Analysis and Damping Control of Power System Low-Frequency Oscillations

Power Electronics and Power Systems
Haifeng Wang
Wenjuan Du
Analysis and
Damping Control
of Power System
Low-frequency
Oscillations
Series editors
Joe H. Chow, Rensselaer Polytechnic Institute, Troy, New York, USA
Alex M. Stankovic, Tufts University, Medford, Massachusetts, USA
David Hill, The University of Hong Kong, Sydney, New South Wales, Australia
More information about this series at http://www.springer.com/series/6403
Haifeng Wang Wenjuan Du
•
Analysis and Damping

Control of Power System
Low-frequency Oscillations
123
Haifeng Wang Wenjuan Du
Beijing Beijing
China China
ISSN 2196-3185 ISSN 2196-3193 (electronic)

ISBN 978-1-4899-7694-9 ISBN 978-1-4899-7696-3 (eBook)
DOI 10.1007/978-1-4899-7696-3
Library of Congress Control Number: 2016931424
© Springer Science+Business Media New York 2016

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained herein or
for any errors or omissions that may have been made.
Printed on acid-free paper
This Springer imprint is published by SpringerNature

The registered company is Springer Science+Business Media LLC New York
Preface
Power system electromechanical low-frequency oscillations threaten the stable

operation of power systems. A great effort has been spent by many researchers for
over half-century in order to understand and solve this complicated engineering
problem. There are two main methods for the analysis and damping control of
power system low-frequency oscillations. They are the damping torque analysis and
modal analysis based on the linearized models of power systems for the study of
small-signal angular stability. In this book, they are named as linearized methods or
small-signal methods. The damping torque analysis is simple and of clear physical
meaning. The phase compensation method developed on the basis of damping
torque analysis for the design of power system stabilizers (PSSs) has been widely
used in practice. The modal analysis is based on the modal control theory and has
been the mostly used method in large-scale multi-machine power systems. This
book is devoted to the introduction of those two linearized methods and their
applications in the design of conventional PSSs and more recently appeared FACTS
stabilizers. It is written as a textbook for postgraduate research students and a
reference book for power system researchers. In addition to the introduction of
modal analysis, great attention has been paid in the book to elaborate the principle
and applications of the damping torque analysis, which we have applied in our
research on the subject in recent 20 years.
We have used the main materials in this book in teaching and tutoring our
research students for years. Some examples in the book are the results of their
exercise work before starting their research projects. We would like to acknowledge
their contributions to the preparation of examples in the book. They are Yifu Lin,
Xiangfeng Wang, Yan Zhang, Shen Yang, Qianhui Wan, Ximin Li (Sect. 2.4), Chen
Lv (Sect. 3.3), Yi Ge (Sect. 4.4), Tianyu Su (Sect. 5.3), Zhijin Zhen (Sect. 6.3), and
Xiaobo Hu (Sect. 7.4.2).
Beijing Haifeng Wang

October 2015 Wenjuan Du
v
Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ........... 1
1.1 Power System Low-frequency Oscillations . . . . . . ........... 1
1.2 Linearized Methods for the Analysis and Damping
Control of Power System Oscillations . . . . . . . . . ........... 3
1.3 FACTS and Grid-Connected ESS . . . . . . . . . . . . ........... 5
1.4 Controllers to Damp Power System Oscillations . . ........... 7
1.5 Design of Damping Controllers to Suppress Power
System Oscillations . . . . . . . . . . . . . . . . . . . . . . ........... 10
1.6 Organization of the Book . . . . . . . . . . . . . . . . . . ........... 12
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ........... 13
2 A Single-Machine Infinite-Bus Power System Installed
with a Power System Stabilizer. . . . . . . . . . . . . . . . . . . . ........ 17
2.1 Linearized Model of a Single-Machine Infinite-Bus
Power System Installed with a Power System Stabilizer . . . . . . . . 17
2.1.1 General Linearized Mathematical Model. . . . . . . . . . . . . . 17
2.1.2 Heffron–Phillips Model . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.2 Modal Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.2.1 Basis of Modal Analysis Theory . . . . . . . . . . . . . . . . . . . 32
2.2.2 Applications of Modal Analysis . . . . . . . . . . . . . . . . . . . 37
2.3 Damping Torque Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.3.1 Damping Torque and Synchronizing Torque . . . . . . . . . . . 42
2.3.2 Damping Torque Analysis and Design
of PSS by Phase Compensation. . . . . . . . . . . . ..... ... 47
2.4 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... ... 53
2.4.1 Linearized Mathematical Models of an Example Power
System. . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... ... 53
2.4.2 Modal Analysis of Example Power System. . . . ..... ... 59
2.4.3 Damping Torque Analysis of Example
Power System . . . . . . . . . . . . . . . . . . . . . . . . ........ 66
vii
viii Contents
2.4.4 Equivalence Between the Damping Torque

and Modal Analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
3 Damping Torque Analysis of Thyristor-Based FACTS
Stabilizers Installed in Single-Machine Infinite-Bus
Power Systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 81
3.1 A Single-Machine Infinite-Bus Power System Installed
with an SVC Stabilizer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 81
3.1.1 Extended Heffron—Phillips Model
of a Single-Machine Infinite-Bus Power System
Installed with an SVC Stabilizer . . . . . . . . . . . . . . . . . .. 81
3.1.2 Damping Torque Analysis of SVC
Stabilizer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 90
with a TCSC or TCPS Stabilizer . . . . . . . . . . . . . . . . . . . . . . .. 97
3.2.1 Extended Heffron–Phillips Model
of a Single-Machine Infinite-Bus Power System Installed
with a TCSC or TCPS Stabilizer . . . . . . . . . . . . . . . . . .. 97
3.2.2 Damping Torque Analysis of TCSC
and TCPS Stabilizers . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
3.3 An Example Power System Installed with an SVC
Stabilizer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
3.3.1 Linearized Model of Example Power
System. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
3.3.2 Design of SVC-Based Stabilizer . . . . . . . . . . . . . . . . . . . 114
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
4 Single-Machine Infinite-Bus Power Systems Installed
with VSC-Based Stabilizers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
4.1 Damping Torque Analysis of a Shunt VSC-Based
Stabilizer Installed in a Single-Machine Infinite-Bus
Power System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
4.1.1 Extended Heffron–Phillips Model of a Single-Machine
Infinite-Bus Power System Installed with a Shunt
VSC-Based Stabilizer. . . . . . . . . . . . . . . . . . . . . . . . . . . 121
4.1.2 Damping Torque Analysis of Shunt VSC-Based Stabilizer
Installed in Single-Machine Infinite-Bus
Power System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
4.2 Damping Function of a Stabilizer Added on a Static Synchronous
Series Compensator (SSSC) Installed in a Single-Machine
Infinite-Bus Power System . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
4.2.1 Damping Torque Analysis of a SSSC Stabilizer Installed
in a Single-Machine Infinite-Bus Power System . . . . . . . . 133
4.2.2 Design of a SSSC Stabilizer . . . . . . . . . . . . . . . . . . . . . . 139
Contents ix
4.3 Damping Function of a Unified Power Flow Controller (UPFC)

Installed in a Single-Machine Infinite-Bus Power System . . . . . . . 143
4.3.1 Mathematical Model of a Single-Machine
Infinite-Bus Power System Installed
with a UPFC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
4.3.2 Design of a UPFC Stabilizer Installed
in a Single-Machine Infinite-Bus Power System . . . . . . . . 155
4.4 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
4.4.1 An Example Single-Machine Infinite-Bus Power System
Installed with a BESS Stabilizer . . . . . . . . . . . . . . . . . . . 163
4.4.2 An Example Single-Machine Infinite-Bus Power System
Installed with a UPFC Stabilizer . . . . . . . . . . . . . . . . . . . 172
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
5 A Multi-machine Power System Installed with Power System
Stabilizers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
5.1 Mathematical Model of a Multi-machine Power System Installed
with Power System Stabilizers. . . . . . . . . . . . . . . . . . . . . . . . . . 183
5.1.1 A Two-Machine Power System Installed
with Power System Stabilizers . . . . . . . . . . . . . . . . . . . . 183
5.1.2 A Multi-machine Power System Installed
with Power System Stabilizers . . . . . . . . . . . . . . . . . . . . 192
5.2 Modal Analysis and Control of Power System Oscillations
in a Multi-machine Power System Installed with Power System
Stabilizers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
5.2.1 Eigensolution for the Analysis of Power System
Oscillations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
5.2.2 Design of Power System Stabilizers
in a Multi-machine Power System . . . . . . . . . . . . . . . . . . 204
5.2.3 Fixed Modes Associated with PSS Control . . . . . . . . . . . . 209
5.3 An Example Three-Machine Power System. . . . . . . . . . . . . . . . . 214
5.3.1 Example Power System and Its Linearized
Heffron–Phillips Model . . . . . . . . . . . . . . . . . . . . . . . . . 214
5.3.2 Modal Analysis of Example Power System. . . . . . . . . . . . 221
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
6 Multi-machine Power System Installed with Thyristor-Based
FACTS Stabilizers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
with Thyristor-Based FACTS Stabilizers. . . . . . . . . . . . . . . . . . . 235
6.1.1 Heffron–Phillips Model . . . . . . . . . . . . . . . . . . . . . . . . . 235
6.1.2 General Linearized Model of an N-Machine
Power System Installed with Multiple
Thyristor-Based FACTS Stabilizers . . . . . . . . . . . . . . . . . 242
x Contents
6.2 Analysis and Damping Control of Thyristor-Based FACTS

Stabilizers Installed in a Multi-machine Power System . . . . . . . . . 266
6.2.1 Damping Torque Analysis in a Multi-machine Power
System. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
6.2.2 Selection of Installing Location and Feedback Signal of a
Stabilizer in a Multi-machine Power System . . . . . . . . . . . 272
6.2.3 Selection of Robust Installing Locations
and Feedback Signals of a Stabilizer by an
Eigensolution-Free Method . . . . . . . . . . . . . . . . . . . . . . . 279
6.2.4 Stabilizer Design in a Multi-machine Power System
Considering Robustness and Interaction of Stabilizers . . . . 282
6.3 An Example Two-Area Four-Machine Power System . . . . . . . . . . 289
6.3.1 Linearized Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
6.3.2 Selection of Installing Locations of Stabilizers . . . . . . . . . 301
6.4 Example Three-Machine Power System . . . . . . . . . . . . . . . . . . . 320
6.4.1 Dynamic Interactions Among PSSs Installed
in Example Three-Machine Power System . . . . . . . . . . . . 320
6.4.2 Design of Non-negatively Interactive PSSs Installed in
Example Power System . . . . . . . . . . . . . . . . . . . . . . . . . 324
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328
7 Multi-machine Power Systems Installed with VSC-Based
Stabilizers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329
with VSC-Based Stabilizers . . . . . . . . . . . . . . . . . . . . . . . . . . . 329
7.1.1 Mathematical Model of a Multi-machine Power System
Installed with a Shunt VSC-Based Stabilizer . . . . . . . . . . . 329
Installed with a UPFC-Based Stabilizer . . . . . . . . . . . . . . 339
7.2 Design of a Shunt VSC-Based Stabilizer by Localized
Phase Compensation Method to Suppress Inter-area Line Power
Oscillations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349
7.2.1 Localized Small-Signal Model of a VSC-Based Unit in a
Multi-machine Power System . . . . . . . . . . . . . . . . . . . . . 350
7.2.2 Design of VSC-Based Stabilizer
by Localized Phase Compensation Method . . . . . . . . . . . . 358
7.2.3 Robustness of an ESS-Based Stabilizer
to Variation of Line-Loading Conditions . . . . . . . . . . . . . 362
7.3 An Example of Multi-machine Power System
with a Grid-Connected FC Power Plan . . . . . . . . . . . . . . . . . . . . 367
with a Grid-Connected FC Power Plant . . . . . . . . . . . . . . 367
7.3.2 Design of a Stabilizer Attached to the VSC
of FC Power Plant by Localized Phase Compensation
Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 375
Contents xi
7.4 Damping of Multi-mode Oscillations by Multiple

Stabilizers Attached to a Single UPFC . . . . . . . . . . . . . . . . . . . . 380
7.4.1 Coordinated Design of Multiple Stabilizers
by Artificial Fish Swarm Algorithm (AFSA) . . . . . . . . . . . 381
7.4.2 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393
Chapter 1
Introduction
1.1 Power System Low-frequency Oscillations
Power system low-frequency oscillations are the oscillations of active power

delivered along particular transmission corridors in a power system with the
oscillation frequency from 0.1 Hz up to a couple of Hz. Once started, the oscilla-
tions can continue for a while and then disappear, or grow continuously to cause
power system collapse. In this book, they are called power system oscillations or
power oscillations for short.
Manifestation of a power oscillation is the oscillation of relative movement of
angular positions of generators in the power system. The oscillation can be triggered
by severe faults, such as a three-phase to-earth short circuit along or tripping of a
transmission line. It can also occur under normal operating conditions when the
power system is only subject to small disturbances. Hence if the power system
collapse is caused by the power oscillation, it could belong to the problem of power
system large-signal rotor angle (angular) stability or small-signal angular stability [1].
This book is devoted to the introduction of the linearized (small-signal) methods
for the analysis and damping control of the power oscillation. Hence strictly
speaking, the power oscillation discussed in the book should belong to the problem
of power system small-signal angular stability. If the power oscillation exhibits
increasing magnitude, the power oscillation is said to have negative damping and
obviously the power system is unstable in terms of power system small-signal
angular stability. If the power oscillation shows a sustained constant magnitude, or
continues for a certain period of time (over several to tens of seconds) and sets
down eventually, it is reckoned that the oscillation is of poor damping though the
power system is still stable. If the oscillation subsides quickly within several sec-
onds with a damping ratio greater than 0.1 (sometimes it could be 0.05 for
large-scale power systems), it is considered that the power oscillation is well
damped.
© Springer Science+Business Media New York 2016 1

H. Wang and W. Du, Analysis and Damping Control of Power System
Low-frequency Oscillations, Power Electronics and Power Systems,
DOI 10.1007/978-1-4899-7696-3_1
2 1 Introduction
Power system oscillation was first observed in the Northern American power
network in Oct. 1964 during a trial interconnection of the Northwest Power Pool
and the Southwest Power Pool [2]. The power oscillation occurred on the tie line
and was of a frequency of 0.1 Hz. Since then, many incidents of power system
oscillations have been reported in power transmission networks around the world.
Examples are as follows:
1. Between the late 1970s and early 1980s, power oscillations were observed in the
power transmission corridor from Scotland to England in the Great Britain
power network. Operational experience indicated that those oscillations were
related to the relatively high loading level of power transmission lines from
Scotland to England. A series of tests were carried out between 1980 and 1985.
Those tests demonstrated that the oscillations occurred when the power transfer
from Scotland to England reached a certain level and the typical oscillation
frequency was around 0.5 Hz [3].
2. Sustained power oscillations were reported in the Taiwan power network in
1984. Those oscillations happened typically when a large amount of power was
transferred along some particular high-voltage transmission lines. It was found
that reduction of the amount of power delivered along the high-voltage trans-
mission lines can effectively improve the damping of power oscillations.
Installation of the PSSs (power system stabilizers) at selected locations worked
successfully to suppress the power oscillations. Further investigation indicated
that other factors, such as the gain value of automatic voltage regulator
(AVR) and characteristics of load, also affected the damping of the power
oscillations [4].
3. An outage of the Northern American Western Systems Coordinating Council
(WSCC) network on August 10 1996 was directly due to the power oscillations.
The power oscillations (with frequency between 0.2 and 0.3 Hz) were triggered
by the loss of a 500-kV line when the system operated under depressed con-
ditions (with overloaded lines and at low voltage level). The incident spread
quickly, tripping more lines, and generating units, which eventually led to the
separation of the WSCC network into four islands. The outage affected
7.5 million customers for up to 9 h, causing considerable economic loss [5].
Over the last half-century, many power system researchers and engineers have
worked on and contributed to the understanding and solution of the problem of
power system oscillations. It is now well recognized that the main cause of power
system oscillations is the poor damping of the so-called electromechanical oscil-
lation modes of the power system. Poor damping could be brought about by the
(1) large amount of long-distance power transmission, (2) weak interconnection of
large power sub-networks, and/or (3) negative damping due to the fast-acting
high-gain AVRs. Power system low-frequency oscillations can be classified,
according to the electromechanical oscillation modes of the power system, into two
types: (1) local power oscillations (associated with local oscillation modes) and
(2) inter-area power oscillations (associated with inter-area oscillation modes).
1.1 Power System Low-frequency Oscillations 3
Local power oscillations associated with the local oscillation modes normally are
the oscillations of one or a group of local generators against a large power network.
Local generators send power over a long distance to the large power network.
Frequency of the oscillations often is about one or several Hz. Inter-area power
oscillations related to the inter-area oscillation modes are the oscillations between
two or more sub-networks in a large-scale power system. A typical inter-area
oscillation is the tie-line power oscillation between two weakly connected areas in
the power system. The inter-area power oscillation could involve many
sub-networks to oscillate against each other (which sometimes is referred to as an
intra-area oscillation). Normally, the oscillation frequency is lower, from 0.1 to 1 Hz.
A power oscillation could be engaged by one oscillation mode only (local or
inter-area oscillation mode). This is the case referred to as the “single-mode power
oscillation” when there is only one electromechanical oscillation mode in the power
system being poorly or negatively damped. In the case that there are multiple poorly
or negatively damped electromechanical oscillation modes in the power system, the
power oscillation is participated by multiple oscillation modes. The power oscil-
lation in this case is called the “multi-mode power oscillation”.
1.2 Linearized Methods for the Analysis and Damping

Control of Power System Oscillations
Study of power system oscillations is in order to investigate the phenomena and to

develop prevention measures from the occurrence of the problem. This is to
understand why the oscillations happen and to recommend how they can be
effectively suppressed. There are two main categories of linearized methods
(small-signal methods) for the analysis and damping control of power system
oscillations: the damping torque analysis and modal analysis.
The technique of damping torque analysis (DTA) was well explained in [6] for a
single-machine infinite-bus power system to investigate the effect of excitation
control on power system small-signal angular stability. Reference [6] is the mile-
stone contribution by deMello and Concordia to the field. Proposal of the DTA is
based on the linearized Heffron–Phillips model of the single-machine infinite-bus
power system [7, 8]. It is used to examine the electric torque contribution from a
particular source in the power system to the so-called electromechanical oscillation
loop of a generator, which is in fact the linearized swing equation of rotor motion of
the generator. The electric torque is decomposed into two components, synchro-
nizing torque and damping torque. Synchronizing torque is in phase with rotor
angle deviation and damping torque in phase with rotor speed deviation of the
generator. Synchronizing torque is the force to keep the generator in synchronism
with the rest of the power system. Lack of damping torque or contribution of
negative damping torque may lead to poor damping of power system oscillations or
even power system instability.
4 1 Introduction
The DTA is established on the physical concept of electric torque and classic
control theory with the description of the system in frequency and Laplace s
domain. It is easy to be understood and simple to be applied. Often analytical
conclusions can be obtained under certain reasonable assumptions. In particular
later, an effective method, the phase compensation method, for the design of a
power system stabilizer (PSS) was developed on the basis of the DTA [9–11],
which has made the DTA a well-accepted method for the analysis and damping
control of power system oscillations. Since the 1970s, considerable effort has been
spent to extend the DTA for the application in general multi-machine power
systems.
Modal analysis (MA) is based on the modal control theory, a special branch in
modern control theory. It is established on the state-space representation of a power
system, i.e. so-called ABCD state-space model [12]. Applications of modal analysis
in the study of power system oscillations include the following: (1) computation of
eigenvalues and associated eigenvectors (so-called eigensolution) of the state
matrix of the system; (2) computation of controllability index, observability index,
and their product, residue. There is no much difference in the procedure to apply the
MA in a single-machine infinite-bus and a multi-machine power system.
Eigensolution is a very specific topic in mathematics. Examples of applying
eigensolution in studying power system oscillations are computation of power
system oscillation modes (i.e. computation of eigenvalues and identification of
oscillation modes among all eigenvalues), eigenvalue sensitivity, and participation
factor [13, 14]. For a large-scale power system, dimension of the state matrix could
become very high. This could lead to the extremely high computational complexity
and even numerical difficulty to calculate the eigenvalues of the state matrix. Hence,
it has been a special research topic to compute eigenvalues of interests of a
high-dimensional state matrix considering the features of the large-scale power
system.
Computation of controllability index, observability index, and residue is mainly
applied for the selection of installing locations and feedback signals of stabilizers in
the multi-machine power system. Because computation of controllability index,
observability index, and residue usually needs to perform eigensolution, various
methods to reduce computational cost have been proposed in the literature.
Examples are the selective modal analysis [13–15], partial modal analysis [16], and
eigensolution-free modal analysis [17].
The MA is a method based on the results of numerical computation. It can tell
whether a power system is stable at given system operating conditions. However, it
is difficult to be used to draw general conclusions by performing numerical
eigensolution. Hence in many occasions, effort has to be spent to establish the
connections between the computational results from the MA and the physical
explanations in order to understand and explain the results of numerical
computation.
1.3 FACTS and Grid-Connected ESS 5
1.3 FACTS and Grid-Connected ESS
Flexible AC transmission systems (FACTS) are the technologies emerged around

the end of last century to reinforce power systems. Foundation of the FACTS is the
application of power-electronic based control in high-voltage power transmission
systems. It is the result of rapid advance in high-power electronics technology over
the last several decades. Though the FACTS controllers are still considered
expensive compared to the conventional power system controllers, they have been
installed in many real power systems in the world due to their superior control
performance.
The basic operational principle of the FACTS controllers is shown in Fig. 1.1
[18, 19]. The active power delivered along the transmission line is as follows:
V1 V2
P12 ¼ sin h ð1:1Þ
x
where h is the phase difference between the voltage at the sending and receiving end
of the transmission line, V 1 and V
2 . A FACTS controller is to change in real time
the line impedance x, and phase and magnitude of line voltage V 1 and V 2 indi-
vidually or simultaneously to regulate the power flow along the transmission line,
such that a great flexibility of power flow regulation is achieved.
In fact, the operational principle of the FACTS was known as early as in the 1920s.
Since then, mechanically controlled series and shunt capacitors have been installed in
power systems to control line power flow and regulate system voltage profile. Power
electronics applications in high-voltage power systems were dated from early 1970s
when converters for high-voltage direct-current systems began using thyristors to
replace mercury arc valves. This was the time when the FACTS technology really
emerged and was applied in power systems. The concept of the FACTS was formerly
introduced by Prof. Hingarani [18] when work to use power electronic switching
devices to replace thyristors began [19]. Hence normally, the FACTS controllers are
classified into two groups: (1) the conventional thyristor-based FACTS controllers
and (2) new generation of converter (power electronic switching devices)-based
FACTS controllers. The following are three main types of thyristor-based FACTS
controllers which have been applied in power systems:
Fig. 1.1 Illustration on the V1 V2

operational principle of x
FACTS controllers
P12
6 1 Introduction
1. Thyristor-controlled static VAR compensator (SVC), which controls the mag-

nitude of line voltage;
2. Thyristor-controlled series compensator (TCSC), which changes the equivalent
impedance of transmission line and hence can be used for power flow control;
3. Thyristor-controlled phase shifter (TCPS), which shifts the phase of line voltage
and normally can be used for power flow control.
New generation of FACTS controllers employs self-commutated, voltage-soured
switching power converters to realize rapidly controllable, static synchronous AC
voltage or current sources. A new-generation FACTS controller is constructed
mainly on the synchronous voltage source (SVS) which is an ideal machine with no
rotation and inertia. The magnitude and phase of the SVS can be controlled
instantaneously to generate reactive power (both capacitive and inductive) and/or to
regulate active power flow independent of power system variables (voltage and
current). For example, Fig. 1.2 shows an SVS constructed by a GTO-based voltage
source converter (VSC), which employs the algorithm of the pulse width modu-
lation (PWM). Through controlling the modulation ratio m and phase /, the
magnitude and phase of the AC SVS can be controlled to realize various functions
of power system regulation.
There are three main types of VSC-based FACTS controllers:
1. The static synchronous compensator (STATCOM). This is the controller real-
ized by installing the VSC-based SVS in shunt along a transmission line.
2. The static synchronous series compensator (SSSC). This is the VSC-based SVS
installed in series in the transmission line.
3. The unified power flow controller (UPFC). This is the most powerful FACTS
controller, a combined device of a STATCOM and a SSSC with a common DC
capacitor.
The development of power systems in the recent years is towards meeting the
requirement of clean power generation. This will lead to the fundamental changes
Three-phase diagram of GTO-based Single-line diagram of GTO-based

voltage source converter voltage source converter SVS
Vc Vdc
transformer
Variable AC DC
voltage voltage
Power switching A DC m φ
circuit capacitor
Fig. 1.2 An SVS realized by a VSC circuit

1.3 FACTS and Grid-Connected ESS 7
Step-down transformer
Vc Vc
φ φ
ac/dc VSC
m m
Cdc ESS
nd
2 converter
Single stage circuit
ESS
Two stage circuit
Fig. 1.3 VSC-based power electronic circuit for grid-connected ESS or renewable generation
plant
of the systems in the near future with a fairly large percentage of grid-connected
renewable power generation. It can be foreseen that the high-power electronic VSC
will play even more important role in shaping the future power systems. Not only
the VSC will be used for the grid connection of many types of renewable power
generation, such as wind, solar, and fuel cell, but also it can be applied for the
energy storage systems (ESSs) to assist power system operation and control in
accommodating variable grid-connected renewable power generation. Hence in
future power systems, the VSC could become one of the mostly installed power
components like transformers to meet various purposes of power regulation, stor-
age, and generation.
Figure 1.3 shows the configuration of grid connection of an ESS device or a
renewable power generation plant by use of the VSC-based power electronics
circuits. The ESS device can be a battery ESS (BESS). The configuration is also
applicable for the grid connection of a renewable generation source, such as a
photovoltaic (PV) power plant. In the two-stage circuit, the second converter can be
a DC/DC converter to connect to a fuel cell (FC) power plant or a DC/AC converter
to a wind farm. In both circuits, the key component is the AC/DC converter
(VSC) which is connected to the high-voltage busbar via a step-down transformer.
1.4 Controllers to Damp Power System Oscillations
To increase the damping of power system oscillations, installation of a supple-

mentary damping controller, the PSS (power system stabilizer), on the side of
excitation system of a generator is a simple, effective and economical method.
8 1 Introduction
Through the effort of and experience accumulated by power engineers and

researchers in the last half-century, the PSS has become a standard device installed
in the generator to supply extra damping to power system oscillations.
However, it has been found that installation of the PSSs cannot always provide
satisfactory solution to the problem of power system oscillations. Typical example
is the inter-area power oscillations on the tie lines or along a particular
long-distance power transmission corridor in a large-scale power system. If there is
no power plant close to the tie lines and the transmission corridor where power
oscillations are observed, it may need complex coordinated design of multiple PSSs
installed at different remote power plants to jointly suppress the power oscillations.
There is no guarantee that the coordinated design of those multiple PSSs is always
able to supply sufficient damping to the power oscillations. Hence, any new option
to the PSS for the effective suppression of power system oscillations is always
welcome.
With the advent of FACTS technology, FACTS-based stabilizers (referred to as
FACTS stabilizers in this book) have been investigated as an alternative type of
controllers to damp power system oscillations. Main difference of the FACTS
stabilizers to PSSs is that the FACTS devices, which the FACTS stabilizers are
attached to, usually are installed at key locations of transmission network in power
systems, such as that close to the tie lines and long-distance transmission corridors.
Hence in certain cases, FACTS stabilizers could be simpler, easier to be imple-
mented, and more effective to suppress power system oscillations.
In fact, the idea of applying variable series compensation to damp power system
oscillations was proposed many decades ago [20, 21], though at that time, variable
compensation was assumed to be realized by switching in or out the series
capacitors using mechanical circuit breakers. Only with the thyristor-controlled
high-power electronic circuits being available after the 1980s, high-speed variable
reactive power compensation has become a feasible technology in power system
application to suppress power system oscillations. Between the late 1980s and
1990s, case study of numerical calculation, computer simulation, and field tests
confirmed the capability of thyristor-based FACTS stabilizers to damp power
system oscillations. Theoretical analysis was carried out to obtain insight into and to
understand the basic principles of thyristor-based FACTS stabilizers. Similarity of
the thyristor-based FACTS stabilizers to the PSSs has helped considerably the
understanding on why and how the thyristor-based FACTS stabilizers can suppress
power system oscillations.
A FACTS stabilizer is superimposed on the normal control function of a FACTS
controller. For example, Fig. 1.4 shows a TCR (thyristor-controlled reactor)-FC
(fixed capacitor) type of SVC with a normal function of voltage control. A damping
controller (the SVC stabilizer) is superimposed on the SVC voltage control func-
tion. In Fig. 1.4, the variable inductive current Il is controlled by the thyristor firing
circuit. Hence, the variable equivalent reactance of the TCR-FC SVC is as follows
1.4 Controllers to Damp Power System Oscillations 9
Power system
Vs
Is
Il Ic
x svcl
α0
Vs π
Vsref - x svcc
voltage α firing circuit
controller +
+
π/2
ysvc SVC
stabilizer
Fig. 1.4 An SVC stabilizer superimposed on the SVC voltage control function
s
V s
V
jxsvc ¼ ¼ ð1:2Þ
Is ðIl þ Ic Þ
where and in Fig. 1.4, V s is the voltage measurement at the location where the SVC

is installed, Vsref the reference signal of voltage control, jxsvc the equivalent reac-
tance of the SVC, and ysvc the feedback signal of the SVC stabilizer. The SVC
voltage controller is to maintain the magnitude of voltage at the SVC installing
location close to V sref by controlling the variable equivalent reactance of the SVC.
Attachment of the SVC stabilizer contributes the control of variable equivalent
reactance jxsvc, hence the exchange of reactive power between the SVC and rest of
the power system. Thus, it affects the variations of active power flow along the
transmission line where the SVC locates. If designed properly, the SVC stabilizer
can help to suppress power oscillations. Attachment of the SVC stabilizer to SVC
normal voltage control function is very similar to the arrangement of a PSS being
superimposed on the voltage control loop of an AVR.
It was well reckoned that VSC-based FACTS stabilizers work as effectively as
the thyristor-based FACTS stabilizers in damping power system oscillations, as
confirmed by case study of numerical calculation, computer simulation, and field
tests [22, 23]. There is no fundamental difference between the mechanism of the
thyristor-based and VSC-based FACTS stabilizers, though examination of the
VSC-based FACTS stabilizers usually is sometimes more complicated as it
involves the VSC functions.
10 1 Introduction
Occurrence of power system oscillations is essentially due to the response of

generators to system disturbances in order to maintain the balance of supply and
consumption of active power. Because generators need time to either accelerate or
decelerate in responding to disturbances and they are at different geographic
positions in the power system, power oscillations occur as the result of the relatively
slow correcting actions of generators to keep the balance of active power. The
slower the correcting actions are, the worse the power oscillations. That is why in
the power system, the power oscillations are of low oscillation frequency and hence
called low-frequency oscillations. Gird-connected ESSs do not have rotating
components involved for the exchange of active power with the rest of power
system. They can respond instantaneously to system disturbances with correcting
actions to maintain the balance of active power in the power system. Hence in
principle, applications of the ESSs in the power system can effectively provide extra
damping to power system oscillations.
Research on the capability of ESS controllers to damp power system oscillations
has been mainly about two types of energy storage devices, the superconducting
magnetic energy storage (SMES) and the battery energy storage systems (BESS)
that are also classified as FACTS devices [24]. The function and effectiveness of a
SMES to damp power system oscillations have been well investigated since several
decades ago. By regulating the active and reactive power exchange of the SMES
with the power system, the SMES can effectively suppress power system oscilla-
tions, which have been demonstrated by simulation and laboratory experiment [25,
26].
In [27], a case of field application of the BESS in improving power oscillation
damping was reported. A 10-MW BESS unit was installed in 1988 at the Chino
substation owned by the Southern California Edison (SCE), USA. Six years later,
an energy source power system stabilizer (ESPSS) was installed by taking the
advantage of very fast response rate of the installed BESS to provide extra damping
to the power system. The ESPSS was designed to modulate the active power output
from the BESS to damp power system oscillations rather than through controlling
voltage or reactive power as a conventional stabilizer does.
1.5 Design of Damping Controllers to Suppress Power

