Tk) , which represents its maximum relative rotor angle within the kth

control period, Tk, is expressed in:

( ) ( ) ( ) ( )

( ) termsorder-higher,

,,,,

1 1 1

1 110

+

+++=

∑ ∑ ∑

∑ ∑∑

=≥=

≥=

=≥==

lnmL

m

L

mnn

L

nll

mnl

nmL

m

L

mnn

mnmL

mmki

uuukia

uukiaukiakiaTf u

(7.8)

In (7.8): i is the ith generator

k is the kth control period

L is number of control variables

( ),,0 kia ( )kiam , , ( )kiamn , , ( )kiamnl , are coefficients of the polynomial function for

the ith generator and kth control period (for m = 1, 2, …, L; n = 1, 2, …, L; l = 1, 2,

…, L).

um, un, ul are the mth, nth and lth elements of control vector u respectively.

The coefficients of the polynomial in (7.8) are to be identified, using the results of the

time-domain transient stability simulations. The identification process is developed in

the following. To achieve more compact notations in the development, the following

vectors are first defined:

Page 150: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

131

( ) ( ) ( ) ( ) ( )( ) ( ) ( )( ) ( ) ( ) ),,,,,,

,,,,,,,,,,,,,,,,(,

112111

1211

210

kiakiakiakiakiakia

kiakiakiakiaki

LLL

LL

Lt =α

(7.9)

),,,,,,,,,,,,,1(

211111

211121

LLL

LLLt

uuuuuuuuuuuuuuuuuu=z (7.10)

With the definitions in (7.9) and (7.10), equation (7.8) becomes:

( ) ( )kit Tfki ,, uαz = (7.11)

Various values of the control vector u within its range are specified and used in the

time-domain simulations in period Tk. From each specified control vector u, elements of

vector z in (7.10) are evaluated in a straightforward manner, using their individual

expressions in the RHS of (7.10). Associated with each specified control vector, there is

a set of maximum relative rotor angles within period Tk which are determined from the

time-domain simulation using the specified control vector. On this basis, the following

system of linear equations in terms of coefficient vector ( )ki,α for the ith generator is

formed, using the relation in (7.11):

( ) ( )kifki sp

tp ,, =αz p = 1, 2, …, NP (7.12)

In (7.12):

NP is the number of specified control vectors

zp is Vector z as defined in (7.10) and evaluated using the pth specified control vector

fsp(i,k) is the maximum relative rotor angle of the ith generator determined from the

results of the time-domain simulation for control period Tk with the pth specified control

vector.

Individual linear equations in (7.12) are assembled in a vector/matrix form as follows:

( ) ( )kiki ,, FαA =⋅ (7.13)

in which:

Page 151: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

132

[ ]NPt zzzA 21= (7.14)

and

( ) ( ) ( ) ( )[ ]kifkifkifki sNPsst ,,,, 21 =F (7.15)

The A matrix defined in (7.14) has the dimension of (NP×NC) where NC is the number

of polynomial coefficients (i.e. elements of vector ( )ki,α ). To obtain a unique solution

for vector ( )ki,α , NP is chosen to be greater than, or at least equal to NC. In general, the

inverse of A does not exist (when NP>NC). Vector ( )ki,α is determined by minimising

the error function defined in [156]:

( ) ( )( ) ( ) ( )( )kikikiki t ,,,,E FαAFαA −⋅−⋅= (7.16)

Minimising E in (7.16) based on the solution of ( ) 0

,E

=∂∂

kiα leads to:

( ) ( )kiki ,, FAα += (7.17)

In (7.17): ( ) tt AAAA1−

=+ (7.18)

Matrix A+ in (7.18) is referred to as the pseudo-inverse of matrix A.

Equation (7.17) applies to each and every generator (except the reference generator) for

evaluating the coefficients of individual polynomials expressing the nonlinear

relationships between the maximum relative rotor angles and FACTS devices

controllers input references, for each control period.

7.6 CONTROL COORDINATION FLOWCHART

Applicable to both options 1 and 2, Fig. 7.3 show the flowchart of the overall control

scheme for each control cycle, as described in Section 7.4.1 and 7.4.2., included in the

flowchart is the block representing the communication channels between the wide-area

measurement system and the control coordination system which receives the inputs in

terms of circuit-breakers and isolators statuses, and PMUs’ outputs. Drawing on the

Page 152: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

133

database held in the computer system of the control centre, network data, and dynamic

models of generators and controllers together with their data are available for forming

the network nodal admittance matrix, differential equations and initial values of state

variables for subsequent time-domain transient stability simulations. There are also

communication channels between the output of the control coordination system and the

FACTS devices controllers which receive the optimal input references as referred to in

the flowchart of Fig.7.3.

Wide-area measurement system

Communication channels

Circuit-breaker and isolator statuses

Synchronised PMUs’ outputs:Nodal voltages and currents

Database

Forming network nodal admittance matrix and its LU

factorisation

Forming nodal powers and state

variables

Time-domain transient stability simulations

Nonlinear function syntheses

Solving constrained optimisation problem

Optimal values of control variables

Communication channels

To FACTS device controllers

Fig.7.3 Flowchart of control coordination scheme

Page 153: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

134

7.7 CONCLUSIONS

Starting from the property that relative rotor angle transients following a disturbance

depend on FACTS device input reference settings, an online control coordination

method has been developed with the objective of maintaining or enhancing power

system transient stability. The method draws on transient stability time-domain

simulations and constrained optimisation for deriving the control algorithm by which

optimal input references for FACTS device controllers are determined to maximise

transient stability margins, taking into account communication channel time delays in

obtaining power system operating state and computing time required in executing the

control algorithm. The dynamic performance of the control method together with its

feasibility in terms of computing time requirement will be investigated and reported in

the next chapter.

Page 154: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

135

Chapter 8 Online Control Coordination of

FACTS Devices for Power System

Transient Stability: Computing Time

Requirement Analysis and Case-

Study

8.1 INTRODUCTION

The previous chapter has presented online transient stability control method, based on

time-domain simulations and constrained optimisation for real-time and optimal

adjustment of FACTS devices controllers input references. This chapter focuses on the

analysis of required computing time in implementing the method, and presents the

results of study cases. The individual components making up the computing time for

executing the control algorithm are identified, determined, and then combined to form

overall feasibility constraints. The control algorithms and feasibility constraints must

then be satisfied by computer systems used for control method implementation.

Page 155: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

136

Computational tasks which are independent of one another are identified so that they

can be performed using parallel computing systems. The effectiveness of the control

coordination in maintaining transient stability is verified by the simulation study, with

the control scheme having realistic timing parameters applied to a representative

multimachine power system.

8.2 COMPUTING TIME REQUIRMENTS

rawing on transient-stability time-domain simulation (TDS) and constrained

optimisation, previous chapter has derived a control coordination method by which

FACTS device input references can be optimally adjusted in real-time to maintain or

enhance transient stability of a power system following a large disturbance.

A key aspect related to the feasibility in implementing the method is that of the

computing time requirements in executing the control algorithm developed in the

previous chapter. The present research performs an analysis of the computing time

requirements which leads to a set of feasibility constraints to be satisfied by a computer

system used for implementing the control method. In the analysis, the computing time

component required for each step of executing the control algorithm is identified and

determined. From this the overall computing time for each control cycle is formed, and

combined with the selected control time window, where relative rotor angles are

controlled to lead to constraints in computing time allocation. To achieve feasibility,

these constraints are to be satisfied by the computer system adopted for the control

method implementation.

Conditions for meeting the feasibility requirements which include ‘faster-than-real-

time’ simulation and the application of parallel computing systems are derived and

discussed in this chapter. As previously developed in Section 7.4, each control cycle

needs several transient stability TDSs. However, they can be performed independently

of one another, using a parallel computing system. In addition, state-of-the-art TDS

technique has reached a high level of maturity, and ‘faster-than-real-time’ simulation is

now feasible [157], with detailed dynamic models for items of plant. Another important

feature of the TDS technique is that the computing time does not increase significantly

with power system size [157]. Based on these state-of-the-art technologies, the research

D

Page 156: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

137

shows that it is feasible to implement, and apply the control coordination method to the

10-generator, 39-node New England test system [158] modified with two thyristor-

controlled series compensators (TCSCs), using a cluster of PCs or high-performance

processors. The feasibility in terms of computing time is achieved with a substantial

margin. Offline simulations of the control scheme with realistic timing parameters

derived from the feasibility constraints are performed to verify the dynamic

performance of the scheme following a large disturbance in the power system.

8.3 ANALYSIS OF COMPUTING TIME REQUIREMENT

8.3.1 Control coordination structure

In Fig. 8.1 is shown the structure in the form of a block diagram for the control

coordination of FACTS devices as developed in the previous chapter. The control

structure applies for each control cycle. To facilitate the subsequent discussion, the time

instants at the inputs and outputs of individual blocks 2 to 8 are indicated in Fig. 8.1, for

the kth control cycle. There are ten functional blocks in the structure of Fig. 8.1, starting

with the wide-area measurement systems (WAMS) identified as block 1. Block 9

represents the FACTS device controllers which receive their input reference signals

formed by the control system comprising blocks 3 to 7. Block 10 represents the power

system to which FACTS devices are connected. Blocks 2 and 8 represent the time

delays in the communication channels described as follows:

Block 2: This represents the time delay of TD1 in sending the results of the

measurements by WAMS to the control centre in which optimal FACTS device input

references are determined for individual control cycles.

Block 8: This represents the time delay TD2 in sending the input reference signals from

the control centre to the FACTS device controllers.

Functional blocks 3 to 7 together with their computing time requirements are discussed

in the following section.

Page 157: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

138

8.3.2 Computing time

The total computing time requirement for each control period is the sum of individual

computing times for the calculation steps as described before. These computing times

are discussed in the following.

8.3.2.1 Power system modelling (block 3)

This calculation step, being a preparatory one for subsequent time-domain simulations,

is performed only once for each control cycle. The computing time required for this step

depends on the size of the power system for which control coordination is to be carried

out, and the processing speed of the computer system used. To facilitate the subsequent

discussion and development, the computing time required for this step is denoted by Ta.

8.3.2.2 Time domain simulations (block 4)

For each control cycle, there are a number of transient stability TDSs associated with

this functional block. The first simulation in the kth control cycle starts from time instant

txk, the measured system state which is received at the control centre at time instant tyk

after the time delay of TD1. This measured system state provides the initial condition for

the simulation. The duration of the simulation is (TD1+TC+TD2+T) where TC is the

computing time allocation, and T is the control time window. The purposes of the first

simulation are threefold, using the existing values of FACTS device input references.

• To assess the system transient stability with respect to the control stopping criterion

described in Section 7.4.1(e).

• To provide the data related to maximum relative rotor angles within the control time

window T if existing FACTS device input references are used. This data constitutes a

subset of the complete data set used for the syntheses of the polynomial functions

representing the relationships between maximum relative rotor angles and FACTS

device input references.

Page 158: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

139

• To provide the initial condition for subsequent simulations required for forming the

complete data set used for polynomial function syntheses, as described in the following.

As derived in Section 7.5, to obtain the solution for the polynomial coefficient vector

with NC elements, it is required to perform NP (with NP ≥ NC) TDSs to provide NP

subsets of data for maximum relative rotor angles occurring within the control time

window T, with t1k = txk + TD1+TC +TD2 as indicated in the structure of Fig.8.1 (i.e. from

t1k to t1k+T). On this basis, there are, in addition to the first simulation, (NP-1)

simulations which, for each, the solution time is T. In performing these simulations, the

FACTS device input references (i.e. control variables) are perturbed within their ranges.

The perturbed input references form the set of specified control vectors. The initial

condition for starting these simulations is provided by the solution at time instant t1k

obtained in the first simulation.

The computing time for the first simulation is β∙(TD1+TC+TD2+T) where β is a constant

depending on the size of the power system, time step length and processing capability of

the computer system used.

The computing time required for (NP-1) simulations subsequent to the first is β∙(NP-1)T

if they are to be performed sequentially. Therefore, the total computing time required,

Tb, for time-domain simulations related to each control cycle is:

[ ]CDDb TTNPTTT +⋅++= 21β (8.1)

or, if NP simulations each with solution time T are performed in parallel, using a cluster

of processing systems.

[ ]CDDb TTTTT +++= 21β (8.2)

8.3.2.3 Polynomial function syntheses (block 5)

The computing time required for this calculation step for each control cycle depends on

the number of generators, number of specified control vectors, number of control

Page 159: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

140

variables, order of the polynomials adopted in representing the nonlinear functions and

processing capability of the computer system. For convenience in the subsequent

discussion, the computing time required for this step is denoted by Tn.

8.3.2.4 Constrained optimisation (block 6)

With a given processing capability of the computer system used, and constrained

optimisation algorithm, the computing time (denoted by T0) required for this calculation

step in determining the optimal values of the control variables, would depend on the

number of control variables, number of generators and order of the polynomials.

8.3.2.5 Overall requirement

From the individual components identified in 8.3.2.1 to 8.3.2.4, the total computing

time required for the control coordination for each control period is:

0TTTTT nbatotal +++= (8.3)

In (8.3), Tb is given by (8.1) or (8.2), depending on the computer systems used.

8.3.2.6 Feasibility constraints

It is required that, for feasibility, the total computing time formed in (8.3) be less than,

1DTse−

txk tyk

Database

WAMSPower system

modelling.Computing

time:Ta

One TDS from txk to t1k+T.

(NP-1) TDSs from t1k to

t1k+T.Computing

time: Tb

Polynomial function

syntheses. Computing

time: Tn

Constrained optimization. Computing

time:To

FACTS device

controllers

Power system

tyk + Ta+ Tb+Tn

tyk + Ta+ Tb+ Tn+ To

tyk + Ta+ Tbtyk + Ta

mTse− 2DTse−

tzk + TD2 = t1ktzk

(1) (2)(3)

(4)TC

(5) (6)

(7) (8)(9) (10)

Time instants

Fig.8.1. Control coordination block diagram

TDS: time-domain simulation Subscript k identifies the control cycle

s: Laplace transform operator TC: Computing time allocation

Page 160: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

141

or at most equal to the computing time allocation, TC:

Ctotal TT ≤ (8.4)

Block 7 in Fig. 8.1 represents the margin between the completion of the computation

and initiating the transmission of the input references to FACTS device controllers. The

time margin, Tm, shown in block 7 is:

Tm=TC – Ttotal (8.5)

For control option 1 as described in Section 7.4.1 (e), it is also required that:

TTC ≤ (8.6)

However, if control option 2 as referred to in Section 7.4.2 is adopted, the condition in

(8.6) becomes:

xC TT ≤ (8.7)

where Tx is defined in Section 7.3.4 and Fig. 7.2.

For given communication channel time delays, the option adopted for control

coordination, and computer system used, the conditions in (8.4), (8.6) and (8.7) provide

a basis for proper coordination among time intervals T, Tx and TC.

8.4 CASE-STUDY SIMULATION RESULTS

To verify the proposed controller coordination scheme, various case study simulations

were carried out. The first experiment was carried out on smaller power system network

of 4 generators and 12 node system. Being smaller sized power system network, only

one TCSC was introduced and the algorithm was tested for various fault locations and

for different control window sizes. The communication and computation delays were

considered while developing required control law for TCSC. The simulation results

Page 161: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

142

proved its effectiveness in maintaining first swing rotor angle stability. To verify the

robustness of this coordination scheme, a larger power system network with increase in

number of TCSCs. The selected results of this representative case study are presented in

next section.

8.5 REPRESENTATIVE POWER SYSTEM NETWORK

The 39-bus New England power system having 10 generators as shown in Fig. 8.2 is

adopted in the study for the purpose of illustrating the performance of the proposed

transient stability control coordination scheme, and investigating the feasibility of its

implementation. The system has two thyristor-controlled series compensators (TCSCs) -

the first TCSC in transmission line L11 between nodes N11 and N15, and the second

TCSC in transmission line L8 between nodes N12 and N16.

11

10

8

2

3

4 5

7

6

9

W

N12

N10

N11

N2

N8

N25

N26

N28

N29

N9N24N27N38

N37

N13

N15

N19

N18

N17

N1N16

N14

N31

N3

N20

N32

N33

N34

N35

N36

N21

N39

N30

N4 N5

N7

N23

N22

N6

TCSCL10

L1L2

L3

L4

L5

L6

L7

L8

L9

L11

L12

L13

L14

L15

L16L17

L18L19

L20

L21

L22

L23L24

L25

L26

L27

L28

L29

L30

L31

L32

L34

L33

W W

W

W W

W

W

W

W

W

W

TCS

C

N40

N41

Fig. 8.2. 10-Generator 39-Node New England test system[158]

The actual reactance output limits of TCSC are of a dynamic form which depends on

the TCSC operating current [23]. The TCSC data used in the study is given in Table 8.1.

In the present case-study the transmission lines in which TCSC is to be inserted are

Page 162: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

143

chosen based on the location and length which will benefit the overall system stability

improvement. With various case-studies carried out with different locations of

placement of TCSC, it is shown that, inserting TCSC in line 11 and line 8 gives the

most effective compensation to favour the entire power system network faster recovery

for transient stability performance.

Table 8.1: Controller parameters

No. Controller Parameters 1

TCSC in line L11

Main Controller: TC= 0.01s; KC=1pu;

pu0.0313Xpu0.05 ref ≤≤− SDC: K1=0.1pu; TW=0.2s; T1=0.2s; T2=0.1s;T3=0.05s;T4=0.2s

pu0.0025Xpu0.0025 SDC ≤≤−

2

TCSC in line L8

Main Controller: TC= 0.01s; KC=1pu;

pu0.0543Xpu0.0868 ref ≤≤− SDC: K1=0.1pu; TW=0.2s; T1=0.2s; T2=0.1s;T3=0.05s;T4=0.2s

pu0.0043Xpu0.0043 SDC ≤≤−

8.6 SYSTEM RESPONSE WITHOUT ONLINE CONTROL COORDINATION

OF TCSCS

The disturbance condition is that of a three-phase-to-earth fault on transmission line L6

near node 37, with a fault clearing time of 160ms. Subsequent to the fault disturbance,

the TCSC reactance input references remain constant at their pre-fault values which

were set at zero. With generator 1 nominated as the reference, the relative rotor angles

of generators 2–10 with respect to the reference generator are shown in Fig. 8.3. The

time step length in the time-domain simulation is 10ms. The fault onset time instant is

200 ms with respect to the origin of the time axis in Fig. 8.3. The responses confirm

that, without online adjustment of TCSCs input references, transient stability is lost

following the fault and fault clearance.