System Oscillations
The prime specification of designing a damping controller (a PSS or a FACTS

stabilizer) is the maximum effectiveness of the stabilizer at the minimum control
cost. Two main issues in the design are as follows: (1) selection of installing
locations and feedback signals and (2) parameters setting of the stabilizer.
In a large-scale multi-machine power system, there are many power stations that
can be or have been equipped with the PSSs. To damp a particular oscillation mode
of interests, it would be impractical if all the PSSs in the power system are set
1.5 Design of Damping Controllers to Suppress Power System Oscillations 11
accordingly. Hence, the selection of installing locations of the PSSs can mean to
choose the locations to install PSSs or to choose the PSSs among those installed to
be set, in order to damp the particular oscillation mode. As far as a FACTS sta-
bilizer is concerned, justification of installing an expensive FACTS device usually
is based on other applications rather than the damping function. Hence, the
selection of installing location of the FACTS stabilizer is carried out among the
installed FACTS devices and/or FACTS normal control functions to attach the
damping function. Even if a location in the power system is the best to have the
FACTS stabilizer, it often needs further justification to consider the installation of
the FACTS stabilizer there.
Leaving the practical consideration about the candidate locations aside, there is
no difference in the strategy to select installing locations for the PSSs and FACTS
stabilizers. On the other hand, selection of feedback signals for the PSSs and
FACTS stabilizers normally is to determine the most effective and locally available
feedback signals for the design of stabilizers. Those signals can be the rotor speed
deviation of generators, active power, line current, etc. In the case that remote
signals can be used, of course, it also includes the choice of signals at different
locations in the power system.
Methods which have been proposed and developed so far for the selection of
installing locations and feedback signals are mostly based on the DTA and modal
analysis (MA). Examples are the induced damping and synchronizing torque
coefficients (IDSTCs) index [28], damping torque coefficient (DTC) index [29],
simple index [30], and damping index [31] derived by using the DTA. Sensitivity
calculation [32], the participation factors [13–15], the LIED (location index for
effective damping) [33], residue index [34, 35], the partial modal analysis [16], and
eigensolution-free modal analysis [17] are established by use of the modal analysis.
Methods to set parameters of multiple stabilizers in a multi-machine power
system can be classified into two categories, sequence setting and simultaneous
tuning. Sequence setting is the strategy to set stabilizers one by one in a sequence.
For example, suppose two stabilizers, stabilizers A and B, to be set. In sequence
setting, stabilizer A is installed in the multi-machine power system firstly and its
parameters are set. Afterwards, stabilizer B is installed with its parameters being set.
The strategy of sequence setting fits the practice of installing stabilizers in the
multi-machine power system. It is simple and easy to be used. The well-known
phase compensation method [9–11] is considered to be an approach of sequential
setting.
However, it was found that this strategy of sequential setting can cause a
problem of so-called “eigenvalue drifting” [36] due to the dynamic interactions of
multiple stabilizers.“Eigenvalue drifting” is that, for example, setting of stabilizer B
affects the setting results of stabilizer A which is designed previously. The con-
sequence is that the sequential setting cannot complete the required setting of two
stabilizers, A and B, to move oscillation modes to their required positions simul-
taneously. This phenomenon of “eigenvalue drifting” is also referred to as the
problem of dynamic interactions between multiple stabilizers.
12 1 Introduction
In order to solve the problem of “eigenvalue drifting” caused by interactions

between multiple stabilizers, the simultaneous tuning strategy has been proposed
and developed for the design of multiple stabilizers. In the simultaneous tuning,
parameters of all the stabilizers are tuned simultaneously. Hence, once the tuning is
completed, all the oscillation modes involved in the tuning are at the target posi-
tions. Simultaneous tuning is often called the coordinated design of multiple sta-
bilizers. So far there have been many methods of simultaneous tuning for different
types of stabilizers proposed and developed. Those methods to simultaneously tune
parameters of multiple stabilizers can be classified into two groups. The first group
are the analytical methods developed either from the DTA technique or by applying
the modern control theory. The second group are the methods of parameter opti-
mization. They usually convert the simultaneous tuning to the solution of an
optimization problem by establishing an objective function in the space of stabi-
lizers’ parameters. Then, an optimization algorithm is employed to find the solution
of the objective function. Normally, the second group of methods heavily relies on
the eigensolution of the power system.
Other two important issues in the design of stabilizers are the robustness of and
interactions between multiple stabilizers. The robustness is about the effectiveness
of stabilizers when power system operating conditions vary. Because the linearized
model of a power system is established at a chosen operating condition for the
design of stabilizers, the effectiveness of stabilizers is ensured only at the operating
point by the design. Hence, the robustness of stabilizers to the variations of power
system operating conditions is an issue which needs special consideration in order
to guarantee the effectiveness of the stabilizers over a certain range of system
operating conditions. The interactions are those among multiple stabilizers them-
selves and with other control functions on which they are superimposed. The
well-known examples are the “eigenvalue drifting” and interaction between a PSS
and the AVR it is attached to.
1.6 Organization of the Book
Linearized methods for the analysis and damping control of power system oscil-
lations covered by this book are based on the linearized models of power systems.
Hence, a comprehensive introduction on establishing linearized models of power
systems is presented in the book. This includes the introduction of the Heffron–
Phillips model, a well-known and relatively simpler linearized model of power
system for the study of power system oscillations. In addition, the establishment of
a more general linearized model of power systems is introduced where the full-scale
mathematical model of synchronous generators is used.
Two groups of linearized methods for the analysis and damping control of power
system oscillations are introduced in the book. They are the damping torque
1.6 Organization of the Book 13
Table 1.1 Chapter arrangement of the book

Type of stabilizers Single-machine infinite-bus Multi-machine power
Type of power systems power systems systems
The PSSs Chapter 2 Chapter 5
The thyristor-based FACTS Chapter 3 Chapter 6
stabilizers
The VSC-based FACTS Chapter 4 Chapter 7
stabilizers
analysis (DTA) and the modal analysis (MA). The basic concepts and theory of the
DTA and the MA are presented firstly for a simpler case of power systems, a
single-machine infinite-bus power system. The introduction of the applications and
extensions of the DTA and the MA in a more complicated case of a multi-machine
power system is then followed in the book.
Power system damping controllers are classified into three groups in the book,
the power system stabilizer (PSS), the thyristor-based FACTS stabilizers, and the
VSC-based stabilizers. Introduction of linearized methods for the analysis and
damping control of power system oscillations in the book is organized according to
the classification of damping controllers (stabilizers) and their applications in the
simple single-machine infinite-bus and complex multi-machine power system. The
organization is illustrated in Table 1.1.
References
1. IEEE/CIGRE Joint Task Force on Stability Terms and Definitions (2004) Definition and
classification of power system stability. IEEE Trans Power Syst 19(2):1387–1401
2. Schleif FR, White JH (1966) Damping for the northwest-southwest tieline oscillations—an
analogue study. IEEE Trans Power Appar Syst 85(12):1239–1247
3. Gibson CM (1988) Application of power system stabilizers on the Anglo-Scottish
interconnection—programme of system proving tests and operational experience. IEE Proc
Part C 135(3):255–260
4. Hsu Y-Y, Shyue S-W, Su C-C (1987) Low frequency oscillations in longitudinal power
systems: experience with dynamic stability of Taiwan power system. IEEE Trans Power Syst 2
(1):92–98
5. Western Electricity Coordinating Council (1996) Western systems coordinating council
disturbance report. www.bpa.biz
6. deMello FP, Concordia C (1969) Concepts of synchronous machine stability as affected by
excitation control. IEEE Trans Power Appar Syst 88(4):316–329
7. Heffron WG, Phillips RA (1952) Effect of modern amplidyne voltage regulators on
underexcited operation of large turbine generators. AIEE Trans (Power Appar Syst) 71:
692–697
8. deMello FP, Laskowski TF (1975) Concepts of power system dynamic stability. IEEE Trans
Power Appar Syst 94(3):827–833
9. Larsen EV, Swann DA (1981) Applying power system stabilizers part I: general concepts.
IEEE Trans Power Appar Syst 100(6):3017–3024
14 1 Introduction
10. Larsen EV, Swann DA (1981) Applying power system stabilizers part II: performance
objectives and tuning concepts. IEEE Trans Power Appar Syst 100(6):3025–3033
11. Larsen EV, Swann DA (1981) Applying power system stabilizers part III: practical
considerations. IEEE Trans Power Appar Syst 100(6):3034–3046
12. Rogers G (2000) Power system oscillations. Kluwer Academic Publisher
13. Perez-Arriaga IJ, Verghese GC, Schweppe FC (1982) Selective modal analysis with
applications to electric power systems, Part I. IEEE Trans Power Appar Syst 101(9):3117–3125
14. Perez-Arriaga IJ, Verghese GC, Schweppe FC (1982) Selective modal analysis with
applications to electric power systems, part II. IEEE Trans Power Appar Syst 101(9):3126–3134
15. Sancha JL, Perez-Arriaga IJ (1988) Selective modal analysis of electric power system
oscillatory instability. IEEE Trans Power Syst 3(2):429–438
16. Larsen EV, Sanchez-Gasca JJ, Chow JH (1995) Concept for design of FACTS controllers to
damp power swings. IEEE Trans Power Syst 10(2):948–956
17. Wang HF (1999) Selection of robust installing locations and feedback signals of
FACTS-based stabilizers in multi-machine power systems. IEEE Trans Power Syst 14
(2):569–574
18. Hingorani NG (1988, Aug) High power electronics and flexible ac transmission systems. IEEE
Power Eng Rev 3–4
19. Gyugyi L (1992) Unified power-flow control; concept for flexible AC transmission systems.
IEE Proc Part C 139(4):323–331
20. Smith OJM (1969) Power system transient control by capacitor switching. IEEE Trans Power
Appar Syst 88(1):28–35
21. Webster RH, Mane AP, Smith OJM (1971) Series capacitor switching to quench
electromechanical transient in power systems. IEEE Trans Power Appar Syst 99(2):427–433
22. Song YH, Johns AT (1999) Flexible AC transmission systems. IEE Press
23. Hingorani NG, Gyugyi L (1999) Understanding FACTS. IEEE Press
24. CIGRE TF 30-01-08 Report (1999) Modelling of power electronics equipment (FACTS) in
load flow and stability programs
25. Hsu CS, Lee WJ (1993) Superconducting magnetic energy storage for power system
applications. IEEE Trans Ind Appl 29(5):990–996
26. Xue XD, Cheng KWE, Sutanto D (2005) Power system applications of superconducting
magnetic energy storage systems. In: Fourtieth IAS annual conference meeting on industry
applications, vol 2, Oct 2005, pp 1524–1529
27. Bhargava B, Dishaw G (1998) Application of an energy source power system stabilizer on the
10 MW battery energy storage system at Chino substation. IEEE Trans Power Syst 13(1):
145–151
28. Pourbeik P, Gibbard MJ (1996) Damping and synchronizing torques induced on generators by
FACTS stabilizers in multimachine power systems. IEEE Trans Power Syst 11(4):1920–1925
29. Swift FJ, Wang HF (1995) Static Var compensator to damp system oscillations in
multi-machine power systems. In: Proceedings of IEE APSCOM conference, Hong Kong
30. Swift FJ, Wang HF (1996) Application of the controllable series compensator in damping
power system oscillations. IEE Proc Part C 143(4):359–364
31. Wang HF, Swift FJ (1997) The indexes for selecting the best locations of PSS or
FACTS-based stabilizers in multi-machine power systems: a comparison study. IEE Proc
Part C 144(2):155–159
32. Abdalla OH, Hassan SA, Tweig NT (1984) Co-ordinated stabilisation of a multimachine
power system. IEEE Trans Power Appar Syst 103(3):483–494
33. Okamoto H, Kurita A, Sekine Y (1995) A method for identification of effective locations of
variable impedance apparatus on enhancement of steady-state stability in large-scale power
systems. IEEE Trans Power Syst 10(3):1401–1407
References 15
34. Martins N, Lima LTG (1990) Determination of suitable locations for power system stabilizers
and static VAR compensators for damping electromechanical oscillations in large power
35. Yang X, Feliachi A (1994) Stabilization of inter-area oscillation modes through excitation
36. Gooi HB, Hill EF, Mobarak MA, Thonne DH, Lee TH (1981) Coordinated multimachine
stabilizer setting without eigenvalue drift. IEEE Trans Power Appar Syst 100(12):3879–3887
Chapter 2
A Single-Machine Infinite-Bus Power
System Installed with a Power System
Stabilizer
2.1 Linearized Model of a Single-Machine Infinite-Bus

Power System Installed with a Power System
Stabilizer
2.1.1 General Linearized Mathematical Model
2.1.1.1 Full Mathematical Model of a Synchronous Generator
Fundamental equations describing the dynamics of a synchronous generator are

the well-known Park’s voltage equations. They are given based on a coordinate
system consisting of a d-axis (direct axis) fixed on the field winding axis of syn-
chronous generator and a q-axis (quadrature axis). After Park’s transformation,
three armature phase windings a, b, and c on the stator of synchronous generator are
replaced by two equivalent armature phase windings, d and q. Two damper
windings on the rotor, D and Q, are permanently short-circuited. Field winding f is
DC excited. Park’s voltage equations of those five windings have the simplest form
as follows:
w_ d ¼ x0 ðvtd þ ra id þ xwq Þ
w_ q ¼ x0 ðvtq þ ra iq xwd Þ
w_ ¼ x0 ðvf rf if Þ
f ð2:1Þ
w_ D ¼ x0 rD iD
w_ Q ¼ x0 rQ iQ
where wd ; wq ; wf ; wD ; wQ , id ; iq ; if ; iD ; iQ , vtd ; vtq ; vf , and ra ; rf ; rD ; rQ are the mag-

netic flux linkage, current, voltage, and resistance of each corresponding winding,
respectively, x0 is the synchronous speed, and x is the rotor speed in per unit (p.u.)

DOI 10.1007/978-1-4899-7696-3_2
18 2 A Single-Machine Infinite-Bus Power System …
of synchronous generator. Since there is no static coupling between any d-axis

winding and q-axis winding, the relationship between the magnetic flux linkage and
the current of those windings is as follows:
2 3 2 32 3
wd xd xad xad id
6 7 6 76 7
4 wf 5 ¼ 4 xad xf xad 54 if 5
wD xad xad xD iD ð2:2Þ
" #
wq xq xaq iq
¼
wQ xaq xQ iQ
where xd ; xq ; xf ; xD ; xQ are the self-reactance of winding d, q, f, D, and Q,

respectively. In Eq. (2.2), it is assumed that the mutual reactance of all windings on
a common axis is same, being xad or xaq respectively.
The rotor motion equation of synchronous generator is as follows:
:
d ¼ x0 ðx 1Þ
: 1 ð2:3Þ
x ¼ ½Tm Tt Dðx 1Þ
M
where M is the inertia of the rotor, D the damping coefficient of the rotor motion, d
the rotor angular position of synchronous generator to a reference axis, and Tm and
Tt the mechanical torque and electric torque applied on the rotor of generator,
respectively.
2.1.1.2 Excitation System and the Automatic Voltage Regulator

(AVR)
Voltage control in a power system is closely related to the regulation of reactive

power flow. The objectives of voltage and reactive power control in the power
system are as follows:
1. To maintain the voltage at terminals of all equipment in the power system within
acceptable limits. As far as the system voltage profile is concerned, a constraint
of system voltage deviations is imposed normally to be within ±5 % of the
nominal value of voltage.
2. To improve system stability in order to maximize the utilization of the trans-
mission system.
3. To minimize reactive power flow so as to reduce transmission losses.
Power system voltage control has a hierarchy structure with multiple levels. At
the primary level, control devices attempt to compensate the rapid and random
voltage variations by maintaining their output variables close to the setting refer-
ence values. The highest level of voltage control uses global information of power
2.1 Linearized Model of a Single-Machine … 19
Fig. 2.1 Arrangement of an if

AVR
vf0 vf
Vt
-
Vtref
vf' TE(s)
+
+ u pss
AVR
system and is implemented by solving some optimization problems. A synchronous

generator can generate or absorb reactive power depending on its excitation, which
is controlled by a voltage control device, the automatic voltage regulator (AVR).
Figure 2.1 shows simple illustration on the arrangement of the AVR on the syn-
chronous generator.
In Fig. 2.1, TE(s) denotes the transfer function of the AVR. The excitation
provided to the field winding of synchronous generator is vf which is from two
sources. One is a constant excitation, vf0 , and another is the output from the AVR,
v0f . The AVR measures the terminal voltage of synchronous generator, Vt , and
compares it to a reference setting value Vtref . The AVR responds to the deviation of
terminal voltage to change the excitation of generator and hence regulates the
reactive power supply or absorption by generator. By doing so, the terminal voltage
of generator is effectively maintained close to the reference setting value. Therefore,
the AVR works at the bottom of the hierarchy of voltage control of power system to
eliminate voltage variations at generator’s terminal. On the other hand, the refer-
ence setting value of the AVR is provided by the result of voltage and/or reactive
control optimization of the whole power system such that the voltage profile of the
system is kept within required constraints and the transmission losses are mini-
mized. Hence, Vtref is from the higher level of hierarchy of voltage control of the
power system.
Historically, the role of excitation system to improve power system performance
has been growing. Early excitation systems were for the maintenance of constant
voltage level at generator’s terminals and were implemented manually. Gradually,
fast-acting AVRs were installed in many generators not only providing a satis-
factory voltage control performance, but also considerably improving power system
steady-state and transient stability. However, during 1960s to 1970s, it was found
that fast-acting AVRs have an adverse effect in providing negative damping to
power system oscillations in some occasions. This results in a conflict requirement
in the design of the AVRs. Subsequently, the power system stabilizer (PSS) was
introduced into excitation control to overcome the problem. Nowadays, it is quite
common to have a combination of a fast-acting AVR and a PSS for the excitation
control of synchronous generators. Stabilizing signal, upss , is superimposed on that
of the AVR as shown in Fig. 2.1.
synchronous generator
DC motor
slip ring
field winding of - Vt
dc motor Vtref
AVR
+
Fig. 2.2 Arrangement of a DC excitation system
Based on the difference of the excitation power sources used, excitation systems
can be classified into three major types.
1. DC excitation systems
A DC excitation system uses a DC generator as the source of excitation power to
provide field current and is connected to field winding through slip rings. The
exciter may be driven by a motor or by the generator itself. The DC excitation
system represents the early application of excitation control. Figure 2.2 shows the
arrangement of the DC excitation system.
2. AC excitation systems
An AC excitation system uses an AC machine (alternator) as the source of exci-
tation power to provide field current. The AC excitation current is rectified to
provide the DC excitation to the synchronous generator. Usually, the AC exciter is
on the same shaft of synchronous generator. There are two major types of AC
excitation systems, depending on the difference of the arrangement of AC excitation
from either the stationary or rotating armature winding of the exciter, as shown in
Figs. 2.3 and 2.4, respectively. In a rotating rectifier excitation system, the armature
windings of the AC exciter and the diode rectifiers rotate with the synchronous
generator field. Thus, the need for slip rings and brushes is eliminated. Hence, such
a system is also called a brushless excitation system.
3. Static excitation systems
In a static excitation system, power supply is from the synchronous generator.
The DC excitation is provided to the field of generator through slip rings after being
rectified. There are three major types of static excitation systems. They are
potential-source controlled-rectifier systems, compound-source rectifier systems,
and compound-controlled rectified excitation systems. Figure 2.5 shows the
arrangement of a potential-source controlled-rectifier excitation system.
alternator stationary diode synchronous generator
slip ring
Vt
field winding of -
alternator
Vtref
AVR
+
Fig. 2.3 Arrangement of a stationary rectifier excitation system
rotational armature and

alternator diode rectifier synchronous generator
Vt
field winding of − Vtref
alternator
AVR
+
Fig. 2.4 Arrangement of a rotating rectifier excitation system
controlled rectifier
Vt
−
Vtref
AVR
+
Fig. 2.5 Arrangement of a potential-source controlled-rectifier excitation system

From Fig. 2.1, it can have
vf ¼ vf0 þ TEðsÞðVtref Vt þ upss Þ ð2:4Þ
where vf0 is the constant excitation, upss the stabilizing signal of the PSS, and Vt
and Vtref the terminal voltage of generator and its reference setting value, respec-
tively. Various forms of transfer function have been recommended for different
types of excitation systems and the AVR. In this book, the following simplest form
is used for the purpose of simple presentation,
KA
TEðsÞ ¼ ð2:5Þ
1 þ sTA
where KA is the gain and TA the time constant of the AVR. Hence, from Eqs. (2.4)
and (2.5), mathematical model of the AVR can be written as
vf ¼ vf0 þ v0f
1 KA ð2:6Þ
v_ 0f ¼ v0f þ Vtref Vt þ upss
TA TA
2.1.1.3 A Single-Machine Infinite-Bus Power System
Figure 2.6 shows the configuration of a power system where a generator sends
power to a large network. Capacity of the large network is much greater than that of
the generator such that operation of the large network is not affected at all by any
changes in the part of the power system on the left-hand side of busbar b in Fig. 2.6.
This effectively means that the voltage and frequency at busbar b are constant when
the focus of the study is the part of the left-hand side of the power system. Thus,
from the point of view of operation of the part of left-hand side of the power
system, capacity of the large network is “infinite”. Hence, busbar b is called the
“infinite busbar”, and the part of the power system on the left-hand side of busbar b
is a “single-machine infinite-bus” power system. The single-machine infinite-bus
power system is an approximate representation of a kind of real power systems,
where a power plant with a generator or a group of generators are connected by
transmission lines to a very large power network.
Vt Vb
xt
A large network
Pt It
busbar b
Fig. 2.6 A single-machine infinite-bus power system

Fig. 2.7 d–q coordinate of q

generator Vt
jxt I t
vq Vb
It
iq δ
d
id vd
For the single-machine infinite-bus power system shown in Fig. 2.6, it can have
t ¼ jxtIt þ V
V b ð2:7Þ
In the d–q coordinate of the generator shown in Fig. 2.7, Eq. (2.7) can be written
as follows:
vtd þ jvtq ¼ jxt ðid þ jiq Þ þ vd þ jvq ð2:8Þ
where vtd ; vtq and id ; iq vd ; vq are the d and q components of terminal voltage of
generator, V t , line current, It , and voltage at the infinite busbar, V
b , respectively.
Comparing the real and imaginary part on the both sides of Eq. (2.8), it can have
vtd ¼ xt iq þ vd
ð2:9Þ
vtq ¼ xt id þ vq
where vd ¼ Vb sin d; vq ¼ Vb cos d and the terminal voltage of generator is as

follows:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Vt ¼ v2td þ v2tq ð2:10Þ
In per unit, the mechanical and electric torque, Tm and Tt , in Eq. (2.3), is equal to
the mechanical power input from the prime mover to the electric power supplied by
the generator, respectively, i.e. Tm ¼ Pm and Tt ¼ Pt . While Pt is equal to the
electric power received at the infinite busbar, that is
Pt ¼ vdt id þ vqt iq ¼ vd id þ vq iq ð2:11Þ

Equations (2.1)–(2.3), (2.6) and (2.9)–(2.11) are the complete dynamic model of
the single-machine infinite-bus power system shown in Fig. 2.6 where Vb and Pm
are constant.
2.1.1.4 Linearized Model of Single-Machine Infinite-Bus Power

System
Linearization of Eqs. (2.1) and (2.2) is as follows:
Dw_ d ¼ x0 ðDvtd þ ra Did þ x0 Dwq þ wq0 DxÞ

Dw_ q ¼ x0 ðDvtq þ ra Diq x0 Dwd wd0 DxÞ
Dw_ ¼ x0 ðDvf rf Dif Þ
f ð2:12Þ
Dw_ D ¼ x0 rD DiD
Dw_ Q ¼ x0 rQ DiQ
2 3 2 31 2 3
Did xd xad xad Dwd
6 7 6 7 6 7
4 Dif 5 ¼ 4 xad xf xad 5 4 Dwf 5
DiD xad xad xD DwD ð2:13Þ
1 " #
Diq xq xaq Dwq
¼
DiQ xaq xQ DwQ
where prefix, D, and subscript 0 are used to denote small increment of a variable
(linearized variable) and value of the variable at the power system steady-state
operating condition where the linearization is carried out, respectively. This nota-
tion will be used throughout this book.
Linearization of Eqs. (2.10) and (2.11) is as follows:
vtd0 vtq0
DVt ¼ Dvtd þ Dvtq ð2:14Þ
Vt0 Vt0
DPt ¼ vdt0 Did þ vqt0 Diq þ id0 Dvdt þ iq0 Dvqt ð2:15Þ
By using Eqs. (2.14) and (2.15), linearization of Eqs. (2.3) and (2.6) can be
obtained to be
Dd_ ¼ x0 Dx
1
Dx_ ¼ ðDPt þ DDxÞ ð2:16Þ
M
1
¼ ðvdt0 Did þ vqt0 Diq þ id0 Dvdt þ iq0 Dvqt þ DDxÞ
M
Dvf ¼ Dv0f

1 KA vtd0 vtq0 ð2:17Þ
Dv_ 0f 0
¼ Dvf þ Dvtd Dvtq þ Dupss
TA TA Vt0 Vt0
Arranging Eqs. (2.12), (2.13), (2.16), and (2.17) in matrix form with all lin-
earized current variables be cancelled, it can have
sDXgdq ¼ Agdq DXgdq þ Bgdq DVdq þ bpss Dupss

ð2:18Þ
DIdq ¼ Cgdq DXgdq
where

T
DXgdq ¼ Dd Dx Dvf Dwd Dwq Dwf DwD DwQ ;
T T
DVdq ¼ ½ Dvtd Dvtq ; DIdq ¼ ½ Did Diq
For the single-machine infinite-bus power system, the network voltage equation
is Eq. (2.9). Its linearization is as follows:
Dvtd ¼ xt Diq þ Dvd ¼ xt Diq þ Vb cos d0 Dd

ð2:19Þ
Dvtq ¼ xt Did þ Dvq ¼ xt Did Vb sin d0 Dd
In matrix form, the above equation can be written as follows:
DVdq ¼ Fdq1 DIdq þ Fdq2 DXgdq ð2:20Þ
where

0 xt Vb cos d0 0
Fdq1 ¼ ; Fdq2 ¼
xt 0 Vb sin d0 0
Substituting Eq. (2.20) into (2.18), state-equation model of the single-machine

infinite-bus power system is obtained to be
sDXgdq ¼ Agcdq DXgdq þ bpss Dupss ð2:21Þ
where Agcdq ¼ Agdq þ Bgdq Fdq1 Cgdq þ Bgdq Fdq2 :

2.1.2 Heffron–Phillips Model
2.1.2.1 Simplification
For the study of power system oscillations, full mathematical model of synchronous
generator of Eqs. (2.1)–(2.2) can be simplified based on the following
considerations:
1. Effect of damper windings is not considered or directly included in the damping
coefficient D in the rotor motion equation in Eq. (2.3). Thus, Eq. (2.1) is sim-
plified to be
w_ ¼ x0 ðvtd þ ra id þ xw Þ
d q
w_ q ¼ x0 ðvtq þ ra iq xwd Þ ð2:22Þ

w_ f ¼ x0 ðvf rf if Þ
2. Effect of fast transient and the resistance of d and q armature windings are
neglected. Equation (2.22) is further simplified to be
0 ¼ vtd þ xwq
0 ¼ vtq xwd ð2:23Þ
w_ f ¼ x0 ðvf rf if Þ
3. In small-signal power oscillations, variation of rotor speed is very small, x 1.

Hence, the first two equations in Eq. (2.23) become
vtd ¼ wq
ð2:24Þ
vtq ¼ wd
To transform the third equation in Eq. (2.23) into a different form, it is defined
that
xad xad vf
E0q ¼ w ; Eq ¼ xad if ; Efd ¼ ð2:25Þ
xf f rf
where E0q is called the q-axis transient excitation voltage, Eq the q-axis excitation
voltage, and Efd the excitation voltage. Multiplying both sides of the third equation
in Eq. (2.23) by xradf , it can have
0
T0d0 E_ q ¼ Efd Eq ð2:26Þ
where T0d0 ¼ xx0 rf f , which is the time constant of the field winding.
Equation (2.26) together with Eq. (2.3) forms the simplified third-order model of
synchronous generator. Equation (2.2) becomes

wd xd xad id
¼
wf xad xf if ð2:27Þ
wq ¼ xq iq
From Eqs. (2.24), (2.25), and (2.27), it can have
vtd ¼ wq ¼ xq iq
ð2:28Þ
vtq ¼ wd ¼ xad if xd id ¼ Eq xd id
From Eqs. (2.25) and (2.27), it can be obtained that
xad xad x2
E0q ¼ wf ¼ ðxf if xad id Þ ¼ Eq ad id ¼ Eq xd x0d id ð2:29Þ
xf xf xf
x2
where x0d ¼ xd xadf , which is called the transient d-axis reactance. Thus, Eq. (2.26)
becomes
0
T0d0 E_ q ¼ Efd E0q xd x0d id ð2:30Þ
2.1.2.2 A Simplified Model of Single-Machine Infinite-Bus Power

System
For the single-machine infinite-bus power system shown in Fig. 2.6, from Eqs. (2.9)
and (2.28), it can have
vtd ¼ vd xt iq ¼ xq iq
ð2:31Þ
vtq ¼ vq þ xt id ¼ Eq xd id ¼ E0q x0d id
Thus,
vd ¼ ðxt þ xq Þiq ¼ xqR iq

ð2:32Þ
vq ¼ Eq ðxd þ xt Þid ¼ E0q ðx0d þ xt Þid ¼ Eq xdR id ¼ E0q x0dR id
Fig. 2.8 Equivalent circuit iq id

model of single-machine + +
infinite-bus power system
x dΣ (x dΣ ')
x qΣ vd vq
Eq (Eq ')
d winding
- -
q winding
Fig. 2.9 Phasor diagram of q
single-machine infinite-bus EQ
power system
j(x q − x 'd )I t
Eq '
jx 'd I t
jx t I t
Vt b
vq Vb
δ
iq It
d
id vd
The single-machine infinite-bus power system can be represented by a circuit

model of Fig. 2.8. Figure 2.9 shows the phasor diagram of the system on the d–q
coordinate.
From Eq. (2.32) or Fig. 2.8, it can have
E0q Vb cos d
id ¼
x0dR
ð2:33Þ
Vb sin d
iq ¼
xqR
By substituting Eqs. (2.9) and (2.33) into Eq. (2.11), the electric power supplied
by the generator can be expressed as follows:
0
Vb sin d Eq Vb cos d
Pt = Vb cos d þ Vb sin d
xqR x0dR
0 ð2:34Þ
Eq Vb V2 xq x0d
¼ 0 sin d b 0 sin 2d
xdR 2 xdR xqR
E0q Vb cos d
Eq ¼ E0q þ xd x0d id ¼ E0q þ xd x0d
x0dR
0 0
ð2:35Þ
Eq xdR xd xd Vb cos d
¼ 0
xdR x0dR
where xdR ¼ xd þ xt . From Eqs. (2.31) and (2.33), it can have
Vb sin d xq Vb sin d
vtd = Vb sin d xt iq ¼ Vb sin d xt ¼ ;
xqR xqR
ð2:36Þ
E0q Vb cos d xt E0q Vb x0d cos d
vtq = Vb cos d þ xt id ¼ Vb cos d þ xt ¼ þ
x0dR x0dR x0dR
Hence, the simplified model of single-machine infinite-bus power system is as

follows:
:
d ¼ x0 ðx 1Þ
: 1
x ¼ ½Pm Pt Dðx 1Þ
M
0 1 ð2:37Þ
E_ q ¼ 0 Eq þ Efd
Tdo
0 1 KA
E_ fd ¼ E0fd þ Vtref Vt þ upss
TA TA
where

E0q Vb V2b xq x0d
Pt ¼ 0 sin d sin 2d
xdR 2 x0dR xqR

E0q xdR xd x0d Vb cos d
Eq ¼ 0
xdR x0dR ð2:38Þ
Efd ¼ Efd0 þ E0fd
xq Vb sin d xt E0q Vb x0d cos d qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
vtd ¼ ; vtq ¼ 0 þ 0 ; Vt ¼ v2td þ v2tq
xqR xdR xdR
The model is a group of 4 first-order differential equations plus 6 algebraic

equations.
2.1.2.3 Heffron–Phillips Model [1–3]
By linearizing Eqs. (2.37) and (2.38) at an operating point of power system, where
Vt ¼ Vt0 ; Vtd ¼ Vtd0 ; Vtq ¼ Vtq0 ; d ¼ d0 ; x0 ¼ 1; E0q ¼ E0q0 ; Efd ¼ Efd0 , it can have
Dd_ ¼ xo Dx
1
Dx_ ¼ ðDPt DDxÞ
M
0 1 ð2:39Þ
DE_ q ¼ 0 ðDEq þ DE0fd Þ
Tdo
0 1 KA
DE_ fd ¼ DE0fd ðDVt Dupss Þ
TA TA
DPt ¼ K1 Dd þ K2 DE0q
DEq ¼ K3 DE0q þ K4 Dd ð2:40Þ
DVt ¼ K5 Dd þ K6 DE0q
where
E0q0 Vb x0dR V 2 ð xq x0 Þ
K1 ¼ cos d0 bx0 xqR d cos 2d0
dR
K2 ¼ xV0 b sin d0
dR
K3 ¼ xxdR
0
dR
ðxd x0d ÞVb sin d0
K4 ¼ x0dR
X q Vb cos d0 V V x0 sin d0
K5 ¼ VVt0
td0
xqR Vtq0 b0 xd0
t0 dR
V
K6 ¼ Vtq0 xx0 t
t0 dR
Substituting Eq. (2.40) into Eq. (2.39), it can be obtained that
Dd_ ¼ xo Dx
1
Dx_ ¼ ðK1 Dd K2 DE0q DDxÞ
M
_ 0 1 ð2:41Þ
DEq ¼ 0 ðK3 DE0q K4 Dd þ DE0fd Þ
Tdo
0 1 KA
DE_ fd ¼ DE0fd ðK5 Dd þ K6 DE0q Dupss Þ
TA TA
K1
ΔPt Δω ω0 Δδ
_ 1
Ms + D s
K4 K5
K2
_ _
ΔEq’ 1 + KA + Δupss
Td0's+K3 1+sTA _
K6
Fig. 2.10 Heffron–Phillips model of a single-machine infinite-bus power system
Equation (2.41) is the so-called Heffron–Phillips model of single-machine

infinite-bus power system, which is shown in Fig. 2.10.
The Heffron–Phillips model can be written in the form of state-space repre-
sentation of Eq. (2.21) where
2 3
2 3 0 xo 0 0
Dd 6 7
6 Dx 7 6 K 1
MD K 2
0 7
6 7 6 M M 7
DXgdq ¼6 7; Agcdq ¼6
6 K0 4 1 7;
7
4 DE0q 5 6 0 K0 3
4 Tdo Tdo T0do 7
5
DE0fd TA K5
K 0 TA K6
K T1 ð2:42Þ
2 3 A A A
0
6 7
6 0 7
bpss ¼6
6 0 7
7
4 5
KA
TA
2.2 Modal Analysis
2.2.1 Basis of Modal Analysis Theory
2.2.1.1 Modal Decomposition
State-space representation of a linear system is as follows:
sX ¼ Ao X þ bo u
y ¼ cTo X ð2:43Þ
u ¼ HðsÞy
where Ao , bo and cTo is the state matrix, control vector, and output vector of
open-loop system, respectively, and HðsÞ is the transfer function of feedback
controller. Transfer function of open-loop system is as follow:
y
GðsÞ ¼ ¼ cTo ðsI Ao Þ1 bo ð2:44Þ
u
The system is shown in Fig. 2.11. Transfer function of closed-loop system is as

follows:
y GðsÞ
TðsÞ ¼ ¼ ð2:45Þ
w 1 GðsÞHðsÞ
Eigen solution is one of the basic techniques in the modal analysis, involving the
computation of eigenvalues and eigenvectors of state matrix, Ao . An eigenvalue of
matrix Ao , k, is a scalar parameter, which satisfies the following equation
Ao v ¼ kv ð2:46Þ
with a non-trivial solution ðv 6¼ 0Þ.
Fig. 2.11 Block diagram of a closed-loop control system

2.2 Modal Analysis 33
Obviously, Eq. (2.46) can be written in the following form
ðAo kIÞv ¼ 0 ð2:47Þ
where I is an unity matrix. In order for Eq. (2.47) to have the non-trivial solution, it
should have
jAo kIj ¼ 0 ð2:48Þ
Equation (2.48) is the following polynomial equation if Ao is an M M matrix
ð1ÞM kM þ aM1 kM1 þ þ a1 k þ a0 ¼ 0 ð2:49Þ
which is called the characteristic equation of state matrix Ao . The characteristic

equation should have M solutions; that is, matrix Ao has M eigenvalues, if the
dimension of matrix is M.
For the ith eigenvalue of matrix Ao , ki , if a nonzero vector vi satisfies the
equation
Avi ¼ ki vi ; i ¼ 1; 2; . . . ; M ð2:50Þ
vi is called the right eigenvector of matrix A associated with ki . Equation (2.50)

can be arranged as follows:
2 3
k1 0 0 0
60 k2 0 0 7
6 7
A½ v1 v2 . . . vM ¼ ½ v1 v2 . . . vM 6 .. 7 ð2:51Þ
40 0 . 0 5
0 0 0 kM
That is
V1 AV ¼ K ð2:52Þ
where
2 3
k1 0 0 0
60 k2 0 0 7
6 7
V ¼ ½ v1 v2 ... vM ; K ¼ 6 .. 7
40 0 . 0 5
0 0 0 kM
Denote
2 3
wT1
6 wT 7
6 2 7
V 1
¼6
6 ..
7 ¼ ½ w1
7 w2 . . . w M T ¼ WT ð2:53Þ
4 . 5
wTM
From Eqs. (2.52) and (2.53), obviously it can have
wTi A ¼ wTi ki ; i ¼ 1; 2; . . .; M ð2:54Þ
Hence, wTi is called the left eigenvector corresponding to eigenvalue ki .

If a new state variable vector z is introduced and defined to be
X ¼ VZ ð2:55Þ
From Eqs. (2.43), (2.52), and (2.53), it can be obtained that
sZ ¼ KZ þ WT bo u
ð2:56Þ
y ¼ cTo VZ
That is
szi ¼ ki zi ; þ wTi bo u i ¼ 1; 2; . . .; M
X
M ð2:57Þ
y ¼ cTo vi z i
i¼1
According to Eq. (2.57), the system can also be shown in Fig. 2.12. This is the
modal decomposition representation of state-space model of open-loop system.
1 z1
w1T b 0 c0 T v1
s − λ1
1 z2
w 2Tb0 c0 T v 2
s − λ2
u y
+
1 zM
w M Tb0 c0 T v M
s − λM
Fig. 2.12 Block diagram of modal decomposition representation of state-space model of

open-loop system
2.2.1.2 Stability of Open-Loop System and Closed-Loop System
Considering the open-loop system when u ¼ 0, the state-space representation of

Eq. (2.57) is as follows:
szi ¼ ki zi ð2:58Þ
Solution of Eq. (2.58) is as follows:
zi ðtÞ ¼ zi ð0Þeki t ; i ¼ 1; 2; . . .; M ð2:59Þ
where zi ð0Þ is the initial value of state variable zi ðtÞ; i ¼ 1; 2; . . .; M. From

Eqs. (2.55) and (2.59), it can have
2 3
z1 ð0Þek1 t
6 z2 ð0Þek2 t 7
6 7
X ¼ V6 .. 7 ð2:60Þ
4 . 5
zn ð0ÞekM t
Hence, time response of the kth state variable of the system,

xk ðtÞ; i ¼ 1; 2; . . .; M, is as follows:
X
M
xk ðtÞ ¼ vk1 z1 ð0Þek1 t þ vk2 z2 ð0Þek2 t þ þ vkM zM ð0ÞekM t ¼ vki zi ð0Þeki t
i¼1
ð2:61Þ
Obviously, the time response of system state variables is decided by the

eigenvalues of state matrix Ao . If there is one or more eigenvalues on the right-hand
half of the complex plan (the real part of eigenvalue is equal to or greater than zero),
the system is unstable. If all the eigenvalues of Ao are on the left-hand side of the
complex plane, the system is stable. Hence, eigenvalues of Ao determine the system
stability. They often are called the modes of the system. If a pair of eigenvalues are
conjugate complex number, i.e. k i;i þ 1 ¼ n jxi , the corresponding component in
i
the time response of the kth state variable of the system should be

vki zi ð0Þeki t ¼ vki zi ð0Þeðni þ jxi Þt ¼ vki zi ð0Þeni t ½cos xi t þ j sin xi t
ð2:62Þ
The component is oscillatory in respect of time. The oscillation angular fre-

quency is xi . The decaying and increasing of the oscillation are determined by the
real part of the mode ni . The pair of conjugate eigenvalues of state matrix Ao are
often called the oscillation mode of the system.
The oscillation frequency f i (Hz) and damping fi associated with k i;i þ 1 ¼
ni jxi are normally defined as follows:
xi ni
fi ¼ ; f ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð2:63Þ
2p i n þ x2
2
i i
From Eq. (2.59), it can be seen that zi ðtÞ; i ¼ 1; 2; . . .; M is related only with the
ith mode of the system ki . Hence, zi ðtÞ; i ¼ 1; 2; . . .; M often is seen as the ith mode
of the system. Equation (2.57) is often called the modal decomposition of
state-space representation.
From Eq. (2.61), it can also be seen that the magnitude of vki measures how
much the ith mode ki contributes to the kth state variable xk ðtÞ. Thus, jvki j is a kind
of measurement of the “observability” of the ith mode in the kth state variable.
On the basis of above discussion, from Fig. 2.12, it can be seen that wTi bo is the
weight on how much the control signal u affects the ith mode of the open-loop
system, the so-called controllability index, whereas cTo vi is the weight on how much
the ith mode is observed in the system output, which is called the observability
index. The product of controllability and observability index is called the residue.
That is
Ri ¼ wTi bo cTo vi ð2:64Þ
From Eqs. (2.53) and (2.55), it can have
Z ¼ V1 X ¼ WT X ð2:65Þ
or
zi ðtÞ ¼ w1i x1 ðtÞ þ w2i x2 ðtÞ þ þ wMi xM ðtÞ ð2:66Þ
wki is the ith row kth column element of matrix W. Equation (2.66) indicates that
the magnitude of wki measures the influence of the kth state variable xk ðtÞ on the ith
state variable zi ðtÞ, or the ith mode ki of the system. It is a kind of measurement of
“controllability” of the kth state variable on the ith mode.
Let the realization of the transfer function of feedback controller HðsÞ be
sXf ¼ Af Xf þ bf y
ð2:67Þ
u ¼ cTf Xf
That is HðsÞ ¼ cTf ðsI Af Þ1 bf . Thus, from Eqs. (2.43) and (2.67), the
state-space representation of closed-loop system can be obtained to be

sX Ao bo cTf X X
¼ ¼ Ac ð2:68Þ
sXf bf cTo Af Xf Xf
where Ac is the state matrix of closed-loop system. Obviously, based on the dis-
cussion above, eigenvalues of Ac or modes of closed-loop system determine the
stability of closed-loop system.
From Eq. (2.68), it can be obtained that
Ac ¼ Ao þ HðsÞbo cTo ð2:69Þ
Denote a variable parameter of feedback controller as a. Thus, state matrix is a

function of the parameter. Influence of the parameter on the ith mode of closed-loop
system can be calculated by use of the following equation
@ki @Ac ðaÞ @Hðki ; aÞ T T @Hðki ; aÞ

¼ wTi vi ¼ wi bo co vi ¼ Ri ð2:70Þ
@a @a @a @a
Hence, the residue measures how much the mode of closed-loop system is
affected by the parameter of the controller.
2.2.2 Applications of Modal Analysis
2.2.2.1 Modal Analysis for the AVR
Consider the simple case of a single-machine infinite-bus power system expressed

by
the Heffron–Phillips
model shown in Fig. 2.10 without the PSS installed
Dupss ¼ 0 . The upper part can be considered as the open-loop system and lower
part the feedback controller. Thus, Eq. (2.41) can be rearranged as follows:
" # " #
sDd 0 xo Dd 0
¼ þ DE0q
sDx K 1
MD Dx KM
2
M

Dd ð2:71Þ
Dd ¼ ½ 1 0
Dx
DE0q ¼ Fdelta ðsÞDd
In the above state-space model of the power system, the input to the open-loop
system is DE0q , the output is Dd, and the transfer function of feedback controller is
Fdelta ðsÞ. Obviously, the state-space realization of Fdelta ðsÞ is as follows:
2 3 2 3
K0 3 1 0 K0 4
sDE0q 6 Tdo 0
Tdo 7 DE Tdo 5
¼4 5
q
þ4 Dd
sDE0fd K
TA K6
T 1 DE0
TA K5
K
A A
fd
A ð2:72Þ
0
DEq
DE0q ¼ ½ 1 0
DE0fd
According to Eq. (2.48), the modes of open-loop system can be found by solving
the following characteristic equation

0 xo 1 0 k xo D K1
D k 0 1 ¼ K1 D k ¼ k þ M k þ M xo ¼ 0
2
K1 M M
M M
ð2:73Þ
Solution of Eq. (2.73) gives the oscillation mode of the single-machine

infinite-bus power system when DE0q ¼ 0, that is the case when the generator is
modelled as a constant voltage source. The oscillation mode is as follows:
2 ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
s 3

1;2 14 D D 2 4x0 K1 5
k ¼ ¼ no jxNOF ð2:74Þ
2 M M M
The oscillation mode is related to the rotor motion of generator, i.e. state vari-
ables Dd and Dx. It is often called the electromechanical oscillation mode of the
power system.
From Eq. (2.71), it can have

sDd 0 xo Dd Dd
¼ K K D ¼ Ac ð2:75Þ
sDx M M2 Fdelta ðsÞ
1
M Dx Dx
where Ac is the state matrix of closed-loop system.


2 31 23
s þ K0 3 1
0 K0 4
6 Tdo T 7 4 Tdo 5
Fdelta ðsÞ ¼ ½ 1 0 4 5
do
KA K6
TA sþ T1
TA K5
K
A A
2 32 3
s þ T1 1
K0 4
1 T0do Tdo 5
4 54
A
¼ ½1 0 K K
s þ K0 3 s þ T1 þ KTA K6 10 TA 6 s þ K0 3 TA K5
K
Tdo A A Tdo A Tdo A
2 3
T0do TA h i K0 4
s þ T1 1 4 Tdo 5
¼ T0do
ðsT0do þ K3 ÞðTA s þ 1Þ þ KA K6 A
TA K5
K
A
ðTA s þ 1ÞK4 þ KA K5
¼
ðsT0doþ K3 ÞðTA s þ 1Þ þ KA K6
ð2:76Þ
Hence, according to Eq. (2.70), it can have
ðTA ki þ 1ÞK4 þ KA K5

@ 0
@ ki ð i do þ K3 ÞðTA ki þ 1Þ þ KA K6
k T
¼ Ri ð2:77Þ
@a @a
where a ¼ KA or TA . By using Eq. (2.77), effect of the AVR on the damping of

electromechanical oscillation, i.e. power oscillation, can be examined.
2.2.2.2 Modal Analysis for the PSS
The general linearized model of the single-machine infinite-bus power system with
the PSS installed is Eq. (2.21) which can be rearranged as follows:
2 3 2 32 3 2 3
sDd 0 x0 0 Dd 0
6 7 6 a21 a22 aT238 76 7 6 7
4 sDx 5 ¼ 4 54 Dx 5 þ 4 0 5Dupss
M M M
sDx38 a138 a238 A33 Dx38 bpss3
2 3
Dd ð2:78Þ
6 7
y ¼ Dx ¼ ½ 0 1 0 4 Dx 5
Dx38
Dupss ¼ Tpss ðsÞDx
a 21
1 Δω ω0 Δδ
- Ms + a 22 s
ΔTe
a 23-8 T a 23-8 a13-8
ΔX 3-8
( sI - A 33 ) −1 + bpss-3 Δu pss
Fig. 2.13 General linearized model of single-machine infinite-bus power system with PSS
installed
The model is shown in Fig. 2.13.

i and the
Let the oscillation mode of the system without the PSS installed be k
corresponding left and right eigenvector be
2 3

vi1
wTi ¼ wi1 wi2 wTi3 ; vi ¼ 4 vi2 5 ð2:79Þ
vi3
That is
2 3

0 x0 0
wi1 wi2 wTi3 4 a21 a22 aT238 i wi1

5¼k wi2 wTi3 ð2:80Þ
M M M
a138 a238 A33
It can have
a21 i w
i2
w Ti3 a138 ¼ k
þw i1
M
a22 i w
x0 w i2
i1 w Ti3 a238 ¼ k
þw i2 ð2:81Þ
M
aT i w
i2 238 þ w
w Ti3 A33 ¼ k Ti3
M
Hence,
aT238
Ti3 ¼ w
w i2 ðki I A33 Þ1 ð2:82Þ
M
From Eqs. (2.64), (2.78), and (2.82), the residue can be obtained to be
2 3 2 3
0 vi1

6 7 6 7
Ri ¼ wi1 wi2 Ti3 4
w 0 5½ 0 1 0 4 vi2 5
bpss3 vi3
ð2:83Þ
¼ wTi3 bpss3 vi2
aT i I A33 Þ1 bpss3
¼ w i2 vi2 238 ðk
M
2.2.2.3 Design of PSS by Pole Assignment
If the feedback signal and transfer function of the PSS to be designed is y and
TPSS ðsÞ respectively, Eq. (2.78) can be written more generally as follows:
sX ¼ AX þ bDupss
y ¼ cT X ð2:84Þ
Dupss ¼ Tpss ðsÞy
Transfer function of open-loop system is as follows:
GðsÞ ¼ cT ðsI AÞ1 b ð2:85Þ
Characteristic equation of closed-loop control system is as follows:
1 þ GðsÞTpss ðsÞ ¼ 0 ð2:86Þ
If design of the PSS is to assign the electromechanical oscillation mode of the

single-machine infinite-bus power system to a target position, k c ¼ n jxc ,
c

kc must be the solution of the characteristic equation of closed-loop control system
of Eq. (2.86). Hence, it should have
c ÞTpss ðk
1 þ Gðk c Þ ¼ 0 ð2:87Þ
By separating the real and imaginary part of the above equation, two equations
will be obtained which can be used to determine two parameters of the transfer
function of PSS. If the transfer function of PSS adopts the following format of a
lead–lag block,
ð1 þ saTÞ2
Tpss ðsÞ ¼ Kpss ð2:88Þ
ð1 þ sTÞ2
With a predetermined T, parameters of the PSS, Kpss and a , can be set according
to Eq. (2.87), thus completing the design of PSS via the pole assignment.
2.3 Damping Torque Analysis
2.3.1 Damping Torque and Synchronizing Torque
2.3.1.1 Damping Torque and Synchronizing Torque Derived

from Heffron–Phillips Model
The damping torque analysis (DTA) was firstly introduced on the basis of the
Heffron–Phillips model for a single-machine infinite-bus power system to examine
the effect of excitation control, such as the AVR, on power system small-signal
stability [1–3]. It was developed based on the understanding that the dynamic of the
electromechanical oscillation loop of a generator decides the damping of power
oscillations in the single-machine infinite-bus power system.
In the Heffron–Phillips model shown in Fig. 2.10, the upper part obviously is the
linearized rotor motion equation and lower part is formed from the mathematical
description of dynamic of the field winding of generator and the AVR. Figure 2.14
shows the upper part of the model which is called the electromechanical oscillation
loop. Signal DTe from the lower part in the Heffron–Phillips model is obviously an
electric torque. Hence, from Fig. 2.14, it can have
K1
Electromechanical oscillation loop
Δω ω0 Δδ
_ 1
Ms + D s
ΔTe
Fig. 2.14 Electromechanical oscillation loop of generator

2.3 Damping Torque Analysis 43
D x0 K1 x0
s2 Dd þ sDd þ Dd þ DTe ¼ 0 ð2:89Þ
M M M
If firstly the contribution from the lower part of Heffron–Phillips model, DTe , is
not considered, the electromechanical oscillation loop of generator shown in
Fig. 2.14 is described by the following second-order differential equation
D x0 K1
s2 Dd þ sDd þ Dd ¼ 0 ð2:90Þ
M M
Equation (2.90) in fact is the linearized model of the single-machine infinite-bus

power system when the dynamic of exciter and the AVR are not considered. This is
the case when the generator is modelled only by the rotor motion equation in
Eq. (2.3).
DdðtÞ ¼ a0 e2Mt cos xNOF t þ b0

D
ð2:91Þ
r ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2
where a0 and b0 are two constants and xNOF ¼ 1 D 4x0 K1 .
2 M M
Equation (2.91) describes the behaviour of rotor motion, i.e. the acceleration and
deceleration to store or release electric power. Hence, it determines the variations of
active power supplied by the generator during dynamic transient (electromechanical
transient), when the power system is subject to small disturbances. If 2D M is small or
negative, a poorly damped or magnitude-increasing power oscillations occur. This
is the electromechanical oscillation associated with the rotor motion of synchronous
generator, i.e. the power system low-frequency oscillation.
In Eq. (2.91), xNOF is called the angular frequency of natural oscillation. The
angular oscillation frequency, xs , of power oscillation in the single-machine
infinite-bus power system is normally very close to the angular frequency of natural
oscillation. Equation (2.91) indicates that the damping of power oscillation of the
single-machine infinite-bus power system is determined by the coefficient of the
first-order derivative in the second-order differential equation in Eq. (2.91) M D.
At the angular oscillation frequency xs , the electric torque contributed from the
lower part of Heffron–Phillips model can be decomposed into two components
DTe ¼ Td Dx þ Ts Dd ð2:92Þ
Equation (2.89) becomes

D Td x0 K1 x0 Ts
s2 Dd þ þ sDd þ þ Dd ¼ 0 ð2:93Þ
M Mx0 M M
Obviously, from the discussion on Eq. (2.91), it is easy to understand that the
component in the decomposition of DTe , Td Dx, contributes to the damping of
power oscillation. This component is called the damping torque. In Eq. (2.92), Ts Dd
is called the synchronizing torque.
2.3.1.2 Electric Torque Contributed from the PSS
From Fig. 2.10, it can be seen that the electric torque contributed from the lower
part of Heffron–Phillips model is as follows:
DTet ¼ Fdelta ðsÞDd þ Fpss ðsÞDupss ð2:94Þ
where Fdelta ðsÞ and Fpss ðsÞ are the transfer function from Dd and Dupss respectively,
to form the electric torque contribution to the electromechanical oscillation loop of
generator.
The electric torque contribution from the PSS is as follows:
DTpss ¼ Fpss ðsÞDupss ð2:95Þ
Figure 2.15 shows that the PSS contributes the electric torque, DTpss , to the
electromechanical oscillation loop of generator. Obviously, Fpss ðsÞ is the transfer
function of forward path from the stabilizing signal of the PSS to the
ΔΤpss
K2
ΔE q’ 1 KA + Δupss
Td0's+K3 1+sTA
K6
Fig. 2.15 Forward path of stabilizing signal of the PSS

electromechanical oscillation loop of generator in the Heffron–Phillips model.

From Fig. 2.15, it can be obtained that
1 KA
K3 þ sT0d0 1 þ sTA K
Fpss ðsÞ ¼ K2 ¼ K2 0
A ð2:96Þ
1 þ K6 1
0
K A K3 þ sT d0 ð1 þ sTA Þ þ K6 KA
K3 þ sTd0 1 þ sTA
Hence, at the angular oscillation frequency, xs , the electric torque provided by

the PSS to the electromechanical oscillation loop is as follows:
pss ðjxs ÞDupss

DTpss ¼ F
For example, if the PSS is a pure-gain controller and takes the deviation of rotor
speed of generator as the feedback signal, i.e. Dupss ¼ Kpss Dx, the electric torque
contributed from the PSS to the electromechanical oscillation loop of generator is as
follows:
pss ðjxs ÞDx ¼ Kpss Re½F

DTpss ¼ Kpss F pss ðjxs ÞDx þ jKpss Im½F
pss ðjxs ÞDx ð2:97Þ

where Re F pss ðjxs Þ and Im Fpss ðjxs Þ denote the real and imaginary part of
pss ðjxs Þ, respectively (this notation will be used throughout the book). From the
F
first equation in Eq. (2.41), it can have sDd ¼ x0 Dx, i.e.
jxs
Dx ¼ Dd ð2:98Þ
x0
By substituting Eqs. (2.98) into Eq. (2.97), it can be obtained that

pss ðjxs ÞDx xs Kpss Im F

DTpss ¼ Kpss Re½F pss ðjxs Þ Dd ð2:99Þ
x0
pss ðjxs ÞDx, and the

Hence, the damping torque supplied by the PSS is Kpss Re½F
xs pss ðjxs ÞDd.
synchronizing torque is x0 Kpss Im½F
2.3.1.3 Damping Torque and Synchronizing Torque Derived

from the General Linearized Model
In Eq. (2.78) or Fig. 2.13, denote DTe ¼ aT238 Dx38 . From Eq. (2.78) or
Fig. 2.13, it can have
a22 a21
s2 Dd þ sDd þ x0 Dd þ DTe ¼ 0 ð2:100Þ
M M
Taking the similar procedure of DTA presented above in Sects. 2.3.1.1 and
2.3.1.2, transfer function of the forward path of stabilizing signal of the PSS can be
obtained to be
Fpss ðsÞ ¼ aT238 ðsI A33 Þ1 bpss3 ð2:101Þ
At the angular oscillation frequency, xs , the electric torque contribution from the
PSS is as follows:
DTpss ¼ aT238 ðjxs I A33 Þ1 bpss3 Dupss ð2:102Þ
The electric torque can be decomposed into the damping and synchronizing
torque. The damping torque contribution from the PSS determines its effect on the
damping of power oscillation.
Assume that the installation of PSS brings about a change of damping coefficient
Dpss Dx in the electromechanical oscillation loop of generator. The state-space
representation of power system with the PSS installed can be equivalently written
as follows:
2 3 2 32 3
sDd 0 x0 0 Dd
6 7 6 a21 a22 þ Dpss aT238 76 7
4 sDx 5 ¼ 4 M 54 Dx 5 ð2:103Þ
M M
Dx38 a138 a238 A33 Dx38
i
@k @Ac
Ti
¼w vi
@Dpss @Dpss
2 32 3
0 x0 0 vi1

@ 6 aT238 76 7
¼ w i1 i2
w i3
w T
4 a21
a22 þ Dpss
M 54 vi2 5 ð2:104Þ
@Dpss M M
a138 a238 A33 vi3
i2
w vi2
¼
M
From Eqs. (2.83), (2.102), and (2.104), it can be seen that the residue in fact
measures the effect of the PSS on the electromechanical oscillation mode of the
system. At the complex frequency k i , it is equal to the forward path of the PSS
multiplied by the sensitivity of the mode to the damping torque contribution.
2.3.2 Damping Torque Analysis and Design of PSS

by Phase Compensation
2.3.2.1 Theoretical Basis of the Damping Torque Analysis
This section explains the theoretical basis of the damping torque analysis by use of
the Phillips–Heffron model as follows.
Firstly, the effect of PSS is not considered, i.e. Dupss ¼ 0. From Fig. 2.10, it can
have
2
Ms þ Ds þ x0 K1 DdðsÞ ¼ x0 DTðsÞ
ð2:105Þ
DTðsÞ ¼ Fdelta ðsÞDdðsÞ
where Fdelta ðsÞ is the transfer function from DdðsÞ to DTðsÞ. Combining two
equations above gives
2
Ms þ Ds þ x0 K1 þ x0 Fdelta ðsÞ DdðsÞ ¼ 0 ð2:106Þ
Thus, characteristic equation of the system is as follows:
Ms2 þ Ds þ x0 K1 þ x0 Fdelta ðsÞ ¼ 0 ð2:107Þ
Solutions of the characteristic equation are the eigenvalues of state matrix of

system model given by Eq. (2.42). One of the pair of complex solutions is called the
electromechanical oscillation mode. Its real part defines the damping of power
s ¼ n þ jxs . In the complex frequency domain, it
oscillation. Denote the mode as k s
should have
2 þ Dk
Mk s þ x0 K1 þ x0 F s Þ ¼ 0
delta ðk ð2:108Þ
s
The second equation in Eq. (2.105) expressed in the complex frequency domain
is as follows:
s Þ ¼ F
DTðk s ÞDdðk
delta ðk s Þ ð2:109Þ
Also in the complex frequency domain, the first equation in Eq. (2.41) becomes
s Þ ¼ ns þ jxs s Þ ¼ ns Ddðk
s Þ þ j xs Ddðk
s Þ
Dxðk Ddðk ð2:110Þ
x0 x0 x0
Let the electric torque defined by Eq. (2.109) be decomposed as follows:
s Þ ¼ Ts1 Ddðk
DTðk s Þ þ Td1 Dxðk
s Þ ð2:111Þ
s ÞDdðk
delta ðk
F s Þ þ Td1 ns Ddðk
s Þ ¼ Ts1 Ddðk s Þ þ jT xs Ddðk
s Þ ð2:112Þ
x0 d1
x0
That is
F s Þ ¼ Ts1 þ Td1 ns þ jT xs
delta ðk ð2:113Þ
x0 d1
x0

(
delta ðk
Td1 ¼ xx0s Im½F s Þ
ð2:114Þ
Ts1 ¼ Re½F s Þ Td1 ns
delta ðk
x0
The above derivation indicates that in the complex frequency domain, the
electric torque can be decomposed into damping and synchronizing torque
according to Eq. (2.111). Substituting Eqs. (2.111) into (2.108), it can have
2 þ Dk
ðMk s þ x0 K1 ÞDdðk
s Þ ¼ x0 Ts1 Ddðk
s Þ x0 Td1 Dxðks Þ
s
ð2:115Þ
¼ x0 Ts1 Ddðk s Ddðk
s Þ Td1 k s Þ
Thus,
Mk s þ x0 K1 þ x0 Ts1 ¼ 0
2 þ ðD þ Td1 Þk ð2:116Þ
s
Solution of the above equation is as follows:
D þ Td1
ns ¼ ð2:117Þ
2M
Equation (2.117) indicates that the damping torque affects the real part of
electromechanical oscillation mode, i.e. the damping of power oscillation.
2.3.2.2 Graphical Explanation of the Damping Torque Analysis
Equations (2.16) and (2.39) include the following linearized rotor motion equation
of generator
:
D d ¼ x0 Dx
: 1 ð2:118Þ
D x ¼ ðDPt þ DDxÞ
M
Without referring to any linearized model of single-machine infinite-bus power

system, assume DPt being comprised of contributions from Dd and stabilizing
signal of the PSS, Dupss . That is to let
DPt ¼ DPt ðDdÞ þ DPt ðDupss Þ

ð2:119Þ
DPt ðDdÞ ¼ Fdelta ðsÞDd; DPt ðDupss Þ ¼ Fpss ðsÞDupss
At a known angular oscillation frequency xs , if DPt is decomposed in the

complex frequency domain, then
DPt ðDdÞ ¼ F delta ðjxs ÞDd ¼ Cdelta Dd þ Ddelta Dx;

ð2:120Þ
pss ðjxs ÞDupss ¼ Cpss Dd þ Dpss Dx
DPt ðDupss Þ ¼ F
Obviously, Ddelta Dx and Dpss Dx are the damping torque contributed to the
electromechanical oscillation loop of generator from Dd and Dupss , respectively.
Function of the damping torque component in suppressing the power oscillation can
be explained graphically by the use of the linearized equal-area criterion as follows.
Without affecting the conclusions of following discussion,
it is assumed
that in
Eq. (2.120), Ddelta ¼ 0. When there is no PSS installed DPt ðDupss Þ ¼ 0 ,
DPt ¼ DPt ðDdÞ ¼ Cdelta Dd ð2:121Þ
The linearized DPt Dd curve is line a–f as shown in Fig. 2.16. In Fig. 2.16, the
operating point of system at steady state is d (Pt0 ; d0 ) and it moves to point a
(Pt1 ; d1 ) after the system is subject to a small disturbance. Hence, when the oper-
ating point moves down from the initial point a (Pt1 ; d1 ) along line a–f, it will stop at
point f (Pt20 ; d20 ) with area a–d–c being equal to area d–g–f. Obviously, in this case,
jPt1 Pt0 j ¼ jPt20 Pt0 j; jd1 d0 j ¼ jd20 d0 j, power oscillation is of unchanged
magnitude and not damped at all.
When the PSS is installed to provide a pure positive damping torque,
DPt ðDupss Þ ¼ Dpss Dx ðassuming Dpss [ 0Þ ð2:122Þ
From Eqs. (2.120), (2.121), and (2.122), it should have
DPt ¼ DPt ðDdÞ þ DPt ðDupss Þ ¼ Cdelta Dd þ Dpss Dx; ð2:123Þ
When the operating point moves down from point a (Pt1 ; d1 ), power angle of
generator decreases and thus Dx\0. Dpss Dx\0 is added on Cdelta Dd as shown in
Eq. (2.123). Hence, the operating point should move below line a–f along curve
DPt ¼ Cdelta Dd þ Dpss Dx;. When the operating point stops moving, Dx ¼ 0. Thus,
it should stop on line a–f at point c (Pt2 ; d2 ) with area A1 being equal to area A2 .
Fig. 2.16 Graphical Pt ΔPt=CdeltaΔδ+Dpss Δω a

explanation of the DTA
ΔPt=CdeltaΔδ
A1
Pt0 g d e
A2
c
Direction of
f Δω < 0
δ
δ2' δ2 δ0 δ1
δ
Time response
with the PSS
time
Time response without the PSS
Obviously, jPt2 Pt0 j\jPt1 Pt0 j; jd2 d0 j\jd1 d0 j, indicating extra positive
damping is provided by the PSS to the power oscillation. A similar analysis can be
carried out to examine the case when the operating point moves up from point c
(Pt2 ; d2 ).
The above discussion explains the function of damping provided by the PSS in
suppressing the power oscillation. It is important to note that the explanation relies
only on the linearized rotor motion equation in Eqs. (2.118), (2.119), and (2.120)
without referring to any particular type of model of power system. This means that
for any type of linearized model of power system, including that of a multi-machine
power system, if Eqs. (2.119) and (2.120) can be established on the basis of the
model, the above procedure can be applied.
2.3.2.3 Design of PSS by the Phase Compensation Method [4]
If the rotor speed of generator is taken as the feedback signal of the PSS, transfer
function of the PSS is Tpss ðsÞ, that is
DuPSS ¼ Tpss ðsÞDx ð2:124Þ
From Eqs. (2.95) or (2.102) and above equation, it can have
DTpss ¼ Fpss ðsÞTpss ðsÞDx ð2:125Þ
At the angular oscillation frequency, xs , the decomposition of the electric torque

contributed by the PSS is as follows:
pss ðjxs ÞT
DTpss ¼ F pss ðjxs ÞDx

¼ Re F pss ðjxs ÞT pss ðjxs Þ Dx þ jIm½F

pss ðjxs ÞT
pss ðjxs ÞDx
¼ Re½Fpss ðjxs ÞT pss ðjxs ÞDx xs Im½Fpss ðjxs ÞT pss ðjxs ÞDd ð2:126Þ
x0
¼ Tpssd Dx þ Tpsss Dd
The damping and synchronizing torque provided by the PSS is Tpssd Dx and
Tpsss Dd, respectively. In order to achieve the most efficient design, ideally the PSS
should provide only the damping torque, that is,
DTpss ¼ Dpss Dx; Dpss [ 0 ð2:127Þ
where Dpss is the coefficient of the damping torque which needs to be provided by
the PSS. Hence, from Eqs. (2.126) and (2.127), it can be seen that design of the PSS
should satisfy that
pss ðjxs ÞT
Dpss ¼ F pss ðjxs Þ ð2:128Þ
According to Eq. (2.128), design of the PSS should set the phase of the PSS,
\T pss ðjxs Þ, to be equal to the minus phase of the forward path, \F
pss ðjxs Þ, that is to
design the PSS such that it can compensate the phase lag of the forward path and
ensure it to provide a pure positive damping torque. Hence, the method to design
the PSS based on Eq. (2.128) is called the phase compensation method.
If it is denoted that

pss ðjxs Þ ¼ Fpss \/; T
F pss ðjxs Þ ¼ Tpss \c ð2:129Þ
The phase compensation method requires

Tpssd ¼ Fpss Tpss cosð/ þ cÞ ¼ Dpss
ð2:130Þ
Tpsss ¼ Fpss Tpss sinð/ þ cÞ ¼ 0
This can be achieved by setting

Dpss
c ¼ /; Tpss ¼ ð2:131Þ
Fpss
Often the PSS is constructed as a lead–lag block with its main part of transfer
function to be
ð1 þ sT2 Þ ð1 þ sT4 Þ ð1 þ sT2 Þ ð1 þ sT4 Þ

Tpss ðsÞ ¼ Kpss ¼ Kpss1 Kpss2 ð2:132Þ
ð1 þ sT1 Þ ð1 þ sT3 Þ ð1 þ sT1 Þ ð1 þ sT3 Þ
where Kpss ¼ Kpss1 Kpss2 . Parameters of the PSS then can be set to satisfy
ð1 þ jxs T2 Þ Dpss /
Kpss1 ¼ \
ð1 þ jxs T1 Þ Fpss 2
ð2:133Þ
ð1 þ jxs T4 Þ /
Kpss2 ¼ 1:0\
ð1 þ jxs T3 Þ 2
for the PSS to provide a positive damping torque Dpss Dx.