Page 163: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

144

As the fault happened to be on line 6, near node 37, it can be seen from the relative rotor

angle plots that the entire power system network gets separated in two parts. Generators

4, 5, 6 and 7 form one group or island, while remaining generators (i.e. 2, 3, 8, 9, and

10) form second group with less deviation in their relative rotor angles. It can be

observed from the transient stability response that the machines in the given group

remain in synchronism with each other inside the group. However, the first group of

machines seems to fall out of synchronism with the second group of machines in less

than 1 second’s time. The separation of groups increases as time passes, indicating that

unless corrective action is taken as early as possible, recovering the system stability

may become very difficult or in some cases, even impossible.

Fig.8.3 Relative rotor angle transients without online control coordination of TCSCs

8.7 OUTLINE OF TCSCS CONTROL COORDINATION STUDY

Drawing on the transient stability control schemes developed in Section 7.4 in the

previous chapter, control coordination of TCSCs reactance input references will be

derived and investigated, subject to feasibility constraints in terms of computing time as

discussed in Section 8.3.2.4. The computing time requirements in relation to time-

domain transient stability simulation, nonlinear function syntheses and constrained

0 0.5 1 1.5 2 2.5 3-100

0

100

200

300

400

Time (s)

Rel

ativ

e R

otor

Ang

le (D

eg)

G2 G3 G4 G5 G6 G7 G8 G9 G10

Page 164: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

145

optimisation will be determined. From this the total computing time required for each

control cycle is then calculated, and allocated, taking into account communication

channels delays. The study will also include the investigation of the control

coordination dynamic performance with various control scheme parameters.

8.8 TIME-DOMAIN SIMULATION COMPUTING TIME REQUIREMENTS

Transient stability software systems, particularly those designed for applications in

industry such as Power System Simulator for Engineering (PSS/E), have reached a high

level of maturity. They have been developed and refined over many years, which

implement techniques for speeding up large-scale power system simulation [157].

Those software systems can perform faster-than-real-time transient stability simulation,

even with detailed dynamic models, and, importantly, the computing time does not

significantly increase with power system size [157]. Using a transient stability analysis

function in DIgSILENT PowerFactory implemented on a PC with Intel(R) 2-core CPU

with E6550 processor, a faster-than-real-time transient stability simulation of the power

system in Fig. 8.1 is achieved with factor β ( as defined in Section 8.3.2.2) having the

value of about 0.188. The computing time required for the simulation includes that for

the power system modelling referred to in Section 8.3.2.1. As discussed in Section

8.3.2.2, each control cycle needs multiple transient stability simulations, which means

that a cluster of PCs or processors would be required to perform these individual

simulations in parallel for the control schemes to be feasible, particularly when the

number of separate simulations (i.e. NP as referred to in Section 7.5 in the previous

chapter) required in each control cycle is large.

8.9 COMPUTING TIME REQUIREMENTS FOR NONLINEAR FUNCTION

SYNTHESES

Many studies have been performed in the present work for identifying the nonlinear

relationships between relative rotor angles and TCSCs input references. The results of

the studies indicate that fourth order polynomials can represent with high accuracy the

nonlinear relationships.

Page 165: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

146

With two TCSCs, there are 15 coefficients to be identified for each fourth order

polynomial. The total computing time required in the syntheses of nine polynomials

representing nine relative rotor angles is about 0.014ms, using the algorithm developed

in Section 7.5 with NP = 15, and the PC referred to in Section 8.7.

8.10 COMPUTING TIME FOR CONSTRAINED OPTIMISATION

For the search method described in Section 7.4.1(d), the objective function and nine

constraint functions associated with the relative rotor angles, are to be calculated for

individual values of TCSCs reactance input references specified within their ranges. The

optimal solution is selected from the results of the calculations. For high resolution, a

discrete search space in terms of a 100X100 grid of TCSC reactance reference values,

with uniform spacing, has been used in the calculations. This search method can be

directly implemented by a parallel computing system. For example, the search space in

this case can be subdivided into four equal subspaces, and the search processes for

individual subspaces can be performed in parallel. The global optimal solution is then

obtained from the four optimal solutions for the subspaces. Using the PC as described in

Section 8.7, the computing time for identifying the optimal solution in each subspace is

about 22ms. With a cluster of four processing systems, each with a computing

capability of the PC referred to in Section 8.7, operating in parallel, it is possible to

obtain the global optimal solution in about 22ms. For comparison, the same constrained

optimisation problem has been solved using the quasi-Newton method which achieves

the optimal solution with a computing time of 1.01s. The solution from the search

method with substantially lower computing time is almost identical to that from the

quasi-Newton method.

8.11 CONTROL COORDINATION STUDY RESULTS

8.11.1 Control Option 1

Based on the development presented in Chapter 7, the computing time required for this

calculation step for each control cycle depends on the number of generators, number of

specified control vectors, number of control variables, order of the polynomials adopted

Page 166: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

147

in representing the nonlinear functions and the processing capability of the computer

system. For convenience in the subsequent discussion, the computing time required for

this step is denoted by Tn.

The investigations in Sections 8.7–8.9 have established the value for factor β (0.188)

related to computing time required in transient stability simulation, computing time

requirement for nonlinear function syntheses (0.014ms) and constrained optimisation

(22ms). The computing time component (Ta) for power system modelling referred to in

Section 8.3.2.1 has been included in the estimation of factor β which gives a value on

the conservative side. Based on [41], a communication channel’s time delay of 60ms is

adopted in the study. It remains to select the value of control time window (T) and

computing time allocation (TC) for each control cycle which are subject to:

(i) inequality constraints in (8.4) and (8.6);

(ii) transient stability being maintained following fault and fault clearance.

In terms of the condition in (i) in the above, it is required that, using the values for TD1

(60ms), TD2 (60ms), β (0.188), Tn (0.014ms) and T0 (22ms):

TTT C ≤≤+× .89450.23 (8.8)

The unit for T and TC in (8.8) is ms. In forming (8.8), it is taken that the computing

time for time-domain simulations in each cycle is given by (8.2) where individual

simulations would be performed in parallel as discussed in Section 8.3.2.2, using a

cluster of processing systems. (i.e. NP processing systems in parallel operation). With

the fourth order polynomials and two TCSCs, the requirement is NP=15 as referred

to in Section 8.8.

Various combinations of values of TC and T can satisfy the inequalities in (8.8), to meet

the feasibility requirements related to computing time based on processing systems as

described in Section 8.7. Many case studies have been carried out, using combinations

of TC and T, for investigating the dynamic performance of the control scheme. The

results of two representative cases are presented and discussed in the following.

Page 167: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

148

8.11.1.1 Case 1: T = 80ms and TC = 80ms

The nominated transient stability threshold in terms of relative rotor angle is 1400. The

control coordination is terminated when all of the relative rotor angles are within the

nominated transient stability threshold, and amplitudes of relative rotor angle

oscillations are less than a specified upper limit of 500. The fault disturbance is the same

as that in Section 8.5. With respect to the origin of the time axis in Fig. 8.4, the fault

onset time instant is 200ms; the fault is cleared at 360ms; and the first control period

starts at 560ms, i.e. the adjustment of TCSCs input references commences 200ms after

the fault clearance. Offline simulations are carried out for successive control cycles to

assess the dynamic performance of the control scheme.

Fig. 8.4 shows the transient responses of TCSCs reactance references (i.e. the control

variables) as determined by the control scheme outputs. Their initial values are 0.00 pu.

Using these controllers’ outputs for adjustments of TCSCs input references leads to the

responses in Fig. 8.5 for nine relative rotor angles. The responses confirm that transient

stability is maintained subsequent to fault and fault clearance. Control coordination

stopping criterion is achieved at 6.24s, i.e. after 71 control cycles as indicated by the

vertical line shown in Fig. 8.5. The transient responses of TCSCs input references are

highly oscillatory between their upper and lower limits during the control coordination

duration of 71 control cycles. Without the control coordination of the TCSCs, transient

stability would be lost as indicated in Section 8.5.

Page 168: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

149

Fig.8.4 TCSC input references for case 1

Fig.8.5 Relative rotor angle transients with online control coordination of TCSCs for case 1

8.11.1.2 Case 2: T = 200ms and TC =110 ms

Subject to the constraints in (8.8), the control time window (T) and computing time

allocation (TC) adopted in study case 1 are minimum possible values. The study in case

2 investigates the sensitivity of the control coordination performance with respect to T

0 1 2 3 4 5 6 7-0.1

-0.05

0

0.05

0.1

Time (s)

Rea

ctan

ce R

efer

ence

(pu)

TCSC1(L11)

TCSC2(L8)

0 1 2 3 4 5 6 7-150

-100

-50

0

50

100

150

Time (s)

Rel

ativ

e R

otor

Ang

le (D

eg)

G2 G3 G4 G5 G6 G7 G8 G9 G10

Page 169: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

150

and TC. Increasing T to 200ms and TC to 110ms leads to TCSCs input references

transient responses being less oscillatory as shown in Fig. 8.6 , in comparison with those

in case 1. However, the maximum relative rotor angles in the first swing are similar in

the two cases, as indicated in Fig. 8.7 for case 2 and Fig. 8.5 for case 1. Although

transient stability is maintained in both cases, the dynamic response of case 1 with

shorter T and TC is marginally better than that of case 2, at least in terms of required

control coordination duration. In case 1, the stopping criterion is achieved at 6.24s,

compared with 6.99s for case 2.

Fig.8.6 TCSC input references for case 2

Fig.8.7. Relative rotor angle transients with online control coordination of TCSCs for case 2

0 1 2 3 4 5 6 7 8-0.1

-0.05

0

0.05

0.1

Time (s)

Rea

ctan

ce R

efer

ence

(pu)

TCSC1(L11)

TCSC2(L8)

0 1 2 3 4 5 6 7 8-150

-100

-50

0

50

100

150

Time (s)

Rel

ativ

e R

otor

Ang

le (D

eg)

G2 G3 G4 G5 G6 G7 G8 G9 G10

Page 170: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

151

8.11.2 Control Option 2

Control coordination in case 2 with option 1 in (a) is modified to form a control

sequence based on option 2 described in Section 7.4.2. The control time window (T)

and computing time allocation (TC) remain at 200ms and 110ms respectively. However,

with option 2, the optimal TCSCs input references obtained from each constrained

optimisation are applied for a time duration Tx < T, instead of T. In the present study, Tx

is chosen to be 120ms. This leads to the TCSC input reference transient responses of

Fig. 8.8. In comparison with control option 1 with the same control time window

(200ms) and computing time allocation (110ms), the TCSCs input reference responses

obtained with control option 2 are more oscillatory as indicated in Figs. 8.6 and 8.8.

However, the relative rotor angle transients in the two control options are similar, as

confirmed in the responses shown in Figs. 8.7 and 8.9. With option 2, the stopping

criterion is achieved at 6.59s which is less than 6.99s for case 2 in option 1.

Fig.8.8 TCSC input reference responses Option 2

0 1 2 3 4 5 6 7-0.1

-0.05

0

0.05

0.1

Time (s)

Rea

ctan

ce R

efer

ence

(pu)

TCSC1(L11)

TCSC2(L8)

Page 171: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

152

Fig.8.9 Relative rotor angle transients Option 2

8.11.3 Approximate control

The TCSCs control schemes in Sections (a) and (b) have been investigated on the basis

of subdividing with high resolution the ranges of TCSCs input references for solving the

constrained optimisation problem. The control schemes are effective in maintaining or

enhancing system transient stability, and feasible in terms of computing time

requirements which can be met by a cluster of PCs or processors operating in parallel. It

is proposed in this section to reduce the computing time requirement by developing an

approximate control scheme in which each TCSC input reference is represented by a

low number of discrete levels. If L is the number of TCSCs, and each TCSC input

reference is represented by ND levels between its lower and upper limits in the control,

then the number of TDSs required in each control cycle for determining the optimal

combination of TCSCs input references for the cycle would be NDL. There would be no

requirements for the syntheses of polynomials and constrained optimisation. As the

number of TCSCs in a power system is relatively small, the required number of TDSs

would be, in general, significantly lower than that in the case of control with a high

resolution, particularly when the number of levels, ND, is low.

0 1 2 3 4 5 6 7-150

-100

-50

0

50

100

150

Time (s)

Rel

ativ

e R

otor

Ang

le (D

eg)

G2 G3 G4 G5 G6 G7 G8 G9 G10

Page 172: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

153

For illustrating the effectiveness of this approximate form of control, the input reference

of each TCSC in the system is represented by two levels: upper and lower limits. With

two TCSCs, each control cycle requires only four TDSs which can be performed by a

cluster of four processors in parallel operation. Optimal values of TCSCs input

references are then selected directly from the comparison among the time-domain

solutions, without the need for polynomial syntheses and constrained optimisation. This

approximate control in the discrete form leads to TCSCs input references in Fig. 8.10

and relative rotor angle transients in Fig. 8.11. The control scheme parameters are the

same as those in case 2, except that only the lower and upper limits of TCSCs input

references are used in the control. The responses in Fig. 8.11 confirm that transient

stability is maintained. The time instant at which the stopping criterion is achieved is

6.99s which is the same as that in case 2. However, the benefit of reducing the

computing system requirement is a substantial one.

Fig.8.10 TCSC input reference responses Approximate control

0 1 2 3 4 5 6 7 8-0.1

-0.05

0

0.05

0.1

Time (s)

Rea

ctan

ce R

efer

ence

(pu)

TCSC1(L11)

TCSC2(L8)

Page 173: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

154

Fig.8.11 Relative rotor angle transients Approximate control

8.12 CONCLUSION

Based on the transient stability control algorithm developed in Chapter 7 the present

chapter has performed a comprehensive analysis of the computing time requirement in

implementing the algorithm. The analysis outcome is a set of feasible constraints which

are to be satisfied by computer systems used for online control coordination of FACTS

devices with the aim of enhancing or maintaining power system transient stability

following disturbances. The various case studies carried out on different power system

configurations, for different fault locations, in various scenarios has confirmed the

robustness of proposed algorithm. With reference to a representative multimachine

power system with TCSCs, the study presented in the chapter indicates that it is

feasible, within the current technology, to implement the control coordination algorithm

in real time, using a cluster of processors operating in parallel. Offline simulation, using

realistic timing parameters for the control scheme, confirms its effectiveness in

maintaining transient stability following a fault disturbance.

0 1 2 3 4 5 6 7 8-150

-100

-50

0

50

100

150

Time (s)

Rel

ativ

e R

otor

Ang

le (D

eg)

G2 G3 G4 G5 G6 G7 G8 G9 G10

Page 174: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

155

Chapter 9 Dynamic Modelling Application for

Estimating Internal States of a

Synchronous Generator in Transient

Operating Mode from External

Measurements

9.1 INTRODUCTION

Drawing on the availability of synchronous generator terminal voltage and current, rotor

speed, and field winding voltage measurements, a procedure is derived for estimating in

transient conditions the generator’s internal operating states. These operating states are

rotor angle and flux linkages associated with field winding and damper windings. The

procedure is based on the fifth-order generator dynamic model. By applying the

numerical integration formula based on the trapezoidal rule, the generator model is

described by a set of algebraic equations of a recursive form in the discrete time-

domain. With external measurements, the unknown variables in the equations are those

Page 175: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

156

representing generator internal operating states. The nonlinear equations derived for

successive time instants are solved by applying the Newton-Raphson method. However,

if the number of equations is greater than that of variables, a minimisation technique

based on sequential quadratic programming method is then applied to solve the

nonlinear equation system. The estimation procedure can be applied to any time instants

including those in the transient operating conditions, without the requirement for

specifying the steady-state condition. The effectiveness and accuracy of the procedure

developed are verified by simulation using a representative multimachine power system

operating in the transient mode.

9.2 BACKGROUND THEORY

Power system stability including transient stability is an issue of increased importance

at present due to reduced stability margins arising from the maximisation of system

utilisation by power companies to increase their competitiveness in market

environments. In an attempt to optimise system dynamic performance, real-time control

methods for enhancing or maintaining system stability have been proposed, developed

and reported in the literature [13, 153, 159]. Central to the methods is the assumed

availability of the internal operating states of synchronous generators in the power

system to which real-time stability control is to be applied. However, the internal

operating states which include generator rotor angles and flux linkages are difficult, if

not impossible, to measure directly for control purpose.

There has been very limited research on real-time estimation of synchronous generator

internal states based on available measurements external to, or at, terminals of a

generator. In [160], a procedure based on a simplified generator model (i.e. single-axis

rotor flux model) in which damper windings and rotor speed transients are discounted

was presented for calculating approximate rotor angle and flux linkage established by

the field winding only. With the simplified model, only algebraic equations are required

in the calculation, using the measurements of generator stator terminal voltages,

currents and power, together with the field winding current. Generator dynamic

responses are not taken into account in the calculation. A method based on artificial

neural networks was proposed in [161] for estimating the rotor angles of synchronous

machines. The method requires offline training of neural networks, the inputs of which

Page 176: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

157

are voltage and current phasors obtained from a phasor measurement unit (PMU) on the

high voltage side of the generator transformer, and the outputs of the trained neural

networks are used for forming the rotor angle. A disadvantage of the method is that the

neural networks need to be updated or retrained when there are changes in network

configuration and/or a combination of generators in operation [161]. In addition, flux

linkages are not estimated in the method.

More recently, an algorithm based on a divide-by-difference filter has been proposed for

estimation of generator rotor angle, using the third-order model of the synchronous

generator [162]. Rotor damper windings are not represented in the estimation. Other

disadvantages of the proposed estimation procedure include the need for specifying a

generator’s steady-state condition for initialising the estimation process, and the

requirement for values of input mechanical torque at individual sampling time instants,

which is difficult, if not impossible, to measure.

Given the above background and state-of-the-art methodology in the estimation of

synchronous generator internal operating states, the present research has the objective of

developing a new procedure for estimating generator flux linkages and rotor angle in

which rotor speed transients and rotor field winding together with damper windings are

represented, requiring only measurements external to the generator. The new procedure

has the following features:

• Requiring only measurements that are practically feasible.