With the PSS installed, from Fig. 2.10, it can have
ðMs2 þ Ds þ x0 K1 ÞDdðsÞ ¼ x0 DTdelta ðsÞ x0 DTpss ðsÞ

DTdelta ðsÞ ¼ Fdelta ðsÞDdðsÞ ð2:134Þ
DTpss ðsÞ ¼ Fpss ðsÞTpss ðs)DxðsÞ
c ¼ n þ jxc be the electromechanical oscillation mode of the closed-loop

Let k c
system with the PSS installed. According to the DTA discussed above in
Sect. 2.3.2.1, in the complex frequency domain, it can have
c Þ ¼ F
DTdelta ðk c ðsÞDdðk
delta k c Þ ¼ C1 Ddðkc Þ þ D1 Dxðkc Þ
ð2:135Þ
DTpss ðkc Þ ¼ F c ÞT
pss ðk c ÞDxðk
pss ðk c Þ ¼ Cpss Ddðk
c Þ þ Dpss Dxðk
c Þ
Taking the similar procedure of discussion in Sect. 2.3.2.1, it can easily prove
c Þ affects the damping of the electromechanical oscillation mode.
that Dpss Dxðk
Hence, the PSS can be designed to satisfy
c ÞTpss ðk
Fpss ðk c Þ ¼ Dpss ð2:136Þ
When the PSS is being designed, if k c ¼ n þ jxc is given, the PSS can be
c
designed by using the phase compensation method defined by Eq. (2.136) to move
c ¼ n þ jxc in
the electromechanical oscillation mode strictly to the position at k c
the complex plane. If only the amount of damping torque provision
Dpss Dx; Dpss [ 0 is given, the PSS can also be designed by the phase compensation
method introduced above from Eqs. (2.126) to (2.133).
2.4 Examples 53
2.4 Examples
2.4.1 Linearized Mathematical Models of an Example

Power System
2.4.1.1 Linearized Mathematical Model with Full Model

of Generator Used
Parameters (in p.u.) of an example single-machine infinite-bus power system are as

follows:
1. Generator:
xd ¼ 1:18; xq ¼ 0:78; xad ¼ 1:0; xaq ¼ 0:6; xD ¼ 1:11; xQ ¼ 0:73; xf ¼ 1:13;

ra ¼ 0:005; rf ¼ 0:00075; rD ¼ 0:002; rQ ¼ 0:04; M ¼ 7 s; D ¼ 0; T0d0 ¼ 5 s
2. AVR: KA ¼ 100; TA ¼ 0:01 s

3. Transmission line: xt ¼ 0:15
4. Steady-state operating point: Pt0 ¼ 0:5; Vt0 ¼ 1:05; Vb0 ¼ 1:0
At the steady-state operating point, the complex power received at the infinite
busbar is as follows:

t0 V
V b0

Vb0 ¼ Pt0 þ jQb0
jxt
where Qb0 is the reactive power received at the infinite busbar and * denotes the
conjugate of a complex phasor. By choosing V b0 as the reference phasor, that is,
¼ Vb0 , from the above equation, it can be obtained that
b0 ¼ V
V b0
2sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 3
2
Vb0 4 xt Pt0
Qb0 ¼ V2t0 Vb0 5 ¼ 0:3155 p:u:
xt Vb0
In the above, calculation is in p.u. Throughout the following calculation in p.u.,

p.u. is omitted.
The line current is as follows:
It0 ¼ Pt0 jQb0 ¼ 0:5 j0:3155

Vb0
Terminal voltage of generator is :
t0 ¼ jxtIt0 þ Vb0 ¼ 1:0473 þ j0:075

V
At steady state, iD0 ¼ iQ0 ¼ 0; x0 ¼ 1, Eqs. (2.1) and (2.2) become
0 ¼ vtd0 þ ra id0 þ wq0

0 ¼ vtq0 þ ra iq0 wd0
0 ¼ vf0 rf if0

wd0 xd xad id0
¼
wf0 xad xf if0
wq0 ¼ xq iq0
Thus, according to Eq. (2.28)
vtd0 ¼ ra id0 wq0 ¼ ra id0 þ xq iq0

vtq0 ¼ ra iq0 þ wd0 ¼ ra iq0 þ xad if0 xd id0 ¼ ra iq0 þ Eq0 xd id0
Hence, in d–q coordinate of generator,
t0 ¼ Vtd0 þ jVtq0 ¼ ra id0 þ xq iq0 þ jðra iq0 þ Eq0 xd id0 Þ

V
¼ ra ðid0 þ jiq0 Þ þ xq iq0 jxq id0 þ jðEq0 xd id0 þ xq id0 Þ
Let an imaginary electromotive force (EMF) be
EQ0 ¼ Eq0 xd id0 þ xq id0 ¼ Eq0 ðxd xq Þid0
Thus,
t0 ¼ ðra þ jxq Þðid0 þ jiq0 Þ þ E

V Q0
Q
Thus, the q-axis of generator can be found by calculating the imaginary EMF E
Q0 ¼ V
E t0 þ ðra þ jxq ÞIt0 ¼ 1:2959 þ j0:4634 ¼ 1:38\19:68
Hence, d0 ¼ 19:68 . From the above equation about Eq0 and EQ0 , it can have
Eq0 ¼ EQ0 ðxq xd Þid0 ¼ 1:1506

Eq0
if0 ¼ ¼ 1:5624
xad
2.4 Examples 55
Fig. 2.17 Phasor diagram for

computing initial values of q
state variables
EQ
[ra + j(x q − x d )]I t
Eq
jx d I t
v tq
jx t I t
Vt b
Vb
δ ϕ
It
vtd d
Above computation can be shown by the phasor diagram of Fig. 2.17 where
subscript 0 is omitted. From Fig. 2.17, it can be seen that
id0 ¼ It0 sinðd0 þ u0 Þ ¼ 0:4654; iq0 ¼ It0 cosðd0 þ u0 Þ ¼ 0:3646

vd0 ¼ Vb0 sin d0 ¼ 0:3368; vq0 ¼ Vb0 cos d0 ¼ 0:9416
Because
t0 ¼ vtd0 þ jvtq0 ¼ jxtIt0 þ Vb0 ¼ jxt ðid0 þ jiq0 Þ þ vd0 þ jvq0

V
thus
vtd0 ¼ vd0 xt iq0 ¼ 0:2820

vtq0 ¼ xt id0 þ vq0 ¼ 1:0114
Eq0
if0 ¼ ¼ 1:5624
xad
Because iD0 ¼ 0; iQ0 ¼ 0, thus according to Eq. (2.2), it can be obtained that
2 3 2 32 3 2 3
wd0 xd xad xad id0 1:0132
6 7 6 76 7 6 7
4 wf0 5 ¼ 4 xad xf xad 54 if0 5 ¼ 4 1:3001 5
wD0 xad xad xD iD0 1:0970
" #
wq0 xq xaq iq0 0:2844
¼ ¼
wQ0 xaq xQ iQ0 0:2187
Denote
2 31 2 3 2 3
xd xad xad a11 a12 a13 4:2331 1:8311 2:1640
6 7 6 7 6 7
4 xad xf xad 5 ¼ 4 a21 a22 a23 5 ¼ 4 1:8311 5:1570 2:9963 5
xad xad xD a31 a32 a33 2:1640 2:9963 5:5498
1
xq xaq b11 b12 3:6842 2:8653
¼ ¼
xaq xQ b13 b14 2:8653 3:7249

2 3
0 x0 0 0 0 0 0 0
6 b V b V 7
60 MD 0 a11M
Vtd0 11M tq0 a12M
Vtd0 a13M
Vtd0 12M tq0 7
6 7
60 T1 7
6 0 0 0 0 0 0 7
6 A 7
60 x w a11 x0 ra x0 x a12 x0 ra a13 x0 ra 7
Agdq ¼6
6
0 q0 0 0 7
7
6 0 x0 wd0 0 x0 x b11 x0 ra 0 0 b12 x0 ra 7
6 7
60 x0 a21 x0 rf a22 x0 rf a23 x0 rf 7
6 0 0 0 7
6 7
40 0 0 a31 x0 rD 0 a32 x0 rD a33 x0 rD 0 5
0 0 0 0 b21 x0 rQ 0 0 b22 x0 rQ
2 3
0 314:16 0 0 0 0 0 0
60 0:0738 0:0872 0:4140 7
6 0 0 0:1706 0:5037 7
6 7
60 0 100 0 0 0 0 0 7
6 7
6 0 89:3377 6:6494 7
6 0 314:16 2:8762 3:3992 0 7
¼6 7
6 0 318:3167 0 314:16 5:4760 0 0 4:5008 7
6 7
60 1:2151 7
6 0 314:16 0:4314 0 0:7060 0 7
6 7
40 0 0 1:35797 0 1:8826 3:4871 0 5
0 0 0 0 36:0068 0 0 46:8088
2.4 Examples 57
2 0 0 3 2 3
0 0
6 id0 iq0 7 6
6 M 7 0:0665 0:0521 7
6 M 7 6 7
6 KA Vtd0 KA Vtq0 7 6 2686:1 9632:5 7
6
7
6 T V T V 7 6 7
6 t0 7
6 7
6 x 7 6 314:1593
A t0 A
6 0 7
0 7
Bgdq ¼6 0 ¼
7 6 6 7
6 x 7 0 314:1593 7
6 0 0 7 6 6 7
6 7 6 7
6 0 0 7 6 0 0 7
6 7 4 7
4 5 0 0 5
0 0
0 0 0 0

0 0 0 a11 0 a12 a13 0
Cgdq ¼
0 0 0 0 b11 0 0 b12

0 0 0 4:2331 0 1:8331 2:1640 0
¼
0 0 0 0 3:4862 0 0 2:8653

0 xt 0 0:15
Fdq1 ¼ ¼
xt 0 0:15 0

Vb cos d0 0 0:9416 0
Fdq2 ¼ ¼
Vb sin d0 0 0:3367 0
Thus, state matrix and control vector of linearized state-space model are obtained
to be
Agcdq ¼ Agdq þ Bgdq Fdq1 Cgdq þ Bgdq Fdq2

2 3
0 314:16 0 0 0 0 0 0
6 0:045 0:09 0:104 0:38 7
6 0 0 0:20 0:47 7
6 7
6 714:3 0 100 6116:3 1404:6 2646 3127 1154:5 7
6 7
6 295:8 89:34 6:64 135:03 7
6 0 478:4 2:88 3:39 7
¼6 7
6 105:78 318:3 0 513:64 5:48 86:29 101:9 4:50 7
6 7
6 1:22 7
6 0 0 314:2 0:43 0 0:70 0 7
6 7
4 0 0 0 1:36 0 1:88 3:48 0 5
00 0 0 36:0 0 0 46:8
2 2 3 3
0 0
6 0 7 6 0 7
6 7 6 7
6 KA 7 6 7
6 T 7 6 10000 7
6 A7 6 7
6 0 7 6 0 7
6 7 6 7
bpss ¼6 7¼6 7
6 0 7 6 0 7
6 7 6 7
6 0 7 6 0 7
6 7 6 7
6 7 6 7
4 0 5 4 0 5
0 0
Eigenvalues of above state matrix are computed to be

1 ;k
k 2 ¼ 24:14 j971:56
3 ;k
k 4 ¼ 36:33 j359:63
k5 ¼ 35:57
k6 ; k
7 ¼ 0:71 j8:44
k8 ¼ 5:71
According to Eq. (2.63), the oscillation frequency f i (Hz) and damping fi for the
electromechanical oscillation mode, k 6 and k7 , are as follows:
xi
fi ¼ ¼ 1:34 Hz
2p
ni
fi ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:0838; i ¼ 6; 7
ni þ x2i
2
2.4.1.2 Heffron–Phillips Model of Example Power System
With D and Q damping winding of generator being ignored, from the given
parameters of above example power system and results of calculation, it can have
x2
x0d ¼ xd ad ¼ 0:2951
x
f
E0q0 0
¼ EQ0 xq xd itd0 ¼ 1:1506
K1 ¼ 1:5248; K2 ¼ 0:7602; K3 ¼ 2:9885;

K4 ¼ 0:6727; K5 ¼ 0:0027; K6 ¼ 0:3245
State equation of (2.41) can be obtained to be

2 : 3 2 32 3 2 3
Dd 0 314:16 0 0 Dd 0
6 : 7
6 D x0 7 6 0:2178 0 0:1086 0 7 6 7 6 7
76 Dx0 7 þ 6 0 7Dupss
6 _ 7¼6
4 DEq 5 4 0:1346 0 0:5978 0:2 54 DEq 5 4 0 5
0 0
DE_ 26:4729 0 3245 100 DEfd 10000
fd
Eigenvalues of state matrix are as follows:
2;3 ¼ 0:0114 8:2610; k4 ¼ 7:6008

k1 ¼ 92:9741; k
2.4 Examples 59
According to Eq. (2.63), oscillation frequency f i (Hz) and damping fi for the
electromechanical oscillation mode, k2;3 , are as follows:
xi
fi ¼ ¼ 1:315 Hz
2p
ni
fi ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:0014; i ¼ 2; 3
ni þ x2i
2
2.4.2 Modal Analysis of Example Power System
Heffron–Phillips model established above is used to demonstrate the modal analysis

of example power system in this section.
2.4.2.1 Modal Decomposition and Stability of Example Power System
Right eigenvectors corresponding to each of eigenvalues of state matrix of Heffron–

Phillips model are calculated to be
2 3 2 3
8:4781 106 0:9014
6 2:5091 106 7 6 7
6 7 6 j0:0237 7
v1 ¼ 6 7; v2 ¼ 6 7
4 0:0022 5 4 0:0050 þ j0:0050 5
1 0:3868 j0:1930
2 3 2 3
0:9014 0:0077
6 j0:0237 7 6 0:0002 7
6 7 6 7

v3 ¼ 6 7; v4 ¼ 6 7
4 0:0050 j0:0050 5 4 0:0284 5
0:3868 þ j0:1930 0:9996
Hence,
¼ ½ v1
V v2 v3 v4
2 3
8:4781 106 0:9014 0:9014 0:0077
6 2:5091 106 j0:0237 7
6 j0:0237 0:0002 7
¼6 7
4 0:0022 0:0050 þ j0:0050 0:0050 j0:0050 0:0284 5
1 0:3868 j0:1930 0:3868 þ j0:1930 0:9996
From Eq. (2.53),
T ¼ ½ w1 1
W 2
w 3
w w4 T ¼ V
2 3
0:2524 0:5554 j0:0001 0:5554 þ j0:0001 0:1773
6 0:8529 0:0312 j21:1226 0:0312 þ j21:1226 7:3269 7
6 7
¼6 7
4 38:0214 0:1626 þ j0:1475 0:1626 j0:1475 38:2206 5
1:0823 0:0003 þ j0:0003 0:0003 j0:0003 0:0827
Modal decomposition thus is obtained to be
szi ¼ ki zi ; þ wTi bo u, i ¼ 1; 2; 3; 4

Without considering the PSS Dupss ¼ 0 , solution of modal decomposition is
obtained to be
zi ðt) ¼ zi ð0Þeki t ; i ¼ 1; 2; 3; 4
Zð0Þ ¼ V1 Xð0Þ ¼ WT Xð0Þ
Time response of state variables can be written as follows:
X
4 X
4
xk ðt) ¼ vki zi ð0Þeki t ¼ vki zi ð0Þeni t ½cos xi t þ j sin xi t
i¼1 i¼1
It can be seen that when time approaches infinity ðt ! 1Þ,
X
4 X
4
lim xk ðt) ¼ lim vki zi ð0Þeki t ¼ lim vki zi ð0Þeni t ½cos xi t þ j sin xi t
t!1 t!1 t!1
i¼1 i¼1
92:9741t
¼ lim fe vk1 z1 ð0Þ þ e0:0114t vk2 z2 ð0Þ½cosð8:2610tÞ þ j sinð8:2610tÞ
t!1
þ e0:0114t vk3 z3 ð0Þ½cosð8:2610tÞ j sinð8:2610tÞ þ e7:6008t vk4 z4 ð0Þg ¼ 0

T
Since XðtÞ ¼ DdðtÞ DxðtÞ DE0q ðtÞ DE0fd ðtÞ and
DdðtÞ ¼ dðtÞ d0 ; DxðtÞ ¼ xðtÞ x0 ;

DE0q ðtÞ ¼ E0q ðtÞ E0q0 ðtÞ; DE0fd ðtÞ ¼ E0fd ðtÞ E0fd0 ðtÞ
2.4 Examples 61
lim xk ðt) ¼ 0; k ¼ 1; 2; 3; 4 should give

t!1
lim dðtÞ ¼ d0 ; lim xðtÞ ¼ x0 ; lim E0q ðtÞ ¼ E0q0 ðtÞ; lim E0fd ðtÞ ¼ E0fd0 ðtÞ
t!1 t!1 t!1 t!1
All the state variables return to their initial points Xð0Þ, the equilibrium point of
the system. Hence, the system is stable in terms of small-signal stability.
2.4.2.2 Modal Analysis of the AVR
State matrix, control vector, and output vector of Eq. (2.71) are as follows:
" #
0 w0 0 314:16
A0 ¼ D ¼ 0:2178
KM M
1
0
" #
0 0
b0 ¼ K ¼ cT0 ¼ ½ 1 0
M 2
0:1086
1;2 ¼ j8:2725 with the corresponding

Its eigenvalues are calculated to be k
eigenvectors to be

0:9997 0:9997
v1 ¼ ; v2 ¼
j0:0263 j0:0263
Thus
T ¼ ½w
W 1 w 1 ¼ ½ v1 v2 1
2 T ¼ V
T
0:9997 0:9997 1 0:5002 0:5002
¼¼ ¼
j0:0263 j0:0263 j18:9947 j18:9947
Hence, the controllability and observability index, respectively, can be obtained

to be

0
T1 b0
w ¼ ½ 0:5002 j18:9914 ¼ j2:0628
0:1086

0
T2 b0
w ¼ ½ 0:5002 j18:9914 ¼ j2:0628
0:1086

0:9997
cT0
v1 ¼ ½1 0 ¼ 0:9997
j0:0263

0:9997
cT0
v2 ¼ ½1 0 ¼ 0:9997
j0:0263
The residue is as follows:
1 ¼ w
R T1 b0 cT0 v1 ¼ j2:0622
2 ¼ w
R T2 b0 cT0 v2 ¼ j2:0622
From Eq. (2.77), when a ¼ KA , it can have
ðTA ki þ 1ÞK4 þ KA K5

i
@k @ 0 i þ 1Þ þ KA K6
ðki Tdo þ K3 ÞðTA k
¼ Ri
@KA @KA
i T0 þ K3 ÞðTA k
K5 ½ðk i þ 1Þ þ KA K6 ½ðTA k
i þ 1ÞK4 þ KA K5 K6
¼ Ri do
½ðki T0 þ K3 ÞðTA ki þ 1Þ þ KA K6 2
do
¼ 0:00018158 j0:00005839 i ¼ 1; 2
When a ¼ TA it can have

i
@k @ 0ðTA ki þ 1ÞK4 þ KA K5
ðki Tdo þ K3 ÞðTA ki þ 1Þ þ KA K6
¼ Ri
@TA @TA
i K4 ½ðk
k i T0 þ K3 ÞðTA k i þ 1Þ þ KA K6 ½ðTA k i þ 1ÞK4 þ KA K5 ðk
i T0 þ K3 Þk
i
¼ Ri do do
½ðki T þ K3 ÞðTA k
0 i þ 1Þ þ KA K6 2
do
¼ 0:0357 j0:1533 i ¼ 1; 2
Above results indicate that increase of the gain value and time constant will
move the electromechanical oscillation mode towards the right on the complex
plane and hence is detrimental to the small-signal angular stability of the power
system. It has been well known that the fast-acting high-gain AVR may be detri-
mental to the damping of power system electromechanical oscillation modes. This
means that increase of gain value of the AVR could move the oscillation mode to
the right. However, increase of the time constant (slower action of the AVR) should
not.
In order to further clarify the results of derivative of the oscillation mode in
respect to the parameters of the AVR obtained above, Fig. 2.18 presents the loci of
movement of the electromechanical oscillation mode on the complex plane with the
change of gain value and time constant of the AVR. In Fig. 2.18, KA increases from
KA ¼ 50 to KA ¼ 100 with TA ¼ 0:01 and TA increases from TA ¼ 0:01 to TA ¼
0:1 with KA ¼ 100. From Fig. 2.18, it can be seen that although at the point
KA ¼ 100; TA ¼ 0:01 where the derivatives are calculated, the trend of loci
movement is towards the right with the increase of the gain value and time constant,
and the oscillation mode in fact moves towards left when the time constant of the
AVR increases. The trend of the loci with the increase of the time constant of the
AVR actually changes the direction at the point KA ¼ 100; TA ¼ 0:01.
2.4 Examples 63
Fig. 2.18 Loci of the movement of the oscillation mode on the complex plane with the changes of
parameters of the AVR
2.4.2.3 Design of the PSS by Pole Assignment for the Example Power
System
A PSS can be installed to increase the damping of example power system by

assigning the electromechanical oscillation mode, k2;3 ¼ 0:0114 j8:2611, to a
target position on the complex plane, kc ¼ 0:9 j8:2611, which is damping over
10 %.
Let the transfer function of the PSS be (see Eq. (2.88)) with T ¼ 0:1 s
ð1 þ saTÞ2
Tpss ðsÞ ¼ Kpss
ð1 þ sTÞ2
State-space model of example power system with the PSS to be installed is

2 3 2 32 3 2 3
Dd_ 0 314:16 0 0 Dd 0
6 7 6 76 Dx 7 6 7
6 Dx_ 7 6 0:2178 0 0:1086 76 0 7 6 7 0
6 7¼6 76 7 þ 6 7Dupss
6 DE_ 0 7 4 0:1346 0:5978 0:2 54 DE0q 5 4 0 5
4 q 5 0
DE_
0
26:4729 0 3245 100 DE0fd 10000
fd
2 3
Dd
6 Dx 7
6 7
Dx ¼ ½ 0 1 0 0 6 7
4 DE0q 5
DE0fd
Dupss ¼ Tpss ðsÞDx
Thus, the transfer function of open-loop system can be obtained to be
238:8s
GðsÞ ¼ cT ðsI AÞ1 b ¼
1:1s4 þ 110:6s3 þ 854:5s2 þ 7563s þ 53020
From the characteristic equation of closed-loop control system of Eq. (2.85), for
the electromechanical oscillation mode k c , it should have
c Þ ¼ 1 1
Tpss ðk c Þ ¼ 0:0753 j0:0996
Gðk
That is
c aTÞ2
ð1 þ k 1
Kpss ¼

ð1 þ kc TÞ 2 0:0753 j0:0996
By solving the above equations, parameters of the PSS can be obtained to be

Kpss ¼ 2:6151
a ¼ 2:4256
In order to establish the state-space model of closed-loop system with the PSS
installed, let
ð1 þ saTÞ
Dx1 ¼ Dx
ð1 þ sTÞ
ð1 þ saTÞ
Dupss ¼ Kpss Dx1
ð1 þ sTÞ
2.4 Examples 65
Thus, state equation of the PSS is as follows:
1 1
sDx1 ¼ Dx þ a sDx Dx1
T T
K1 1 D K2 1
¼ aDd þ a Dx aDE0q Dx1
M T M M T
1 1
sDupss ¼ Kpss Dx1 Dupss þ Kpss a sDx1
T T
K1 2 1 D K2 0
¼ Kpss a Dd þ Kpss að aÞDx Kpss a2 DEq
M T M M
1 1
þ ð1 aÞ Kpss Dx1 Dupss
T T
By writing the state equation of open-loop system and the PSS together, the
state-space model of closed-loop system is obtained to be
2 : 32 32 3
Dd 0 314:159 0 0 0 0 Dd
6 : 7 6
6 Dx 7 6 0:218 0 0:109 0 0 0 7 6
76 Dx0 7
7
6 0 7 6
6 DE_ q 7 6 0:134 0 0:597 0:200 0 0 7 76
6 DEq 7
7
6 0 7 ¼ 6 26:582 10000 7 6 0 7
6 DE_ 7 6
6 fd 7
0 3245:044 100 0 76 DEfd 7
4 Dx_ 1 5 4 0:528 10 0:263 0 10 0 54 Dx1 5
Du_ pss 3:352 63:432 1:671 0 37:281 10 Dupss
Eigenvalues of state matrix are calculated to be
1 ¼ 93:4535
k
2;3 ¼ 0:8995 j5:2621
k
4;5 ¼ 9:4136 j5:0918
k
k6 ¼ 6:5176
Hence, the electromechanical oscillation mode is successfully assigned to the

target position.
Figure 2.19 shows the simulation result of example power system without and
with the PSS installed. At 1.0 s of simulation, a three-phase to-earth short circuit
occurred on the transmission line which was cleared in 100 ms. From Fig. 2.19, it
can be seen that the low-frequency oscillation is damped effectively by the PSS
designed by use of the method of pole assignment.
rotor angle δ (degree)

50
without PSS with PSS installed
40
30
20
10
-10
0 1 2 3 4 5 6 7 8 9 10
time (second)
Fig. 2.19 Simulation result of example power system without and with PSS installed
2.4.3 Damping Torque Analysis of Example Power

System
Heffron–Phillips model established is again used in this section to demonstrate the

DTA of example power system.
2.4.3.1 Damping Torque Provided by the AVR in the Example Power

System
From Fig. 2.10, it can be seen that the electric torque provided by the AVR to the
electromechanical oscillation loop of generator is as follows:
K5 1 KA
sT0do þ K3 sTA þ 1
DTavr ¼ K2 Dd
1 þ 0 K6 sTKAþ 1
sTdo þ K3 A
K2 K5 KA
¼ Dd ¼ Favr ðsÞDd
K6 KA þ ðsT0do þ K3 ÞðsTA þ 1Þ
At the complex frequency of the electromechanical oscillation with

s ¼ n þ jxs ¼ k1 ¼ 0:0114 þ j8:2610, let the decomposition of the electric
k s
torque provided by the AVR be
2.4 Examples 67

s ¼ Tsavr Dd k
DTavr k s þ Tdavr Dx k
s
Because
s Þ ¼ ns þ jxs s Þ ¼ ns Ddðk
s Þ þ j xs Ddðk
s Þ
Dxðk Ddðk
x0 x0 x0
thus
DTavr ðk s Þ þ Tdavr ns Ddðk

s Þ ¼ Tsavr Ddðk s Þ þ jT xs s Þ
Ddðk
x0 davr
x0
That is
xs s Þ
Tdavr ¼ ImFavr ðk
x0
Since
s Þ ¼ K2 K5 KA
Favr ðk s TA þ 1Þ
K6 KA þ ðks T0do þ K3 Þðk
K2 K5 KA
¼
K6 KA þ K3 þ ðK3 TA þ T0d0 Þk 2
s þ T0 TA k
d0 s
K2 K5 KA
¼
K6 KA þ K3 þ ðK3 TA þ T0d0 Þðns þ jxs Þ þ T0d0 TA ðns þ jxs Þ2
K2 K5 KA K2 K5 KA
¼ ¼ 2 ða jbÞ ¼ 0:0024 þ j0:0031
a þ jb a þ b2
where

a ¼ K6 KA þ K3 þ K3 TA þ T0d0 ns þ T0d0 TA n2s x2s ¼ 31:9695

b ¼ K3 TA þ T0d0 xs þ 2T0d0 TA ns xs ¼ 41:5426
thus it can be obtained
x0 s Þ ¼ x0 K2 K5 KA b
Tdavr ¼ ImFavr ðk
xs xs a2 þ b2
¼ 0:0988
Since
@a @a

@ KA
¼ K6 ¼ 0:3245; @ TA
¼ K3 ns þ T0d0 n2s x2s ¼ 341:2561
@b
@ KA
¼ 0; @b
@ TA
¼ K3 xs þ 2T0d0 ns xs ¼ 23:7488
sensitivity of the damping torque provided by the AVR to its parameters can be
obtained to be
h i
@Tdavr x0 K2 K5 b þ KA @@Kb a2 þ b2 KA b 2a @@Ka þ 2b @@Kb
¼ A A A
¼ 0:0003
@KA xs ða2 þ b2 Þ2
h i
@b @a @b
x0 K2 K5 KA @ TA ða þ b Þ b 2a @ TA þ 2b @ TA
2 2
@Tdavr
¼ ¼ 0:9055
@TA xs ða2 þ b2 Þ2
The above results indicate that (1) with the increase of the AVR gain, less
damping torque will be provided by the AVR, detrimental to the system
small-signal angular stability and (2) with the increase of the AVR time constant,
less damping torque will be provided by the AVR, also detrimental to the damping
of low-frequency power oscillations.
2.4.3.2 Design of PSS Installed in the Example Power System

by the Phase Compensation Method
The PSS to be designed is to provide a damping torque DTpss ¼ Dpss Dx; Dpss ¼ 15.
From Eq. (2.96), the forward path of stabilizing signal of the PSS can be obtained to
be ðjxs ¼ j8:44Þ
KA
Fpss ðjxs Þ ¼ K2 0
ðK3 þ jxs Td0 Þð1 þ jxs TA Þ þ K6 KA
¼ 0:8598 j1:1451 ¼ 1:4320\53:0989
The PSS adopts the deviation of rotor speed of generator as the feedback signal,
and its transfer function is as follows:
ð1 þ sT2 Þ ð1 þ sT4 Þ
Tpss ðsÞ ¼ K1 K2 with T1 ¼ 0:09 s; T3 ¼ 0:09 s
ð1 þ sT1 Þ ð1 þ sT3 Þ
2.4 Examples 69
According to Eq. (2.133), parameters of the PSS are set to compensate the phase
of the forward path and thus obtained to be
Kpss ¼ Kpss1 Kpss2 ¼ 3:2271; T2 ¼ 0:2405 s; T4 ¼ 0:2405 s
Let
ð1 þ sT4 Þ
Dx1 ¼ Kpss2 Dx ¼ ð9:3706 þ j4:6819ÞDx
ð1 þ sT3 Þ
ð1 þ sT2 Þ
Dupss ¼ Kpss1 Dx1 ¼ ð0:8946 þ j0:4470ÞDx1
ð1 þ sT1 Þ
Thus, state-space realization of the PSS is as follows:
1 Kpss2
sDx1 ¼ Dx1 þ ðT4 sDx þ DxÞ
T3 T3

1 Kpss2 T4
¼ Dx1 þ ðK1 Dd K2 DE0q DDxÞ þ DxÞ
T3 T3 M

Kpss2 T4 K1 Kpss2 T4 Kpss2 T4 K2 0 1
¼ Dd þ 1 D Dx DEq Dx1
T3 M T3 M T3 M T3
¼ 0:3230Dd þ 6:1672Dx 0:1224DE0q 11:1111Dx1

1 Kpss1 T2 Kpss1 Kpss1 T2 Kpss2 T4 K1
sDupss ¼ Dupss þ Dx1 þ Dd
T1 T1 T3 T1 T1 T3 M

Kpss2 T4 K2 0 Kpss2 T4
DEq þ 1 D Dx
T3 M T3 M

Kpss T2 T4 K1 Kpss T2 T4 Kpss T2 T4 K2 0
¼ Dd þ 1 D Dx DEq
T1 T3 M T1 T3 M T1 T3 M

Kpss1 T2 Kpss1 1
+ Dx1 Dupss
T1 T3 T1 T1
¼ 5:0182Dd þ 95:8026Dx 2:5018DE0q 108:0008Dx1 11:1111Dupss
By writing the state equation of open-loop system and the PSS together, state
matrix of closed-loop system is obtained to be
2 0 x0 0 0 0 0 3
6 K1 D K 0 7
6 0 0 7
2
6 M M M 7
6 K 7
6 K0 4 0 03 1
0 0 7
6 Td0 Td0 T0d0 7
6 7
A¼6 KA 7
6 KTA K5 0 KTA K6 K A
0 TA 7
6 A
A TA 7
6 7
6 Kpss2 T4 K1 Kpss2
1 TTM KT 20 T4 K2
K2 0 T1 0 7
6 7
4
4 T 3M T 3M
5

3 3 3
Kpss1
T 2 T 4 K1
T1 T3 M K PSS
T2 1 T4 D K
T1 T3 M PSS T 2 T 4 K2
T1 T3 M KPSS 0 T1 1 T2
T3 T1
1
2 3
0 314:16 0 0 0 0
6 0:2178 0:1086 7
6 0 0 0 0 7
6 7
6 0:1345 0 0:5977 0:2 0 0 7
¼6
6 26:5821
7
6 0 3245 100 0 10000 7
7
6 7
4 0:3230 6:1672 0:1224 0 11:1111 0 5
5:0182 95:8026 2:5018 0 108:0008 11:1111
Eigenvalues of state matrix are calculated to be
k1 ¼ 93:6838
2;3 ¼ 1:2125 j8:0051
k
k4;5 ¼ 8:7171 j6:2080
k6 ¼ 9:2771
Hence, damping of the electromechanical oscillation mode is successfully

increased by the PSS designed above via the phase compensation method.
Figure 2.20 shows the simulation result of the example power system without
and with the PSS installed. At 1.0 s of the simulation, a three-phase to-earth short
Fig. 2.20 Simulation result of power system without and with the PSS designed by use of the
phase compensation method
2.4 Examples 71
circuit occurred on the transmission line which was cleared in 100 ms. From
Fig. 2.20, it can be seen that the low-frequency oscillation is damped effectively by
the PSS designed by use of the phase compensation method.
2.4.3.3 Theoretical Basis and Graphical Explanation of the Damping

Torque Analysis
From Fig. 2.10 and Eq. (2.105), it can be obtained that

0 1
K4 1
K5 KA
1
0 0
B sT þ K3 sT sTA þ 1C
þ K3
Fdelta ðsÞ ¼ K2 @ þ
KA A
do do
K
1 þ 0 6 sTKAþ 1 1þ 0 K 6
sTdo þ K3 A sTdo þ K3 sTA þ 1
K2 ½K4 ðsTA þ 1Þ þ K5 KA
¼
K6 KA þ sT0do þ K3 ðsTA þ 1Þ
At the complex frequency of electromechanical oscillation with

s ¼ n þ jxs ¼ k1 ¼ 0:0114 þ j8:2610,
k s

delta ks ¼ 0:0042 þ j0:0042
F
According to Eq. (2.114)

(
Td1 ¼ xx0s Im F s Þ ¼ 0:1591

delta ðk

Ts1 ¼ Re F s Þ Td1 ns ¼ 0:0042

delta ðk
x0
Substituting the above result into Eq. (2.116), it can have
2 þ 0:1553k
7k s þ 477:7128 ¼ 0
s
Solution of above equation is ks ¼ 0:0114 þ j8:2610. It is the electrome-

chanical oscillation mode of example power system without the PSS being
installed.
s ÞDd ¼ K1 Dd þ Ts1 Dd þ Td1 Dx ¼ 1:5207Dd þ 0:1553Dx

DPt ¼ K1 Dd þ Fdelta ðk
This is the first equation in Eq. (2.120). Figure 2.21 presents the Pt d curve
from simulation. At 1 s of simulation, the mechanical power input to the generator
without PSS
with PSS installed
Fig. 2.21 Pt d curve, corresponding variation of rotor angle and power output of the generator
of example power system
increased by 1 % and then returned to its original value in 10 ms. Pt d curve

depicted in Fig. 2.20 is the first cycle of power oscillation starting from its first
peak. Dashed curve is the case without the PSS installed.
With the PSS designed in Sect. 2.4.3.2 being installed,
c ÞDd þ F
DPt ¼ K1 Dd þ Fdelta ðk c ÞT
pss ðk c ÞDxðk
pss ðk c Þ
¼ K1 Dd þ Ts1 Dd þ Td1 Dx þ Dpss Dxðk c Þ
¼ 1:5207Dd þ 15:1553Dx
where kc ¼ 1:2279 þ j8:0264 is the electromechanical oscillation mode of

example power system with the PSS being installed. Solid Pt d curve in Fig. 2.21
is the case that the PSS is installed in the example power system. Figure 2.21
confirms the graphical explanation of the DTA illustrated in Fig. 2.16.
2.4 Examples 73
2.4.4 Equivalence Between the Damping Torque and Modal

Analysis
i , the residue is equal to

It is concluded in Sect. 2.3.1.3 that at a complex frequency k
the forward path of the PSS multiplied by the sensitivity of the mode to the
damping torque contribution. In this section, this conclusion is to be demonstrated
by example power system.
2.4.4.1 Demonstration by Use of Heffron–Phillips Model of Example

Power System
In Sect. 2.4.2.3, state-space model of the example power system is obtained to be

2 : 2 3 32 3 2 3
Dd 0 314:16 0 0 Dd 0
6 : 7 6
6 D x0 7 6 0:2178 0 0:1086 0 7 6 7 6 7
76 Dx0 7 þ 6 0 7Dupss
6 _ 7¼4 5 4 5 4
4 DEq 5 0:1345 0 0:5977 0:2 DEq 0 5
0 0
DE_ 26:5821 0 3245 100 DEfd 10000
fd 2 3
Dd
6 Dx 7
Dx ¼ ½ 0 1 0 0 6 4 DE0 5
7
q
DE0fd
The electromechanical oscillation modes are as follows:
k2;3 ¼ 0:0114 j8:2610
In Sect. 2.4.2.1, matrix formed by right eigenvectors is as follows:
¼ ½ v1
V v2 v3 v4
Thus, the right eigenvectors related to the electromechanical oscillation modes

are as follows:
2 3 2 3
0:9014 0:9014
6 j0:0237 7 6 j0:0237 7
2 ¼ 6
v 7 6 7
4 0:0050 + j0:0050 5 v3 ¼ 4 0:0050 j0:0050 5
0:3868 j0:1930 0:3868 + j0:1930
Matrix formed by left eigenvectors is as follows:
T ¼ ½ w1 1
W 2
w 3
w w 4 T ¼ V
The left eigenvectors related to the electromechanical oscillation modes are as

follows:
T2 ¼ ½ 0:5554 j0:0001 0:0312 j21:1226 0:1626 þ j0:1475 0:0003 þ j0:0003

w
T3 ¼ ½ 0:5554 þ j0:0001 0:0312 þ j21:1226 0:1626 j0:1475 0:0003 j0:0003
w
Thus, according to Eq. (2.64), for the electromechanical oscillation modes, the
residue is calculated to be
2 ¼ w
R T2 bo cTo
v2
¼ ½ 0:5554 j0:0001 0:0312 j21:1226 0:1626 þ j0:1476 0:0003 þ j0:0003
2 3 2 3
0 0:9014
6 0 7 6 7
6 7 6 j0:0237 7
6 7½ 0 1 0 0 6 7
4 0 5 4 0:0050 + j0:0050 5
10000 0:3868 j0:1930
¼ 0:0633 þ j0:0822
3 ¼ w
R T3 bo cTo
v3
¼ ½ 0:5554 þ j0:0001 0:0312 þ j21:1226 0:1626 j0:1475 0:0003 j0:0003
2 3 2 3
0 0:9014
6 0 7 6 j0:0237 7
6 7 6 7
6 7½ 0 1 0 0 6 7
4 0 5 4 0:0050 j0:0050 5
10000 0:3868 + j0:1930
¼ 0:0633 j0:0822
From Eq. (2.104), sensitivity of the electromechanical oscillation modes to the

coefficient of damping torque provided by the PSS can be computed to be
2
@k 22 v22
w ð0:0312 j21:1226Þðj0:0237Þ
¼ ¼ ¼ 0:0715
@Dpss M 7
3
@k 32 v32
w ð0:0312 þ j21:1226Þðj0:0237Þ
¼ ¼ ¼ 0:0715
@Dpss M 7
2.4 Examples 75
At the complex frequency k 2;3 ¼ 0:0114 j8:2610, the forward path can be
calculated from Eq. (2.96) as
2;3 Þ ¼ K2 KA
Fpss ðk
2;3 T0 Þð1 þ k2;3 TA Þ þ K6 KA
ðK3 þ k d0
0:7602 100
¼
½2:9885 þ ð0:0114 j8:2611Þ 5½1 þ ð0:0114 j8:2611Þ 0:01 þ 0:3245 100
¼ 0:8845 j1:1493
Hence,
@k2;3
R2;3 ¼ Fpss ðk2;3 Þ ¼ ð0:0715Þ ð0:8845 j1:1493Þ
@Dpss
¼ 0:0633 j0:0822
Thus, it is conformed that the residue is equal to the forward path of the PSS
multiplied by the sensitivity of electromechanical oscillation modes to the coeffi-
cient of damping torque contribution from the PSS.
2.4.4.2 Demonstration by Use of General Linearized Model

of Example Power System
In Sect. 2.4.1.1, state matrix and control vector of state-space model of example
power system are obtained to be
2 3
0 314:16 0 0 0 0 0 0
6 0:045 0:09 0:104 0:39 7
6 0 0 0:20 0:47 7
6 7
6 714:29 0 100 6116:3 1404:6 2646 3127 1154:5 7
6 7
6 295:81 89:35 6:65 135:03 7
6 0 478:4 2:88 3:40 7
Agcdq ¼6 7
6 105:79 318:31 0 513:64 5:48 86:29 101:98 4:50 7
6 7
6 1:22 7
6 0 0 314:16 0:43 0 0:71 0 7
6 7
4 0 0 0 1:36 0 1:88 3:49 0 5
0 0 0 0 36:0 0 0 46:8
2 3
0
6 0 7
6 7
6 7
6 10000 7
6 7
6 0 7
6 7
bpss ¼6 7
6 0 7
6 7
6 0 7
6 7
6 7
4 0 5
0
Eigenvalues of state matrix are as follows:
1;2 ¼ 24:12 j971:56

k
3;4 ¼ 36:32 j359:63
k
k5 ¼ 35:57
6;7 ¼ 0:71 j8:44
k
k8 ¼ 5:71
The pair of electromechanical oscillation modes are as follows:
k6;7 ¼ 0:71 j8:44
For each of eigenvalues, right eigenvector is calculated to be

2 3 2 3
0:000005 j0:000011 0:000005 þ j0:000011
6 0:000032 þ j0:000016 7 6 0:000032 j0:000016 7
6 7 6 7
6 7 6 7
6 0:9507 7 6 0:9507 7
6 7 6 7
6 0:0001 þ j0:0182 7 6 0:0001 j0:0182 7
6 7 6 7
v1 ¼ 6 7v2 ¼ 6 7
6 0:0369 þ j0:0014 7 6 0:0369 j0:0014 7
6 7 6 7
6 0:0072 j0:3072 7 6 0:0072 þ j0:3072 7
6 7 6 7
6 7 6 7
4 0:000569 þ j0:000026 5 4 0:000569 j0:000026 5
j0:0014 j0:0014
2 3 2 3 2 3
0:000215 j0:000026 0:000215 þ j0:000026 0:0527
6 0:000055 j0:000244 7 6 0:000055 þ j0:000244 7 6 0:006 7
6 7 6 7 6 7
6 7 6 7 6 7
6 0:7251 7 6 0:7251 7 6 0:0539 7
6 7 6 7 6 7
6 0:0163 j0:2196 7 6 0:0163 þ j0:2196 7 6 0:1055 7
6 7 6 7 6 7
v3 ¼ 6 7v4 ¼ 6 7 v5 ¼ 6 7
6 0:1654 þ j0:025 7 6 0:1654 j0:025 7 6 0:2563 7
6 7 6 7 6 7
6 0:0615 j0:6274 7 6 0:0615 þ j0:6274 7 6 0:4925 7
6 7 6 7 6 7
6 7 6 7 6 7
4 0:0041 þ j0:0006 5 4 0:0041 j0:0006 5 4 0:0334 5
0:003 j0:0165 0:003 þ j0:0165 0:8209
2 3 2 3 2 3
0:7004 0:7004 0:000104
6 0:0016 þ j0:0188 7 6 0:0016 j0:0188 7 6 0:000002 7
6 7 6 7 6 7
6 7 6 7 6 7
6 0:000061 j0:000914 7 6 0:000061 þ j0:000914 7 6 0:0095 7
6 7 6 7 6 7
6 0:1482 þ j0:0043 7 6 0:1482 j0:0043 7 6 0:000116 7
6 7 6 7 6 7
v6 ¼ 6 7v7 ¼ 6 7 v8 ¼ 6 7
6 0:5516 þ j0:0283 7 6 0:5516 j0:0283 7 6 0:000082 7
6 7 6 7 6 7
6 0:0317 þ j0:0041 7 6 0:0317 j0:0041 7 6 0:7636 7
6 7 6 7 6 7
6 7 6 7 6 7
4 0:0077 þ j0:0284 5 4 0:0077 j0:0284 5 4 0:6456 5
0:413 þ j0:0977 0:413 j0:0977 0:000072
2.4 Examples 77
They form the following matrix
¼ ½ v1
V v2 ... v8
¼ ½w
T 1
; left eigenvectors corresponding to
As W 1 w 2 . . . w8 T ¼ V
eigenvalues are calculated to be
2 3T 2 3T
1:2419 j0:4483 1:2419 þ j0:4483
6 0:0486 j0:262 7 6 0:0486 þ j0:262 7
6 7 6 7
6 0:4494 þ j0:0412 7 6 0:4494 j0:0412 7
6 7 6 7
6 0:2153 j3:8659 7 T 6 0:2153 þ j3:8659 7
T1
w ¼6 7 w ¼ 6
6 1:9686 þ j0:7712 7 2 6 1:9686 j0:7712 7
7
6 7 6 7
6 0:0188 þ j1:3997 7 6 0:0188 j1:3997 7
6 7 6 7
4 0:0288 þ j1:6542 5 4 0:0288 j1:6542 5
0:6025 j0:5408 0:6025 þ j0:5408
2 3T 2 3T 2 3T
1:6552 þ j0:1633 1:6552 j0:1633 0:7411
6 0:1388 þ j0:9214 7 6 0:1388 j0:9214 7 6 6:8756 7
6 7 6 7 6 7
6 0:1003 j0:0038 7 6 0:1003 þ j0:0038 7 6 0:000219 7
6 7 6 7 6 7
6 0:2788 þ j1:9563 7 T 6 0:2788 j1:9563 7 T 6 0:0934 7
T3 6
¼6 7 6
4 ¼ 6 7 6 7
w 7 w 7 w5 ¼ 6 0:0106 7
6 2:5805 j0:126 7 6 2:5805 þ j0:126 7 6 7
6 0:0246 þ j0:1141 7 6 0:0246 j0:1141 7 6 0:000045 7
6 7 6 7 6 7
4 0:0277 þ j0:1353 5 4 0:0277 j0:1353 5 4 0:00006 5
0:7412 j0:2714 0:7412 þ j0:2714 1:3311
2 3T 2 3T
0:7422 þ j0:06 0:7422 j0:06
6 0:2586 j27:632 7 6 0:2586 þ j27:632 7
6 7 6 7
6 0:000019 þ j0:00094 7 6 0:000019 j0:00094 7
6 7 6 7
6 0:0034 þ j0:0146 7 T 6 0:0034 j0:0146 7
T6
w ¼6 7 w ¼ 6
6 0:0004 þ j0:000131 7 7 6 0:0004 j0:000131 7
7
6 7 6 7
6 0:000019 þ j0:000298 7 6 0:000019 j0:000298 7
6 7 6 7
4 0:000139 þ j0:0002727 5 4 0:000139 j0:0002727 5
0:0499 þ j0:2027 0:0499 j0:2027
2 3T
0:0803
6 2:7921 7
6 7
6 0:0021 7
6 7
6 0:003 7
wT8 ¼6
6 0:0283 7
7
6 7
6 0:000637 7
6 7
4 1:5496 5
0:079
According to Eq. (2.64), for the pair of electromechanical oscillation modes, the
residue is calculated to be
6;7 ¼ w
R T6;7 bo cTo v6;7
2 3T 2 3
0:7422 j0:06 0
6 0:2586
j27:632 7 6 0 7
6 7 6 7
6 7 6 7
6 0:000019 j0:00094 7 6 10000 7
6 7 6 7
6 0:0034 j0:0146 7 6 0 7
6 7 6 7
¼6 7 6 7
6 0:0004 j0:000131 7 6 0 7
6 7 6 7
6 0:000019 j0:000298 7 6 0 7
6 7 6 7
6 7 6 7
4 0:000139 j0:0002727 5 4 0 5
0:0499 þ j0:2027 0
2 3
0:7004
6 0:0016 j0:0188 7
6 7
6 7
6 0:000061
j0:000914 7
6 7
6 0:1482 j0:0043 7
6 7
½0 1 0 0 0 0 0 0 6 7
6 0:5516 j0:0283 7
6 7
6 0:0317 j0:0041 7
6 7
6 7
4 0:0077 j0:0284 5
0:413 j0:0977
¼ 0:1772
j0:0114
From Eq. (2.104), the sensitivity of electromechanical oscillation modes to the

coefficient of damping torque provided by the PSS can be computed to be
6;7
@k 62;72 v62;72
w ð0:2586
j27:632Þð0:0016 j0:0188Þ
¼ ¼
@Dpss M 7
¼ 0:0743
j0:0056
At the complex frequency k6;7 ¼ 0:71 j8:44, the forward path can be cal-
culated to be
2.4 Examples 79
6;7 Þ ¼ aT ðsI A33 Þ1 bpss3

Fpss ðk 238
¼ ½ 0 1:425 3:283 0:617 0:729 2:698
0 2 3 11
1
B 6 7 C
B 6 1 7 C
B 6 7 C
B 6 1 7 C
B ð0:71 j8:44Þ6 7 C 2 3
B 6 7 C
B 6 1 7 C 10000
B 6 7 C 6
B 4 1 5 C 6 0 7 7
B C 6
B C 6 0 7 7
B 1 C
B 2 3C 6 7
B 100 6116:3 1404:6 2646 3127 1154:5 C 6 6 0 7
7
B C 6 7
B 6 0 6:65 7 C
135:03 7 C 4 0 5
B 6 478:4 2:88 3:40
B 6 7C
B 6 0 513:64 5:48 86:29 101:98 4:50 7 C 0
B 6 7C
B 6 314:16 1:22 7C
B 6 0:43 0 0:71 0 7C
B 6 7C
@ 4 0 1:36 0 1:88 3:49 0 5A
0 0 36:0 0 0 46:8
¼ 2:3823 j0:0253
Hence,
6;7
@k
R6;7 ¼ Fpss ðk6;7 Þ ¼ ð0:0743
j0:0056Þð2:3823 j0:0253Þ
@Dpss
¼ 0:1772
j0:0114
It is thus confirmed that the residue is equal to the forward path of the PSS
multiplied by the sensitivity of oscillation modes to the coefficient of damping
torque contribution from the PSS.
References
1. Demello FP, Concordia C (1969) Concepts of synchronous machine stability as affected by

excitation control. IEEE Trans Power Appar Syst 88(4):316–329
2. Heffron WG, Phillips RA (1952) Effect of modern amplidyne voltage regulators on
underexcited operation of large turbine generators. AIEE Trans Power Apparatus Syst
71:692–697
3. Demello FP, Laskowski TF (1975) Concepts of power system dynamic stability. IEEE Trans
Power Apparatus Syst 94(3):827–833
4. Larsen EV, Swann DA (1981) Applying power system stabilizers part I: general concepts. IEEE
Trans Power Appar Syst 100(6):3017–3024
Chapter 3
Damping Torque Analysis
of Thyristor-Based FACTS Stabilizers
Installed in Single-Machine Infinite-Bus
Power Systems

with an SVC Stabilizer
3.1.1 Extended Heffron—Phillips Model of a Single-Machine

Infinite-Bus Power System Installed with an SVC
Stabilizer
3.1.1.1 Nonlinear Mathematical Model of a Single-Machine

Infinite-Bus Power System Installed with an SVC Stabilizer
Installation of a shunt thyristor-controlled reactive power compensator, the static

VAR compensator (SVC), is for the voltage support at a key location in a power
system. An additional stabilizing signal can be superimposed on the voltage control
loop of the SVC to provide extra damping to the power system. This supplementary
damping controller is named the SVC stabilizer in this book. Figure 3.1 shows the
configuration of a thyristor-controlled reactor and fixed capacitor (TCR-FC) type of
SVC with an additional damping control loop, i.e., the SVC stabilizer.
Figure 3.2 shows the configuration of a single-machine infinite-bus power
system installed with an SVC, where bsvc is the equivalent admittance of the SVC.
For the TCR-FC, SVC [1]
jbsvc ¼ ½1 CðaÞ=jxsvcl 1=jxsvcc ð3:1Þ
where CðaÞ ¼ 2asin

2p
2a
1 and a is the firing angle.
Isb ¼ Its Is ¼ Its jbsvc V

s
ð3:2Þ
V s ¼ jxsbIsb þ V b ¼ jxsbIts þ xsb bsvc V
s þ V
b

DOI 10.1007/978-1-4899-7696-3_3
82 3 Damping Torque Analysis of Thyristor-Based FACTS Stabilizers …
x svcl
α0
Vs π
Vsref - x svcc
voltage α firing circuit
controller +
+
π/2
ysvc SVC
stabilizer
Fig. 3.1 Configuration of a TCP-FC SVC with damping control function
Fig. 3.2 A single-machine Vt Vs Vb

infinite-bus power system xts xsb
installed with an SVC
Its I sb
Is jb svc
t ¼ jxtsIts þ V
V s ð3:3Þ

s ¼ jxsb Its þ Vb
V ð3:4Þ
csvc
where csvc ¼ 1 xsb bsvc :

By substituting Eq. (3.4) into Eq. (3.3) it can have
t ¼ jxlRIts þ V
V b =csvc ð3:5Þ
where xlR ¼ xts þ xsb =csvc .

Equation (3.5) indicates that the power system installed with the SVC is
equivalent to a power system without the SVC installed. The equivalent system is

of a line impedance xlR connected to an infinite bus with a voltage being cVsvcb .
3.1 A Single-Machine Infinite-Bus Power System Installed … 83
Therefore, the mathematical model of the single-machine infinite-bus power system

installed with the SVC as shown in Fig. 3.2 can be obtained simply by modifying
the model of the system without the SVC of Eqs. (2.37) and (2.38). The modifi-
cation is to replace xt and Vb in Eqs. (2.37) and (2.38) by xlR and cVsvcb , respectively,
to obtain that
:
d ¼ xo ðx 1Þ
: 1
x ¼ ½Pm Pt D(x 1Þ
M
0 1 ð3:6Þ
E_ q ¼ 0 ðEq þ Efd Þ
Tdo
0 1 KA
E_ fd ¼ E0fd þ ðVtref Vt Þ
TA TA
E0q Vb V2b ðxq x0d Þ

Pt ¼ sin d sin 2d
csvc x0dR c2svc 2 x0dR xqR
E0q xdR ðxd x0d Þ Vb cos d
Eq ¼ 0
xdR csvc x0dR ð3:7Þ
Efd ¼ Efd0 þ E0fd
xq Vb sin d xt E0q Vb x0d cos d qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
vtd ¼ ; vtq ¼ 0 þ 0 ; Vt ¼ v2td + v2tq
csvc xqR xdR csvc xdR
where x0dR ¼ x0d þ xlR ; xdR ¼ xd þ xlR ; xqR ¼ xq þ xlR :

Similar modification of Eq. (2.33) gives
E0q Vb cos d
itsd ¼
x0dR csvc x0dR
ð3:8Þ
Vb sin d
itsq ¼
csvc xqR
where itsd and itsq are the d and q component of line current Its , respectively.
3.1.1.2 Extended Heffron–Phillips Model
By linearizing the first two equations in Eq. (3.7), it can have
DPt ¼ K1 Dd þ K2 DE0q þ Kp Dbsvc

ð3:9Þ
DEq ¼ K3 DE0q þ K4 Dd þ Kq Dbsvc
where
E0q0 Vb V2b ðxq x0d Þ

K1 ¼ cos d0 cos 2d0
csvc0 x0dR c2svc0 x0dR xqR
Vb
K2 ¼ sin d0
csvc0 x0dR
0

@Pt Vb Eq0 sin d0 xsb xts þ x0d V2b ðxq x0d Þsin(2d0 Þxsb xts þ x0d
Kp ¼ ¼
@bsvc 0 ðcsvc0 x0dR Þ2 2c3svc0 ðx0dR Þ2 xqR

V2 ðxq x0d Þsin(2d0 Þxsb xts þ xq
b
2c3svc0 x0dR ðxqR Þ2

xsb xts þ x0d Vb E0q0 sin d0 V2b ðxq x0d Þsin(2d0 Þ
¼
csvc0 x0dR csvc0 x0dR 2c2svc0 x0dR xqR

Vb ðxq x0d Þsin(2d0 Þ xsb xts þ xq
2

2c2svc0 x0dR xqR csvc0 xqR
xdR
K3 ¼
x0dR
ðxd x0d Þ Vb sin d0
K4 ¼
csvc0 x0dR
E0q0 xdR x2 E0q0 x2 ðxd x0d Þ Vb cos d0
Kq ¼ 0 2 2sb þ 0 2sb 0 2
xsb xts þ x0d
ðxdR Þ csvc0 x dR csvc0 ðcsvc0 xdR Þ
In deriving Kp and Kq above, following equations are used
csvc x0dR ¼ csvc ðx0d þ xts þ xsb =csvc Þ ¼ csvc ðx0d þ xts Þ þ xsb
¼ ð1 xsb bsvc Þðx0d þ xts Þ þ xsb ð3:10Þ
csvc xq ¼ ð1 xsb bsvc Þðxq þ xts Þ þ xsb
By linearizing the last three equations in Eq. (3.7), it can have
xq Vb cos d0 xq Vb sin d0
Dvtd ¼ Dd þ xsb ðxsb þ xq ÞDbsvc
csvc0 xqR ðcsvc0 xqR Þ2
Vb x0d sin d0 xLR
Dvtq ¼ Dd þ 0 DE0q ð3:11Þ
csvc0 x0dR xdR
" #
xlR E0q0 x2 E0q0 x2 Vb x0d cos d0
þ 0 2 2sb þ 0 2sb þ xsb ðxsb þ x0d Þ Dbsvc
ðxdR Þ csvc0 xdR csvc0 ðcsvc0 x0dR Þ2
Hence,
vtd0 vtq0
DVt ¼ Dvtd þ Dvtq ¼ K5 Dd þ K6 DE0q þ Kv Dbsvc ð3:12Þ
Vt0 Vt0
where
vtd0 xq Vb cos d0 vtq0 Vb x0d sin d0

K5 ¼
Vt0 csvc0 xqR Vt0 csvc0 x0dR
vtq0 xLR
K6 ¼
Vt0 x0dR
vtd0 xq Vb sin d0
Kv ¼ xsb ðxsb þ xq Þ
Vt0 ðcsvc0 xqR Þ2
" #
vtq0 xlR E0q0 x2sb E0q0 x2sb Vb x0d cos d0 0
þ 0 2 2 þ 0 2 þ x ðx þ xd Þ
0 Þ2 sb sb
Vt0 ðxdR Þ csvc0 xdR csvc0 ðcsvc0 xdR
By substituting Eqs. (3.9) and (3.12) into the linearized equation of Eq. (3.6),
it can have
Dd_ ¼ xo Dx
1
Dx_ ¼ ðK1 Dd DDx K2 DE0q Kp Dbsvc Þ
M
_ 0 1 ð3:13Þ
DEq ¼ 0 ðK3 DE0q K4 Dd Kq Dbsvc þ DEfd Þ
Td0
0 1 0 KA
DE_ fd ¼ DE_ fd ðK5 Dd þ K6 DE0q þ Kv Dbsvc Þ
TA TA
Equation (3.13) is the extended Heffron–Phillips model of the single-machine

infinite-bus power system installed with the SVC, which is shown in Fig. 3.3.
3.1.1.3 Extended Heffron–Phillips Model with Both the SVC Voltage

and Damping Control Function Included
By substituting Its ¼ itsd þ jitsq into Eq. (3.4), it can have
Vb sin d itsq xsb

vsd ¼
csvc
ð3:14Þ
Vb cos d þ itsd xsb
vsq ¼
csvc
K1
1 ω0
- D + sM s
Δbsvc
K2 K4 K5
Kp Kq Kv
SVC
1
- - KA
K 3 + sTd0 ' + 1 + sTA
-
K6
Fig. 3.3 Extended Heffron–Phillips model of the single-machine infinite-bus power system
installed with an SVC stabilizer
s , respectively. By using
where vsd and vsq are the d and q component of voltage V
Eq. (3.8), from Eq. (3.14), it can be obtained that
Vb sin d V b sin d
csvc xqR xsb 1 xsb Vb sin d
vsd ¼ ¼ Vb sin d
csvc csvc csvc xqR
0
1 xsb Eq Vb cos d ð3:15Þ
vsq ¼ Vb cos d þ 0 xsb
csvc xdR csvc x0dR
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Vs ¼ v2sd þ v2sq
Linearization of Eq. (3.15) is as follows:
DVs ¼ C1 Dd þ C2 DE0q þ C3 Dbsvc ð3:16Þ
where

@Vs vsd0 @vsd vsq0 @vsq
C1 ¼ ¼ þ
@d 0 Vs0 @d 0 Vs0 @d 0

vsd0 1 xsb Vb cos d0 vsq0 1 Vb sin d0
¼ Vb0 cos d0 þ Vb0 sin d0 þ x sb
Vs0 csvc0 csvc0 xqR Vs0 csvc0 csvc0 x0dR

@Vs vsd0 @vsd vsq0 @vsq vsq0 1 xsb
C2 ¼ 0 ¼ þ ¼
@Eq Vs0 @E0q Vs0 @E0q Vs0 csvc0 x0dR

0 0
0

@Vs vsd0 @vsd vsq0 @vsq
C3 ¼ ¼ þ
@bsvc 0 Vs0 @bsvc 0 Vs0 @bsvc 0
" #
vsd0 xsb vsd0 x2sb Vb sin d0
¼ ðxq þ xts Þ
Vs0 csvc0 csvc0 ðcsvc0 xqR Þ2

vsq0 xsb xsb E0q0 Vb cos d0
þ V b cos d 0 þ x
x0dR csvc0 x0dR
sb
Vs0 c2svc0

1 E0q0 x3 Vb cos d0
þ 0 2 2sb 2 0 2 x2sb ðx0d þ xts Þ
csvc0 xdR csvc0 csvc0 xdR
For the simplicity of discussion, it is assumed that both the SVC voltage and
damping controller adopt the proportional control law with a gain CV and CS ,
respectively. If the feedback signal of the SVC stabilizer is the rotor speed of
generator, from Fig. 3.1, it can have
Da ¼ CV DVs þ CS Dx ð3:17Þ
1 1 C(aÞ
bsvc ¼ ð3:18Þ
xsvcc xsvcl
Hence,
_ 0Þ
Cða _ 0Þ
Cða
Dbsvc ¼ Da ¼ ðCV DVs þ CS DxÞ
xsvcl xsvcl ð3:19Þ
where
4 _ 0 Þ ¼ ð2 2cos 2a0 Þ [ 0
[ Cða
p p
By substituting Eq. (3.16) into Eq. (3.19), the following equation can be
obtained
:
Cða0 Þ
Dbsvc ¼ ðCV C1 Dd CV C2 DE0q CV C3 Dbsvc þ CS DxÞ ð3:20Þ
Xsvcl
Thus,
CV C1 Dd CV C2 DE0q þ CS Dx
Dbsvc ¼ x ð3:21Þ
: svcl þ CV C3
Cða0 Þ
K1
1 ω0
- D + sM s
SVC Cs
K2 C + C3 C V
−C 2 C V −C1CV
+
C + C3 C V C + C3 C V
Δbsvc
K4 K5
Kp Kq Kv
1
- - KA
K 3 + sTd0 ' 1 + sTA
-
+
K6
Fig. 3.4 Extended Philip–Heffron model with SVC voltage and damping control function
included
By taking account of Eq. (3.21), Fig. 3.3 can be further extended to include the
SVC voltage and damping control function as shown in Fig. 3.4, where C ¼ :xsvcl .
Cða0 Þ
3.1.1.4 Calculation of Initial Compensation
This section demonstrates how the initial compensation of the SVC installed in the
single-machine infinite-bus power system shown in Fig. 3.2 can be calculated to
satisfy the requirement to maintain the voltage profile of the system. The initial
conditions to compute the initial compensation, bsvc0 , of the SVC are the magnitude
of busbar voltage, Vt0 ; Vb0 ; Vs0 , and the active power supplied by the generator, Pt0 .
Figure 3.5 shows the circuit model of the system as shown in Fig. 3.2. From
Fig. 3.5, it can be obtained that
s0 V
V b
¼ Psb0 jQsb0
V ð3:22Þ
s0
jxsb
Since the resistance of the transmission lines is ignored, Psb0 ¼ Pt0 . By choosing
s0 as the reference phasor on x-axis, it can have V
V ¼ Vs0 . From
s0 ¼ V
s0
Eq. (3.14), the following equation can be obtained
b ¼ Vs0 xsb ðQsb0 þ jPsb0 Þ ¼ Vs0 xsb Qsb0 j xsb Psb0

V
Vs0 Vs0 Vb
The above equation gives

2 2
xsb xsb
V2b ¼ Vs0 Q þ Psb0 ð3:23Þ
Vs0 sb0 Vs0
Solution of Eq. (3.23) is as follows

2 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 3
2ffi
Vs0 4 x
Psb0 5
sb
Qsb0 ¼ Vs0 V2b ð3:24Þ
xsb Vs0
Pt 0 + jQ t 0 jx ts Pts0 + jQ ts0 Psb0 + jQsb0 jx sb Pb0 + jQ b0

Vt0 V s0 Vb
I ts0 Isb0
Ps0 + jQs0
bsvc0
j Is
Fig. 3.5 Circuit model of the single-machine infinite-bus power system installed with SVC
Similarly, from the following equation
t0 V
V s0
¼ Pts0 jQts0
V s0
jxts
it can have
2 2
xts xts
V2t0 ¼ Vs0 þ Q þ Pts0 ð3:25Þ
Vs0 ts0 Vs0

2sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 3
2
Vs0 4 x
Pts0 Vs0 5
ts
Qts0 ¼ V2t0 ð3:26Þ
xts Vs0
Hence,
Ps0 þ jQs0 ¼ 0:0 þ j(Qts0 Qsb0 Þ
Since
s0 jbsvc0 ¼ Is0
V
it can have
jbsvc0 ¼ Is0 V
s0 V
V ¼ Ps0 jQs0
s0 s0
Finally from the above equation, the initial compensation is calculated to be
ðPs0 jQs0 Þ Q Q Q
bsvc0 ¼ ¼ s0 ¼ sb0 2 ts0 ð3:27Þ
jV2s0 V2s0 Vs0
3.1.2 Damping Torque Analysis of SVC Stabilizer
3.1.2.1 Electric Torque Provided by the SVC Stabilizer
From Fig. 3.4, the forward path of the SVC stabilizer to the electromechanical
oscillation loop of generator can be established as shown in Fig. 3.6.
From Fig. 3.6, it can be seen that the SVC stabilizer contributes the electric
torque to the electromechanical oscillation loop through two paths. The electric
Δω
-
Cs
K2 C + C3 C V
−C 2 C V
+
C + C3 C V
Δbsvc
Kp Kq Kv
1
- KA
K 3 + sTd0 ' 1 + sTA
-
+
K6
Fig. 3.6 Forward path of the SVC stabilizer
torque through the block, KP , is called the direct electric torque, denoted by DTed .
That through the blocks, Kq and KV , is named the indirect electric torque, denoted
by DTei . According to the damping torque analysis introduced in Chap. 2, electric
torque can be decomposed into a damping torque and a synchronizing torque at the
angular oscillation frequency, xs , that is
DTed ¼ DTsd þ DTdd ¼ Tsd Dd þ Tdd Dx

ð3:28Þ
DTei ¼ DTsi þ DTdi ¼ Tsi Dd þ Tdi Dx
where Tsd ; Tsi ; Tdd ; Tdi are direct, indirect synchronizing torque and direct, indirect
damping torque coefficients, respectively. From Fig. 3.6, the transfer function of
forward path of the SVC stabilizer to the electromechanical oscillation loop of
generator can be derived to be
1 1 KA
K3 þ sT0d0 K3 þ sT0d0 1 þ sTA
Fsvc ðs) ¼ Kp K2 Kq K2 Kv
1 þ K6 1 KA 1 þ K6 1 KA
K3 þ sT0d0 1 þ sTA K3 þ sT0d0 1 þ sTA
K2 ½Kq ð1 þ sTA Þ þ Kv KA
¼ Kp
ðK3 þ sT0d0 Þð1 + sTA Þ + K6 KA
ð3:29Þ
Hence, the electric torque provided by the SVC stabilizer is as follows
DTet ¼ Fsvc ðjxs ÞDbsvc ð3:30Þ
From Fig. 3.6, it can be seen that the signals are attenuated by the exciter and the
AVR before they form the indirect damping toque. Hence, normally Tdd Tdi .
From Fig. 3.6 and Eqs. (3.29) and (3.30), it can be seen that the coefficient Kp is the
dominant weight parameter in determining the amount of damping torque contri-
bution from the SVC stabilizer. Hence, in the following section, how the effec-
tiveness of the SVC stabilizer is affected by various factors is examined by
checking the dominant weight parameter KP .
3.1.2.2 Damping Control as Affected by the Load Conditions

Pt0 ¼ sin d0 sin 2d0 ¼ Pt10 Pt20 ð3:31Þ
csvc0 x0dR c2svc0 2 x0dR xqR

where Pt10 ¼ csvc0 x0dR sin d0 ; Pt20 ¼ c2svc0 2 x0dR xqR sin 2d0 :
From Eq. (3.9) and by using Eq. (3.31), the weight parameter Kp can be obtained
to be

xsb xts þ x0d Vb E0q0 sin d0 V2b ðxq x0d Þsin(2d0 Þ
Kp ¼
csvc0 x0dR csvc0 x0dR 2c2svc0 x0dR xqR
0

Vb ðxq xd Þsin(2d0 Þ xsb xts þ xq
2
ð3:32Þ
2c2svc0 x0dR xqR csvc0 xqR
0

xsb xts þ xd xts þ xq
¼ Pt0 Pt20
csvc0 x0dR xqR
From Fig. 3.7 and Eq. (3.32), it can be seen that at a higher load condition,
difference between Pt0 and Pt20 is bigger. Thus, the higher the load condition is, the
bigger Kp is, the more damping torque is provided by the SVC stabilizer. Hence,
effectiveness of the SVC stabilizer should increase with system load conditions.
Fig. 3.7 Pt d curve of the Pt 0 = Pt10 − Pt 20

power system
Pt10
Pt 20
π/2 δ0
π/4 π
3.1.2.3 Influence of Parameters of the Generator
Since xq [ x0d , it should have

xts þ x0d 1 1 xts þ xq
¼ \ ¼ ð3:33Þ
x0dR 1 þ c xxsbþ x0 1 þ c xxsbþ x xqR
svc0 ð ts dÞ svc0 ð ts qÞ
Equations (3.32) and (3.33) indicate that Kp is affected not only by the load
conditions of the power system, but also by the parameters of the generator, x0d and
xq , such that:
1. If the difference between the values of x0d and xq is small (for example, for the
generator in a hydropower station, value of x0d is around 0.2–0.35 and that of xq
is around 0.45–0.7) or if the system operates at a higher load condition
ðxts þ x0 Þ ðxts þ xq Þ
(Pt0 Pt20 ), x0 d Pt0 [ xqR Pt20 such that Kp [ 0, the SVC stabilizer
dR
will provide the power system with positive damping.
2. If xq is much greater than x0d such that x0 d xqR ; (for example, for a
dR
generator in a thermal power plant, x0d is around 0.15 to 0.24 and xq is around
1.2–2.2) and when the system operates at a lower load condition such that the
difference between Pt0 and Pt20 is small, there is a possibility that
ðxts þ x0d Þ ðxts þ xq Þ
x0dR Pt0 \ xqR Pt20 such that Kp \0. This means that the SVC stabilizer
which supplies positive damping torque at a higher load condition may provide
the power system with negative damping torque at a lower load condition.
3. Around the operating point where x0 d Pt0 ¼ xqR Pt20 such that Kp ¼ 0,
dR
the damping torque provided by the SVC stabilizer is zero. At this operating
point, the SVC stabilizer will lose its capability to suppress power system
oscillation. This load condition is called the “dead point” of the SVC stabilizer’s
function.
3.1.2.4 Electric Length of the Transmission Line
Define z to be an index of the length of the transmission line and take z0 ¼ 1. The
reactance of the transmission line is proportional to its electric length, i.e., the index z.
Kp can be written as a function z as (see Eq. (3.32))

zxsb zxts þ x0d zxts þ xq
Kp ðz) ¼ Pt0 Pt20 ð3:34Þ
csvc0 x0d þ z(xts þ xsb =csvc0 Þ xq þ z(xts þ xsb =csvc0 Þ
For the simplicity of discussion, it is assumed that at the steady-state operation,

bsvc0 ¼ 0 such that csvc0 ¼ 1. Hence,

zxts þ x0d zxts þ xq
Kp ðz) ¼ zxsb Pt0 Pt20 ð3:34Þ
x0d þ z(xts þ xsb Þ xq þ z(xts þ xsb Þ
It can have