• Elimination of the need for specifying generator steady-state conditions for

initialising the estimation process.

• Representation of generator dynamic responses. The field winding and damper

winding flux linkage transients are included in the estimation process.

• Flexibility of starting the estimation process at any time instant including in the

transient periods.

• Low computing time requirements. This allows real-time estimation, with potential

applications in real-time transient stability control and monitoring.

• High accuracy and numerical stability.

• Robustness. The estimation process is independent of power network configuration

and/or operating conditions.

Page 177: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

158

Based on the standard practice in transient stability analysis and control, the fifth-order

dynamic model of the generator is adopted for deriving the estimation procedure in

which generator dynamic responses are represented. Through the use of the trapezoidal

rule of numerical integration, the relationships among stator terminal voltages, currents

and rotor flux linkages together with rotor angle and speed are transformed into a set of

recursive equations in the discrete time-domain. The time step length to be used in the

numerical integration of the differential equations in the generator model is a typical

one for transient stability analysis, and compatible with the sampling rate of phasor

measurements achieved at present. Combining the generator dynamic model in the

discrete time-domain with the available measurements at successive sampling time

instants of generator terminal voltage and current phasors, together with rotor field

winding voltage and rotor speed, leads to a system of nonlinear equations in which the

unknown variables are the rotor flux linkages and rotor angles at individual sampling

time instants. Values of input mechanical torque are not required in the estimation

procedure.

The estimation procedure can start from any nominated time instant. There is no need

for specifying the steady-state condition for the starting the procedure. The number of

successive sampling time instants of measurements is chosen, subject to the constraints

that the number of nonlinear equations formed at these individual time instants, to be

greater than, or equal to the number of unknown variables. The system of nonlinear

equations where the number of unknown variables is less than that of equations is

solved using an unconstrained minimisation algorithm. In each iteration, an objective

function of a quadratic form in the variables is derived and then minimised by the

Newton’s method. In the particular case where the number of unknown variables is

equal to that of equations, the Newton-Raphson method is applied for solving the

nonlinear equation system. The necessary conditions for the feasibility of solving the

estimation problem are derived and discussed in this chapter.

Although the specific context of the present research is that of estimating synchronous

generator operating states, the estimation procedure developed is of general validity and

applicable to any dynamical system where unmeasured internal state variables are

required for monitoring and control purpose. Given this general nature, the development

Page 178: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

159

of the estimation procedure in its dynamic form will commence with a general nonlinear

dynamical system described by a set of differential and algebraic equations.

Synchronous generators are interpreted, for the purpose of estimating their internal

states, as a particular case of nonlinear dynamical system.

Within the context of this research, the estimation procedure developed is verified for its

effectiveness and accuracy by simulation applied to a synchronous generator in a

representative multimachine power system subject to a fault disturbance. The simulation

study confirms that the estimation procedure has low computing time requirement, and

is suitable for real-time application.

9.3 DEVELOPMENT OF ESTIMATION PROCEDURE

9.3.1 Continuous-time nonlinear dynamical system and estimation requirement

In a general form, a continuous-time nonlinear dynamical system is modelled by the

following set of differential and algebraic equations as derived in the Chapter 2 and

represented as:

))(),(),(()( tttt uyxfx = (9.1)

0))(),(),(( =ttt uyxg (9.2)

In (9.1) and (9.2):

t : independent continuous-time variable

x(t) : vector of state variables

y(t) : vector of non-state variables

u (t) : vector of input variables

f, g : nonlinear vector functions of x(t), y(t), and u(t)

The total number of equations in (9.1) and (9.2) is denoted by N. The total number of

individual variables in x(t) and y(t) is equal to L.

It is taken that the system described by (9.1) and (9.2) operates, in general, in transient

condition, and it is, therefore, not valid to assume that 0)( =tx .

Page 179: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

160

The dynamical system described in (9.1) and (9.2) can be a subsystem operating within

a system, and some of the variables in x and/or y in (9.1) and (9.2) are the interface

variables between the system and its subsystem. An example is that of a synchronous

generator operating in a multimachine power system. The interface variables in this case

are the voltage and current at the generator terminal to which the power network is

connected.

There are direct measurements in the discrete-time domain of some of the variables in

vectors x and/or y. It is required that the remaining variables in x and/or y which cannot

be directly measured be estimated, using the set of available measurements at individual

sampling time instants.

The sampling rate of the measurement is denoted by fs corresponding to the sampling

time interval of ∆t = 1/fs.

It is taken that the sampling rate of the measurement system has been selected, taking

into account the transient phenomena of interest arising from the responses of the

dynamical system following disturbances.

9.3.2 Discrete-time domain system model

With measurements available in the form of time series obtained via data acquisition

systems and PMUs, it is required to transform (9.1) and (9.2) into discrete-time

equations for deriving the estimation algorithm. The time step length used in

transforming into the discrete-time domain is chosen to be the same as the sampling

time interval, ∆t, of the measurement system. By applying the trapezoidal rule of

numerical integration, the differential equation in (9.1) is transformed into:

[ ]))1-(),1-(),1-(())(),(),((2

)1()( nnnnnntnn uyxfuyxfxx +∆

+−= (9.3)

In (9.3), integer n is the discrete time variable from which the actual time is given by

n.∆t.

Page 180: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

161

The algebraic equation in (9.2) transforms directly to, in the discrete-time domain:

0))(),(),(( =nnn uyxg (9.4)

The set of algebraic equations in (9.3) and (9.4) is combined to:

0))1-(),1-(),1-(),(),(),(( =nnnnnn uyxuyxh (9.5)

Vector function h in (9.5) is defined in:

−−⋅−

=))(),(),((

))1-(),1-(),1-(()1-())(),(),(()(

))1-(),1-(),1-(),(),(),((nnn

nnnnnnnn

nnnnnnuyxg

uyxfxuyxfx

uyxuyxhα

α

(9.6)

where 2t∆

=α (9.7)

The total number of individual discrete-time equations in (9.5) is N.

9.3.3 Estimation problem formulation

With the input vector u being specified, and a subset of the variables in vectors x and/or

y being measured at individual time steps, it is required to estimate all of the

unmeasured variables in vectors x and/or y. With time instant n=0 nominated as the

time reference, the estimation of the unmeasured variables is to be carried out for

successive time instants, starting from n=0. If the total number of measured variables is

denoted by M, then it is required to estimate (L-M) variables at each time instant n ≥ 0.

The problem of estimation does not arise if M=L.

If the dimension of vector g is NG, then there are NG equations obtained from (9.4) at

time n=0:

0))0(),0(),0(( =uyxg (9.8)

Page 181: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

162

With successive time instants j = 1, 2, . . ., K applied to (9.5), the following set of

equations are obtained:

0))1-(),1-(),1-(),(),(),(( =jjjjjj uyxuyxh j=1, 2,..., K (9.9)

As the dimension of h is N, there are K.N individual equations in (9.9).

With (K+1) time instants, there are (K+1).(L-M) unmeasured elements to be estimated.

From (9.8) and (9.9), there are (K.N+NG) equations. A necessary condition for the

estimation to be possible is:

(K.N + NG) ≥ (K+1) (L – M) (9.10) or , if (N – L + M ) > 0 :

)()(

MLNNGMLK

+−−−

≥ (9.11)

It is not feasible to solve the estimation problem when (N – L + M ) ≤ 0.

If there does not exist a non-negative integer K that satisfies (11), then it is not possible

to solve the estimation problem. If the inequality in (11) is satisfied for some non-

negative integer K, then the problem is that of solving the set of algebraic equations in

(8) and (9) to obtain the values for unmeasured variables in x and/or y at time instants 0,

1, 2, . . . , K. In the present work, the smallest non-negative integer K which satisfies

(11) is chosen. Once the values of unmeasured variables at sampling time instants 0, 1,

2, . . . , K have been determined, the estimation process for subsequent time instants is

carried out as described in Section 9.3.5.

9.3.4 Solution method

Equations (9.8) and (9.9) are assembled in:

H (XK) = 0 (9.12)

In (9.12):

Page 182: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

163

XK = unmeasured elements in vectors x(i) and/or y(i) for i = 0,1,2, . . ., K.

and

=−−−

=

Kjjjjjjj

K

,1,2,3,for ))1()1()1()()(),((

))0()0(),0(()(

u,y,x,u,yxh

uyxgXH (9.13)

Based on (9.11), if there is a non-negative integer K such that K = (L – M – NG) / (N – L

+ M), then the number of equations in (9.12) is equal to the number of unknown

variables in vector XK.

The Newton-Raphson method is applied to solve the set of nonlinear equations in (9.12)

for XK. The iterative solution sequence is given in:

−=

−1-

11-1- p

KpK

pK

pK XHXJXX (9.14)

In (9.14):

p = the Newton-Raphson iteration step counter 1-

,pK

pK XX : values of the elements in vector XK at iteration steps (p-1) and p

respectively

J = Jacobian matrix of vector function H:

1-1- )(

pK

K

KpK

XXXH

XJ

∂=

(9.15)

A necessary condition for achieving the solution using the iterative sequence in (9.14) is

that the Jacobian matrix formed in (9.15) is nonsingular for each iteration. The iterative

sequence in (9.14) is terminated when the following convergence criterion is achieved:

ε≤

pi KH X i = 1,2, . . . , (K.N + NG) (9.16)

Page 183: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

164

Hi in (9.16) is the ith element of vector function H, and ε is a pre-set tolerance for

convergence checking.

On the other hand, from (9.11), if, K is greater than (L – M – NG) / (N – L + M), then

there are more individual equations in (9.12) than unknown variables in vector XK. In

this case, the solution for XK is determined by minimising the following objective

function with respect to XK:

( ) ( )KKtE XHXH ⋅= (9.17)

On linearising H(XK) around the tentative solution at the previous iteration step (p-1)

where p is the current step:

( )

+

−−− 111 pKK

pK

pKK XXXJXHXH (9.18)

In (9.18), J is the Jacobian matrix of vector function H as given in (9.15). However, J in

(9.18) is not a square matrix as there are now more equations than unknown variables.

Combining (9.17) and (9.18) gives:

+

+

=

−−−

−−−

111

111

pKK

pK

pK

tpKK

pK

pKE

XXXJXH

XXXJXH (9.19)

Minimising the objective function in (9.19) as a quadratic form in XK by Newton’s

method leads to the following optimal solution for XK, at iteration step p:

−=

−−−−− 11111 pK

pK

tppK

pK XHXJAXX (9.20)

In (9.20):

=

−−− 111 pK

pK

tpXJXJA (9.21)

Page 184: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

165

A necessary condition for the feasibility of solving the estimation problem is that

−− 11 pK

pK

t XJXJ is nonsingular for each iteration step. Convergence is achieved

when, based on (9.17):

ε≤

pK

pK

t XHXH (9.22)

In (9.22), ε is a pre-set tolerance for convergence checking.

9.3.5 Estimation process for subsequent time instants

The formulation and solution method developed in Sections 9.3.3 and 9.3.4 are for the

set of time instants {0, 1, 2, . . . , K}. The procedure is that which starts the estimation

process. It leads to the set of values {x(i), y(i) for i = 0,1,2,. . . , K} which allows the

estimation for the subsequent time instants greater than K to be carried out.

At time instant (K+1), there are N equations derived from (9.5) with n = K+1:

0))(),(),(),1(),1(),1(( =+++ KKKKKK uyxuyxh (9.23)

There are (L – M) unknown (i.e unmeasured) variables in vectors x(K+1) and y(K+1).

The values of all of the elements in vectors x(K) and y(K) are completely known, using

the estimation results at time instant K, combined with the measured data at that time

instant.

As (N – L + M) > 0, there are more equations in (9.23) than unknown variables in

x(K+1) and y(K+1). Therefore, the unconstrained minimisation algorithm developed

and described in (9.17) to (9.22) is applied for solving the estimation problem for time

instant (K+1). The vector function H in (9.17) to (9.22) is now replaced by h in (9.23),

and XK in (9.17) to (9.22) by the vector of unknown variables in x(K+1) and y(K+1).

The above estimation process is then repeated successively for individual time instants

(K + l )’s for l ≥ 2.

Page 185: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

166

9.4 APPLICATION TO SYNCHRONOUS GENERATOR

9.4.1 Generator dynamic model

The fifth-order dynamic model of a 3-phase synchronous generator [163] formed in the

rotor d-q axes is adopted in the present work. On discounting the stator flux transients,

the following stator algebraic equations have been derived [163]:

0))((cos)())((sin)(

))((sin)())((cos)())((

)()()(

)())((cos)())((sin)(

))((sin)())((cos)(

=

⋅+⋅−

⋅+⋅⋅+⋅

+

⋅⋅−

⋅+⋅−

⋅+⋅

ttIttI

ttIttIt

ttt

tttVttV

ttVttV

QD

QD

R

kq

kd

fd

R

QD

QD

δδ

δδω

ψψψ

ωδδ

δδ

ss

s

RL

P

(9.24)

In (9.24):

VD(t), VQ(t): D- and Q- components of stator terminal voltage in the network D-Q axes

ID(t), IQ(t): D- and Q- components of stator current in the network D-Q axes

δ(t): angular separation between the rotor d-axis and the network D-axis (rotor angle)

ωR(t): rotor angular speed

ψfd(t): field winding flux linkage

ψkd(t): d-axis damper winding flux linkage

ψkq(t): q-axis damper winding flux linkage

Matrices Ps, Ls and Rs are defined in the Appendix in terms of generator parameters.

The rotor flux linkage transients are described by the following set of differential

equations [163]:

⋅+⋅−

⋅+⋅⋅

+

+

=

))((cos)())((sin)(

))((sin)())((cos)(

00

)(

)()()(

)()()(

ttIttI

ttIttI

tV

ttt

ttt

QD

QD

fd

kq

kd

fd

kq

kd

fd

δδ

δδ

ψψψ

ψψψ

m

m

F

A

(9.25)

Page 186: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

167

In (9.25), Vfd is the field winding voltage. The coefficient matrices Am and Fm are

formed from generator parameters as shown in the Appendix.

The relationship between the rotor angle and angular speed is expressed in the following

differential equation:

0RR ωωδ −= (9.26)

In (9.26), ωR0 is the nominal rotor angular speed.

9.4.2 Procedure for generator internal state estimation

The generator dynamic model given in (9.24), (9.25) and (9.26) is that of a continuous-

time nonlinear dynamical system of the form described in (9.1) and (9.2), with the

following definition for the state variable vector, x, and non-state variable vector, y:

[ ]t)(),(),(),()( ttttt kqkdfd δψψψ=x (9.27)

[ ]t)(),(),(),(),(),()( tVttItItVtVt fdRQDQD ω=y (9.28)

The total number of variables in vector x is four, and that in vector y is six. The

superscript t in (9.27) and (9.28) denotes vector transpose. The input vector u(t) in this

case is zero. The generator state equation set is that given in (9.25) and (9.26) are

assembled to (9.29), using the definitions in (9.27) and (9.28):

))()()( ttt y,f(xx = (9.29)

where f is the vector function described in (9.25) and (9.26). The total number of

individual differential equations in (9.29) is four. The complete generator model is

described by the state equation in (9.29) and the stator algebraic equation set in (9.24)

which is expressed in the following compact form:

Page 187: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

168

0))()( =tt y,g(x (9.30)

where g is the vector function defined in (9.24). The number of individual algebraic

equations in (9.30) is NG = 2. The total number of individual equations in (9.29) and

(9.30) is N = 6, and the total number of individual variables in x(t) and y(t) is L = 10.

Equations (9.29) and (9.30) are of the general form in (9.1) and (9.2) with u(t) = 0. The

estimation procedure developed in a general form in Section 9.3 applies directly for

estimating the generator internal operating states, using external measurements.

9.4.3 Measurement requirements

As derived from Section 9.3.3, a necessary condition for estimation feasibility is (N – L

+ M) > 0, which means in this case the number of measured variables, M, is to be 5 or

greater. It is not practically feasible to measure the state variables in vector x of (9.27)

which comprises rotor flux linkages and rotor angle. However, the measurements of the

non-state variables in vector y of (9.28) present no difficulty. These six variables are the

generator terminal voltage and current phasors, rotor speed and field winding voltage.

Key aspects relevant to the measurement are discussed in the following:

9.4.3.1 Voltage and current phasor measurements

There are synchronised PMUs at terminals of individual generators operating in the

power system. Phasors representing stator terminal voltage and current at the supply

frequency for each generator are formed in the positive-phase sequence.

The power system waveforms from which the phasors are formed are expressed as

discrete time series with the time origin nominated to be at the time reference n = 0 at

which the estimation process starts. All of the phase angles of the generator terminal

voltage phasors formed by the synchronised PMUs at any sampling time instants are

relative to the time origin as nominated. For the purpose of defining the network D-axis,

the voltage phasor at time n = 0 of the terminal of a generator is selected. The angular

separation between the voltage phasor and the D-axis is the phase angle of the voltage

Page 188: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

169

phasor. In this way, the D-axis is defined, and the Q-axis is that separated from the D-

axis by 900. With the time origin common to all of the power system waveforms, all of

the network voltage and current phasors at any sampling time instants formed from the

waveforms are relative to the network D-Q axes as established.

VD and VQ in Section 9.4.1 are the components of the stator voltage phasor VD + j.VQ in

the positive-phase sequence formed in the network D-Q axes. Similarly, ID and IQ are

the components of stator current phasor ID + j.IQ in the positive-phase sequence formed

in the network D-Q axes.

The measured values for VD, VQ, ID and IQ at successive sampling time instants are used

in the estimation process described in Sections 9.4.1 and 9.4.2.

9.4.3.2 Availability of rotor speed measurements

Rotor speed measurement is required if there is a power system stabiliser (PSS) with a

rotor speed input signal. In general, rotor speed measurement is also required as an

input to the governor system. Therefore, advantage can be taken of the available rotor

speed measurement which is used for PSS and/or governor input in forming the set of

measured variables needed in the estimation procedure of Section 9.4.2.

With six measured variables, the necessary condition identified in Section 9.3.3 for

estimation feasibility is satisfied. On this basis together with the practicality of

measuring the variables in vector y of (9.28), the present work applies the procedure for

estimating generator internal states with measurements of stator voltage and current

phasors, rotor speed and field winding voltage.

9.4.4 Discussion

9.4.4.1 Representing magnetic saturation

When it is required to represent magnetic saturation, the d- and q-axis magnetising

inductances (Lmd and Lmq respectively) in the generator dynamic model of Section 9.4.1

Page 189: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

170

are modified, based on the generator magnetisation characteristic and air-gap flux

linkage [6]. The calculation of air-gap flux linkage, taking into account the contribution

of the negative-phase-sequence component of the stator current in unbalanced

operation, is described in [6]. The estimation procedure in Section 9.4.2 would be

iterated, using modified magnetising inductances at each iteration.

9.4.4.2 Initial values

The Newton-Raphson iterative solution method presented in Section 9.3.4 requires

starting or initial values of the variables. The following scheme for forming the starting

values is proposed, in the context of a synchronous generator. With rotor speed set to its

nominal value or measured value, and discounting the flux linkages in d-axis and q-axis

damper windings, equation (9.24) is solved for the rotor angle for each sampling time

instant.

With rotor angle and speed being known, equations (9.24) and (9.25) are linear in terms

of rotor flux linkages. These linear continuous time-domain equations transform into a

linear equation set in the discrete time domain, using the trapezoidal rule of integration.

For (K+1) sampling time instants, individual linear equation sets are formed, and then

solved simultaneously to give approximate solution values for rotor winding flux

linkages.

With respect to the estimation process based on a minimisation method for subsequent

time instants described in Section 9.3.5, the initial values for the unknown variables at

time instant (K + l) (for l ≥ 1) are set to be equal to the values obtained in the previous

time instant (K + l - 1). The above procedure for forming starting values for the iterative

solution methods based on the Newton-Raphson algorithm and/or minimisation

algorithm does not require any specification or knowledge of the steady-state condition

of the generator and power network.

Extensive case studies have been carried out to verify the effectiveness of the above

procedure for forming the starting values of the variables required in the estimation

procedure of Section 9.4.2. They lead to the convergence of the iterative sequence

Page 190: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

171

derived from the estimation procedure. The results of a representative study are

presented in Section 9.5.

9.5 REPRESENTATIVE CASE STUDY

The estimation procedure developed in Sections 9.3 and 9.4 is applied to generator 9 in

the 10-generator, 39-node New England power system [158] augmented with two

TCSCs (thyristor-controlled series compensators) as shown in Fig.F1 of Appendix F.

The disturbance condition is that of a 3-phase-to-earth fault on transmission line L6 and

close to node N37, with a fault clearing time of 160ms. The estimation starts at 20ms

subsequent to the fault clearance. For the purpose of illustration, the results and

computing time of the following case study by simulation are presented in the following

sections.

9.5.1 Results

In the study, it is taken that the generator terminal voltage and current phasors, rotor

speed and field winding voltage are measured and available at successive sampling time

instants. The set of measurements satisfies the necessary condition as discussed in

Section 9.4.3. In the simulation study, time-domain transient stability analysis with the

time step length of 10ms is carried out to give the values of the generator terminal

voltage and current phasors, rotor speed, and field winding voltage at successive

sampling time instants. The field, d-axis damper and q-axis damper winding flux

linkages and rotor angle are to be estimated.

With time instants denoted by 0 and 1 used for starting the process, there are eight

individual equations in (9.8) and (9.9) (when applied to generator model) with eight

unknowns in rotor flux linkages and rotor angle at time instants 0 and 1. The Newton-

Raphson method described in Section 9.3.4 is applied to solve the estimation problem.

For estimation of each subsequent time instant greater than 1, there are four unknowns

(i.e. rotor flux linkages and rotor angle at each time instant) and six equations. The

unconstrained minimisation method developed in Section 9.3.5 is applied to estimate

the values for the unknown variables.

Page 191: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

172

In Figs. 9.1 to 9.4 are shown the comparisons between the estimated rotor flux linkages

and rotor angle and their actual values obtained from time-domain transient stability

simulation. The comparison indicates that very high accuracy has been achieved with

the estimation procedure. In addition, the procedure has high numerical stability. There

is no evidence of error accumulation throughout the estimation process performed in

many successive sampling time instants. For each estimation, the typical number of

iterations required for convergence is three, with the tolerance of 10-6 pu. The Jacobian

matrix in (9.15) or the A matrix formed in (9.21) is nonsingular in each iteration.

Fig.9.1 Comparison between actual and estimated rotor angle

Fig. 9.2 Comparison between actual and estimated main field flux

(pu on generator rating)

0.4 0.5 0.6 0.7 0.8 0.9 1 1.160

70

80

90

100

110

Time (s)

Rot

or A

ngle

(Deg

)

ActualEstimated

0.4 0.5 0.6 0.7 0.8 0.9 1 1.13.8

3.9

x 10-3

Time (s)

d-a

xis M

ain

Fiel

d Fl

ux (p

u)

ActualEstimated

Page 192: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

173

Fig.9.3 Comparison between actual and estimated d-axis damper winding flux

(pu on generator rating)

Fig.9.4 Comparison between actual and estimated q-axis damper winding flux

(pu on generator rating)

9.5.2 Computing time

With Intel (R) 2-core CPU having E6550 processor, the computing time required for the

estimation of sampling time instants 0 and 1 is about 5ms, and that for each subsequent

time instant is about 3.6ms. This computing time requirement confirms that real-time

0.4 0.5 0.6 0.7 0.8 0.9 1 1.13

3.1

3.2

3.3

3.4

3.5

3.6x 10-3

Time (s)

d-a

xis D

ampe

r Win

ding

Flu

x (p

u)

ActualEstimated

0.4 0.5 0.6 0.7 0.8 0.9 1 1.1-10

-8

-6

-4

-2x 10-4

Time (s)

q-a

xis D

ampe

r Win

ding

Flu

x (p

u)

ActualEstimated

Page 193: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

174

estimation is feasible, even with a relatively high sampling rate of 100Hz (i.e. the

sampling time interval of 10ms) in the measurement system.

9.6 CONCLUSIONS

The chapter has derived in a general form a systematic procedure based on dynamic

modelling and numerical techniques for solving nonlinear equations for estimating the

internal states of a nonlinear dynamical system, using available measurements of

variables external to the system. The procedure is applied to estimate the synchronous

generator rotor flux linkages and rotor angle, using available measurements of generator

terminal voltage and current phasors, rotor speed and field winding voltage. The key

advantages of the procedure developed include:

(i) Initialising the estimation process. The estimation process can be started at any

time instant, without the need of specifying or knowing the steady-state operating

condition.

(ii) Generator dynamic responses. They are taken into account in the procedure where

transients in rotor flux linkages of the field and damper windings, rotor angle and

speed, and field winding voltage are fully represented.

(iii) Accuracy and numerical stability. The high accuracy and numerical stability of

the procedure are verified by many simulation studies, the representative results of

which are presented in the chapter.

(iv) Computing time requirement. The low computing time requirement of the

procedure would allow real-time estimation, even when the measurement system

has a high sampling rate of 100Hz. Drawing on this key advantage, the potential

applications of the estimation procedure include real-time transient-stability

control and monitoring.

Page 194: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

175

Chapter 10

Conclusions and Future Work

10.1 CONCLUSIONS

This chapter brings together and summarises the original contributions and advances

that have been made in the research and presented in the body of the thesis.

A new MPC based TCSC controller is developed for a single-machine-infinite-bus

system. The control scheme is developed using a detailed dynamic model of a

synchronous generator and is coordinated with exciter, prime-mover and governor

controllers. The MPC scheme developed generates a required control output (i.e. TCSC

input reference for series compensation requirement) based on present and future

operating scenarios. The control scheme has shown its effectiveness in restoring rotor

angle stability even when the fault is cleared after critical clearing time. The research

and results presented in the thesis have confirmed that it is possible to form a control

law which is adaptive to power system operating conditions, and effective in improving

or maintaining its transient stability. This is achieved by directly deriving the

relationship between the relative rotor angle and the control variable through

linearisation in individual time horizons, which leads to the objective function to be

minimised for forming successive optimal values for the control variable.

Page 195: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

176

As a further extension of transient stability improvement using FACTS devices, the

single machine problem is extended to a multimachine system. Considering the wide-

area network requirements, an online control coordination scheme is successfully

developed for a multimachine system which is adaptive to the changing power system

topology.

The developed method draws on transient stability time-domain simulations and

constrained optimisation for deriving the control algorithm. Optimal input references for

FACTS device controllers are determined to maximise the transient stability margin,

taking into account communication channel time delays in obtaining the power system

operating state, and computing time required in executing the control algorithm.

An optimisation problem is formulated using available data, predicted performance, and

future system states, including the future possible control actions and solved for

obtaining the optimised value of necessary TCSC reactance. The time required for the

entire process, of getting the data, predicting, formulating the optimisation problem and

solving it, is considered as computation delay. Even after solving the optimisation

problem and getting the output of necessary compensation, there can be delay in

communication in passing on this signal to TCSC and actual insertion of this TCSC

reactance in the system. The computation delay for the controller to check with the

future predicted system state, formulate the optimisation problem with given constraints

and give output as a required value of TCSC reactance compensation to make the

system stable, is considered in terms of the computing time and the form of the

computing constraints. The dynamic performance of the control method together with

its feasibility in terms of computing time requirements is investigated in detail. The

analysis outcome is a set of feasible constraints which are to be satisfied by computer

systems used for online control coordination of FACTS devices with the aim of

enhancing or maintaining power system transient stability following disturbances.

With reference to a 10 generator New England system, having TCSCs, the study

presented in this thesis has indicates that it is feasible, within the current technology, to

implement the control coordination algorithm in real time, using a cluster of processors

operating in parallel. Offline simulation, using realistic timing parameters for the

Page 196: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

177

control scheme, confirms its effectiveness in maintaining transient stability following a

fault disturbance.

The online controller coordination scheme is developed with two scenarios based on the

system model updating strategy. With option 2 the system model used in control law

formation, is updated more frequently to reflect changed system topology considering

the fast dynamics after severe disturbance. With many experimental simulation studies

it is found that the length of the control window can be increased to generate a feasible

control law if the communication and computation delays are greater.

The last part of the thesis gives a procedure for estimation of internal states of

synchronous generator using the available measurements. As in practical scenario, it is

not possible to measure the internal state variables of the synchronous generator and the

only measurements available are voltage, current phasors and speed. Considering all

voltage, current measurements availability, along with rotor speed and field

measurements as complete measurements, a case study is carried out for estimation of

internal variables of rotor fluxes and rotor angle. The simulation results are validated

with a representative machine from 10-generator power system network of the New

England system. The plots of estimation and actual values show the close tracking and

correctness of the method developed. The key features of the developed method is that,

it does not need to start with a steady-state assumption and that it can be started at any

instant, including before or after fault in post-fault scenarios.

10.2 FUTURE WORK

With the foundation of work provided by the new concepts and developments presented

in the thesis, further research is envisaged and outlined in the following sections.

10.2.1 MPC based controller for transient stability using multiple FACTS devices

This proposal is that the MPC algorithm will be further developed for accommodating

various shunt and series types of FACTS devices coordinated together for transient

stability improvement. The proposed method can be converted to general form to handle

Page 197: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

178

multimachine problems of any size. The method can be extended to accommodate

various time delays to make it adaptive to wide-area network requirements for a real-

time controller.

10.2.2 Online control coordination scheme for multiple FACTS devices

This proposal is that of re-hosting the software systems which have been developed in

the thesis on a cluster of high-performance processors, and then carrying out extensive

testing in the real-time environment to provide effective compensations using shunt as

well as series compensations with various FACTS devices. In the testing, real-time

dynamic simulation of power systems will be performed, and provide the interactions

between the FACTS devices and the power systems together with their controllers.

10.2.3 Control coordination for power system stability improvements in case of

loss of communication signal information/data

The online control coordination considered in the thesis is for dynamic mode i.e.

transient state of system operation with an assumption that all information is available

through WAM without any loss of signal. However, in reality, because of the

widespread nature of power system networks, where huge data information is travelling

over many channels through different modes of communication, there are possibilities

of loss of communication signal. In such circumstances the controller will have to

generate a correct control law even with this incomplete input data information. With

high-speed computing facilities and FACTS controllers, it is proposed to investigate the

feasibility of deriving the control law and its implementation for FACTS devices

coordination in a shorter time frame related to system stability even with loss of

communication signal, with the objective of preserving system stability following a

large disturbance.

Page 198: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

179

10.2.4 Estimation of internal state variables for synchronous generator with

incomplete/inaccurate measurements

The case study represented in Chapter 9 shows that using rotor speed, field voltage

measurements and terminal voltage and current phasors, the internal state variables of a

synchronous generator can be estimated accurately. However, it is not possible to have

all the measurements available all the time or the available measurements might have

some measurement errors. To consider such a scenario, the present estimation algorithm

is proposed to be developed for two cases: firstly, with incomplete measurements and

secondly with errors in measurements. In the case of errors in measurement, based on

the standard accuracy and acceptable errors in speed, voltage and current measurements

the robustness of the controller will be verified. In the case of incomplete

measurements, as the number of known inputs is less than the unknown variables to be

calculated from the same set of equations, it is necessary to consider a sufficient number

of time instants to form a complete, feasible set of equations which can be solved.

Depending on the availability of speed or field voltage, the algorithm is divided in two

cases- Case 1 deal with calculation of unknown field voltage along with rotor flux and

rotor angle while case 2 considers a scenario where the field voltage is known and rotor

speed is unknown.

The problem is formed using dynamic equations of rotor flux and rotor angle equations

combined with algebraic equations of system voltages. The procedure for forming this

set of four main equations is the same as explained in Chapter 9 with a case of eight

equations and eight unknowns. However, in this particular case the number of

unknowns is more, due to the equations for the next successive instants of two and three

being added to previous model, making it twenty equations and twenty variables.

Although the main set of equations to be solved remains the same, the problem

formulation procedure for two cases will differ based on unknown variables.

10.2.5 Estimation of internal state variables for exciter, turbine and governor

system

The algorithm developed in the Chapter 9 is very general and can be applied to any

dynamical system at any instant for estimation of internal variables. Considering the

Page 199: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

180

effect and importance of exciter, turbine and governor systems in dynamic studies,

when a complete dynamic model is formed including these state variables, the

developed algorithm can be used for estimation of internal state variables of exciter and

turbine system.

Page 200: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

181

Bibliography

[1] B. Fardanesh, "Future trends in power system control," IEEE Computer Applications in Power, vol. 15, pp. 24-31, 2002.

[2] D. P. Kothari and I. Nagrath, Modern Power System Analysis: Tata Mcgraw Hill, 2003.

[3] "IEEE Guide for Synchronous Generator Modeling Practices and Applications in Power System Stability Analyses," IEEE Std 1110-2002 (Revision of IEEE Std 1110-1991), pp. 0_1-72, 2003.

[4] W. D. Humpage, J. P. Bayne, and K. E. Durrani, "Multinode-power-system dynamic analysis," Proceedings of the Institution of Electrical Engineers, vol. 119, pp. 1167-1175, 1972.

[5] B. Adkins, "Transient theory of synchronous generators connected to power systems," Journal of the Institution of Electrical Engineers, vol. 1951, pp. 244-246, 1951.

[6] P. Kundur, Power System Stability and Control: Tata McGraw-Hill, 2007. [7] "IEEE recommended practice for excitation system models for power

systemstability studies," IEEE Std 421.5-2005 (Revision of IEEE Std 421.5-1992), pp. 0_1-85, 2006.

[8] "Dynamic models for fossil fueled steam units in power system studies," IEEE Trans. Power Systems, vol. 6, pp. 753-761, 1991.

[9] I. C. Report, "Dynamic models for steam and hydro turbines in power system studies," IEEE Trans. Power Apparatus and Systems, vol. PAS-92, pp. 1904-1915, 1973.

[10] P. Saure and M. A. Pai, Power System Dynamics and Control: Prentice Hall, Upper Saddle River, New Jersey 0758, 1998.

[11] J. Machowski, J. W. Bialek, and J. R. Bumby, Power System Dynamics and Stability: John Wiley and sons, 1997.

[12] J. J. Ford, G. Ledwich, and Z. Y. Dong, "Nonlinear control of single-machine-infinite-bus transient stability," in IEEE Power Engineering Society General Meeting, 2006.

[13] J. J. Ford, G. Ledwich, and Z. Y. Dong, "Efficient and robust model predictive control for first swing transient stability of power systems using flexible AC transmission systems devices," IET Generation, Transmission and Distribution, vol. 2, pp. 731-742, 2008.

Page 201: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

182

[14] M. Tavahodi, G. Ledwich, and E. Palmer, "Mixed model predictive control/ energy function control design for power systems," in Proc. Australasian Universities Power Engineering Conference AUPEC 2006, Australia, 2006.

[15] Y. N. Yu, Electrical power System Dynamics-modes and coherent generators Academic Press, INC, 1983.

[16] D. G. Taylor, "Analysis of synchronous machines connected to power-system networks," Proc. of the IEE - Part C: Monographs, vol. 109, pp. 606-610, 1962.

[17] N. G. Hingorani and L. Gyugyi, Understanding FACTS: Concepts and Technology of Flexible AC Transmission Systems: IEEE Press 2000.

[18] L. Kirschner, D. Retzmann, and G. Thumm, "Benefits of FACTS for power system enhancement," in IEEE/PES Transmission and Distribution Conference and Exhibition: Asia and Pacific, 2005, pp. 1-7.

[19] M. Noroozian, N. A. Petersson, B. Thorvaldson, A. B. Nilsson, and C. W. Taylor, "Benefits of SVC and STATCOM for electric utility application," in IEEE PES Transmission and Distribution Conference and Exposition, 2003, pp. 1143-1150 vol.3.

[20] V. L. Nguyen, "Modeling and control cooridnation of power systems with FACTS devices in stead-state operating mode," PhD Thesis Energy Systems Centre, School of Electrical, Electronic and Computer Engineering The University of Western Australia, Perth, 2008.

[21] "Static VAr compensator models for power flow and dynamic performance simulation," IEEE Trans. Power Systems vol. 9, pp. 229-240, 1994.

[22] R. Gianto, "Coordination of power system controllers for optimal damping of electromechanical oscillations," PhD Thesis Energy Systems Centre, School of Electrical, Electronic and Computer Engineering The University of Western Australia, Perth, 2008.