@ zxts þ x0d zxts þ xq
Kp ðz) ¼ xsb 0 Pt0 Pt20
@z xd þ z(xts þ xsb Þ xq þ z(xts þ xsb Þ
(
xts zxts þ x0d ðxts þ xsb Þ
þ zxsb 0 Pt0 Pt0
xd þ z(xts þ xsb Þ ½x0d þ z(xts þ xsb Þ2
)
xts zxts þ xq ðxts þ xsb Þ
Pt20 þ Pt20
xq þ zðxts þ xsb Þ ½xq þ z(xts þ xsb Þ2
Hence,

@Kp ðz) xts þ x0d xts þ xq
¼ x P P
@z z¼z0 ¼1
sb to t02
x0d þ ðxts þ xsb Þ xq þ ðxts þ xsb Þ
(
xts xts þ x0d ðxts þ xsb Þ
þ xsb 0 Pt0 0 Pt20
xd þ ðxts þ xsb Þ ½xd þ ðxts þ xsb Þ2
)
xts xts þ xq ðxts þ xsb Þ
Pt20 þ Pt20
xq þ ðxts þ xsb Þ ½xq þ ðxts þ xsb Þ2

xsb xts þ x0d xsb xts xsb xts þ x0d ðxts þ xsb Þ
¼ 0 Pt0 þ 0 Pt0 Pt0
xd þ ðxts þ xsb Þ xd þ ðxts þ xsb Þ ðx0d þ xts þ xsb Þ2

xsb xts þ xq xsb xts xsb xts þ xq ðxts þ xsb Þ
Pt20 Pt20 þ Pt20
xq þ ðxts þ xsb Þ xq þ ðxts þ xsb Þ ðxq þ xts þ xsb Þ2

xsb xts þ x0d xsb xts xsb xts þ x0d ðxts þ xsb Þ
¼ 0 Pt0 þ 0 Pt0 Pt0
xd þ ðxts þ xsb Þ xd þ ðxts þ xsb Þ ðx0d þ xts þ xsb Þ2

xsb xts þ xq xsb xts xsb xts þ xq ðxts þ xsb Þ
Pt20 Pt20 þ Pt20
xq þ ðxts þ xsb Þ xq þ ðxts þ xsb Þ ðxq þ xts þ xsb Þ2

xsb xts þ x0d x0d þ xsb xts x0d þ xts þ xsb
¼ Pt0
ðx0d þ xts þ xsb Þ2

xsb xts þ xq xq þ xsb xts xq þ xts þ xsb
2 Pt20
xq þ xts þ xsb

xsb x0 xts þ x0d xsb xts xsb xq xts þ xq xsb xts
¼ 0 d P t0 þ 0 P t0 P
2 t20
Pt20
ðxd þ xts þ xsb Þ2 xd þ xts þ xsb xq þ xts þ xsb x q þ xts þ xsb

xsb x0d xts þ x0d xsb xq xts þ xq xsb xts xsb xts
¼ Pt0 Pt20 þ 0 Pt0 Pt20
x02
dR x2qR xdR xqR

xsb x0d xts þ x0d xsb xq xts þ xq Pt0 Pt20
¼ 02
Pt0 2
Pt20 þ xsb xts 0
xdR xqR xdR xqR
ð3:35Þ
Thus,

@Kp ðz) xsb x0d xts þ x0d þ xsb xts x0dR xsb xq xts þ xq þ xsb xts xqR
¼ P Pt20
@z z¼z0 ¼1
t0
X02
dR X2qR
2 2
xsb xts þ x0d þ xts x2sb xsb xts þ xq þ xts x2sb
¼ Pt0 Pt20
x02
dR x2qR
ð3:36Þ
When Kp [ 0

xts þ x0d xts þ xq
Pt0 [ Pt20 ð3:37Þ
x0dR xqR
Hence, normally it should have

2 2
xts þ x0d xts þ xq
Pt0 [ Pt20 ð3:38Þ
x0dR xqR
Using Eqs. (3.36) and (3.38), from Eq. (3.35), it can be obtained that
@Kp ðz)
[0 ð3:39Þ
@z
Equation (3.39) means that for the SVC stabilizer designed to supply positive
damping to power system oscillations, it will be more effective when the electric
length of transmission lines increases.
3.1.2.5 Installing Location of the SVC
If it is assumed that the total length of the transmission line is fixed, that is,
xts xsb ¼ xt , from Eq. (3.32), it can be obtained that (assuming csvc0 ¼ 1)

xsb xt xsb þ x0d xsb xt xsb þ xq
Kp ðxsb Þ ¼ P P ð3:40Þ
xt þ x0d
t0 t20
xq þ xt
Define the following two functions of xsb

xsb xt xsb þ x0d
g(xsb Þ ¼ Pt0
x0 þ xt
d ð3:41Þ
xsb xt xsb þ xq
h(xsb Þ ¼ Pt20
xq þ xt
Thus,
Kp ðxsb Þ ¼ g(xsb Þ h(xsb Þ ð3:42Þ
By solving the following two equations
@ xt 2xsb þ x0d
g(xsb Þ ¼ Pt0 ¼ 0
@xsb x0d þ xt
ð3:43Þ
@ xt 2xsb þ xq
h(xsb Þ ¼ Pt20 ¼ 0
@xsb xq þ xt
ðx0d þ xt Þ ðx þ x Þ
It can have xsb ¼ ¼ xsb1 and xsb ¼ q 2 t ¼ xsb2 . Hence, g(xsb Þ and
2
0
ðx þ x Þ ðx þ x Þ
h(xsb Þ achieve its maximum value at xsb ¼ d 2 t ¼ xsb1 and xsb ¼ q 2 t ¼ xsb2 ,
respectively.
K p (x sb )
g(x sb )
K p (x sb )
h(x sb )
x sb
0 xt x sb1 x sb2 xt
2
Fig. 3.8 Illustration on the most effective installing location of the SVC stabilizer
From Eqs. (3.42) and (3.43), the curve of Kp ðxsb Þ can be sketched as shown in
Fig. 3.8 (Pt0 Pt20 ). Figure 3.8 demonstrates that when xsb x2t , Kp ðxsb Þ achieves
its maximum value. This means that when the SVC is installed near the middle
point of the transmission line, the SVC stabilizer contributes the most amount of
damping torque and hence is most effective in suppressing power system
oscillations.

with a TCSC or TCPS Stabilizer

Infinite-Bus Power System Installed with a TCSC
or TCPS Stabilizer
3.2.1.1 Extended Heffron–Phillips Model of a Single-Machine

Infinite-Bus Power System Installed with a TCSC
It was discovered decades ago that power oscillations along transmission lines can
be effectively suppressed by switching series capacitors in and out from the lines
after a fault [2, 3]. Nowadays, the high-speed thyristor-controlled series compensator
(TCSC) has made it practically possible. A TCSC can be simply represented as a

capacitor whose capacitance is adjustable within a certain range as shown in Fig. 3.9,
which is a single-machine infinite-bus power system installed with a TCSC.
Obviously, installation of the TCSC only changes the equivalent reactance of the
transmission line connecting the generator and the infinite busbar. Therefore, the
mathematical model of the system can be established by simply modifying Eqs. (2.
37) and (2.38). The modification is just to change the expressions of the equivalent
reactance of the transmission line as follows
d_ ¼ xo ðx 1Þ
1
x_ ¼ ½Pm Pt D(x 1Þ
M
0 1 ð3:44Þ
E_ q ¼ 0 ðEq þ Efd Þ
Tdo
0 1 KA
TA TA

E0q Vb V2b xq x0d
Pt ¼ 0 sin d sin 2d
xdR 2 x0dR xqR

E0q xdR xd x0d Vb cos d
Eq ¼ 0
xdR x0dR ð3:45Þ
Efd ¼ Efd0 þ E0fd
xq Vb sin d xt E0q Vb x0d cos d qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
vtd ¼ ; vtq ¼ 0 þ ; Vt ¼ v2td þ v2tq
xqR xdR x0dR
where
xdR ¼ xd þ xL xtcsc ; x0dR ¼ x0d þ xt xtcsc

ð3:46Þ
xqR ¼ xq þ xL xtcsc
Vt Vb
− jx tcsc
jx t
Fig. 3.9 A single-machine infinite-bus power system installed with a TCSC

By linearizing Eq. (3.45), it can be obtained that
DPt ¼ K1 Dd þ K2 DE0q þ Kp Dxtcsc

DEq ¼ K3 DE0q þ K4 Dd þ Kq Dxtcsc ð3:47Þ
DVt ¼ K5 Dd þ K6 DE0q þ Kv Dxtcsc
where

E0q0 Vb V2b xq x0d
x0dR x0dR xqR
Vb
K2 ¼ 0 sin d0
xdR
0
@Pt Vb Eq0 sin d0 V2b ðxq x0d Þ sin 2d0 0
Kp ¼ ¼ 02
02
xdR þ xqR
@x tcsc 0 x dR 2x x 2
dR qR
xdR
K3 ¼ 0
xdR

xd x0d Vb sin d0
K4 ¼
x0dR
0
@Eq Eq0 xdR xd x0d Vb cos d0
Kq ¼ ¼ 02
@xtcsc 0 xdR x02
dR
0
vtd0 xq Vb cos d0 vtq0 Vb xd sin d0
K5 ¼
Vt0 xqR Vt0 x0dR
vtq0 xLR
K6 ¼
Vt0 x0dR

@Vt vtd0 xq Vb sin d0 vtq0 xL 1 0 Vb x0d cos d0
Kv ¼ ¼ þ 0 Eq0 þ
@xtcsc 0 Vt0 x2qR Vt0 x02
dR xdR x02dR
By substituting Eq. (3.46) into the linearized equation of Eq. (3.44), it can have
:
D d ¼ xo Dx
: 1
D x ¼ ðK1 Dd DDx K2 DE0q Kp Dxtcsc Þ
M
0 1 ð3:48Þ
DE_ q ¼ 0 ðK3 DE0q K4 Dd Kq Dxtcsc þ DE0fd Þ
Td0
1 KA
DE0fd ¼ DE0fd ðK5 Dd þ K6 DE0q þ Kv Dxtcsc Þ
TA TA
K1
1 ω0
- D + sM s
Δx t csc
K2 K4 K5
Kp Kq Kv
TCSC
1
- - KA
K 3 + sTd0 ' + 1 + sTA
-
K6
installed with a TCSC stabilizer

infinite-bus power system installed with the TCSC stabilizer, which is shown in
Fig. 3.10.
3.2.1.2 Extended Heffron–Phillips Model of a Single-Machine

Infinite-Bus Power System Installed with a TCPS
The phase shifters have been used to regulate the steady-state power flow in power
systems. The idea of applying online control of phase shifters was proposed dec-
ades ago. However, the low-speed mechanical tap changers precluded the use of
online control of phase shifters to improve power system dynamic performance.
With the advances in power electronics, high-power high-speed electronic switches
make it possible to realize real-time control of a phase shifter. The feasibility of a
thyristor-controlled phase shifter (TCPS) as a means to adjust the value of
quadrature component of line voltage has been well recognized.
Figure 3.11 illustrates the arrangement of a TCPS installed along a transmission
line. Its function can be simply expressed as a phase shifting of line voltage as
shown in Fig. 3.11. Hence, for a single-machine infinite-bus power system installed
with a TCPS as shown in Fig. 3.12, system dynamic Eqs. (2.37) and (2.38) need to
be modified only in the phase relationship as follows.
a V V'
Vector diagram
Symbol
V V'
V
φ
V'
Converter
TCPS
Fig. 3.11 Arrangement of a TCPS and its simple representation
Vt ∠δ Vt ∠δ + φ Vb
TCPS jx t
Fig. 3.12 The single-machine infinite-bus power system installed with a TCPS
:
d ¼ xo ðx 1Þ
: 1
x ¼ ½Pm Pt D(x 1Þ
M
ð3:49Þ
_E0 ¼ 1 ðEq þ Efd Þ
q
T0do
0 1 KA
TA TA
E0q Vb V2b ðxq x0d Þ

Pt ¼ sin(d þ /Þ sin2(d þ /Þ
x0dR 2 x0dR xqR
E0q xdR ðxd x0d ÞVb cos(d þ /Þ
Eq ¼ 0
xdR x0dR
ð3:50Þ
xq Vb sin(d þ /Þ
vtd ¼
xqR
xL E0q Vb x0d cos(d þ /Þ
vtq ¼ 0 þ
xdR x0dR
By linearizing Eq. (3.50), it can have
DPt ¼ K1 Dd þ K2 DE0q þ Kp D/
DEq ¼ K3 DE0q þ K4 Dd þ Kq D/ ð3:51Þ
DVt ¼ K5 Dd þ K6 DE0q þ Kv D/
where

x0dR x0dR xqR
Vb
K2 ¼ 0 sin d0
xdR
0
@Pt Eq0 Vb V2 ðxq x0d Þ
Kp ¼ ¼ 0 cos d0 b 0 cos 2d0
@/ 0 x dR x x dR qR
xdR
K3 ¼ 0
xdR
ðxd x0d ÞVb sin d0
K4 ¼
x0dR

@Eq ðxd x0d ÞVb sin d0
Kq ¼ ¼
@/ 0 x0dR
vtd0 xq Vb cos d0 vtq0 Vb x0d sin d0
K5 ¼
Vt0 xqR Vt0 x0dR
vtq0 xLR
K6 ¼
Vt0 x0dR

@Vt vtd0 xq Vb cos d0 vtq0 Vb x0d sin d0
Kv ¼ ¼
@/ 0 Vt0 xqR Vt0 x0dR
By substituting Eq. (3.51) into the linearized equation of Eq. (3.49), it can be
obtained that
:
D d ¼ xo Dx
: 1
D x ¼ ðK1 Dd DDx K2 DE0q Kp D/Þ
M
0 1 ð3:52Þ
DE_ q ¼ 0 ðK3 DE0q K4 Dd Kq D/Þ
Td0
1 KA
DE0fd ¼ DE0fd ðK5 Dd þ K6 DE0q þ Kv D/Þ
TA TA

infinite-bus power system installed with the TCSC stabilizer, which is shown in
Fig. 3.13.
K1
1 ω0
- D + sM s
Δφ
K2 K4 K5
Kp Kq Kv
TCPS
1
- - KA
K 3 + sTd0 ' + 1 + sTA
-
K6
installed with a TCPS stabilizer
3.2.2 Damping Torque Analysis of TCSC and TCPS

Stabilizers
3.2.2.1 General Expression of Damping Torque Contributed

by the TCSC and TCPS Stabilizers
For the simplicity of discussion, it is assumed that both the TCSC and TCPS
stabilizers adopt the deviation of rotor speed of the generator as the feedback signal.
Denote the transfer function of the stabilizers by Ktc Ttc ðs). The output control signal
of the TCSC and TCPS stabilizers is as follows
Dutc ¼ Ktc Ttc ðs)Dx ð3:53Þ
where Dutc is Dxtcsc or D/ for the TCSC stabilizer or TCPS stabilizer, respectively.
From Figs. 3.10 and 3.13, the forward path of the stabilizer can be obtained and
shown in Fig. 3.14.
From Fig. 3.14, it can be seen that the direct damping torque contribution, DTdd ,
is through Kp . It should be much greater than the indirect damping torque, DTdi ,
ΔTd
-
ΔTdd
ΔTdi
Δu tc
K2
Kp Kq Kv
1
- KA
K 3 + sTd0 ' + 1 + sTA
-
K6
Fig. 3.14 Forward path of the TCSC or TCPS stabilizer
because Dutc is attenuated by two first-order lag blocks before it forms the indirect
damping torque. That is
DTd Tdd Dx ¼ Re[Kp Ktc Ttc ðjxs ÞDx ð3:54Þ
From Eq. (3.54), it can be seen that coefficient Kp weighs the amount of damping
torque contributed by the TCSC and TCPS stabilizers to the power system.
3.2.2.2 Damping Torque Provided by the TCSC Stabilizer [4]

E0q0 Vb V2b xq x0d
Pt0 ¼ 0 sin d0 sin 2d0 ¼ Pt10 Pt20 ð3:55Þ
xdR 2 x0dR xqR
where
E0q0 Vb
Pt10 ¼ sin d0
x0dR
ð3:56Þ
V2 ðxq x0d Þ
Pt20 ¼ b 0 sin 2d0
2 xdR xqR
Hence, for the TCSC stabilizer, from Eq. (3.47), it can be obtained that
0
@Pt Vb Eq0 sin d0 V2b ðxq x0d Þ sinð2d0 Þ 0
Kp ¼ ¼ 02
02
xdR þ xqR
@xtcsc 0 xdR 2
2xdR XqR
0
1 Vb Eq0 sin d0 Vb ðxq xd Þsin(2d0 Þ
2 0
1 V2b xq x0d sinð2d0 Þ
¼ 0
xdR x0dR 2x0dR xqR xqR 2x0dR xqR
1 1
¼ 0 Pt10 Pt20
xdR xqR
ð3:57Þ
Since xqR [ x0dR and Pt0 [ Pt20 as shown by Fig. 3.15,
Kp [ 0 ð3:58Þ
Equation (3.58) indicates that when the load conditions of the power system
change, a properly designed TCSC stabilizer can always provide positive damping
torque to the power system.
At a higher load condition, difference between Pt0 and Pt20 is greater as shown
in Fig. 3.15 and hence Kp is bigger as shown in Eq. (3.57). This means that
Fig. 3.15 Pt d curve of the Pt 0 = Pt10 − Pt 20

power system installed with a
TCSC
Pt10
Pt 20
π/2 δ0
π/4 π
effectiveness of the TCSC stabilizer in damping power system oscillations increases

with the increase in system load conditions.
!
@Kp Pt0 Pt20
¼ 02 2 \0 ð3:59Þ
@xt xdR xqR
Equation (3.59) indicates that the TCSC stabilizer is less effective when the
equivalent reactance of transmission line increases if the load condition keeps
unchanged.
3.2.2.3 Damping Torque Provided by the TCPS Stabilizer [5]
For the TCPS stabilizer, from Eq. (3.51), it can have

@Pt
Kp ¼ ¼ Kp1 Kp2 ð3:60Þ
@/ 0
where
Vb E0q0 cos d0
Kp1 ¼
x0
dR 0 ð3:61Þ
V xq xd cos 2d0
2
Kp2 ¼ b
2x0dR xqR
Kp1 and Kp2 are shown in Fig. 3.16. From Fig. 3.16, it can be seen that Kp [ 0.
Hence, when the load conditions of the power system change, a properly designed
TCPS stabilizer can always supply positive damping torque to the power system.
!
@Kp Kp1 Kp2
¼ 02 2 \0 ð3:62Þ
@xt xdR xqR
Equation (3.62) indicates that the TCPS stabilizer is less effective when the
equivalent reactance of the transmission line increases if the loading condition is
unchanged.
3.3 An Example Power System Installed with an SVC Stabilizer 107
Pt 0
a
K p1
K p2 b
π/2 δ0
π/4 π
Vb E q0 'sin δ0 Vb 2 (x q − x d ') sin 2 δ0

a= ,
b=
x dΣ ' 2x dΣ ' x qΣ
Fig. 3.16 Examination of the damping effectiveness of the TCPS stabilizer
3.3 An Example Power System Installed with an SVC

Stabilizer
Parameters of an example single-machine infinite-bus power system with an SVC

installed in Fig. 3.2 are (in p.u. except indicated)
0
Generator: xd ¼ 1:0; xq ¼ 0:8; x0d ¼ 0:15; M ¼ 6:0 s:; D ¼ 0; Td0 ¼ 5:044 s:
The AVR: KA ¼ 20:0; TA ¼ 0:01 s:
Transmission line: Xts ¼ 0:3; Xsb ¼ 0:3:
The SVC: xsvcl ¼ 1; xsvcc ¼ 1; Kvp ¼ 1; Kvi ¼ 8:
Steady-state operating condition: Pt0 ¼ 0:5; Vt0 ¼ 1:0; Vs0 ¼ 1:0; Vb ¼ 1:0
3.3.1 Linearized Model of Example Power System
3.3.1.1 Linearized Model with the Transfer Function of the SVC

Voltage Controller Included
The SVC is equipped with a proportional–integral (PI) voltage controller. The added
damping controller adopts the locally available deviation of line active power as the
feedback signal and the transfer function of a conventional PSS. That is (see Fig. 3.1)

Kvi
a ¼ a0 þ Kvp þ ðVsref Vs Þ þ usvcs ð3:63Þ
s
where usvc-s is the output stabilizing signal of the SVC stabilizer, i.e.,
ð1 þ sT2 Þð1 þ sT4 Þ

usvcs ¼ Ks ðPtref Pt Þ
Linearization of Eq. (3.63) is as follows

Kvi
Da ¼ Kvp þ DVs þ Dusvcs ð3:64Þ
s
:
From Eq. (3.19) with Dbsvc ¼ Cxða 0Þ

svcl
Da, the above equation can be converted to
: :
Cða0 Þ Kvi Cða0 Þ
Dbsvc ¼ Kvp þ DVs þ Dusvcs ð3:65Þ
xsvcl s xsvcl
Substituting Eq. (3.16) into the above equation, it can have

: i C: ða Þ
Cða0 Þ Kvi h
C1 Dd þ C2 DE0q þ C3 Dbsvc þ
0
Dbsvc ¼ Kvp þ Dusvcs
xsvcl s xsvcl
ð3:66Þ
The above equation can give
Dbsvc ¼ Fsvc1 ðs)(C1 Dd þ C2 DE0q Þ þ Fsvc2 ðs)Dusvcs ð3:67Þ
where
:
:
Cxða 0Þ
K vp þ Kvi
s 1 Cða0 Þ
Fsvc1 ðs) ¼ ; Fsvc2 ðs) ¼
svcl
: :
1 þ C3 Cxða 0Þ
K þ Kvi 1 þ C Cða0 Þ K þ Kvi xsvcl
svcl
vp s 3 xsvcl vp s
K1
1 ω0
-
D + sM s
SVC
Fsvc2 (s)
K2 Δu svc −s
C2 Fsvc1 (s) + C1Fsvc1 (s)
Δbsvc
K4 K5
Kp Kq Kv
1
- - KA
K 3 + sTd0 ' 1 + sTA
-
+
K6
Fig. 3.17 Linearized model with SVC voltage control function included
From Eq. (3.67) and Fig. 3.3, linearized Heffron–Phillips model of the power
system with the SVC installed is shown in Fig. 3.17, where the SVC voltage
controller adopts a PI controller and usvcs is the stabilizing signal of the SVC-based
stabilizer.
:
In Eq. (3.66), denote Dzsvc ¼ Cxða 0 Þ Kvi

svcl
0
s ðC1 Dd þ C2 DEq þ C3 Dbsvc Þ, that is
:
: Cða0 Þ
D z ¼ Kvi ðC1 Dd þ C2 DE0q þ C3 Dbsvc Þ ð3:68Þ
svc xsvcl
Equation (3.66) becomes

:
_ 0Þ
Cða Cða0 Þ
Dbsvc ¼ Kvp C1 Dd þ C2 DE0q þ C3 Dbsvc þ Dz þ Dusvcs ð3:69Þ
xsvcl xsvcl

_
Cxða 0Þ
Dd þ DE 0 C_ ða0 Þ
K vp C1 C 2 q þ Dz þ xsvcl Dusvcs
Dbsvc ¼ ð3:70Þ
svcl
_
1 þ Cxða 0Þ
svcl
Kvp C3
Substituting Eq. (3.70) into Eq. (3.68), it can have
D_zsvc ¼ CA Dd þ CE DE0q þ CZ Dz þ CU Dusvcs ð3:71Þ
where
0: 1 :
: Cða0 Þ K C Cða0 Þ K C
Cða0 Þ B xsvcl vp 3 C xsvcl vi 3
CA ¼ Kvi C1 @1 : A; CZ ¼ : ;
xsvcl
1 þ Cxða 0Þ
svcl
K C
vp 3 1 þ Cxða 0Þ
svcl
K C
vp 3
0 1 : 2
:
: Cða0 Þ K C Cða0 Þ K C
Cða0 Þ B xsvcl vp 3 C xsvcl vi 3
CE ¼ Kvi C2 @1 : A; CU ¼ :
xsvcl
1 þ Cxða 0Þ
svcl
K C
vp 3 1 þ Cxða 0Þ
svcl
Kvp C3

:
D d ¼ xo Dx
: 1
D x ¼ ðK01 Dd DDx K02 DE0q Kpz Dz K0p Dusvcs Þ
M
0 1 ð3:72Þ
DE_ q ¼ 0 ðK03 DE0q K04 Dd Kqz Dz K0q Dusvcs þ DEfd Þ
Td0
0 1 0 KA 0
DE_ fd ¼ DE_ fd ðK Dd þ K06 DE0q þ Kvz Dz þ K0v Dusvcs Þ
TA TA 5
where
: :
0
Cxða 0Þ
Kvp C1 0
Cxða 0Þ
Kvp C2
K1 ¼ K1 þ Kp :
svcl
; K2 ¼ K2 þ Kp :
svcl
;
1 þ Cxða 0Þ
svcl
K vp C3 1 þ Cxða 0Þ
svcl
K vp C3
:
C ða0 Þ
Kp xsvcl
Kp
Kpz ¼ 1
; K0p ¼ 1
: :
1þ
Cða0 ÞK C 1þ
Cða0 ÞK C
xsvcl vp 3 xsvcl vp 3
: :
0
Cxða 0Þ
Kvp C2 0
Cxða 0Þ
Kvp C1
K3 ¼ K3 þ Kq :
svcl
; K4 ¼ K4 þ Kq :
svcl
;
1 þ Cxða 0Þ
svcl
K vp C3 1 þ Cxða 0Þ
svcl
K vp C3
:
C ða0 Þ
Kq xsvcl
Kq
Kqz ¼ 1
; K0q ¼ 1
: :
1þ
Cða0 ÞK C 1þ
Cða0 ÞK C
xsvcl vp 3 xsvcl vp 3
: :
Cxða 0Þ
Kvp C1 Cxða 0Þ
Kvp C2
K05 ¼ K5 þ Kv svcl
: ; K06 ¼ K6 þ Kv svcl
: ;
1 þ Cxða 0Þ
svcl
Kvp C3 1 þ Cxða 0Þ
svcl
Kvp C3
:
Kv Kv Cxða 0Þ
Kvz ¼ 1
; K0v ¼ svcl
1
: :
Cða Þ
1 þ x 0 Kvp C3
Cða Þ
1 þ x 0 Kvp C3
svcl svcl
By writing Eqs. (3.71) and (3.72) together in matrix form, the state equation of
the system can be obtained to be
X_ ¼ AX þ bDusvcs ð3:73Þ
where
3 2
2 0 3
Dd 6 K 70
6 7 6 p 7
6 Dx 7 6 M 7
6 7 6 K0q 7
6 DE0 7 6 7
X¼6 q 7; b ¼ 6 0 7;
6 7 6 T do 7
6 DE0fd 7 6 7
4 5 6 KA K0v 7
4 T 5
Dzsvc A
CU
2 3
0 xo 0 0 0
6 K 0
K 0
K 7
6 1
MD M 2
0 Mpz 7
6 M 7
6 0 0 7
6 K K Kqz 7
A ¼ 6 T0 4 0 03 1
T0do
0 7
6 Tdo Tdo 7
6 7
do
6 KA K05 K K0
TA 6 T1 K
TA Kvz 7
4 TA 0
A A A
5
CA 0 CE 0 CZ
Linearization Eq. (3.63) is

Dusvcs ¼ Ks ðDPt Þ ¼ Tsvcs ðsÞðDPt Þ
ð1 þ sT1 Þð1 þ sT3 Þ ð3:74Þ
¼ Tsvcs ðsÞðK01 Dd þ K02 DE0q þ Kpz Dz þ K0p Dusvcs Þ
Hence, the output equation of the system is as follows:
y ¼ cX þ dDusvcs ð3:75Þ
where

y ¼ DPt ; c ¼ K01 0 K02 0 Kpz ; d ¼ K0p
3.3.1.2 Linearized Model of the Example Power System
At the given steady-state operating condition, the voltage at the busbar where the
SVC is installed is Vs0 ¼ 1:0 p:u: This is achieved by an initial compensation from
the SVC at the operating condition, bsvc0 . This initial compensation can be calcu-
lated as follows as introduced in Sect. 3.1.1.4.
From Eqs. (3.24) and (3.26), the reactive power injection at the busbar with the
SVC installed is as follows
2 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 3
2ffi
Vs0 4 x
Psb0 5 ¼ 0:0377
sb
Qsb0 ¼ Vs0 V2b
xsb Vs0
2sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 3
2
Vs0 4 x
Pts0 Vs0 5 ¼ 0:0377
ts
Qts0 ¼ V2t0
xts Vs0
Hence, the initial compensation is calculated to be
Qsb0 Qts0
bsvc0 ¼ ¼ 0:0754
V2s0
This gives csvc0 ¼ 1 xsb bsvc0 ¼ 0:9774.

bsvc0 ¼ ½1 C(a0 Þ=Xsvcl 1=Xsvcc

Thus, it can be obtained that at the steady-state operating condition,
2a0 sin 2a0

C(a0 Þ ¼ 1 ¼ 0:0754
2p
Hence, a0 ¼ 42:0727 .
From the circuit model of the power system installed with the SVC in Fig. 3.5, it
can be calculated that
Is0 ¼ Vs0 jbSVC0 ¼ j0:0754; Its0 ¼ ðPt0 jQts0 Þ ¼ 0:5 þ j0:0377

Vs0
Isb0 ¼ Its0 Is0 ¼ 0:5 þ j0:0377; Vb0 ¼ Vs0 jIsb0 xsb ¼ 0:9887 j0:15

Psb0 þ jQb0 ¼ Isb0 Vb0 ¼ 0:4887 j0:1123
With the reference voltage, recalculation of the value of various variables is as

follows
Isb0 ¼ ðPt0 jQb0 Þ ¼ 0:5 þ j0:0377; V s0 ¼ V b0 þ jXsbIsb0 ¼ 0:9887 þ j0:15

b0
V
Is0 ¼ V
s0 jbsvc0 ¼ 0:0113 þ j0:0746; Its0 ¼ Isb0 þ Is0 ¼ 0:4887 þ j0:1123
t0 ¼ V
V s0 þ jIts0 xts ¼ 0:955 þ j0:2966; E Q0 ¼ V t0 þ jxqIts0 ¼ 0:8652 þ j0:6876
d0 ¼ 38:4741

Id0 ¼ ðsin d0 þ j cos d0 ÞIts0 ¼ 0:2161; Eq0 ¼ EQ0 þ xd xq Id0 ¼ 1:1483

E0q0 ¼ EQ0 þ x0d xq Id0 ¼ 0:9646; Isbd0 ¼ ðsin d0 þ j cos d0 ÞIsb0 ¼ 0:2816
Isbq0 ¼ ðcos d0 j sin d0 ÞIsb0 ¼ 0:4149
From Eq. (3.7) it can have

xts
xlR ¼ xts þ ¼ 0:6069; x0dR ¼ x0d þ xlR ¼ 0:7569;
cSVC0
xdR ¼ xd þ xlR ¼ 1:6069; xqR ¼ xq þ xlR ¼ 1:4069
Thus, from Eqs. (3.9) and (3.12), parameters of the Heffron–Phillips model of
the example power system can be calculated to be
K1 ¼ 0:8765; K2 ¼ 0:8410; K3 ¼ 2:1229; K4 ¼ 0:7148; K5 ¼ 0:0473;

K6 ¼ 0:7475; Kp ¼ 0:0166; Kq ¼ 0:2990; Kv ¼ 0:0806
CA ¼ 0:0358; CZ ¼ 0:4259; CE ¼ 0:7615; CU ¼ 0:1218;

K01 ¼ 0:8766; K02 ¼ 0:8394; Kpz ¼ 0:0175; K0p ¼ 0:0050; K03 ¼ 2:1514;
K04 ¼ 0:7134; Kqz ¼ 0:3158; K0q ¼ 0:0903; K05 ¼ 0:0476;
K06 ¼ 0:7398; Kvz ¼ 0:0852; K0v ¼ 0:0243:
State matrix of Eq. (3.73) thus is obtained to be

2 3
0 314:2 0 0 0
6 0:1461 0 0:1399 0 0:0029 7
6 7
A¼6
6 0:1415 0 0:4299 0:1983 0:0626 77
4 95:2539 0 1479:6 100:0 170:3097 5
0:0358 0 0:7614 0 0:4259
Eigenvalues of the state matrix are calculated to be
k1 ¼ 96:9595; k2;3 ¼ 0:1349 j6:7166

4 ¼ 3:2758; k
k 5 ¼ 0:3474
Hence, the electromechanical oscillation mode of the power system is

2;3 ¼ 0:1349 j6:7166.
k
3.3.2 Design of SVC-Based Stabilizer
3.3.2.1 SVC Stabilizer Designed by Using the Phase Compensation

Method
From Fig. 3.17 and Eq. (3.74), the forward path of the stabilizing signal of the
SVC-based stabilizer, D usvc-s, can be drawn as shown in Fig. 3.18. Transfer
function of the forward path of the SVC-based stabilizer can be obtained to be
8 9
< Kq þ Kv 1 +KAsT C2 Fsvc1 ðs)Kp þ K2 =
Fsvcs ðs) ¼ Fsvc2 ðs) Kp A
: K3 þ sT0d0 þ K6 1 þKsT
A
þ C2 Fsvc1 ðs) Kq þ Kv 1 þKsT
A ;
A A
ð3:76Þ
Hence, from Fig. 3.17, the electric torque contribution from the SVC-based
stabilizer can be obtained to be
DTet ¼ Fsvcs ðs)Tsvcs ðs)( DPt Þ ¼ Fsvcs ðs)Tsvcs ðs)(D þ sM)Dx ð3:77Þ
K1
−ΔPt Δω ω0
1
-
D + sM s
Tsvc −s (s)
K2
Δu svc −s
C2 Fsvc1 (s) + Fsvc2 (s)
Δbsvc
Kp Kq Kv
1
- KA
K 3 + sTd0 ' 1 + sTA
-
+
K6
Fig. 3.18 Forward path of the SVC-based stabilizer
2;3 ¼ 0:1349 j6:7166. By taking

Oscillation modes of the power system are k
xs ¼ 6:7166, from Eqs. (3.67) and (3.76), it can have
svc1 ðjxs Þ ¼ 0:2899 þ j0:3040;

F svc2 ðjxs Þ ¼ 0:2696 þ j0:0171;
F

Fsvcs ðjxs Þ ¼ 0:0912 j0:0114
Take the transfer function of the SVC stabilizer to be