[23] CIGRE, "Modeling of power electronics equipment (FACTS) in load flow and stability programs: a representative guide for power system planning and analysis," 1999 August.

[24] S. Meikandasivam, R. K. Nema, and S. K. Jain, "Behavioral Study of TCSC device- A MATLAB/ Simulink Implementation " Proc. World Academy of Science, Engineering and Technology, vol. 35, pp. 695-700, 2008.

[25] S. Meikandasivam, R. K. Nema, and S. K. Jain, "Performance of installed TCSC projects," in India International Conference, Power Electronics (IICPE),, 2011, pp. 1-8.

[26] B. H. Li, Q. H. Wu, D. R. Turner, P. Y. Wang, and X. X. Zhou, "Modelling of TCSC dynamics for control and analysis of power system stability," International Journal of Electrical Power & Energy Systems, vol. 22, pp. 43-49, 2000.

[27] B. H. Li, Q. H. Wu, P. Y. Wang, and X. X. Zhou, "Influence of the transient process of TCSC and MOV on power system stability," IEEE Trans.Power Systems, vol. 15, pp. 798-803, 2000.

[28] C. Gama, L.Angquist, G.Ingestrom, and M.Noroozian, "Commissioning and Operation Experiance of TCSC for Damping Power Oscillation in the Brazilian North-South Interconnection," 2000.

[29] A. D. Del Rosso, C. A. Canizares, and V. M. Dona, "A study of TCSC controller design for power system stability improvement," IEEE Trans. Power Systems, vol. 18, pp. 1487-1496, 2003.

[30] "Load representation for dynamic performance analysis [of power systems]," IEEE Trans. Power Systems, vol. 8, pp. 472-482, 1993.

Page 202: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

183

[31] "Bibliography on load models for power flow and dynamic performance simulation," IEEE Trans. Power Systems, vol. 10, pp. 523-538, 1995.

[32] "Standard load models for power flow and dynamic performance simulation," IEEE Trans. Power Systems,, vol. 10, pp. 1302-1313, 1995.

[33] I. R. Navarro, "Dynamic Load Models for Power Systems: Estimatiom of time-varying parametrs during normal operation ", Lund University, Sweden, 2002.

[34] B. Chaudhuri and B. C. Pal, "Robust damping of multiple swing modes employing global stabilizing signals with a TCSC," IEEE Trans. Power Systems, vol. 19, pp. 499-506, 2004.

[35] Y.-q. Li, Tenglin, W.-s. Liu, and J.-f. Liu, "The study on real-time transient stability emergency control in power system," in IEEE CCECE Canadian Conference on Electrical and Computer Engineering, 2002, pp. 138-143 vol.1.

[36] K. S. Chandrashekhar and D. J. Hill, "Cutset stability criterion for power systems using a structure-preserving model," International Journal of Electrical Power & Energy Systems, vol. 8, pp. 146-157, 1986.

[37] W. U. Hongxia and G. T. Heydt, "Design of delayed-input wide area power system stabilizer using gain scheduling method," in Proc. IEEE power engineering society general meeting, Toronto, 2003, pp. 1704-9.

[38] M. Amin, "Evolving energy enterprise-"grand challenges": possible road ahead and challenges for R&D," in Power Engineering Society Winter Meeting, 2002. IEEE, 2002, pp. 1456-1458 vol.2.

[39] T. Geyer, M. Larsson, and M. Morari, "Hybrid emergency voltage conrtol in power systems," in The European Control Conference Cambridge, UK, 2003.

[40] M. Begovic, "Trends in power system wide area protection," in Power Systems Conference and Exposition, 2004. IEEE PES, 2004, pp. 1612-1613 vol.3.

[41] M. Begovic, D. Novosel, D. Karlsson, C. Henville, and G. Michel, "Wide-Area Protection and Emergency Control," Proc. IEEE, vol. 93, pp. 876-891, 2005.

[42] Q. Lu, Y. Sun, and S. Mei, Nonlinear Control Systems and Power System Dynamics Kluwer Academic Publishers, 2001.

[43] M. Zima and G. Andersson, "Model Predictive Control Employing Trajectory Sensitivities for Power Systems Applications," in 44th IEEE Conference on Decision and Control and European Control Conference. CDC-ECC '05., 2005, pp. 4452-4456.

[44] J. Arrillaga and N. Watson, Power Systems Electromagnetic Transients Simulation. London, UK: The Institution of Engineering and Technology, Nov. 2002.

[45] O. I. Elgard, Electric Energy Systems Theory: An Introduction McGraw-Hill Book Company 1971.

[46] R. Avila-Rosales and J. Giri, "Wide-area monitoring and control for power system grid security," presented at the 15th PSCC, Liege, 2005.

[47] D. Jiang and X.Lie, "A nonlinear TCSC control strategy for power system stability enhancement " in International conference on advances in power system control, operation and management, APSCOM 2000, Hong Kong, Oct.2000, pp. 576-581.

[48] P. Kundur, J. Paserba, V. Ajjarapu, G. Andersson, A. Bose, C. Canizares, N. Hatziargyriou, D. Hill, A. Stankovic, C. Taylor, T. Van Cutsem, and V. Vittal, "Definition and classification of power system stability IEEE/CIGRE joint task force on stability terms and definitions," IEEE Trans.Power Systems, vol. 19, pp. 1387-1401, 2004.

Page 203: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

184

[49] P. Kundur, J. Paserba, and S. Vitet, "Overview on definition and classification of power system stability," in CIGRE/IEEE PES International Symposium Quality and Security of Electric Power Delivery Systems 2003, pp. 1-4.

[50] "IEEE guide for synchronous generator modeling practices in stability analyses," in IEEE Std 1110-1991, ed, 1991, p. 1.

[51] "Benifits of FACTS for transmission systems " in Siemens H. V. D. Siemens Power Tranmission and Distribution Ed., ed. 91050 Erlangen Germany Siemens 2003.

[52] X. Chu and Y. Liu, "On-line learning applied to power system transient stability prediction " presented at the IEEE International Symposium Circuits and Systems, ISCAS, 2005.

[53] S. Bruno, M. D. Benedictis, and M. L. Scala, "Integrating Dynamic Optimization Methodologies with WAMS technologies," in IEEE Power Engineering Society General Meeting 2007, pp. 1-8.

[54] M. Ishimaru, R. Yokoyama, G. Shirai, and T. Niimura, "Robust thyristor-controlled series capacitor controller design based on linear matrix inequality for a multi-machine power system," International Journal of Electrical Power & Energy Systems, vol. 24, pp. 621-629, 2002.

[55] D. Chatterjee and A. Ghosh, "TCSC control design for transient stability improvement of a multi-machine power system using trajectory sensitivity," Electric Power Systems Research, vol. 77, pp. 470-483, 2007.

[56] D. Chatterjee and A. Ghosh, "Application of Trajectory Sensitivity for the evaluation of the effect of TCSC placement on transient stability," International Journal of Emerging Electrical Power Systems vol. 8 2007.

[57] A. Ghosh, D. Chatterjee, P. Bhandiwad, and M. A. Pai, "Trajectory sensitivity analysis of TCSC compensated power systems," in IEEE Power Engineering Society General Meeting 2004, pp. 1515-1520 Vol.2.

[58] X. Zhou and J. Liang, "Overview of control schemes for TCSC to enhance the stability of power systems," IEE Proc. Generation, Transmission and Distribution, vol. 146, pp. 125-130, 1999.

[59] N. Martins, H. J. C. P. Pinto, and J. J. Paserba, "Using a TCSC for line power scheduling and system oscillation damping-small signal and transient stability studies," in IEEE Power Engineering Society Winter Meeting, 2000, pp. 1455-1461 vol.2.

[60] H. Sun, S. Cheng, and J. Wen, "Dynamic response of TCSC and reactance control method study," presented at the International Conference on Power System Technology 2006.

[61] J. B. Rawlings, "Tutorial: model predictive control technology," in Proc. The American Control Conference, 1999, pp. 662-676 vol.1.

[62] J. B. Rawlings, "Tutorial overview of model predictive control," IEEE Control Systems Magazine, vol. 20, pp. 38-52, 2000.

[63] M. Zima and G. Andersson, "Stability assessment and emergency control method using trajectory sensitivities," in IEEE Bologna Power Tech Conference Proceedings, 2003, p. 7 pp. Vol.2.

[64] M. Zima, P. Korba, and G. Andersson, "Power systems voltage emergency control approach using trajectory sensitivities," in Proc. IEEE Conference on Control Applications 2003, pp. 189-194 vol.1.

[65] M. Larsson and C. Rehtanz, "Predictive frequency stability control based on wide-area phasor measurements," in Power Engineering Society Summer Meeting, 2002 IEEE, 2002, pp. 233-238 vol.1.

Page 204: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

185

[66] C. Xiaodong and L. Yutian, "Transient stability emergency control based on hybrid systems modeling," in Power Engineering Conference, 2007. IPEC 2007. International, 2007, pp. 877-882.

[67] V. Rajkumar and R. R. Mohler, "Nonlinear control methods for power systems: a comparison," IEEE Trans. Control Systems Technology, vol. 3, pp. 231-237, 1995.

[68] Y. Wang, R. R. Mohler, R. Spee, and W. Mittelstadt, "Variable-structure facts controllers for power system transient stability," IEEE Trans. Power Systems, vol. 7, pp. 307-313, 1992.

[69] V. Rajkumar and R. R. Mohler, "Bilinear generalized predictive control using the thyristor-controlled series-capacitor," IEEE Trans. Power Systems, vol. 9, pp. 1987-1993, 1994.

[70] T. Jiang and C. Chen, "A design method of nonlinear optimal predictive controller for thyristor controlled series compensation," Electric Power Systems Research, vol. 76, pp. 753-759, 2006.

[71] E. F. Camacho and C. Bordons, Model Predictive Control, Second ed.: Springer-Verlag London Limited 2004.

[72] J. M. Maciejowski, Predictive Control with Constraints Printice Hall, 2000. [73] W.H.Kwon and S. Han, Receding Horizon Control : Model Precitive Control for

State Models Springer 2005. [74] P. D. Roberts, "A brief overview of model predictive control," in IEE Seminar

on Practical experiences with predictive control (Ref. No. 2000/023), 2000, pp. 1/1-1/3.

[75] S. C. Savulescu, Real-Time Stability in Power Systems: Techniques for Early Detection of the Risk of Blackout: Springer.

[76] R. Zhengyun, Z. Hong, and S. Huihe, "Comparison of PID control and PPI control," in Decision and Control, 2003. Proceedings. 42nd IEEE Conference on, 2003, pp. 133-138 Vol.1.

[77] M. Iancu, M. V. Cristea, and P. S. Agachi, "MPC vs. PID. The advanced control solution for an industrial heat integrated fluid catalytic cracking plant," in Computer Aided Chemical Engineering. vol. Volume 29, M. C. G. E.N. Pistikopoulos and A. C. Kokossis, Eds., ed: Elsevier, 2011, pp. 517-521.

[78] A. Bemporad, Ferrari-Trecate, D. M. G., M. Morari, and T. F, "Model predictive control: Ideas for the next generation," presented at the Proc. European Control Conference, 1999.

[79] A. Bemporad, N. L. Ricker, and J. G. Owen, "Model predictive control - new tools for design and evaluation," in Proc. The American Control Conference, 2004, pp. 5622-5627 vol.6.

[80] J. B. Froisy, "Model predictive control: Past, present and future," ISA Trans., vol. 33, pp. 235-243, 1994.

[81] M. Morari and J. H. Lee, "Model predictive control: past, present and future," Computers & Chemical Engineering, vol. 23, pp. 667-682, 1999.

[82] M. Morari, C. E. Garcia, and D. M. Prett, "Model predictive control: Theory and practice - A survey " Automatica, vol. 25, pp. 335-348, 1989.

[83] D. Q. Mayne, J. B. Rawlings, C. V. Rao, and P. O. M. Scoksert, "Constrained model predictive control: Stability and optimality," Automatica, vol. 36, pp. 789-814, 2000.

[84] R. Findeisen and F. Allgower, "An Introduction to Nonlinear Model Predictive Control " presented at the 21st Benelux Meeting on Systems and Control Veldhoven, 2002.

Page 205: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

186

[85] W.-H. Chen, D. J. Ballance, and P. J. Gawthrop, "Optimal control of nonlinear systems: a predictive control approach," Automatica, vol. 39, pp. 633-641, 2003.

[86] A. Mills, A. Wills, and B. Ninness, "Nonlinear model predictive control of an inverted pendulum," in American Control Conference, 2009. ACC '09., 2009, pp. 2335-2340.

[87] V. Rajkumar and R. R. Mohler, "Nonlinear predictive control for the damping of multimachine power system transients using FACTS devices," in Proc. 33rd IEEE Conference on Decision and Control, 1994, pp. 4074-4079 vol.4.

[88] W. H. Chen, D. J. Ballance, and J. O'Reilly, "Model predictive control of nonlinear systems: computational burden and stability," IEE Proc. Control Theory and Applications, vol. 147, pp. 387-394, 2000.

[89] A. A. Tyagunov, "High-Performance Model Predictive Control for Process Industry," PhD, Technische Universiteit Eindhoven, Rusland, 2004.

[90] E. Camponogara, D. Jia, B. H. Krogh, and S. Talukdar, "Distributed model predictive control," IEEE Control Systems Magazine, vol. 22, pp. 44-52, 2002.

[91] A. N. Venkat, I. A. Hiskens, J. B. Rawlings, and S. J. Wright, "Distributed Output Feedback MPC for Power System Control," in 45th IEEE Conference on Decision and Control, 2006, pp. 4038-4045.

[92] M. Pavella, "Generalized one-machine equivalents in transient stability studies," IEEE Power Engineering Review, vol. 18, pp. 50-52, 1998.

[93] M. Pavella, D. Ernst, and D. Ruiz-Vega, Transient stability of power systems: A Unified approach to assessment and control Boston/Dordrecht/London: Kluwer Academic, 2000.

[94] R. Zarate-Minano, T. Van Cutsem, F. Milano, and A. J. Conejo, "Securing transient stability using time-domain simulations within an optimal power flow," IEEE Trans.Power Systems, vol. 25, pp. 243-253, 2010.

[95] E. W. Kimbark, "Improvement of system stability by switched series capacitors," IEEE Trans. Power Apparatus and Systems, vol. PAS-85, pp. 180-188, 1966.

[96] D. Chatterjee and A. Ghosh, "Evaluation of transient stability margin of a power system containing multiple FACTS devices," in IEEE Power Engineering Society General Meeting, , June 2005, pp. 1787-1794.

[97] P. Korba, M. Larsson, B. Chaudhuri, B. Pal, R. Majumder, R. Sadikovic, and G. Andersson, "Towards real-time implementation of adaptive damping controllers for FACTS devices," in IEEE Power Engineering Society General Meeting, 2007, pp. 1-6.

[98] H. Xin, D. Can, Y. Li, T. S. Chung, and J. Qiu, "Transient stability preventive control and optimization via power system stability region analysis," in IEEE Power Engineering Society General Meeting, 2006, p. 8.

[99] L. Angquist, G. Ingestrom, and H.-A. Jonsson, "Dynamic performance of TCSC schemes," CIGRE 1996:14-302.

[100] G. C. Goodwin, M. M. seron, and J. A. D. Dona, Constrained control and estimation: an optimization approach: Springer, London, 2004.

[101] G. Shackshaft, "General-purpose turbo-alternator model," Proc. The Institution of Electrical Engineers, , vol. 110, pp. 703-713, 1963.

[102] M. Zima, M. Larsson, P. Korba, C. Rehtanz, and G. Andersson, "Design aspects for wide-area monitoring and control systems," Proc. The IEEE, vol. 93, pp. 980-996, 2005.

[103] M. Larsson, P. Korba, and M. Zima, "Implementation and applications of wide-area monitoring systems," in IEEE Power Engineering Society General Meeting, 2007, pp. 1-6.

Page 206: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

187

[104] M. G. Adamiak, A. P. Apostolov, M. M. Begovic, C. F. Henville, K. E. Martin, G. L. Michel, A. G. Phadke, and J. S. Thorp, "Wide area protection-technology and infrastructures," IEEE Trans. Power Delivery, vol. 21, pp. 601-609, 2006.

[105] D. H. Wilson, "Wide-area measurement and control for dynamic stability," in IEEE Power Engineering Society General Meeting, 2007, pp. 1-5.

[106] J. Bertsch, M. Zima, A. Suranyi, C. Carnal, and C. Rehtanz, "Experiences with and perspectives of the system for wide area monitoring of power systems," in CIGRE/IEEE PES International Symposium Quality and Security of Electric Power Delivery Systems, 2003, pp. 5-9.

[107] J. Quanyuan, Z. Zhenyu, and C. Yijia, "Wide-area TCSC controller design in consideration of feedback signals' time delays," in Power Engineering Society General Meeting, 2005, pp. 1676- 1680

[108] B. Naduvathuparambil, M. C. Valenti, and A. Feliachi, "Communication delays

in wide area measurement systems," in Proc. The Thirty-Fourth Southeastern Symposium on System Theory, 2002, pp. 118-122.

[109] I. Kamwa, R. Grondin, and Y. Hebert, "Wide-area measurement based stabilizing control of large power systems-a decentralized/hierarchical approach," IEEE Trans. Power Systems, vol. 16, 2001.

[110] A. Rubaai and F. E. Villaseca, "Transient stability hierarchical control in multimachine power systems," IEEE Trans. Power Systems, vol. 4, pp. 1438-1444, 1989.

[111] F. Okou, L. A. Dessaint, and O. Akhrif, "Power systems stability enhancement using a wide-area signals based hierarchical controller," IEEE Trans. Power Systems, vol. 20, pp. 1465-1477, 2005.

[112] C. W. Taylor, D. C. Erickson, K. E. Martin, R. E. Wilson, and V. Venkatasubramanian, "WACS-wide-area stability and voltage control system: R&D and online demonstration," Proc. The IEEE, vol. 93, pp. 892-906, 2005.

[113] E. A. Leonidaki, G. A. Manos, and N. D. Hatziargyriou, "An effective method to locate series compensation for voltage stability enhancement," Electric Power Systems Research, vol. 74, pp. 73-81, 2005.