Tsvcs ðs) ¼ Ks ; T1 ¼ 0:09 s:; T3 ¼ 0:09 s:
The SVC stabilizer is to be designed to provide a damping torque

Dsvc Dx ¼ 1:6Dx, i.e., Dsvc ¼ 1:6. Parameters of the SVC-based stabilizer can be
set by using the phase compensation method to provide a pure damping torque
DTet ¼ Fsvcs ðjxs ÞðD þ jxs M)Tsvcs ðjxs ÞDx ¼ Dsvc Dx ð3:78Þ
The results of computation by the use of Eq. (3.78) are as follows
Ks ¼ K1 K2 ¼ 5:8540; T2 ¼ 1:7369 s; T4 ¼ 1:7369 s
Let
1 þ sT4 1 þ sT2
Dx1 ¼ Dx; Dusvcs ¼ K Dx1
1 þ sT3 1 þ sT1
Thus,
h
T4 K01
sDx1 ¼ T1 Dx1 þ T1 M Dd DT M
4
1 Dx
3
0
3
0

T4 K2 0 T4 Kpz T4 K
M DEq M Dzsvc M p Dusvcs ð3:79Þ
sDusvc s ¼ Ks1ð1þþsT
sT2 Þ Dx
1
1
From Eqs. (3.73) and (3.79), state equation of the closed-loop system can be
obtained to be
2 : 3 2 3
Dd Dd
6 : 7
6 Dx 7 6 Dx 7
6 0 7 6 7
6 DE_ q 7 6 DE0q 7
6 7 6 7
6
0
DE_ fd 7 ¼ Ac 6 DE0fd 7 ð3:80Þ
6 7 6 7
6 D_zsvc 7 6 Dzsvc 7
6 7 6 7
4 Dx_ 1 5 4 Dx1 5
Du_ svcs Dusvcs
where
2 3
0 314:1593 0 0 0 0 0
6 0:1461 0 0:1399 0 0:0029 0 0:0008 7
6 7
6 0:1415 0 0:4265 0:1983 0:0626 0 0:0179 7
6 7
Ac ¼ 6
6 95:2539 0 1479:6 100 170:3097 0 48:6868 7
7
6 0:0358 0 0:7615 0 0:4259 0 0:1218 7
6 7
4 0:7364 524:6004 0:2133 0:3170 0:0005 1:0411 0:0205 5
7:4877 5333:9 2:1683 3:2233 0:0056 4:7312 0:7913
Fig. 3.19 Simulation of the δ /degree Without SVC stabilizer

example power system 42
without and with the SVC With SVC stabilizer
stabilizer installed 41
40
39
38
37
36
t/s
35
0 1 2 3 4 5 6 7 8 9 10
Eigenvalues of the state matrix of the closed-loop system are as follows
k1 ¼ 95:2487; 2;3 ¼ 0:6870 j6:6669;

k k4 ¼ 3:7484;
k5 ¼ 1:2281; k6 ¼ 0:2937; k7 ¼ 0:7921
The electromechanical oscillation mode of the power system is now k 2;3 ¼

0:6870 j6:6669 with sufficient damping.
Figure 3.19 gives the simulation results of the power system without and with
the SVC stabilizer installed. At 0.5 s of the simulation, a three-phase to-earth short
circuit occurred on the transmission line of the single-machine infinite-bus power
system in Fig. 3.2. The fault was cleared in 0.1 s. From Fig. 3.19, it can be seen that
the SVC stabilizer effectively suppresses the power oscillation.
3.3.2.2 Damping Control Effectiveness of the SVC Stabilizer

as Affected by Various Factors
Effectiveness of the SVC stabilizer installed is affected by various factors as

discussed in Sect. 3.1.2. Figure 3.20 shows the computational results of the
damping torque provided by the SVC stabilizer when the load condition of the
power system varies. From Fig. 3.20, it can be seen that
1. The direct damping torque provided by the SVC stabilizer is much greater than
the indirect damping torque;
2. When the active power supplied by the generator increases, more damping
torque is provided by the SVC stabilizer.
Table 3.1 shows the computational result of electromechanical oscillation mode
of the power system with the SVC stabilizer installed at three different load
Fig. 3.20 Computational 45 Damping torque provided

results of the damping torque by the SVC stabilizer (p.u.)
provided by the SVC 40
stabilizer installed 35 Total damping torque
30
25
20
15 Direct
damping torque
10
Indirect damping torque
5
0
Pt0 (p.u.)
-5
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Table 3.1 Electromechanical oscillation mode

Load condition Pt0 (p.u.) 0.3 0.5 0.7
Electromechanical 0:1732 j5:4744 0:6870 j6:6669 1:0831 j6:0391
oscillation mode
conditions. Figures 3.21 and 3.22 show the simulation results. All those results in
Figs. 3.20, 3.21, and 3.22 and in Table 3.1 confirm the analytical conclusion that
the effectiveness of the SVC stabilizer increases with the load condition of power
system.
Fig. 3.21 Simulation results δ /degree

Without the SVC
at Pt0 ¼ 0:3 27 stabilizer
With the SVC stabilizer
26
25
24
23
22
21
t/s
20
0 1 2 3 4 5 6 7 8 9 10
Fig. 3.22 Simulation results δ /degree Without the SVC stabilizer

at Pt0 ¼ 0:7 58 With the SVC stabilizer
56
54
52
50
48
t/s
46
0 1 2 3 4 5 6 7 8 9 10
Discussions in Sect. 3.1.2 conclude that the effectiveness of the SVC stabilizer is
also affected by the length of the transmission line and the installing locations of the
SVC stabilizer. To confirm the conclusion, Fig. 3.23 presents the computational
results of the damping torque provided by the SVC stabilizer at variable load
condition when the system operates with single and double transmission lines
connecting the generator and the infinite busbar. From Fig. 3.23, it can be seen that
when the electric length of the line is longer (single transmission line connecting the
Damping torque provided by Single line

180 the SVC stabilizer(p.u.)
160
140
120
100
Double
80 lines
60
40
20
Pt0 (p.u.)
0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Fig. 3.23 Computational results of the damping torque provided by the SVC stabilizer
Damping torque provided Direct damping

12 torque
by the SVC stabilizer
10 Total damping
torque
2
Indirect damping
torque
0
Xts+Xsb
-2
0.1 0.2 0.3 0.4 0.5 0.6 0.7
Fig. 3.24 Computational results of the damping torque provided by the SVC stabilizer installed at
various locations
generator and the infinite busbar), the SVC stabilizer provides more damping torque
and hence is more effective in suppressing the power oscillation.
Keeping the operating condition of the example power system unchanged at
Pt0 ¼ 0:5; Vb0 ¼ 1:0; Vt0 ¼ 1:0; Vs0 ¼ 1:0, the computational results of the
damping torque provided by the SVC stabilizer installed at various locations along
the transmission line are presented in Fig. 3.24. From Fig. 3.24, it can be seen that
the SVC stabilizer is most effective when it is installed around the middle point of
the line as it provides the most damping torque at the point.
References
1. Wang HF, Swift FJ (1996) Capability of the static VAr compensator in damping power system
oscillations. IEE Proc Part C 143(4):353–358
2. Smith OJM (1969) Power system transient control by capacitor switching. IEEE Trans Power
Apparatus Syst 88(1):28–35
3. Smith OJM, Webster RH (1971) Series capacitor switching to quench electromechanical
transients in power systems. IEEE Trans Power Apparatus Syst 90(2):427–433
4. Wang HF, Swift FJ (1996) Application of the Controllable Series Compensator in damping
power system oscillations. IEE Proc Part C 143(4):359–364
5. Wang HF, Swift FJ (1997) Analysis of thyristor-controlled phase shifter applied in damping
power system oscillations. Int J Electr Power Energ Syst 19(1):1–9
Chapter 4
Single-Machine Infinite-Bus Power
Systems Installed with VSC-Based
Stabilizers
4.1 Damping Torque Analysis of a Shunt VSC-Based

Stabilizer Installed in a Single-Machine Infinite-Bus
Power System

Infinite-Bus Power System Installed with a Shunt
VSC-Based Stabilizer
4.1.1.1 A Shunt VSC Installed in a Single-Machine Infinite-Bus Power

System
Figure 4.1 shows the configuration of a shunt voltage source converter (VSC)-based
unit connected to high-voltage transmission line through a step-down transformer.
xs is the equivalent reactance of the step-down transformer. The unit can be an
energy storage system (ESS), such as a battery energy storage system (BESS) or a
superconducting magnetic energy storage (SMES) device. It can also be a renew-
able power generation plant, such as a grid-connected photovoltaic (PV) or fuel cell
(FC) power plant. In case that Idc2 ¼ 0, it is a static synchronous compensator
(STATCOM) [1].
There are two commonly used algorithms to control the voltage at the AC
terminal of the converter Vc , the pulse width modulation (PWM), and the pulse
amplitude modulation (PAM). When the PWM is used, Vdc is kept constant by
controlling the modulation phase, ϕ (angle between Vc and Vs ), to charge or dis-
charge the capacitor. The amplitude of voltage at the AC terminal is regulated by

DOI 10.1007/978-1-4899-7696-3_4
122 4 Single-Machine Infinite-Bus Power Systems …
Fig. 4.1 Configuration of a transmission line Vs

shunt VSC-based unit
Is
xs
Vc
φ
m
Idc1
Idc
Cdc Idc2
Vdc
the modulation ratio m, i.e. Vc ¼ mkVdc , where k is the ratio between the AC and
DC voltages of the converter (including the ratio of the step-down transformer),
dependent of the structure of the converter circuit.
Figure 4.2 shows the configuration of a single-machine infinite-bus power
system, where a shunt VSC-based unit is installed at a location along the trans-
mission line. From Fig. 4.2, it can be obtained that
Vt ¼ jxts Its þ Vs
Vs ¼ jxs Is þ Vc ð4:1Þ
Vs Vb ¼ jxsb ðIts Is Þ
Fig. 4.2 A single-machine Vt Vs Vb

infinite-bus power system x ts x sb
installed with a VSC-based
unit
Its Isb
Is
xs
Vc
VSC-based unit
4.1 Damping Torque Analysis of a Shunt VSC-Based … 123
Hence
jxs Is þ Vc Vb ¼ jxsb ðIts Is Þ

ð4:2Þ
Vt ¼ jxts Its þ jxsb ðIts Is Þ þ Vb
In the d–q coordinate of generator, as shown in Fig. 4.3, Eq. (4.2) gives
jxs ðisd þ jisq Þ þ Vc ðcos w þ j sin wÞ Vb ðsin d þ j cos dÞ

¼ jxsb [(itsd isd Þ þ j(itsq isq Þ
ð4:3Þ
vtd þ jvtq
¼ jðxts þ xsb Þðitsd þ jitsq Þ jxsb ðisd þ jisq Þ þ Vb ðsin d þ j cos dÞ
By equating the real and imaginary parts on both sides of Eq. (4.3), it can have
xs isq þ Vc cos w Vb sin d ¼ xsb ðitsq isq Þ

vtd ¼ ðxts þ xsb Þitsq þ xsb isq þ Vb sin d
ð4:4Þ
xs isd þ Vc sin w Vb cos d ¼ xsb ðitsd isd Þ
vtq ¼ ðxts þ xsb Þitsd xsb isd þ Vb cos d
Because
vtd ¼ xq itsq
ð4:5Þ
vtq ¼ E0q x0d itsd
Fig. 4.3 Phasor diagram of y q

system shown in Fig. 4.2
Vt
jxts Its
jxs Is
Vs
δ
φ
Vc jxsb Isb x
Vb
ψ
d

xsb xs xsb itsq Vc cos w þ Vb sin d
¼
xq þ xts þ xsb xsb isq Vb sin d
" # ð4:6Þ
xsb xs xsb Itsd Vc sin w Vb cos d
¼
x0d þ xts þ xsb xsb Isd E0q Vb cos d

1
itsq xsb xs xsb Vc cos w þ Vb sin d
¼
isq xq þ xts þ xsb xsb Vb sin d
1 " # ð4:7Þ
itsd xsb xs xsb Vc sin w Vb cos d
¼ 0
isd xd þ xts þ xsb xsb E0q Vb cos d
The active power supplied by the generator can be expressed as
Pt ¼ vtd itsd þ vtq itsq ¼ xq itsq itsd þ ðE0q x0q itsd Þitsq
ð4:8Þ
¼ E0q itsq þ ðxq x0q Þitsd itsq
Eq ¼ E0q þ ðxd x0q Þitsd

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð4:9Þ
Vt ¼ v2td þ v2tq ¼ ðxq itsq Þ2 þ ðE0q x0q itsd Þ2
Magnitude of Vc is mkVdc . In the d–q coordinate of generator as shown in Fig. 4.3,

its phase is w. Hence,
Vc ¼ mkVdc ðcos w þ j sin wÞ ¼ mkVdc \w ð4:10Þ
Active power received by the DC capacitor of VSC-based unit is
Vdc Idc1 ¼ isd vcd þ isq vcq ¼ isd mkVdc cos w þ isq mkVdc sin w ð4:11Þ

Idc ¼ isd mk cos w þ isq mk sin w ð4:12Þ
Dynamic equation on the DC side of the converter is
1 1 Idc2
V_ dc ¼ ðIdc1 þ Idc2 Þ ¼ ðisd mk cos w þ isq mk sin wÞ þ ð4:13Þ
Cdc Cdc Cdc
From the phasor diagram of Fig. 4.3, it can have

vsq
w ¼ tan1 / ð4:14Þ
vsq
Equations (4.7)–(4.9), (4.13), and (4.14), together with the differential equations of
synchronous generator of Eq. (2.37), form the mathematical model of the
single-machine infinite-bus power system with the shunt VSC-based unit installed.
In Eqs. (4.10) and (4.14), the modulation ratio and phase of PWM algorithm, m and
/, are regulated by two controllers. Hence, m and / are two control variables in the
mathematical model of the power system.
4.1.1.2 Extended Heffron–Phillips Model
Linearization of Eq. (4.7) is

Ditsq c11 c12
¼
Disq c21 c22

m0 k cos w0 DVdc kVdc0 cos w0 Dm þ m0 kVdc0 sin w0 Dw þ Vb cos d0 Dd

Vb cos d0 Dd

Ditsd d11 d12
¼
Disd d21 d22
" #
m0 k sin w0 DVdc þ kVdc0 sin w0 Dm þ m0 kVdc0 cos w0 Dw þ Vb sin d0 Dd

DE0q þ Vb sin d0 Dd
ð4:15Þ
The second equation in Eq. (4.1) is
vsd þ jvsq ¼ jxs ðisd þ jisq Þ þ mkVdc cos w þ jmkVdc sin w
Hence, it can have

vsd ¼ xs isq þ mkVdc cos w
vsq ¼ xs isd þ mkVdc sin w
By using Eq. (4.15), linearization of above equations can be obtained as
Dvsd ¼ ðxs þ 1Þm0 k cos w0 DVdc þ ðxs þ 1ÞkVdc0 cos w0 Dm

ðxs þ 1Þm0 kVdc0 sin w0 Dw xs Vb cos d0 Dd
¼ ad1 DVdc þ ad2 Dm þ ad3 Dw þ ad4 Dd
Dvsq ¼ ðxs þ 1Þm0 k sin w0 DVdc þ ðxs þ 1ÞkVdc0 sin w0 Dm
þ ðxs þ 1Þm0 kVdc0 cos w0 Dw þ Vb sin d0 D
¼ aq1 DVdc þ aq2 Dm þ aq3 Dw þ aq4 Dd
By using above equations, linearization of Eq. (4.14) can be obtained as
Dw ¼ a01 DVdc þ a02 Dm þ a03 Dw þ a04 Dd D/
where
1
a01 ¼ ðvsd0 a1q vsq0 a1d Þ
v2sd0 þ v2sq0
1
a02 ¼ 2 ðvsd0 a2q vsq0 a2d Þ
vsd0 þ v2sq0
1
vsd0 þ v2sq0
1
vsd0 þ v2sq0
That is
a01 a02 a04 1

Dw ¼ 0 DVdc þ 0 Dm þ Dd D/
1 a3 1 a3 1 a03 1 a03
¼ a1 DVdc þ a2 Dm þ a3 Dd þ a4 D/ ð4:16Þ
By using Eqs. (4.15) and (4.16), linearization of Eqs. (4.8) and (4.9) can be
obtained as
DPt ¼ K01 Dd þ K2 DE0q þ K0pdc DVdc þ K0pm Dm þ K0pw Dw

¼ ðK01 a3 K0pw ÞDd þ K2 DE0q þ ðK0pdc a1 K0pw ÞDVdc
þ ðK0pm a2 K0pw ÞDm þ a4 K0pw D/
¼ K1 Dd þ K2 DE0q þ Kpdc DVdc þ Kpm Dm þ Kpw D/
DEq ¼ K04 Dd þ K3 DE0q þ K0qdc DVdc þ K0qm Dm þ K0qw Dw
¼ ðK04 a3 K0qw ÞDd þ K3 DE0q þ ðK0qdc a1 K0qw ÞDVdc
ð4:17Þ
þ ðK0qm a2 K0qw ÞDm þ a4 K0qw D/
¼ K4 Dd þ K3 DE0q þ Kqdc DVdc þ Kqm Dm þ Kqw D/
DVt ¼ K05 Dd þ K6 DE0q þ K0vdc DVdc þ K0vm Dm þ K0vw Dw
¼ ðK5 a3 K0vw ÞDd þ K6 DE 0 q þ ðK0vdc a1 K0vw ÞDVdc
þ ðK0vm a2 K0vw ÞDm þ a4 K0vw D/
¼ K5 Dd þ K6 DE0q þ Kvdc DVdc þ Kvm Dm þ Kvw D/
where
C0 ¼ m0 k
h i
K01 ¼ E0q0 itsd0 x0d xq ðc11 þ c12 ÞVb cos d0

itsq0 x0d xq ðd11 þ d12 ÞVb sin d0

K2 ¼ itsq0 itsq0 x0d xq d12
h i
K0pdc ¼ c11 C0 cos w0 E0q0 itsd0 x0d xq d11 C0 sin w0 Itsq0 x0d xq
h i
K0pm ¼ c11 kVdc0 cos w0 E0q0 itsd0 x0d xq d11 kVdc0 sin w0 Itsq0 x0d xq
h i
K0pw ¼ c11 C0 Vdc0 sin w0 E0q0 itsd0 x0d xq d11 C0 Vdc0 cos w0 Itsq0 x0d xq
K4 ¼ ðd11 þ d12 ÞVb sin d0 ðxd x0d Þ
K03 ¼ d12 ðxd x0d Þ þ 1
K0qdc ¼ d11 C0 sin w0 ðxd x0d Þ
K0qm ¼ d11 kVdc0 sin w0 ðxd x0d Þ
K0qw ¼ d11 C0 Vdc0 cos w0 ðxd x0d Þ

0 vtd0 Xq ðc11 þ c12 ÞVb cos d0 vtq0 x0d ðd11 þ d12 ÞVb sin d0
K5 ¼
Vt0

vtq0 1 x0d d12
K6 ¼
Vt0

0 v td0 xq c11 C0 cos w0 vtq0 x0d d11 C0 sin w0
Kvdc ¼
Vt0

0 v td0 xq kc 11 V dc0 cos w0 vtq0 x0d d11 kVdc0 sin w0
Kvm ¼
Vt0

0 vtd0 xq c11 C0 Vdc0 sin w0 vtq0 x0d d11 C0 Vdc0 cos w0
Kvw ¼
Vt0
Cdc DV_ dc ¼ ðisd0 k cos w0 þ isq0 k sin w0 ÞDm

þ m0 kðisd0 sin w0 þ isq0 cos w0 ÞDw
þ m0 kðcos w0 Disd þ sin w0 Disq Þ þ DIdc2
By using Eqs. (4.15) and (4.16), it can have
1
DV_ dc ¼ K07 Dd þ K8 DE0q þ K09 DVdc þ K0dm Dm + K0dw Dw þ DIdc2
Cdc
¼ ðK07 a3 K0dw ÞDd þ K8 DE0q þ ðK09 a1 K0dw ÞDVdc
1
þ ðKdm a2 K0dw ÞDm + a4 K0dw Df + DIdc2
Cdc
1
¼ K7 Dd þ K8 DE0q þ K9 DVdc þ Kdm Dm + Kdw D/ þ DIdc2 ð4:18Þ
Cdc
where
½m0 k cos w0 ðd21 þ d22 ÞVb sin d0 þ m0 k sin w0 ðc21 þ c22 ÞVb cos d0
K07 ¼
Cdc
m0 k cos w0 d22
K8 ¼
C
2 2 dc
0 m k cos w0 d21 sin w0 m20 k2 sin w0 c21 cos w0
K9 ¼ 0
Cdc
1
K0dm ¼ ðkisd0 cos w0 þ kisq0 sin w0 þ m0 k2 cos w0 d21 Vdc0 sin w0
Cdc
m0 k2 sin w0 c21 Vdc0 cos w0 Þ
1
K0dw ¼ ½m0 kðisd0 sin w0 isq0 cos w0 Þ þ m20 k2 cos w0 d21 Vdc0 cos w0
Cdc
þ m20 k2 sin w0 c21 Vdc0 sin w0
By substituting Eq. (4.17) into the linearized differential equation of syn-

chronous generator of Eq. (2.37) (for the simplicity of expression, the PSS is not
considered), the extended Heffron–Phillips model of the single-machine infinite-bus
power system installed with the shunt VSC-based unit can be obtained as
Dd_ ¼ xo Dx
1
Dx_ ¼ ðK1 Dd DDx K2 DE0 q Kpdc DVdc Kpm Dm Kpw D/Þ
M
1
DE_ 0q ¼ 0 ðK4 Dd K3 DE0 q þ DE0fd þ Kqdc DVdc þ Kqm Dm þ Kqw D/Þ
Td0
1 KA
DE_ 0fd ¼ DE0fd ðK5 Dd þ K6 DE0 q þ Kvdc DVdc þ Kvm Dm þ Kvw D/Þ
TA TA
1
DV_ dc ¼ K7 Dd þ K8 DE0 q þ K9 DVdc þ Kdm Dm þ Kdw D/ þ DIdc2
Cdc
ð4:19Þ
The model is shown in Figs. 4.4 and 4.5.
ΔIdc2
1
Cdc
ΔE q' K8
1
+ ΔVdc
Δδ K7 s − K9
K dm K dψ
Δm Δφ
Fig. 4.4 Extended Heffron–Phillips model—part of the VSC-based unit
K 1 − K Pψ
1 Δω ω0 Δδ
- sM + D s
K 4 − K qψ K 5 − K vψ
K2 ΔVdc
[K pdc K pψ K pm ] [Δψ ] [K vdc K vψ K vm ]
Δm
[K qdc K qψ K qm ]
1 KA
ΔE q ' Td0 's + K 3 - + 1 + sTA -
K6
Fig. 4.5 Extended Heffron–Phillips model—part of the power system

4.1.2 Damping Torque Analysis of Shunt VSC-Based

Stabilizer Installed in Single-Machine Infinite-Bus
Power System
4.1.2.1 Damping Torque Contributed by the Shunt VSC-Based

Stabilizer [2, 3]
A damping control function can be superimposed on the modulation ratio m of the

PWM algorithm. This forms a VSC-based stabilizer which functions through reg-
ulating the exchange of reactive power between the VSC-based unit and rest of the
power system. In this book, it is named as the VSC-based reactive power stabilizer
for the convenience of discussion. If it is assumed that the feedback signal and
transfer function of the stabilizer are y and Tvsc ðsÞ, respectively, and only the
stabilizer’s function is considered, it can have Dm ¼ Tvsc ðsÞDy. At the angular
oscillation frequency of the power system, xs , stabilizing signal of the VSC-based
reactive power stabilizer Dm ¼ Tvsc ðjxs ÞDy can always be decomposed into a
synchronizing and damping torque, that is
Dm ¼ Tvsc ðjxs ÞDy ¼ Kvscd Dx þ Kvscs Dd ð4:20Þ
For the simplicity of discussion and presentation, it is assumed that the feedback
signal of the stabilizer is the rotor speed of generator and the stabilizer adopts a
proportional control law, it has
Dm ¼ Krs Dx ð4:21Þ
From Eq. (4.20), it can be seen that the above assumption will not change the
following discussion and conclusions about the effectiveness of stabilizer.
From Eq. (4.21), Figs. 4.4, and 4.5, the direct damping torque contribution from
the VSC-based reactive power stabilizer can be obtained as

K9
DTrsd ¼ Kpm Kpdc Kdm Krs Dx ð4:22Þ
x2s þ K29
If the shunt VSC-based unit (such as an energy storage system or a renewable

power generation plant) can exchange active power with the rest of power system, a
damping control function can also be superimposed on the modulation phase of
PWM algorithm, /. This VSC-based stabilizer functions by controlling the
exchange of active power between the VSC-based unit and rest of the power
system. In this book, it is named as the VSC-based active power stabilizer. Again, it
can be assumed that the feedback signal of stabilizer is the rotor speed of generator
and the stabilizer is a pure-gain controller. That is
D/ ¼ Kas Dx ð4:23Þ
From Eq. (4.23), Figs. 4.4, and 4.5, the direct damping torque contribution from the
VSC-based active power stabilizer can be obtained as

K9
DTasd ¼ Kpw Kpdc Kdw 2 Kas Dx ð4:24Þ
xs þ K29
In the following section, difference between the damping effectiveness of the

VSC-based reactive and active power stabilizers is examined using Eqs. (4.22) and
(4.24).
4.1.2.2 Difference of Damping Control Effectiveness Between

the VSC-Based Reactive and Active Power Stabilizers
jxsb xsb Vb
VS ¼ x Its þ x Vc þ ð4:25Þ
1 þ xs sb
xs ð1 þ xs Þ
sb
1 þ xxsbs
That gives
Vt ¼ jxts Its þ Vs

xs xsb xsb xs ð4:26Þ

¼ j xts þ Its þ Vc þ Vb ¼ jxIts þ Va
xs þ xsb xs þ xsb xs þ xsb
where

xs xsb
x¼ xts þ
xs þ xsb
xsb xs
Va ¼ Vc þ Vb ¼ aVc þ bVb
xs þ xsb xs þ xsb
Hence, from Eq. (4.26), the active power supplied by the generator can be
expressed as
E0q Va 0 V2a ðxq x0d Þ

Pt ¼ sin d sin 2d0 ð4:27Þ
x0dR 2 x0dR xqR
where d0 is the angle between E0q and Va and x0dR ¼ x þ x0d ; xqR ¼ x þ xq . From the
phasor diagram of Fig. 4.6, it can have
Fig. 4.6 Phasor diagram for q

Eq. (4.27)
E’
q
δ Va
δ aVc
bVb ψ
d
d
Va sin d0 ¼ bVb sin d þ aVc cos w

ð4:28Þ
Va cos d0 ¼ bVb cos d þ aVc sin w
Thus, from Eqs. (4.27) and (4.28), the active power supplied by the generator can
be expressed alternatively to be
E0q
Pt ¼ ðbVb sin d þ aVc cos wÞ
x0dR
ðxq x0d Þ
0 ðbVb sin d þ aVc cos wÞðbVb cos d þ aVc sin wÞ ð4:29Þ
xdR xqR
E0q
0 ðbVb sin d þ aVc cos wÞ
xdR
where Vc ¼ mkVdc . Equation (4.29) is the explicit expression of the active power
supplied by the generator in the single-machine infinite-bus power system installed
with the shunt VSC-based unit. By using Eqs. (4.16) and (4.17), from Eq. (4.29), it
can be obtained that

@Pt
Kpm ¼ a4 ¼ a4 akVdc0 cos w0
@m0

@Pt
Kpw ¼ a4 ¼ a4 akm0 Vdc0 sin w0 ð4:30Þ
@w 0

@Pt
Kpdc ¼ a4 ¼ a4 akm0 cos w0
@Vdc 0
From Eqs. (4.22) and (4.30), the damping torque contributed by the VSC-based
reactive power stabilizer is obtained as

K9
DTrsd ¼ a4 ak cos w0 Vdc0 m0 Kdm Krs Dx ð4:31Þ
x2s þ K29
From Eqs. (4.24) and (4.30), the damping torque contributed by the VSC-based
active power stabilizer is derived as

K9
DTasd ¼ a4 akm0 Vdc0 sin w0 þ cos w0 Kdw Kas Dx ð4:32Þ
x2s þ K29
From Fig. 4.3, it can be seen that d0 increases when more active power is supplied
by the generator. Hence, at a heavier load condition, w0 is smaller, and more
damping torque is provided by the VSC-based reactive power stabilizer as indicated
by Eq. (4.31). This means that the VSC-based reactive power stabilizer is more
effective in damping power oscillation when the single-machine infinite-bus power
system operates at the heavier load condition. However, with the increase of load
condition, cos w0 increases, but sin w0 decreases in Eq. (4.32). This means that with
variations of power system load conditions, the damping torque provided by the
VSC-based active power stabilizer changes less than that provided by the reactive
power stabilizer. Hence, the VSC-based active power stabilizer is more robust to the
variations of power system load conditions in damping power oscillations.
4.2 Damping Function of a Stabilizer Added on a Static

Synchronous Series Compensator (SSSC) Installed
in a Single-Machine Infinite-Bus Power System
4.2.1 Damping Torque Analysis of a SSSC Stabilizer

Installed in a Single-Machine Infinite-Bus Power
System
4.2.1.1 A Single-Machine Infinite-Bus Power System Installed

with a Static Synchronous Series Compensator
A static synchronous series compensator (SSSC) is a “series STATCOM”, a

solid-state voltage source inverter generating a controllable AC voltage source, and
connected in series to power transmission lines in a power system. The injected
voltage is in quadrature with the line current and emulates an inductive or capac-
itive reactance so as to influence the power flow along the transmission lines. With
the capability to change its reactance characteristic from capacitive to inductive, the
SSSC is very effective in controlling power flow in the power system. An auxiliary
stabilizing signal can be superimposed on the power flow control function of the
SSSC so as to improve the damping of power system oscillations.
Figure 4.7 shows a single-machine infinite-bus power system installed with a
SSSC, which consists of a series coupling transformer with a leakage reactance,
xsct , and a VSC connected to a DC capacitor. The exchange of reactive power
between the SSSC and rest of the power system is achieved by controlling the
magnitude and phase of the inserted voltage which is kept in quadrature with the
line current (the inverter losses are ignored). Phasor diagrams in Fig. 4.8 show the
basic operation principle of the SSSC, from which it can be seen that the com-
pensation level can be controlled dynamically by regulating the magnitude of Vinv .
Mathematical description of the SSSC is
Vt It xt Vb
Pt
+ V sssc -
I ts
V1 + V inv - V2
x sct
x ts x sb
Pts
m
VSC
φ
Idc Vdc
+ -
Fig. 4.7 A single-machine infinite-bus power system with a SSSC
V inv V inv
V1 V2
jx cst I ts jx cst I ts
V2 V1
I ts I ts
SSSC in inductive mode SSSC in capacitive mode
Fig. 4.8 Phasor diagram of operation of the SSSC

4.2 Damping Function of a Stabilizer Added on a Static … 135
Its ¼ itsd þ jitsq ¼ Its \u

Vinv ¼ mkVdc ðcos / þ j sin /Þ ¼ mkVdc \/; u þ / ¼ 90 ð4:33Þ
dVdc Idc mk
¼ ¼ ðitsd cos / þ itsq sin /Þ
dt Cdc Cdc
where k is the fixed ratio between the inverter’s AC and DC voltages and m is the
modulation ratio of the PWM algorithm implemented by the series VSC.
According to Fig. 4.8, a variable of equivalent reactance, xdc , can be introduced
to represent the AC voltage, Vinv , to be
Vinv ¼ jxdc Its ð4:34Þ
From the circuit equation of power system of Fig. 4.7, it can have
Vt Vb ¼ j(xts þ xsb þ xsct ÞIts þ Vinv ¼ jxt It ð4:35Þ
In the d–q coordinate, the above equation becomes
ðxts þ xsb þ xsct Þitsq þ j(xts þ xsb þ xsct Þitsd xdc itsq þ jxdc itsd
ð4:36Þ
¼ xt itq þ jxt itd
Hence,
xts þ xsb þ xsct þ xdc
itq ¼ itsq
xt
ð4:37Þ
itd ¼ itsd
xt
Considering the voltage equation of generator of Eq. (4.35), it can have
Vt j(xts þ xsb þ xsct ÞIts ¼ xq ðitsq þ itq Þ þ j½E0q x0d ðitsd þ itd Þ jðxts þ xsb þ xsct ÞIts

xts þ xsb þ xsct þ xdc xts þ xsb þ xsct þ xdc
¼ xq 1 þ itsq þ j E0q x0d 1 þ itsd
xt xt
þ ðxts þ xsb þ xsct Þitsq j(xts þ xsb þ xsct Þitsd
¼ Vinv þ Vb ¼ xdc itsq þ jxdc itsd þ Vb sin d þ jVb cos d ð4:38Þ
From the above equation, it can be obtained that
Vb sin d
itsq ¼
xts þ xsb þ xsct þ cs xq þ xdc
ð4:39Þ
E0q Vb cos d
itsd ¼
xts þ xsb þ xsct þ cs x0d þ xdc
where
cs ¼ 1 þ
xt
Thus,
Pt ¼ E0q ðitq þ itsq Þ þ ðxq x0d Þðitd þ itsd Þðitq þ itsq Þ

¼ E0q cs itsq þ ðxq x0d Þc2s itsd itsq
Eq ¼ E0q þ ðxd x0d Þðitd þ itsd Þ ¼ E0q þ ðxd x0d Þcs itsd
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð4:40Þ
Vt ¼ ½E0q x0d ðitd þ itsd Þ2 þ ½xq ðitq þ itsq Þ2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
¼ ðE0q x0d cs itsd Þ2 þ ðxq cs itsq Þ2
Vinv ¼ jxdc Its ¼ xdc itsq þ jxdc itsd ¼ mkVdc ðcos / þ j sin /Þ ð4:41Þ

xdc itsq
cos / ¼
mkVdc
ð4:42Þ
xdc itsd
sin / ¼
mkVdc
By using Eq. (4.42), dynamic equation in Eq. (4.33) is converted to the following
equivalent form
dVdc
¼0 ð4:43Þ
dt
Equation (4.43) shows that indeed, there is no exchange of active power between
the SSSC and rest of the power system.
4.2.1.2 Extended Heffron–Phillips Model of the Single-Machine

Infinite-Bus Power System Installed with the SSSC
and Damping Torque Analysis [4]
DPt ¼ K1 Dd þ K2 DE0q þ Kpx Dxdc

DEq ¼ K4 Dd þ K3 DE0q þ Kqx Dxdc ð4:44Þ
DVt ¼ K5 Dd þ K6 DE0q þ Kvx Dxdc
xdc Its ¼ mkVdc ð4:45Þ
Hence,
Its0 Dxdc þ xdc0 DIts ¼ kVdc0 Dm ð4:46Þ
Because only the damping control function will be considered, it can be assumed
that xdc0 ¼ 0 as far as the damping control function at the steady-state operation is
concerned. Hence, Eq. (4.44) becomes
DPt ¼ K1 Dd þ K2 DE0q þ Kpm Dm

DEq ¼ K4 Dd þ K3 DE0q þ Kqm Dm ð4:47Þ
DVt ¼ K5 Dd þ K6 DE0q þ Kvm Dm
where
Kpx kVdc0
Kpm ¼
Its0
Kqx kVdc0
Kqm ¼
Its0
Kvx kVdc0
Kvm ¼
Its0
By substituting Eq. (4.47) into Eq. (2.39) (without the PSS being considered), the
extended Heffron–Phillips model of the single-machine infinite-bus power system
with the SSSC installed can be obtained as
Dd_ ¼ xo Dx
1
Dx_ ¼ ½K1 Dd DDx K2 DE0q Kpm Dm
M
1 ð4:48Þ
DE_ 0q ¼ 0 ðK4 Dd K3 DE0q þ DE0fd Kqm DmÞ
Td0
1 KA
DE_ 0fd ¼ DE0fd ðK5 Dd þ K6 DE0q þ Kvm DmÞ
TA TA
The model is shown in Fig. 4.9.