[114] L. Rouco and F. L. Pagola, "An eigenvalue sensitivity approach to location and controller design of controllable series capacitors for damping power system oscillations," IEEE Trans. Power Systems, vol. 12, pp. 1660-1666, 1997.

[115] P. Korba, "Real-time monitoring of electromechanical oscillations in power systems: first findings," IET Generation, Transmission & Distribution, vol. 1, pp. 80-88, 2007.

[116] P. Korba, M. Larsson, and C. Rehtanz, "Detection of oscillations in power systems using Kalman filtering techniques," in Proc. IEEE Conference on Control Applications CCA 2003, pp. 183-188 vol.1.

[117] N. Mithulananthan, C. A. Canizares, J. Reeve, and G. J. Rogers, "Comparison of PSS, SVC, and STATCOM controllers for damping power system oscillations," IEEE Trans. Power Systems, vol. 18, pp. 786-792, 2003.

[118] M. J. Gibbard, D. J. Vowles, and P. Pourbeik, "Interactions between, and effectiveness of, power system stabilizers and FACTS device stabilizers in multimachine systems," IEEE Trans. Power Systems, vol. 15, pp. 748-755, 2000.

[119] M. A. Abido, "Pole placement technique for PSS and TCSC-based stabilizer design using simulated annealing," International Journal of Electrical Power & Energy Systems, vol. 22, pp. 543-554, 2000.

Page 207: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

188

[120] L. Rouco, "Coordinated design of multiple controllers for damping power system oscillations," International Journal of Electrical Power & Energy Systems, vol. 23, pp. 517-530, 2001.

[121] X. Y. Li, "Nonlinear controller design of thyristor controlled series compensation for damping inter-area power oscillation," Electric Power Systems Research, vol. 76, pp. 1040-1046, 2006.

[122] M. Ghandhari, G. Andersson, M. Pavella, and D. Ernst, "A control strategy for controllable series capacitor in electric power systems," Automatica, vol. 37, pp. 1575-1583, 2001.

[123] R. Majumder, B. C. Pal, C. Dufour, and P. Korba, "Design and real-time implementation of robust FACTS controller for damping inter-area oscillation," IEEE Trans. Power Systems, vol. 21, pp. 809-816, 2006.

[124] J. H. Chow, J. J. Sanchez-Gasca, R. Haoxing, and W. Shaopeng, "Power system damping controller design-using multiple input signals," IEEE Trans. Control Systems, vol. 20, pp. 82-90, 2000.

[125] S. M. Bamasak and M. A. Abido, "Robust coordinated design of PSS & statcom controllers for damping power system oscillation," presented at the 15th PSCC, Liege, Belgum, 2005.

[126] L. D. Colvara, E. B. Festraits, and S.C.B.Araujo, "Effect of FACTS devices on power system perofrmance in view of small-signal or great disturbances," IEEE Bologna Power Tech Conf., June 23-26 2003.

[127] K. R. Padiyar and K. Uma Rao, "Discrete control of TCSC for stability improvement in power systems," in Proc. The 4th IEEE Conference on Control Applications,, 1995, pp. 246-251.

[128] K. R. Padiyar and K. U. Rao, "Discrete control of series compensation for stability improvement in power systems," Elsevier Science , Electrical Power and Energy systems vol. 19, pp. 311-319, 1997.

[129] A. Ghosh and D. Chatterjee, "Transient stability assessment of power systems containing series and shunt compensators," IEEE Trans. Power Systems, vol. 22, pp. 1210-1220, 2007.

[130] T. Athay, R. Podmore, and S. Virmani, "A practical method for the direct analysis of transient stability," IEEE Trans. Power Apparatus and Systems, vol. PAS-98, pp. 573-584, 1979.

[131] Y. Guo, D. J. Hill, and Y. Wang, "Global transient stability and voltage regulation for power systems," IEEE Trans. Power Systems, vol. 16, pp. 678-688, 2001.

[132] W. Zhang, Y. Li, X. Gu, B. Zhao, and Y. Zhang, "A new fast method of assessing transient stability," in International Conference on Intelligent System Design and Engineering Application (ISDEA), 2010, pp. 827-832.

[133] Y. Yare and G. K. Venayagamoorthy, "Real-time transient stability assessment of a power system during energy generation shortfall," in Innovative Smart Grid Technologies (ISGT), 2010, 2010, pp. 1-9.

[134] P. Kundur, "Techniques for power system stability limit search," IEEE (PES) Catalog Number 99TP138, 1999.

[135] J. Hauer, D. Trudnowski, G. Rogers, B. Mittelstadt, W. Litzenberger, and J. Johnson, "Keeping an eye on power system dynamics," IEEE Computer Applications in Power, vol. 10, pp. 50-54, 1997.

[136] H. Wu, H. Ni, and H. G.T., "The impact of time delay on robust control design in power systems," in IEEE Power Engineering Society Winter Meeting 2002, pp. 1511 - 1516.

Page 208: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

189

[137] R. Majumder, B. Chaudhuri, and B. C. Pal, "Implementation and test results of a wide-area measurement-based controller for damping interarea oscillations considering signal-transmission delay," IET Generation, Transmission & Distribution, vol. 1, pp. 1-7, 2007.

[138] H. Wu, K. S. Tsakalis, and G. T. Heydt, "Evaluation of time delay effects to wide-area power system stabilizer design," IEEE Trans. Power Systems, vol. 19, pp. 1935 - 1941, 2004.

[139] G. L. Yu, B. H. Zhang, H. Xie, and C. G. Wang, "Wide-area measurement-based nonlinear robust control of power system considering signals' delay and incompleteness," in IEEE Power Engineering Society General Meeting, 2007, pp. 1-8.

[140] S. Mei, J. Chen, Q. Lu, A. Yokoyama, and M. Goto, "Coordinated nonlinear robust conrtol of TCSC and excitation for multi-machine systems," Journal of Control Theory and Applications, vol. 2, pp. 35-42, 2004.

[141] B. Chaudhuri, R. Majumder, and B. C. Pal, "Wide-area measurement-based stabilizing control of power system considering signal transmission delay," IEEE Trans. Power Systems, vol. 19, pp. 1971-1979, 2004.

[142] H. Jia, X. Yu, Y. Yu, and C. Wang, "Power system small signal stability region with time delay," International Journal of Electrical Power & Energy Systems, vol. 30, pp. 16-22, 2008.

[143] H. A. Othman and L. Angquist, "Analytical modeling of thyristor-controlled series capacitors for SSR studies," IEEE Trans. Power Systems, vol. 11, pp. 119-127, 1996.

[144] P. C. Srivastava, A. Ghosh, and S. V. Jayaram Kumar, "Model-based control design of a TCSC-compensated power system," International Journal of Electrical Power & Energy Systems, vol. 21, pp. 299-307, 1999.

[145] I. A. Hiskens, "Time-delay modelling for multi-layer power systems," in International Symposium on Circuits and Systems, 2003, pp. 316-319.

[146] H. Xin, D. Gan, and J. Qiu, "Transient stability preventive control and optimization," in IEEE PES Power Systems Conference and Exposition, PSCE '06, 2006, pp. 428-435.

[147] M. Ali, Z. Y. Dong, and P. Zhang, "Adoptability of grid computing technology in power systems analysis, operations and control," IET Generation, Transmission & Distribution, vol. 3, pp. 949-959, 2009.

[148] M. H. Haque, "Improvement of first swing stability limit by utilizing full benefit of shunt FACTS devices," IEEE Tran. Power Systems, vol. 19, pp. 1894-1902, 2004.

[149] T. T. Nguyen and R. Gianto, "Neural networks for adaptive control coordination of PSSs and FACTS devices in multimachine power system," IET Generation, Transmission & Distribution, vol. 2, pp. 355-372, 2008.

[150] D. Ruiz-Vega and M. Pavella, "A comprehensive approach to transient stability control. I. Near optimal preventive control," IEEE Trans. Power Systems, vol. 18, pp. 1446-1453, 2003.

[151] T. T. Nguyen, V. L. Nguyen, and A. Karimishad, "Transient stability-constrained optimal power flow for online dispatch and nodal price evaluation in power systems with flexible AC transmission system devices," Generation, Transmission & Distribution, IET, vol. 5, pp. 332-346, 2011.

[152] M. La Scala, M. Trovato, and C. Antonelli, "On-line dynamic preventive control: an algorithm for transient security dispatch," IEEE Trans. Power Systems, vol. 13, pp. 601-610, 1998.

Page 209: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________

190

[153] D. Ernst, A. Bettiol, Y. Zhang, L. Wehenkel, and M. Pavella, "Real time transient stability emergency control of the south-southeast Brazilian system," in The Proc. VI Symposium of Specialists in Electric Operational and Expansion Planning (SEPOPE 1998), Salvador, Brazil, May 1998, p. 9.

[154] D. Q. Mayne and E. Polak, "Feasible directions algorithms for optimization problems with equality and inequality constraints," Mathematical programming 11, pp. 67-80, 1976.

[155] E. R. Panier and A. L. Tits, "On combining feasibility, descent and superlinear convergence in inequality constrained optimization," Mathematical Programming 59, pp. 261-276, 1993.

[156] T. T. Nguyen and X. J. Li, "Application of a z-transform signal model and median filtering for power system frequency and phasor measurements," IET Generation, Transmission & Distribution, vol. 1, pp. 72-79, 2007.

[157] V. Jalili-Marandi and V. Dinavahi, "SIMD-based large-scale transient stability simulation on the graphics processing unit," IEEE Trans. Power Systems, vol. 25, pp. 1589-1599, 2010.

[158] M. A. Pai, Energy Function Analysis for Power System Stability Kluwer Academics Publishers, Boston, 1989.

[159] J. W. Chapman, M. D. Ilic, C. A. King, L. Eng, and H. Kaufman, "Stabilizing a multimachine power system via decentralized feedback linearizing excitation control," IEEE Trans. Power Systems, , vol. 8, pp. 830-839, 1993.

[160] V. Venkatasubramanian and R. G. Kavasseri, "Direct computation of generator internal dynamic states from terminal measurements," in Proc. The 37th Annual Hawaii International Conference on System Sciences, 2004, p. 6

[161] A. Del Angel, P. Geurts, D. Ernst, M. Glavic, and L. Wehenkel, "Estimation of rotor angles of synchronous machines using artificial neural networks and local PMU-based quantities," Neurocomputing, vol. 70, pp. 2668-2678, 2007.

[162] P. Tripathy, S. C. Srivastava, and S. N. Singh, "A Divide-by-difference-filter based algorithm for estimation of generator rotor angle utilizing synchrophasor measurements," IEEE Trans. Instrumentation and Measurement, vol. 59, pp. 1562-1570, 2010.

[163] W. D. Humpage, "Structure for multinode-power-system dynamic-analysis methods," Proc. The Institution of Electrical Engineers, vol. 120, pp. 853-859, 1973.

Page 210: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.1

Appendix A

Generator Rotor Dynamics

A.1 GENERATOR STATOR VOLTAGE EQUATIONS

Assuming a synchronous machine with one main field winding, one direct-axis and one

quadrature-axis damper winding, the total voltage and current vector of synchronous

machine v and current vector i can be given as follows:

=

=r

s

kq

kd

fd

q

d

vv

vvvvv

v ;

=

=r

s

kq

kd

fd

q

d

ii

iiiii

i

(A1)

The relation between this voltage vector and current vector can be given by

RiiGdtdiLv r ++= ω (A2)

where, the coefficient matrices L, G, and R are as follows in voltage equation v in

terms of current i

Page 211: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.2

=

−−−

=rrrs

srss

kqkdfdqd

kqmq

kdmdmd

mdfdmd

mqq

mdmdd

s

s

LLLL

LLLLLLLL

LLLLL

kqkdfdqd

L

ss

0000000

00000

[ ]srss

kkfqd

mqmdd

mqq

s

s GGLLL

LLqd

G

qddss

=

−=

00000

(A3)

=

=rr

ss

kkfqd

kq

kd

fd

a

a

q

d

d

s

s

RR

RR

RR

R

kkfqd

R

qddss

00

00000000000000000000

Equation (A1) can be expressed explicitly in complete matrix form as

×

+

×

−−−

=

kq

kd

fd

q

d

kq

kd

fd

a

a

rmqmdd

mqq

kq

kd

fd

q

d

kqmq

kdmdmd

mdfdmd

mqq

mdmdd

fd

q

d

iiiii

RR

RR

R

LLLLL

pipipipipi

LLLLLLLL

LLLLL

vvv

ω00

000

0000000

00000

00

Partitioning (A4) in terms of stator and rotor circuit equations

(A4)

[ ]

+

+

=

r

s

r

s

r

ssrssr

r

s

rrrs

srss

r

sii

RR

ii

GGpipi

LLLL

vv

00

ω

(A5)

ssrsrrsssrrsrssss iRiGiGpiLpiLv ++++= ωω

(A6)

rrrrrsrsr iRpiLpiLv ++=

(A7)

Using flux linkages expression

Page 212: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.3

=

r

s

rrrs

srss

r

sii

LLLL

ψψ

(A8)

rsrssss iLiL +=ψ (A9)

and

rrrsrsr iLiL +=ψ (A10)

Substituting (A9) in (A6) for rsrssss piLpiLp +=ψ and (A10) into (A7)

[ ] ssrsrrsrssrrrrrsrss iRLLGGLGpv ×+−++= −− 11 ωψωψ (A11)

At this stage, advantage can often be taken, in practical stability studies based on the

synchronously rotating reference frame, of reducing the computing time expanded in

analysis. This is achieved by eliminating the terms of stator-voltage transients

corresponding to the rate of change of stator flux linkages with respect to time. These

terms contribute very little dynamic response analyses that are dominated by the inertial

characteristics of rotating machines and they may be readily eliminated. Hence,

eliminating stator transients and simplifying it using:

1−= rrsrrm LGp ω ; ( )[ ]srsrrsrssrm RLLGGz +−−= −1ω (A12)

smrms izpv −= ψ (A13)

Using notations of (A12) and expanding (A13) to separate d-axis and q-axis voltage

equations as:

qmdmkqmkdmfdmd izizpppv 1211131211 −−++= ψψψ (A14)

qmdmkqmkdmfdmq izizpppv 2221232221 −−++= ψψψ (A15)

However, examining the structure of matrices pm and zm as given in (A17) - (A18),

taking advantage of sparsity, vd and vq equations can be further simplified as (A19-

A20).

Page 213: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.4

−+

−−−

−−

=

q

d

a

a

mq

md

md

kq

kdmd

mdfd

mdmd

mq

d

qr

kq

kd

fd

kq

kdmd

mdfd

mdd

qr

q

d

ii

RR

LLL

LLLLL

LLL

LL

LLLLL

LLL

vv

00

000

0000

000

00

0000

000

1

1

ω

ψψψ

ω (A16)

where,

1

0000

000

=

kq

kdmd

mdfd

mdd

qrm

LLLLL

LLL

p ω (A17)

−+

−−−

−−

=

a

a

mq

md

md

kq

kdmd

mdfd

mdmd

mq

d

qrm R

R

LLL

LLLLL

LLL

LL

z0

0

000

0000

000

00

1

ω

(A18) as pm11 = 0; pm12 = 0; pm23 =0, eliminating these three terms from (A14) and (A15), final

expression for stator voltages in terms of d-axis and q-axis can be obtained as:

qmdmkqmd izizpv 121113 −−= ψ (A19)

qmdmkdmfdmq izizppv 22212221 −−+= ψψ (A20)

A.2 GENERATOR ROTOR FLUX EQUATIONS

Using (A11) rrrsrsr piLpiLp +=ψ and ( )srsrrrr iLLi −= − ψ1 (A21)

simplifying (A7) with (A21) will lead to

[ ]sssrrrrrr iLLRpv −+= − ψψ 1 (A22) Rearranging with 1−−= rrrm LRA and rsrrrm LLRF 1−−= rotor flux equations for one

main field winding and two damper windings, one on d- axis and one on q-axis can be

given as

rsmrmr viFAp ++= ψψ (A23)

Page 214: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.5

which can be explicitly written for expressing rotor dynamics as follows:

fdmqmdmkqmkdmfdmfd EKiFiFAAA 111211131211 +++++=•

ψψψψ (A24)

qmdmkqmkdmfdmkd iFiFAAA 2221232221 ++++=•

ψψψψ (A25)

qmdmkqmkdmfdmkq iFiFAAA 3231333231 ++++=•

ψψψψ (A26)

A.3 ELECTROMAGNETIC TORQUE EXPRESSION

[ ]

=

r

ssrsstrse I

IGGIIT

00refω (A27)

[ ] [ ]

=

r

ssrss

ts I

IGGIrefω (A28)

rsrtssss

ts IGIIGI refref ωω += (A29)

Substituting for ( )srsrrrr iLLi −= − ψ1 from (A15)

( )( )srsrrrsrtssss

ts iLLGIIGI −+= − ψωω 1

refref (A30)

[ ] rsrtssrsrrsrss

ts GIILLGGI ψωω ref

1ref +−= − (A31)

Let, rsrrsrss LLGGAA 1−−= and rsrGBB ψ=

BBIIAAIT tss

tse refref ωω += (A32)

Page 215: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.6

Appendix B

Exciter and Prime-Mover

Modelling

B.1 EXCITER AND AUTOMATIC VOLTAGE REGULATOR MODELLING

Assuming a synchronous machine with one main field winding, one direct-axis and one

quadrature-axis damper winding, the total voltage and current vector of synchronous

machine v and current vector i can be given as follows:

The complete exciter system consists of five major components:

(i) Main exciter: being a power stage of the excitation system, exciter provides dc power

to the synchronous machine field winding;

(ii) Regulator: processes and amplifies input control signals to a level and form

appropriate for control of the exciter;

(iii) Terminal voltage transducer: This senses generated terminal voltage, rectifies and

filters it to dc quantity and compares it with a desired terminal voltage set as

reference;

(iv) Power system stabilizer: PSS provides additional input signal to the regulator for

damping power system oscillations. The input given to PSS can be rotor speed

deviation, accelerating power, and frequency deviation;

Page 216: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.7

(v) Limiters and protective circuits: these include a wide array of control and protective

functions which ensures that the capability limits of exciter and synchronous

generator are not exceeded;

Generator Exciter Excitation Controller Regulator

Terminal voltage transducer and

load compensator

Power system stabilizer

Vref

VUEL

VOEL

VR

Vs

Ifd

Efd

|Vt|It

Fig.B1: Functional block diagram for synchronous machine exciter control system

Fig.B1 summarises the main blocks for synchronous machine exciter controller

system. VUEL and VOEL are the upper and lower exciter limits while Vt and It is the

output voltage and current which is feedback to control and adjust the regulator and

exciter output.

In the present research, a standard IEEE Type I exciter is used as shown in Fig B2.

Practically, Efd is the output of the exciter which is input to the generator as shown

in Fig B1. However, it should be noted that independent selection of per unit system

is necessary for modelling the excitation system and for proper interfacing between

the low voltage exciter system and high voltage generator circuit. The exciter output

Efd and generator input Vfd are related to each other by a factor of Km11 (= Rfd/Lmd).

RV

FV

F

F

sTsK+1

( )fdE EfS =

fdE+

+ +

−maxRV

minRV

refV

tV −

+

PSSV

∑ ∑ ∑EE sTK +

1

A

A

sTK+1

Fig.B2: IEEE Type I Excitation system model

Page 217: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.8

The dynamic equations describing the excitation control system can be derived from the

above block diagram.

( )( ) RfdfdEEE

fd VEESKT

E ++−=• 1

(B1)

( )

−+−+−=

•trefAfd

F

FAfAR

AR vVKE

TKK

RKVT

V 1

maxminRRR VVV ≤≤

(B2)

+−=

fdF

Ff

Ff E

TKR

TR 1

(B3)

where, FVETKR fd

F

Ff −= and

F

Ffd T

sKEV+

=1F

and, Efd is known as exciter field voltage, VR as automatic voltage regulator output and

Rf as rate feedback. The meaning and explanation of all other symbols is as explained in

the list of symbols.

Simplifying the block diagram of Fig.B2 and with the meaning of other symbols

explained in the list of symbols, the set of above dynamic equations can be rewritten in

compact notation as:

ExcExcExcExcExc VBXAX +=•

(B4)

where,

=

f

f

ExcRVE

X R

d;

−−

+−

=

FFF

AAFA

EEE

TTTK

TK

TTTKK

TTSK

A

10

1

01

F

AF

EE

Exc

(B5)

=

0

0A

ExcAT

KB ; [ ]tVVV −= refExc

(B6)

Page 218: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.9

B.2 PRIME-MOVER AND GOVERNOR COMBINED SYSTEM

In the present research, a steam turbine with nonreheat type is chosen as the prime

mover. Figure B3 shows the block diagram representation of turbine and governor

combined model.

dR1

CHsT+11

+

-PSV

Psv(max)

Psv(min)

PC

ω(pu) TM∑ SVsT+1

1

Fig.B3: Turbine and governor model

The steam chest dynamics and the effect of the steam valve position (PSV) on the

synchronous machine torque (TM) are of major concern from transient stability

modeling point of view. Equations (B7)-(B9) gives a complete model of turbine and

governor system which can be represented in compact notations as shown in (B10)

( )SVM

CHM PT

TT +−=• 1

(B7)

(B8)

( )( )puSVCSV PPP ω−−=•

where, ( )ref

refωωω

ω−

=pu (B 9)

CGovGovGovGovGovGov PDCBXAX +++=•

ωω ref (B10)

where,

=

SV

MGov P

TX and AGov, BGov, CGov, and DGov are matrices dependent on

system design for a given gain and time constants of controller.

Page 219: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.10

B.2 POWER SYSTEM STABILIZER (PSS)

The PSS is one the most commonly used controllers for damping rotor oscillations.

Using rotor speed as input denoted by X in the PSS block diagram given in Fig. B4, the

equations can be derived from three major blocks as:

From block 1:

XKXT

X PSSPSSPSS

PSS += 111 (B11)

From block 2:

PSSPSS

PSSPSS

PSSPSS

PSSPSS X

TTX

TX

TX 1

2

12

21

22

11 +−= (B12)

From block 3:

PSSPSS

PSSPSS

PSSPSS

PSSPSS X

TTV

TX

TV 2

4

3

42

4

11 +−= (B13)

which can be simplified and represented in final form as:

KPSS

Vpss(min)

X2pss

pss

psssTsT

2

111+

+

pss

pssT

sT+1 pss

psssTsT

4

311+

+X1pss

Vpss(max)

Vpss⋅

X

Fig.B4: PSS schematic block diagram

X

TTTTTK

TTK

K

VXX

TTTTT

TTTT

TT

TTTTT

T

VXX

PSSPSS

PSSPSSPSSPSSPSS

PSSPSSPSS

PSS

PSS

PSS

PSSPSSPSS

PSSPSS

PSSPSS

PSSPSS

PSS

PSS

PSSPSSPSS

PSSPSSPSS

PSS

PSS

PSS

⋅⋅⋅⋅

⋅+

−⋅−

⋅−

−⋅−

=

42

4312

12

1

42

32

2

1

4

3

22

12

1

1

01

001

(B14)

The complete PSS model can be represented in final compact form as:

XBXAX PSSPSSPSSPSS += (B15)

where, XPSS is the vector of state variables of PSS and, APSS, BPSS are matrices elements

depending on the gain and time constant of the PSS controller while X is the speed

variation.

Page 220: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.11

Appendix C

Dynamic modelling of FACTS

devices

C.1 SVC DYNAMIC MODEL

With reference to the block diagram presented in Fig. 2.5

From block 1:

( ) s

s

SDCtref sTK

XVVX

+=

−− 11

(C1)

( )( )SDCtrefs

sXVVKX

TX −−+−= 11

1 (C2)

From block 2:

( ) s

s

SDCtref sTK

XVVX

+=

−− 11

(C3)

( )( )SDCtrefs

sXVVKX

TX −−+−= 11

1 (C4)

2

1

1 11

sTsT

XB

++

= (C5)

Page 221: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.12

( )111

2

1 XTXBT

B ⋅++−= (C6)

With proper substitution of (C2) in (C4) and simplifying, the final SVC dynamic model

can be represented in compact matrix form as:

ref

s

ss

s

t

s

ss

s

SDC

s

ss

s

s

ss V

TTTK

TK

V

TTTK

TK

X

TTTK

TK

BX

TTTTT

TBX

−+⋅

−+⋅

−+

−−

−=

2

1

2

1

2

11

22

11

1

01

(C7)

The complete model can be expressed in compact notation as:

refSVCtSVCSDCSVCSVCSVCSVC VDVCXBXAX ⋅+⋅+⋅+⋅= (C8)

C.2 DYNAMIC MODELLING OF TCSC

Iline

Inductive limits

Iline

Xmin

Iline Capacitive limits

Iline

Xmax

∑ csc

csc1 t

tsT

K+

Xreftcsc

-

+

XSDC

Xtcsc

Xtcsc(max)

Xtcsc(min)

Xtcsc(min)

Fig.C1: TCSC dynamic model

( ) tcscSDCreftcsc 1

1 XsT

XXc=

+− (C9)

etttt PBXAX csc1csc1csc11csc1 +=

(C10)

Page 222: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.13

ettttt PtBXAXAX csc2csc2csc22csc1csc21csc2 ++=

(C11)

etttttttt PBXAXAXAX csc3cscSDCcsc33csc2csc32csc1csc31cscSDC +++= (C12)

( )cscSDCreftcsctcsc

1t

cXXX

TX −+=⋅

(C13)

Dynamic reactance limits

With reference to the Fig.C1 for transient reactance limits, the TCSC model permits

operation anywhere within the enclosed region. These boundaries are due to a number

of constraints on both the capacitive as well as the inductive side as explained in

Chapter 2.

In the capacitive region, the constraints are due to

(a) limit on the firing angle, expressed as a constant reactance limit (Xmax0);

(b) limit on a voltage across the TCSC;

(c) limit on the line current (ILtran) at which point the TCSC will go into a protective

bypass mode.

XTCSC

Inductive-2

12

Xmin0

Xmax0Capacitive

Xmin VL

Xmax VC

Iline=1.0puXmax ILine

ILine

Thyristor Bypass

3

XTCSC=1.0pu

Xmin ILT

Xbypass

Fig.C6: Dynamic reactance Limits

Once the TCSC is bypassed on this overcurrent constraint, it is subject to a time delay

on reinsertion after line current falls back below ILtran. In a multi-module TCSC, it is

possible that only some of the modules will bypass, since one module has to stay in

Page 223: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.14

capacitive mode. For simplicity in typical stability studies, it is suggested that this

nuance be neglected. The final capacitive reactance limit is the minimum of these

individual constraints.

On the inductive side similar constraints apply:

(a) limit on the firing angle, expressed as a constant reactance limit (Xmin0)

(b) limit on the harmonics, approximated as a constant voltage across the TCSC;

(c) limit on the thyristor current: As an approximation, the fundamental component of

thyristor current is limited to that at which the TCSC can operate in thyristor bypass

for duration of the transient.

C.3 DYNAMIC MODELLING OF STATCOM

The first state equation of STATCOM is started from the capacitor voltage on dc side and given by:

dc

dcdc I

CV 1

=

(C14) The current Idc can be determined using STATCOM active power flow equation as:

dcdcdcC IVPP ⋅== (C15)

Where, ( )*stastaC IVReP ⋅= and (C16)

φstadc VAV = (C17)

SDCstaCqstaTstaTrefsta XEIDVCVBV +++= (C18)

φstaSDCstaTstaTrefstadcstastaC VLXKVJVHVGVFX +++++= (C19)

φNXMφ staCsta += (C20)

Page 224: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.15

C.4 SUPPLEMENTARY DAMPING CONTROLLER (SDC)

S D C

S D C1 sT

sT+ SD C2

SD C111

sTsT

+

+

SD C4

SD C311

sTsT

+

+PeSDC1K XSDCX1SDC X2SDC

XSDC(max)

XSDC(min)

Fig.B5: Supplementary damping control block diagram

ePKXT

X SDC1SDC1SDC

SDC11

+−= (C21)

ePKTT

XTTTT

XT

X SDC1SDC2

SDC1SDC1

SDCSDC2

SDC1SDCSDC2

SDC2SDC2

1+

−+−= (C22)

( )e

SDC

PKTTTT

XTTTTTT

XTT

TTX

TX

SDC1SDC4SDC2

SDC3SDC1SDC1

SDCSDC4SDC2

SDC1SDCSDC3

SDC2SDC4SDC2

SDC3SDC2

SDC4SDC

1

+−

+−

+−= (C23)

Substituting,

SDCSDC11

1T

A −= ; SDCSDC2

SDC1SDCSDC21 TT

TTA

−= ;

SDC2SDC22

1T

A −=

SDC1SDC1 KB = ; SDC1SDC2

SDC1SDC2 K

TT

B = ; SDC1SDC4SDC2

SDC3SDC1 KTTTT

;

SDC1SDC4SDC2

SDC3SDC1SDC3 K

TTTT

B =

Using above symbolic notations

ePBXAX 11111 += (C24)

ePBXAXAX 22221212 ++= (C25)

ePBXAXAXAX 3333232131SDC +++=

(C26)

Page 225: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.16

Appendix D

Optimisation problem

formulation for MPC

Considering a linear, discrete-time, state-space model of the system

( ) ( ) ( )kBukAxkx +=+1 (D1) ( ) ( )kxCky y= (D2)

( ) ( )kxCkz z= (D3)

where,

x is nx dimensional state vector

u is nu dimensional input vector

y is ny dimensional vector of measured outputs

z is nz dimensional vector of output which are to be controlled

The MPC controller will produce ( ) ( ) ( )1−−=∆ kukuku which will be passed on to the

system as input. One of the ways to include this ‘integration’ in a state space model is

by augmenting the state vector.

For example defining the state vector by ( ) ( )( )

=1ku

kxkξ

Page 226: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.17

Modifying the model of (D1) and (D2)

( )( )

( )( ) ( )

110

x k x kA B Bu k

u k u kI I+

= + ∆ − (D5)

( ) [ ] ( )( )

01

x ky k C

u k

= − (D6)

When whole state vector is measured such that

( ) ( ) ( )kykxkkx ==ˆ so C = I (D7)

The components of y and z may overlap and may be the same.

With this assumption y = z, and Cy= Cz

Defining the cost function as

( )( ) ( ) ( )( )

( )( )

21

0

2

1,J

j

CN

jj

N

jk kjkkjkkjkk

RQref uyyux ∑∑

==+∆++−+= (D8)

Starting prediction with assumption that whole state vector is measure, so that in the

circumstances of no information about any disturbances or measurement noise; the only

option is to predict by iterating the model (D1) and (D2).

Predicting by iterating (D1) and (D2)

( ) ( ) ( )kBukAxkkx +=+1 (D9)

( ) ( ) ( )kkBukkAxkkx 112 +++=+

( ) ( ) ( )kkBukkABukxA 12 +++= (D10)

( ) ( ) ( )kNkBukNkAxkNkx 11 −++−+=+

( ) ( ) ( )kNkBukkBuAkkxA NN 11 −++++= −

Summarising it, ( ) ( ) [ ]( )

( )

−++=+ −−

kjku

kkuBIAAkxAkjkx jjj

1

21

(D11)

At the time of computing the prediction ( )ku is unknown so using ( )kku instead of

( )ku

Page 227: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.18

Assuming that input may only change at times 1,,1, −++ CNkkk and will remain

constant after that. This will lead to ( ) ( )1−+=+ CNkukjku for 1−<< NjNC

As the ( )1−ku is already known at time k, the prediction can be expressed in terms of

( )kjku +∆ rather than ( )kjku +

( ) ( ) ( )kjkukjkukjku 1−+−+=+∆ (D12)

With this

( ) ( ) ( )1−+∆= kukkukku (D13)

( ) ( ) ( ) ( )111 −+++∆=+ kukkukkukku

( ) ( ) ( ) ( )111 −+++−+∆=−+ kukkukNkukNku CC

Substituting value of (D14) in (D9) for ( )ku in terms of ( )kku

( ) ( ) ( ) ( )[ ]11 −+∆+=+ kukkuBkAxkkx (D14)

( ) ( ) ( ) ( )[ ] ( ) ( ) ( )[ ]1112 2 −+∆++∆+−+∆+=+ kukkukkuBkukkuABkxAkkx

( ) ( ) ( ) ( ) ( ) ( )112 −+++∆+∆++= kBuIAkkuBkkuBIAkxA

At the end of the control horizon

( ) ( ) ( ) ( )( ) ( ) ( )11 1

1

−++++−++

+∆++++=+−

kBuIAAkNkBu

kkuBIAAkxAkNkx

C

CC

NC

NNC

(D15)

In general, for (k+j) instant predicted at kth instant

( ) ( )( )

( )( )1

1

1

0

1

0−+

−+

+=+ ∑∑

=

=kBuA

kjku

kkuBBAkxAkjkx

j

i

ij

i

ij

(D16)

For CNj ≤

( ) ( ) ( ) ( )( ) ( ) ( ) ( )11

1 1

−++++−+∆++

∆++++=++ +

kBuIAAkNkuBIA

kkuBIAAkxAkNkx

C

CC

NC

NNC

(D17)

( ) ( ) ( ) ( )( ) ( )( ) ( )1

11

1

−++++

−+∆++++

+∆++++=+

kBuIAA

kNkuBIAA

kkuBIAAkxAkNkx

NC

NN

NN

C

Page 228: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.19

Summarising in short for NjNC ≤< ,

( ) ( )( )

( )( )1

1

1

00

1

0−+

−+

+=+ ∑∑∑

=

=

=

kBuAkNku

kkuBABAkxAkjkx

j

i

i

C

Nj

i

ij

i

ijC

Combining all state predictions in one expression to give a matrix-vector form as

( )

( )( )

( )( )

( ) ( )

( )

( )( )

Future

kU

C

G

NN

i

iN

i

i

N

i

i

N

i

i

Past

N

i

i

N

i

i

N

i

i

N

N

N

ky

C

C

kNku

kku

BABA

BABBA

BBA

BABB

ku

BA

BA

BA

B

kx

A

AA

A

kNkx

kNkxkNkx

kkx

y

C

C

C

C

C

C

C

=

=

=

=

Γ

=

=

=

Φ

+

−+∆

+

+

+

+

=

+

+++

+

∑∑

1

00

11

1

0

1

0

0

1

0

1

0

0

1

0

1

(D18)

The prediction of y is now obtained for j = 1… N as

( ) ( )kjkCxkjky +=+ (D19)

Now rewriting the objective function of (D8) as

( ) ( ) ( ) 22RQrefk kUkYkYJ ∆+−= (D20)

where,

( )( )

( )( )

( )

( )( )

( )

( )

−+∆

∆=∆

+

+=

+

+=

kNku

kkukU

kNky

kkykY

kNky

kkykY

Cref

ref

ref1

;11

(D21)

and the weighting matrices Q and R are given by

Page 229: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.20

( )( )

( )

( )( )

( )

=

=

100

010000

;

00

020001

CNR

RR

R

NQ

QQ

Q

(D22)

Y (k) has the form of

( ) ( ) ( ) ( )kUGkukxkY y∆+−Γ+Φ= 1 (D23)

For suitable matrices ΓΦ, and yG error vector E(k) can be written as

( ) ( ) ( ) ( )1−Γ−Φ−= kukxkYkE ref (D24)

E(k) can be thought of as ‘tracking error’ which represents the difference between the

future target trajectory and the ‘free response’ of the system. The free response is

nothing but the response that would occur over the prediction horizon if no input

changes were made that is – if ( ) 0=∆ kU . Now modifying the objective function of

(D16) in terms of tracking error:

( ) ( ) ( ) 22RQyk kUkEkUGJ ∆+−∆= (D25)

( ) ( )[ ] ( ) ( )[ ] ( ) ( )KURkUkEkUGQkEGkU Ty

TTy

T ∆∆+−∆−∆= (D26)

( )[ ] ( ) ( ) ( ) ( ) ( )kQEkEkUQGkEkURQGGkU Ty

Ty

Ty

T +∆−∆+∆= 2 (D27)

This has the form

( ) ( ) ( ) constkUfkUHkUJ TTk +∆+∆∆=

21 (D28)

where,

( ) ( )kQEGfRQGGH Tyy

Ty 22 −=+= (D29)

and neither H nor f depends on - ( )kU∆

Finally writing the simple relation between input increment u∆ and control input u as:

( )( )

( )

( )( )

( )( )1

1

1

1

1−+

−+∆

+∆∆

=

−+

+kfu

kNku

kkukku

M

kNku

kkukku

CC

(D30)

Page 230: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.21

=

=

I

II

f

III

III

M

000

(D31)

Therefore from equation (D21) the following optimisation problem is to be solved

( )( ) ( ) ( )kUfkUHkU TT

kU∆+∆∆

∆ 21min (D32)

It is to be remembered that only the first step of the above solution will be used as per

the receding horizon strategy.

This is a standard optimisation problem known as the Quadratic programming (QP)

problem and standard algorithms are available for its solution.

Page 231: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.22

Appendix E

Data sheet for SMIB system used

in Chapter 5

E.1 BASE QUNATITIES

• Base MVA = 37.5 MVA

• Base stator voltage = 11.8 kV

• Base rotor voltage = 154 kV

E.2 GENERATOR PARAMETERS

• Armature resistances: Ra = 0.0020 pu

• Field resistance: Rfd = 0.00107 pu

• Direct-axis damper resistance: Rkd = 0.00318 pu

• Quadrature-axis damper resistance: Rkq = 0.00318 pu

• Direct-axis magnetising reactance: Xmd (Xad) = 1.859 pu

• Quadrature-axis magnetising reactance: Xmq (Xaq) = 1.560 pu

• Armature leakage reactance: Xl = 0.140 pu

• Field leakage reactance: Xfd = 0.140 pu

Page 232: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.23

• Direct-axis damper leakage reactance: Xkd = 0.140 pu

• Quadrature-axis damper leakage reactance: Xkq = 0.140 pu

• Inertia constant: H=5.3 sec.kW/KVA

E.3 TRANSFORMER PARAMETERS

• Resistance: rT = 0.0056 pu

• Reactance: xT = 0.1328 pu

E.4 TRANSMISSION LINE PARAMETERS

• Positive-sequence resistance: RL1 = 0.0075 pu

• Positive-sequence reactance: XL1= 0.5076 pu

• Zero-sequence resistance: RL0 = 0.0225 pu

• Zero-sequence reactance: XL0 = 0.1458 pu

Page 233: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.24

Appendix F

Data sheet of 10 generator 39

node New England system used

in Chapter 8

F.1 SYSTEM DETAILS USED FOR SIMULATION CASE-STUDY OF

CHAPTER 8

The case study of online controller coordination for multi-machine system is carried out

on 10 generators 39 nodes power system. The IEEE standard New England system is

adapted and the power system network is as shown in Fig. F1. The transmission line

details are given in Table F1, while the generator and load details are given in Table F2.

As shown in Fig.F1, the first TCSC is inserted in branch 11 which is a transmission line

from node 11 to node 15, and the second TCSC is inserted at node 12 and node 16. It is

assumed that this insertion of TCSC in long transmission lines will give rise to

additional intermediate nodes which are numbered in sequence for sake of simplicity as

node 13 and node 14 respectively. Following the numbering sequence as first generator

nodes, followed by FACTS devices node (TCSC in this case) and finally all load nodes

will make the system look like 41 nodes instead of 39 nodes.

Page 234: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.25

11

10

8

2

3

4 5

7

6

9

W

N12

N10

N11

N2

N8

N25

N26

N28

N29

N9N24N27N38

N37

N13

N15

N19

N18

N17

N1N16

N14

N31

N3

N20

N32

N33

N34

N35

N36

N21

N39

N30

N4 N5

N7

N23

N22

N6

TCSCL10

L1L2

L3

L4

L5

L6

L7

L8

L9

L11

L12

L13

L14

L15

L16L17

L18L19

L20

L21

L22

L23L24

L25

L26

L27

L28

L29

L30

L31

L32

L34

L33

W W

W

W W

W

W

W

W

W

W

TCSC

N40

N41

Fig.F1: 10 Generators 39 nodes New England System

Page 235: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.26

Table F1: Transmission line data

Sending end Node1

Receiving end Node2

Resistance (pu)

Reactance (pu)

Suspetance (pu)

37 27 0.0013 0.0173 0.3216 37 38 0.0007 0.0082 0.1319 36 24 0.0003 0.0059 0.068 36 21 0.0008 0.027 0.2548 36 39 0.0016 0.04 0.304 36 37 0.0007 0.018 0.1342 16 36 0.0009 0.0094 0.171 14 16 0.009 0.1085 1.830 33 12 0.0009 0.0101 0.1723 28 15 0.0014 0.0151 0.249 13 15 0.0057 0.0625 1.029 11 28 0.0043 0.0474 0.7802 11 27 0.0014 0.0147 0.2396 25 11 0.0032 0.0323 0.513 23 24 0.0022 0.035 0.361 22 23 0.0006 0.0096 0.1846 21 22 0.0008 0.014 0.2565 20 33 0.0004 0.0043 0.0729 20 31 0.0004 0.0043 0.0729 19 2 0.001 0.025 1.2 18 19 0.0023 0.0363 0.3804 17 18 0.0004 0.0046 0.078 35 31 0.0007 0.0082 0.1389 35 17 0.0006 0.0092 0.113 41 18 0.0008 0.0112 0.1476 41 35 0.0002 0.0026 0.0434 34 12 0.0008 0.0129 0.1382 34 41 0.0008 0.0128 0.1342 29 38 0.0011 0.0133 0.2138 29 34 0.0013 0.0213 0.2214 40 25 0.007 0.0086 0.146 40 29 0.0013 0.0151 0.2572 26 40 0.0035 0.0411 0.6987 26 2 0.001 0.025 0.75

Page 236: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.27

Table F2: Load data

Node No

Pload

(MW) Qload

(MVAr) 1 9.2 4.6 2 1104 250 3 0 0 4 0 0 5 0 0 6 0 0 7 0 0 8 0 0 9 0 0 10 0 0 11 139 17 12 0 0 13 0 0 14 0 0 15 283.5 26.90 16 320 153 17 233.8 84 18 522 176.6 19 0 0 20 0 0

Node No

Pload

(MW) Qload

(MVAr) 21 274 115 22 0 0 23 274.5 84.60 24 308.6 92.20 25 224 47.20 26 0 0 27 281 75.50 28 206 27.60 29 322 2.4 30 680 103 31 0 0 32 8.5 88 33 0 0 34 500 184 35 0 0 36 329.4 32.30 37 0 0 38 158 30 39 0 0 40 0 0 41 0 0

Page 237: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.28

Table F3: Transformer data

Sending end Node1

Receiving end Node2

Resistance (pu)

Reactance (pu)

Suspetance (pu)

39 30 0.0007 0.0138 0 39 5 0.0007 0.0142 0 32 33 0.0016 0.0435 0 32 31 0.0016 0.0435 0 30 4 0.0009 0.0180 0 15 9 0.0008 0.0156 0 25 8 0.0006 0.0232 0 23 7 0.0005 0.0272 0 22 6 0 0.0143 0 20 3 0 0.0200 0 35 1 0 0.0250 0 40 10 0 0.0181 0

Table F4: Generator data

Node Pgen Qgen Qgen(min) Qgen(max) 1 0 0.98 0 0 2 1000 1.03 -500 500 3 650 0.98 -325 325 4 508 1.01 -254 254 5 632 1 -316 316 6 650 1.05 -325 325 7 560 1.06 -280 280 8 540 1.03 -270 270 9 830 1.03 -415 415 10 250 1.05 -125 125

Page 238: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.29

Table F5: Generator parameters

Gen No.

Xd (pu)

Xq (pu)

Xmd (pu)

Xmq (pu)

Xfd (pu)

Xkd (pu)

Xkq (pu)

1 0.1305 0.0474 0.1125 0.0294 0.1254 0.1315 0.03740 2 0.1305 0.0474 0.1125 0.0294 0.1254 0.1315 0.03740 3 0.1864 0.0677 0.1607 0.042 0.1792 0.1878 0.05350 4 0.2175 0.0790 0.1875 0.049 0.2091 0.2191 0.06240 5 0.1864 0.0677 0.1607 0.042 0.1792 0.1878 0.05350 6 0.1864 0.0677 0.1607 0.042 0.1792 0.1878 0.05350 7 0.2175 0.0790 0.1875 0.049 0.2091 0.2191 0.06240 8 0.2175 0.0790 0.1875 0.049 0.2091 0.2191 0.06240 9 0.1450 0.0527 0.1250 0.0327 0.1394 0.1461 0.0416 10 0.4350 0.1580 0.3750 0.0980 0.4181 0.4383 0.1247

Gen No.

Ra (pu)

Rfd (pu)

Rkd (pu)

Rkq (pu)

H

1 0.0003 0.0001 0.0016 0.0012 500 2 0.0003 0.0001 0.0016 0.0012 30.30 3 0.0004 0.0001 0.0022 0.0017 35.80 4 0.0005 0.0001 0.0026 0.002 26 5 0.0004 0.0001 0.0022 0.0017 28.60 6 0.0004 0.0001 0.0022 0.0017 34.80 7 0.0005 0.0001 0.0026 0.0020 26.40 8 0.0005 0.0001 0.0026 0.0020 24.30 9 0.0003 0.0001 0.0017 0.0013 34.5 10 0.001 0.0003 0.0052 0.0040 42.0

Armature resistances: Ra

Field resistance: Rfd

Direct-axis damper resistance: Rkd

Quadrature-axis damper resistance: Rkq

Direct-axis magnetising reactance: Xmd (Xad)

Quadrature-axis magnetising reactance: Xmq (Xaq)

Direct-axis armature reactance: Xd

Quadrature-axis armature reactance: Xq

Field leakage reactance: Xfd

Direct-axis damper leakage reactance: Xkd

Quadrature-axis damper leakage reactance: Xkq

Inertia constant: H (sec.kW/KVA )

Page 239: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.30

Table F6 Excitation controller parameters

KA TA KE TE KF TF SE1 SE2 VRmin VRmax

25 0.06 -0.0445 0.5 0.16 1 0.0011 0.3043 -10 10

Table F7 Governor and prime mover parameters

TCH TSV RD PSV(max) dPmin dpmax

4 2 0.05 10 -1.0 1.0

Table F8 TCSC dynamic modeling data

TC K1(SDC) TW(SDC) T1(SDC) T2(SDC) T3(SDC) T4(SDC)

0.01 0.1 0.2 0.2 0.1 0.05 0.2

Table F9 TCSC details

Node1(tcsc) Node2(tcsc) Line no Xref(tcsc) Xmin(tcsc) Xmax(tcsc)

11 13 11 1e-3 -0.05 0.0075

12 14 8 1e-3 -0.075 0.0113

Note: All resistance, reactance and susceptance data is in pu on 100 MVA.

Page 240: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.31

Appendix G

Problem Formulation for Internal

State Estimation of Synchronous

generator

G.1 SYNCHRONOUS GENERATOR INTERNAL STATE VARIABLE

ESTIMATION

Starting with the dynamic rotor flux linkages equations from Chapter 2 and using the

detailed derivation as given in Appendix A:

fdmqmdmkqmkdmfdmfd EKiFiFAAA 111211131211 +++++=⋅

ψψψψ (G1)

qmdmkqmkdmfdmkd iFiFAAA 2221232221 ++++=⋅

ψψψψ (G2)

qmdmkqmkdmfdmkq iFiFAAA 3231333231 ++++=⋅

ψψψψ (G3)

refr ωωδ −=⋅

(G4)

Page 241: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.32

Using trapezoidal rule of numerical integration, the set of (G1) to (G4) can be

transformed in to as following:

++

+++

+

+

++++

∆+=

−−−

−−−−

)1()1(12

)1(11

)1(13

)1(12

)1(11

)()(12

)(11

)(13

)(12

)(11

)1()(2

nfdm

nqm

ndm

nkqm

nkdm

nfdm

nfdm

nqm

ndm

nkqm

nkdm

nfdm

nfd

nfd

EKiFiF

AAA

EKiF

iFAAA

t

ψψψ

ψψψ

ψψ

(G5)

+

+++

+

+

+++

∆+=

−−

−−−−

)1(22

)1(21

)1(23

)1(22

)1(21

)(22

)(21

)(23

)(22

)(21

)1()(2

nqm

ndm

nkqm

nkdm

nfdm

nqm

ndm

nkqm

nkdm

nfd

nkd

nkd

iFiF

AAA

iFiF

AAAm

t

ψψψ

ψψψ

ψψ (G6)

+

+++

+

+

+++

∆+=

−−

−−−−

)1(32

)1(31

)1(33

)1(32

)1(31

)(32

)(31

)(33

)(32

)(31

)1()(2

nqm

ndm

nkqm

nkdm

nfdm

nqm

ndm

nkqm

nkdm

nfdm

nkq

nkq

iFiF

AAA

iFiF

AAA

t

ψψψ

ψψψ

ψψ (G7)

( ) ( )[ ]refn

refnnn t ωωωωδδ −+−

∆+= −− )1()()1()(

2 (G8)

As it can be seen that (n) and (n-1) states of the system are both unknown, in the above

set of equations, the total number of unknowns are eight while the number of equations

available are just four. To solve this problem, another set of algebraic equations can be

organised using the available bus voltages and currents for (n) and (n-1) instant and

transforming it from system D-Q axis to rotor electrical d-q axis as given in Chapter 2.

)(

12)(

11)(

13)(

12)(

11)( n

qmn

dmn

kqmn

kdmn

fdmn

d iZiZPPPv −−++= ψψψ (G9)

)(22

)(21

)(23

)(22

)(21

)( nqm

ndm

nkqm

nkdm

nfdm

nq iZiZPPPv −−++= ψψψ (G10)

)1(12

)1(11

)1(13

)1(12

)1(11

)1( −−−−−− −−++= nqm

ndm

nkqm

nkdm

nfdm

nd iZiZPPPv ψψψ (G11)

)1(22

)1(21

)1(23

)1(22

)1(21

)1( −−−−−− −−++= nqm

ndm

nkqm

nkdm

nfdm

nq iZiZPPPv ψψψ (G12)

Page 242: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.33

Combining (G5) to (G8) and (G9) to (G12) to form a complete set of equations

considering n = 1 and (n-1) = 0 instants to represent the first two successive states:

++

+++

+

++

+++

∆++−=

0012011

013012011

1112111

113112111

01 21

fdmqmdm

kqmkdmfdm

fdmqmdm

kqmkdmfdm

fdfd

EKiFiF

AAA

EKiFiF

AAA

tfψψψ

ψψψ

ψψ

(G13)

+

+++

+

+

+++

∆++−=

022021

023022021

122121

123122121

01 22

qmdm

kqmkdmfdm

qmdm

kqmkdmfdm

kdkd

iFiF

AAA

iFiF

AAA

tfψψψ

ψψψ

ψψ

(G14)

+

+++

+

+

+++

∆++−=

032031

033032031

132131

133132131

01 23

qmdm

kqmkdmfdm

qmdm

kqmkdmfdm

kqkq

iFiF

AAA

iFiF

AAA

tfψψψ

ψψψ

ψψ

(G15)

( ) ( )[ ]refreftf ωωωωδδ −+−

∆++−= 0101 2

4

(G16)

01201101301201105 qmdmkqmkdmfdmd iZiZPPPvf −−+++−= ψψψ (G17)

02202102302202106 qmdmkqmkdmfdmq iZiZPPPvf −−+++−= ψψψ (G18)

11211111311211117 qmdmkqmkdmfdmd iZiZPPPvf −−+++−= ψψψ (G19)

12212112312212118 qmdmkqmkdmfdmq iZiZPPPvf −−+++−= ψψψ (G20)

where,

))((sin)())((cos)( ttVttVv QDd δδ ⋅+⋅=

))((cos)())((sin)( ttVttVv QDq δδ ⋅+⋅−=

))((sin)())((cos)( ttIttIi QDd δδ ⋅+⋅=

))((cos)())((sin)( ttIttIi QDq δδ ⋅+⋅−=

and, pm and zm are speed dependent as derived in Appendix (A12)

Page 243: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.34

1−= rrsrrm LGp ω ; ( )[ ]srsrrsrssrm RLLGGz +−−= −1ω (G21)

The convergence and accuracy of the optimisation problem depend on the initial values

given for optimisation.

G.2 INITIAL GUESS VALUES FOR STARTING ESTIMATION ALGORITHM

To find the initial values to start the estimation process, the relationship between the

generator terminal voltage and the current expressed in the d-q axis of the rotor at 0

instants can be used.

daqqd iRiXv ⋅−⋅= (G22)

qaddfdmdq iRiXIXv ⋅−⋅−⋅= (G23)

In (G22) and (G23), id and iq are expressed in terms of rotor angle delta for 0 instant.

Using the rotor angle relation with speed measurements, the next successive initial

values for rotor angle at instant 1 is calculated using (G16). Once the rotor angle delta at

0 and 1 instant are calculated, it can be used to solve a set of six algebraic equations

formed by (G13) to (G15) and (G17) to (G20) can be solved to get the initial estimate of

rotor fluxes as explained in Chapter 9.

Page 244: Robust WAM -based Controller Coordination of FACTS Devices ... › files › 3240990 › Wagh_Sush… · Robust WAM -based Controller Coordination of FACTS Devices for Power System

______________________________________________________________________A.35

Appendix H

Publications

1. T.T. Nguyen, S. R. Wagh, “Model Predictive Control of FACTS Devices for

Power System Transient Stability”, the Proceedings of the IEEE Transmission

and Distribution Asia Conference, Seoul, Korea, October, 2009.

2. T.T. Nguyen, S. R. Wagh, “Predictive Control-Based FACTS Devices for Power

System Transient Stability Improvement”, the Proceedings of the 8th IET

International Conference on Advances in Power System Control, Operation and

Management, APSCOM 2009, Hong Kong, November, 2009.

3. T.T. Nguyen, S. R. Wagh, “Application of Dynamic Modelling for Estimating

Internal States of a Synchronous Generator in Transient Operating Mode from

External Measurements”, submitted to IEEE Trans. Power Systems, 2011 (under

review).

4. T.T. Nguyen, S. R. Wagh, “Online Control Coordination of TCSCs for Power

System Transient Stability”, submitted to IET Generation, Transmission and

Distribution, Dec 2011 (under review).