Active power supplied by the generator can be expressed as
Pt ¼ E0q ðitsq þ itq Þ þ ðxq x0d Þðitsd þ itd Þðitsq þ itq Þ

ð4:49Þ
¼ E0q cs itsq þ ðxq x0d Þc2s itsd itsq
K1
1 ω0
- D + sM s
Δm
K2 K4 K5
K pm K qm K vm
1
- - KA
K 3 + sTd0 ' + 1 + sTA
-
K6
Fig. 4.9 Extended Heffron–Phillips model of single-machine infinite-bus power system installed
with a SSSC stabilizer
By using Eq. (4.39), it can have
cs E0q Vb c2s V2b ðxq x0d Þ

Pt ¼ sin d sin 2d ð4:50Þ
x0dR 2 x0dR xqR
where
x0dR ¼ cs x0d þ xts þ xsb þ xsct þ xdc

xqR ¼ cs xq þ xts þ xsb þ xsct þ xdc
From Eq. (4.50), it can be defined that
cs E0q0 Vb c2 V2 ðxq x0d Þ

Pt0 ¼ 0 sin d0 s b 0 sin 2d0 ¼ Pt10 Pt20
xdR 2 xdR xqR
cs E0q0 Vb
Pt10 ¼ sin d0 ð4:51Þ
x0dR
c2s V2b ðxq x0d Þ
Pt20 ¼ sin 2d0
2 x0dR xqR
According to Eqs. (4.41) and (4.51), from Eq. (4.50), it can have
0 0
xd xq xd
@Pt 1 xt þ 1 2 xt þ 1 xt þ 1
Kpx ¼ ¼ Pt10 Pt10 Pt20 þ Pt20 þ Pt20
@xdc 0 xt cs x0dR xt c s xqR x0
0 dR
xd xq
1 1 xL þ 1 xL þ 1
¼ ðPt10 Pt20 Þ Pt20 ðPt10 Pt20 Þ þ Pt20
xt cs xt c s x0dR xqR

1 x0 þ xt 1 xq þ xt
¼ d 0 Pt0 Pt20
xt c s xt xdR xt c s xt xqR
x0 cs x0d cs xt xqR cs xq xt cs
¼ dR Pt0 Pt20
xt cs x0dR xt cs xqR
1 1
¼ Pt0 þ Pt20
cs x0dR cs xqR
ð4:52Þ

kVdc0 1 1
Kpm ¼ Pt0 Pt20 ð4:53Þ
cs Its0 x0dR xqR
Taking the similar procedure of analysis to that from Eqs. (3.58) to (3.59), from Eq.
(4.53), it can be concluded that
1. When the load conditions change, the SSSC stabilizer can always provide
positive damping to power system oscillations if it is designed properly;
2. The heavier the load condition of the power system is, the more effective the
SSSC stabilizer is;
3. When the load condition is unchanged, the SSSC stabilizer is less effective when
the equivalent reactance of transmission line is bigger.
4.2.2 Design of a SSSC Stabilizer
Using a SSSC stabilizer as an example, this section demonstrates how a

thyristor-based or VSC-based stabilizer can be designed effectively to suppress
power system oscillations in a single-machine infinite-bus power system using the
phase compensation method.
From Fig. 4.9, the forward path of the SSSC stabilizer is shown in Fig. 4.10.
From Fig. 4.10, the transfer function of the forward path can be obtained as
Δm
K2
K pm K qm K vm
1
- KA
K 3 + sTd0 ' + 1 + sTA
-
K6
Fig. 4.10 Forward path of stabilizing signal of SSSC stabilizer installed in a single-machine
infinite-bus power system
1
K3 þ sT0d0
Fsssc ðsÞ ¼ Kpm K2 Kqm
1 þ K61 KA
K3 þ sT0d0 1 þ sTA
1 KA
K3 þ sT0d0 1 þ sTA ð4:54Þ
K2 Kvm
1 þ K6 1 KA
K3 þ sT0d0 1 þ sTA
K2 ½Kqm ð1 þ sTA Þ þ Kvm KA
¼ Kpm
ðK3 þ sT0d0 Þð1 þ sTA Þ þ K6 KA
Consider the case that the feedback signal and transfer function of the SSSC
stabilizer are Dx and Tsssc ðs), respectively, that is
Dm ¼ Tsssc ðsÞDx ð4:55Þ
If the amount of damping torque contribution required from the SSSC stabilizer is
Dsssc Dx, the transfer function of SSSC stabilizer should be set by using the phase
compensation method at the angular oscillation frequency, xs , to satisfy
Fsssc ðjxs ÞTsssc ðjxs Þ ¼ Dsssc ð4:56Þ
In a more general case, the state-space representation of linearized model of the
single-machine infinite-bus power system with the SSSC stabilizer installed can
always be arranged in the following form:
2 3 2 3 2
32 3
Dd_ 0 x0 0 Dd 0
6 7 6 76 7 6 7
4 Dx_ 5 ¼ 4 k d A23 54 Dx 5 þ 4 B2 5Dm
x_ A31 A32 A33 x B3
2 3 ð4:57Þ
Dd
6 7
Dy ¼ ½ C1 C2 C3 4 Dx 5 þ dm Dm
x
Equation (4.57) can be written as
Dd_ ¼ x0 Dx
Dx_ ¼ kDd dDx A23 x B2 Dm
ð4:58Þ
x_ ¼ A31 Dd þ A32 Dx þ A33 x þ B3 Dm
Dy ¼ C1 Dd þ C2 Dx þ CT3 x þ dm Dm
sDd ¼ x0 Dx
sDx ¼ kDd dDx A23 x B2 Dm
ð4:59Þ
sx ¼ A31 Dd þ A32 Dx þ A33 x þ B3 Dm
Dy ¼ C1 Dd þ C2 Dx þ CT3 x þ dm Dm
Hence,

s
x ¼ ðsI A33 Þ1 A31 þ A32 Dd þ ðsI A33 Þ1 B3 Dm ð4:60Þ
x0
sDd ¼ x0 Dx

s s
sDx ¼ k þ d þ A23 ðsI A33 Þ1 A31 þ A32 Dd
x0 x0
½A23 ðsI A33 Þ1 B3 þ B2 Dm ¼ KðsÞDd Kc ðsÞDm

s
Dy ¼ C1 Dd þ C2 Dx þ CT3 ðsI A33 Þ1 A31 þ A32 Dd þ ðsI A33 Þ1 B3 Dm
x0
h x i
1 x0
A31 þ A32 Dx þ ½dm þ CT3 ðsI A33 Þ1 B3 Dm
0
¼ C1 þ C2 þ C3 ðsI A33 Þ
T
s s
¼ Ko ðsÞDx þ Kil ðsÞDm
ð4:61Þ
where

s s
KðsÞ ¼ k þ d þ AT23 ðsI A33 Þ1 A31 þ A32
x0 x0
x
0 1 x0
Ko ðsÞ ¼ C1 þ C2 þ C3 ðsI A33 Þ
T
A31 þ A32
s s
Kc ðsÞ ¼ AT23 ðsI A33 Þ1 B3 þ B2
Kil ðsÞ ¼ CT3 ðsI A33 Þ1 B3 þ dm
Hence, according to Eq. (4.61), the system is shown in Fig. 4.11.

Fig. 4.11 Closed-loop Δδ ω0

system installed with the K(s)
s
SSSC stabilizer
Electromechanical oscillation loop
-
1 Δω
s
-
K c (s) K o (s)
+
+
K il (s)
y
Δm
Tsssc (s)
From Fig. 4.11, it can be seen that with the SSSC stabilizer installed, the electric
torque provided by the SSSC stabilizer to the electromechanical oscillation loop of
generator is
Kc ðjxs ÞKo ðjxs ÞTsssc ðjxs Þ

DTet ¼ Dx ð4:62Þ
1 Kil ðjxs ÞTsssc ðjxs Þ
where xs is the angular frequency of power system oscillation. An ideal SSSC

stabilizer should contribute a pure positive damping torque to the electromechanical
oscillation loop, DTdt ¼ Dsssc Dx, that is
Kc ðjxs ÞKo ðjxs ÞTsssc ðjxs Þ

¼ Dsssc ð4:63Þ
1 Kil ðjxs ÞTsssc ðjxs Þ
Hence,
Dsssc ¼ ½Kc ðjxs ÞKo ðjxs Þ þ Dsssc Kil ðjxs ÞTsssc ðjxs Þ
ð4:64Þ
¼ Fsssc ðjxs ÞTsssc ðjxs Þ
According to Eq. (4.64), the phase compensation method can be used to design the
SSSC stabilizer by setting
Dsssc
h ¼ u; Tsssc ¼ ð4:65Þ
Fsssc
where Tsssc ðjxs Þ ¼ Tsssc \hFsssc ðjxs Þ ¼ Fsssc \u.

4.3 Damping Function of a Unified Power Flow Controller … 143
4.3 Damping Function of a Unified Power Flow Controller

(UPFC) Installed in a Single-Machine Infinite-Bus
Power System
4.3.1 Mathematical Model of a Single-Machine Infinite-Bus

Power System Installed with a UPFC
4.3.1.1 Dynamic Model of a UPFC
Figure 4.12 shows the configuration of a single-machine infinite-bus power system

installed with a unified power flow controller (UPFC). The UPFC consists of an
excitation transformer (ET), a boosting transformer (BT), two three-phase
GTO-based voltage source converters (VSCs), and a DC link capacitor. In
Fig. 4.12, me ; mb and de ; db are the amplitude modulation ratio and phase angle of
the pulse width modulation (PWM) implemented by the shunt and series VSC,
respectively[5].
In Fig. 4.12, the shunt and series synchronous voltage source (SVS) can be
denoted as
Ves ¼ me ke Vdc \wes

ð4:66Þ
Vbs ¼ mb kb Vdc \wbs
Vt Vb
Itb x tb
Pt Vbt
I te +
Vet I bs Vbs
+
x te x bt
x bs BT
Pte Ies VSC − E VSC − B

ET
x es
Iedc I bdc
+
Ves Cdc
m e δe UPFC m b δb
Fig. 4.12 A single-machine infinite-bus power system installed with a UPFC

y q y q
Vet Vbt
δ jx es Ies jx bs I bs
δ
δe δb
Ves Vbs
x x
ψ es ψ bs
d d
Fig. 4.13 Phase diagram of shunt and series SVS
where wes and wbs are the angle between d-axis of d–q coordinate of the synchronous
generator and voltage Ves and Vbs , respectively, as shown in Fig. 4.13, which gives
Vet ¼ jxes Ies þ Ves

ð4:67Þ
Vbt ¼ jxbs Ibs þ Vbs
The active power injected into the DC side of the UPFC from the shunt and series
transformer is the active power “consumed” by the shunt and series SVS, respec-

tively, Pes ¼ ReðVes Ies Þ and Pbs ¼ ReðVbs Ibs Þ, where ReðÞ and * denote the real
part and conjugate of a complex variable, respectively. From Eqs. (4.66) and (4.67),
it can be obtained that
Ves Vet me ke Vdc Vet
Pes ¼ sin de ¼ sin de
xes xe
ð4:68Þ
Vbs Vbt mb kb Vdc Vbt
Pbs ¼ sin db ¼ sin db
xbs xbs
The shunt and series active power injection should be equal to that “received” at the
DC capacitor from the shunt and series VSC, respectively, that is
me ke Vdc Vet
Pes ¼ sin de ¼ Iedc Vdc
xes
ð4:69Þ
mb kb Vdc Vbt
Pbs ¼ sin db ¼ Ibdc Vdc
xbs
me ke Vet
Iedc ¼ sin de
xes
ð4:70Þ
mb kb Vbt
Ibdc ¼ sin db
xbs
Alternatively, in the d–q coordinate as shown in Fig. 4.13,
Pes ¼ vesd iesd þ vesq iesq ¼ iesd me ke Vdc cos wes þ iesq me ke Vdc sin wes ¼ Iedc Vdc
Pbs ¼ vbsd ibsd þ vbsq ibsq ¼ ibsd mb kb Vdc cos wbs þ ibsq mb kb Vdc sin wbs ¼ Ibdc Vdc
ð4:71Þ
Iedc ¼ iesd me ke cos wes þ iesq me ke sin wes

ð4:72Þ
Ibdc ¼ ibsd mb kb cos wbs þ ibsq mb kb sin wbs
Hence, from Eq. (4.70), dynamic equation of the UPFC can be obtained as

Iedc þ Ibdc 1 me ke Vet mb kb Vbt

V_ dc ¼ ¼ sin de þ sin db ð4:73Þ
Cdc Cdc xes xbs
Alternatively, from Eq. (4.72), it can have
1
V_ dc ¼ ðiesd me ke cos wes þ iesq me ke sin wes þ ibsd mb kb cos wbs þ ibsq mb kb sin wbs Þ
Cdc
ð4:74Þ
where
vetq
wes ¼ tan1 de
vetd
vbtq ð4:75Þ
wbs ¼ tan1 db
vbtd
4.3.1.2 Nonlinear Model of the Single-Machine Infinite-Bus Power

System Installed with the UPFC
Vt ¼ jxtb Itb þ Vb
Vt ¼ jxte Ite þ Vet ð4:76Þ
Vet ¼ Vbt þ jxbt Ibs þ Vb
which can be expressed in the d–q coordinate as
vtd þ jvtq ¼ jxtb ðitbd þ jitbq Þ þ Vb sin d þ jVb cos d

¼ xtb itbq þ Vb sin d þ jxtb itbd þ jVb cos d
ð4:77Þ
vtd þ jvtq ¼ jxte ðiesd þ ibsd þ jiesq þ jibsq Þ þ vetd þ jvetq
vetd þ jvetq ¼ vbtd þ jvbtq þ jxbt ibsd xbt ibsq þ Vb sin d þ jVb cos d
From the first equation in Eq. (4.77), it can be obtained that
vtq Vb cos d
itbd ¼
xtb
ð4:78Þ
Vb sin d vtd
itbq ¼
xtb
Because
vtd ¼ xq ðiteq þ itbq Þ

ð4:79Þ
vtq ¼ E0q x0d ðited þ itbd Þ
xq xtb iteq þ xq Vb sin d

vtd ¼
xq þ xtb
0 ð4:80Þ
xtb Eq x0d xtb ited þ x0d Vb cos d
vtq ¼
x0d þ xtb
The second and third equations in Eq. (4.77) give
vtd þ jvtq ¼ jxte ðiesd þ ibsd þ jiesq þ jibsq Þ xes iesq þ vesd þ jxes iesd þ jvesq
xes iesq þ vesd þ jxes iesd þ jvesq
¼ xbs ibsq þ vbsd þ jxbs ibsd þ jvbsq þ jxbt ibsd xbt ibsq þ Vb sin d þ jVb cos d
That is
vtd ¼ xte ðiesq þ ibsq Þ xes iesq þ vesd

vtq ¼ xte ðiesd þ ibsd Þ þ xes iesd þ vesq
ð4:81Þ
xes iesq þ vesd ¼ xbs ibsq þ vbsd xbt ibsq þ Vb sin d
xes iesd þ vesq ¼ xbs ibsd þ vbsq þ xbt ibsd þ Vb cos d
½xq xtb þ xte ðxq þ xtb Þ þ xes ðxq þ xtb Þiesq

þ ½xq xtb þ xte ðxq þ xtb Þibsq ¼ ðxq þ xtb Þvesd xq Vb sin d
½xte ðx0d þ xtb Þ þ xes ðx0d þ xtb Þ þ x0d xte iesd
þ ½xte ðx0d þ xtb Þ þ x0d xtb ibsd ¼ xtb E0q þ x0d Vb cos d ðx0d þ xtb Þvesq
xes iesq þ ðxbs þ xbt Þibsq ¼ vbsd vesd þ Vb sin d
xes iesd ðxbs þ xbt Þibsd ¼ vbsq vesq þ Vb cos d
which in the matrix form is

1
iesq xq xtb þ ðxte þ xes Þðxq þ xtb Þ xq xtb þ xte ðxq þ xtb Þ
¼
ibsq xes xes þ xbt

ðxq þ xtb Þme ke Vdc cos wes xq Vb sin d

mb kb Vdc cos wbs me ke Vdc cos wes þ Vb sin d
ð4:82Þ
iesd ðxte þ xes Þðx0d þ xtb Þ þ x0d xte xte ðx0d þ xtb Þ þ x0d xtb 1
¼
ibsd xes ðxbs þ xbt Þ

xtb E0q þ x0d Vb cos d ðx0d þ xtb Þme ke Vdc sin wes

mb kb Vdc sin wbs me ke Vdc sin wes þ Vb cos d
For the single-machine infinite-bus power system with the UPFC installed, it can
have
ited ¼ iesd þ ibsd

iteq ¼ iesq þ ibsq
Pt ¼ vtq ðiteq þ itbq Þ þ vtd ðited þ itbd Þ ð4:83Þ
Eq ¼ E0q þ ðxd x0d Þðited þ itbd Þ
Vt ¼ v2td þ v2tq
Equations (4.74), (4.75), (4.78), (4.80), (4.82), and (4.83), together with the dif-
ferential equations of synchronous generator of Eq. (2.35), form the nonlinear
mathematical model of the single-machine infinite-bus power system with the
UPFC installed.
4.3.1.3 Linearized Model of the Single-Machine Infinite-Bus Power

System Installed with the UPFC

2 3 2 3
2 3 Dme 2 3 Dme
Dd Dd
Diesq 6 Dwes 7 Diesd 6 7
0 6 Dwes 7
¼ F0q 4 DE0q 5 þ G0q 6 7 04 0 5
4 Dmb 5; Dibsd ¼ Fd DEq þ Gd 4 Dmb 5
Dibsq
DVdc DVdc
Dwbs Dwbs
ð4:84Þ
where

1 xbs þ xbt xte ð2xq þ xtb Þ f q11 0 f q13
F0q ¼
xR1 xes xq xte þ ðxte þ xes Þðxq þ xtb Þ f q21 0 f q23

0 1 xbs xbt xte xtb f d11 xL f d13
Fd ¼
xR2 xes ðxte þ xes Þðx0d þ xtb Þ x0d xte f d21 0 f d23
" #
1 x bs þ x bt x te ð2x q þ x tb Þ g g 0 0
G0q ¼
q11 q12
xR1 xes xq xte þ ðxte þ xes Þðxq þ xtb Þ gq21 gq22 gq23 gq24

0 1 xbs xbt xte xtb gd11 gd12 0 0
Gd ¼
xR2 xes ðxte þ xes Þðx0d þ xtb Þ x0d xte gd21 gd22 gd23 gd24
xR1 ¼ ½xq xte þ ðxte þ xes Þðxq þ xtb Þðxbs þ xbt Þ þ x2es ð2xq þ xtb Þ
xR2 ¼ ½ðxte þ xes Þðx0d þ xtb Þ x0d xte ðxbs þ xbt Þ x2te xtb
f q11 ¼ ðxq þ xtb Þme0 ke Vdc0 sin wes0 xq Vb cos d0

f q13 ¼ ðxq þ xtb Þme0 ke cos wes0
f q21 ¼ mb0 kb Vdc0 sin wbs0 me0 ke Vdc0 sin wes0 þ Vb cos d0
f q23 ¼ mb0 kb cos wbs0 me0 ke cos wes0
f d11 ¼ ðx0d þ xtb Þme0 ke Vdc0 cos wes0 x0d Vb sin d0
f d13 ¼ ðx0d þ xtb Þme0 ke sin wes0
f d21 ¼ me0 ke Vdc cos wes0 mb0 kb Vdc0 cos wbs0 Vb sin d0
f d23 ¼ mb0 kb sin wbs0 me0 ke sin wes0
gq11 ¼ ðxq þ xtb Þke Vdc0 cos wes0
gd12 ¼ ðxq þ xtb Þme0 kE Vdc0 sin wes0
gq21 ¼ ke Vdc0 cos wes0
gq22 ¼ me0 ke Vdc0 sin wes0
gq23 ¼ kb Vdc0 cos wbs0
gq24 ¼ mb0 kb Vdc0 sin wbs0
gd11 ¼ ðx0d þ xtb Þke Vdc0 sin wes0
gd12 ¼ ðx0d þ xtb Þme0 ke Vdc0 cos wes0
gd21 ¼ ke Vdc0 sin we0
gd22 ¼ me0 ke Vdc cos wes0
gd23 ¼ kb Vdc0 sin wbs0
gd24 ¼ mb0 kb Vdc0 cos wbs0
vetd þ jvetq ¼ jxes ðiesd þ jiesq Þ þ me ke Vdc cos wes þ jme ke Vdc sin wes
vbtd þ jvbtq ¼ jxbs ðibsd þ jibsq Þ þ mb kb Vdc cos wbs þ jmb kb Vdc sin wbs
The above equations give
vesd ¼ xes iesq þ me ke Vdc cos wes

vesq ¼ xes iesd þ me ke Vdc sin wes
vbsd ¼ xbs ibsq þ mb kb Vdc cos wbs
vbsq ¼ xbs ibsd þ mb kb Vdc sin wbs
By using Eq. (4.84), linearization of the above equation can be obtained as
Dvesd ¼ xes f 0q11 Dd xE f 0q12 DE0q ðxes f 0q13 me0 ke cos wes0 ÞDVdc
ðxes g0q11 ke Vdc0 cos wes0 ÞDme ðxes g0q12 þ me0 ke Vdc0 sin wes0 ÞDwes
xes g0q13 Dmb xes g0q14 Dwbs
¼ aed1 Dd þ aed2 DE0q þ aed3 DVdc þ aed4 Dme þ aed5 Dwes þ aed6 Dmb þ aed7 Dwbs
Dvesq ¼ xes f 0d11 Dd þ xes f 0d12 DE0q þ ðxes f 0d13 þ me0 ke sin wes0 ÞDVdc
þ ðxes g0d11 þ ke Vdc0 sin wes0 ÞDme
þ ðxes g0d12 þ me0 ke Vdc0 cos wes0 ÞDwes þ xes g0d13 Dmb þ xes g0d14 Dwbs
¼ aeq1 Dd þ aeq2 DE0q þ aeq3 DVdc þ aeq4 Dme þ aeq5 Dwes þ aeq6 Dmb þ aeq7 Dwbs
Dvbsd ¼ xbs f 0q21 Dd f 0q22 DE0q ðxbs f 0q23 mb0 kb cos wbs0 ÞDVdc
ðxbs g0q23 kb Vdc0 cos wbs0 ÞDmb ðxbs g0q24 þ mb0 kb Vdc0 sin wbs0 ÞDwbs
xbs g0q21 Dme xbs g0q22 Dwes
¼ abd1 Dd þ abd2 DE0q þ abd3 DVdc þ abd4 Dme þ abd5 Dwes þ abd6 Dmb þ abd7 Dwbs
Dvbsq ¼ xbs f 0d21 Dd þ xbs f 0d22 DE0q þ ðxbs f 0d23 þ mb0 kb sin wbs0 ÞDVdc
þ ðxbs g0d23 þ kb Vdc0 sin wbs0 ÞDmb þ ðxbs g0d24 þ mb0 kb Vdc0 cos wbs0 ÞDwbs
þ xbs g0d21 Dme þ xbs g0d22 Dwes
¼ abq1 Dd þ abq2 DE0q þ abq3 DVdc þ abq4 Dme þ abq5 Dwes þ abq6 Dmb þ abq7 Dwbs
By using the above equations, linearization of Eq. (4.75) can be obtained as
Dwes ¼ a0e1 Dd þ a0e2 DE0q þ a0e3 DVdc þ a0e4 Dme

þ a0e5 Dwes þ a0e6 Dmb þ a0e7 Dwbs Dde
Dwbs ¼ a0b1 Dd þ a0b2 DE0q þ a0b3 DVdc þ a0b4 Dme
þ a0b5 Dwes þ a0b6 Dmb þ a0b7 Dwbs Ddb
where
1
a0ei ¼ ðvesd0 aeqi vesq0 aedi Þ
v2esd0 þ v2esq0
1
a0bi ¼ 2 ðvbsd0 abqi vbsq0 abdi Þ
vbsd0 þ v2bsq0
i ¼ 1; 2; 3; 4; 5; 6; 7
Rearrangement of above equations is

2 3
1 Dd
Dwes 1 a0e5 a0e7 a0e1 a0e2a0e3 6 7
¼ 4 DE0q 5
Dwbs a0b5 1 a0b7 a0b1 a0b2a0b3
DVdc
!
a0e4 a0e6 Dme 1 0 Dde
þ þ
a0b4 a0b6 Dmb 0 1 Ddb
2 3
Dd
6 0 7
Dme Dde
¼ AEB1 4 DEq 5 þ AEB2 þ AEB3 ð4:85Þ
Dmb Ddb
DVdc

3 2
2 3
Dme
Dd 6 Dd 7
Diesq 6 7 6 e7
¼ Fq 4 DE0q 5 þ Gq 6 7
Dibsq 4 Dmb 5
DVdc
Ddb
2 3 ð4:86Þ
2 3 Dme
Dd 6 Dd 7
Diesd 6 7 6 e7
¼ Fd 4 DE0q 5 þ Gd 6 7
Dibsd 4 Dmb 5
DVdc
Ddb

xq Vb sin d0 xq xtb xq xtb Diesq
Dvtd ¼ Dd þ
xq þ xtb xq þ xtb
xq þ xtb Dibsq
ð4:87Þ
0
xd Vb sin d0 xtb 0 x0d xtb x0d xtb Diesd
Dvtq ¼ Dd þ 0 DE þ 0
x0d þ xtb xd þ xtb q xd þ xtb x0d þ xtb Dibsd
Substituting Eq. (4.86) into (4.87) gives

2 3
2 3 Dme
Dd 6 Dd 7
6 7 6 e7
Dvtd ¼ Hd 4 DE0q 5 þ Ld 6 7
4 Dmb 5
DVdc
Ddb
2 3 ð4:88Þ
2 3 Dme
Dd 6 Dd 7
6 7 6 e7
Dvtq ¼ Hq 4 DE0q 5 þ Lq 6 7
4 Dmb 5
DVdc
Ddb
where

xq Vb sin d0 xq xtb xq xtb
Hd ¼ 0 0 þ F
xq þ xtb xq þ xtb xq þ xtb q

xq xtb xq xtb
Ld ¼ Gq
xq þ xtb xq þ xtb
0
xd Vb sin d0 xtb x0d xtb x0d xtb
Hq ¼ 0 þ 0 Fd
x0d þ xtb x0d þ xtb xd þ xtb x0d þ xtb

x0 xtb x0d xtb
Lq ¼ 0 d Gd
xd þ xtb x0d þ xtb

Vb sin d0 1
Ditbd ¼ Dd þ Dvtq
xtb xtb
ð4:89Þ
Vb cos d0 1
Ditbq ¼ Dd Dvtd
xtb xtb
Dited ¼ Diesd þ Dibsd

Diteq ¼ Diesq þ Dibsq
DPt ¼ ðiteq0 þ itbq0 ÞDvtq þ ðited0 þ itbd0 ÞDvtd
þ vtq0 ðDiteq þ Ditbq Þ þ vtd0 ðDited þ Ditbd Þ ð4:90Þ
DEq ¼ DE0q þ ðxd x0d ÞðDited þ Ditbd Þ
vtd0 vtq0
DVt ¼ Dvtd þ Dvtq
Vt0 Vt0
By using Eqs. (4.86), (4.88), and (4.89) and the first two equations in Eq. (4.90), the
last three equations in Eq. (4.90) become
DPt ¼ ðiteq0 þ itbq0 ÞDvtq þ ðited0 þ itbd0 ÞDvtd
þ vtq0 ðDiesq þ Dibsq þ Ditbq Þ þ vtd0 ðDiesd þ Dibsd þ Dibsd Þ

Vb cos d0 Vb sin d0
¼ vtq0 þ vtd0 Dd
xtb xtb

vtq0 vtd0
þ ited0 þ itbd0 Dvtd þ iteq0 þ itbq0 þ Dvtq
xtb xtb

Diesq Diesd
þ ½ vtq0 vtq0 þ ½ vtd0 vtd0
Diesq Diesd
2 3
2 3 Dme
Dd 6 Dd 7
6 7 6 e7
¼ ½ K1 K2 Kpdc 4 DE0q 5 þ ½ Kpme Kpde Kpmb Kpdb 6 7 ð4:91Þ
4 Dmb 5
DVdc
Ddb

Diesd
DEq ¼ DE0q
þ ½ ðxd x0d Þ
ðxd x0d Þ
Dibsd

0 Vb sin d 0 1
þ ðxd xd Þ Dd þ Dvtq
xtb xtb
Vb sin d0 1
¼ ðxd x0d Þ Dd þ DE0q þ ðxd x0d Þ Dvtq
xtb xtb

0 0
Diesd
þ ½ ðxd xd Þ ðxd xd Þ
Dibsd
2 3
2 3 Dme
Dd 6 Dd 7
6 7 6 e7
¼ ½ K4 K3 Kqdc 4 DE0q 5 þ ½ Kqme Kqde Kqme Kqde 6 7
4 Dmb 5
DVdc
Ddb
ð4:92Þ
2 3
2 3 Dme
Dd 6 Dde 7
DVt ¼ ½ K5 K6 Kvdc 4 DE0q 5 þ ½ Kvme Kvde Kvmb Kvdb 6 7
4 Dmb 5 ð4:93Þ
DVdc
Ddb
where
h i
½ K1 K2 Kpdc ¼ vtq0 Vbxcos d2 þ vted0 Vbxsin d0 0 0

tb tb
vteq0 vted0
þ ited0 þ itbd0 Hd þ iteq0 þ itbq0 þ Hq
xtb xtb
þ ½ vtq0 vtq0 Fq þ ½ vtd0 vtd0 Fd
vteq0
½ Kpme Kpde Kpmb Kpdb ¼ ðited0 þ itbd0 ÞLd
xtb
vted0
þ ðiteq0 þ itbq0 þ ÞLq þ ½ vtq0 vtq0 Gq þ ½ vtd0 vtd0 Gd
xtb
h i
½ K4 K3 Kqdc ¼ ðxd x0d Þ Vbxsin d0 1 0
tb
ðxd x0d Þ
þ Hq þ ½ ðxd x0d Þ ðxd x0d Þ Fd
xtb
ðxd x0d Þ
½ Kqme Kqde Kqmb Kqdb ¼ Lq þ ½ ðxd x0d Þ ðxd x0d Þ Gd
xtb
vtd0 vtq0
½ K5 K6 Kvdc ¼ Hd þ Hq
Vt0 E
vtd0 vtq0
½ Kqme Kqde Kqmb Kqdb ¼ Ld þ Lq
Vt0 Vt0

Diesd
Cdc DV_ dc ¼ ½ me0 ke cos wes0 mb0 kb cos wbs0
Dibsd

Diesq
þ ½ me0 ke sin wes0 mb0 kb sin wbs0
Dibsq
þ ke iesd0 cos wes0 Dme þ ke iesq0 sin wes0 Dme þ kb ibsd0 cos wbs0 Dmb
þ kb ibsq0 sin wbs0 Dmb me0 ke iesd0 sin wes0 ðDd þ Dde Þ
þ me0 ke iesq0 cos wes0 ðDd þ Dde Þ mb0 kb ibsd0 sin wbs0 ðDd þ Ddb Þ
þ mb0 kb ibsq0 cos wbs0 ðDd þ Ddb Þ
¼ ðme0 ke iesd0 sin wes0 me0 ke iesq0 cos wes0 þ mb0 kb ibsd0 sin wbs0

Diesd
mb0 kb ibsq0 cos wbs0 ÞDd þ ½ me0 ke cos wes0 mb0 kb cos wbs0
DiBbsd

Diesq
þ ½ me0 ke sin wes0 mb0 kb sin wbs0
Dibsq
þ ðke iesd0 cos wes0 þ ke iesq0 sin wes0 ÞDme
þ ðkb ibsd0 cos wbs0 þ kb ibsq0 sin wbs0 ÞDmb
þ ðme0 ke iesq0 cos wes0 me0 ke iesd0 sin wes0 ÞDde
þ ðmb0 kb ibsq0 cos wbs0 mb0 kb ibsd0 sin wbs0 Ddb ÞDdb
2 3
2 3 Dme
Dd 6 Dd 7
6 7 6 e7
¼ ½ K7 K8 K9 4 DE0q 5 þ ½ Kdme Kdde Kdmb Kddb 6 7
4 Dmb 5
DVdc
Ddb
ð4:94Þ
where
½ K7 K8 K9
2 3T
me0 ke iesd0 sin wes0 me0 ke iesq0 cos wes0 þ mb0 kb ibsd0 sin wbs0 mb0 kb ibsq0 cos wbs0
6 7
¼4 0 5
0
þ ½ me0 ke cos wes0 mb0 kb cos wB0 Fd þ ½ me0 ke sin wE0 mb0 kb sin wB0 Fq
2 3T
ke iesd0 cos wes0 þ ke iesq0 sin wes0
6 me0 ke iesq0 cos w me0 ke iesd0 sin w 7
6 es0 es0 7
½ Kdme Kdde Kdmb Kddb ¼ 6 7
4 kb ibsd0 cos wbs0 þ kb ibsq0 sin wbs0 5
mb0 kb ibsq0 cos wbs0 mb0 kb ibsd0 sin wbs0
þ ½ me0 ke cos wes0 mb0 kb cos wbs0 Gd þ ½ me0 ke sin wes0 mb0 kb sin wbs0 Gq
By substituting Eqs. (4.91)–(4.93) into Eq. (2.37) (without the PSS control) and
arranging the equations obtained together with Eq. (4.94) in matrix form, the
extended Heffron–Phillips model of the single-machine infinite-bus power system
with the UPFC installed can be obtained as
2 3
2 3 0 xo 0 0 0 2 3
Dd_ 6
6 K1 D K2 Kpdc 7 7 Dd
6 _ 7 6 M M M 0 M 76 Dx 7
6 Dx 7 6 Kqdc 7 6 7
6 7 6 K K 0 7 6 7
6 DE_ 0 7 ¼ 6 T0 0 1
0 7 6 0 7
4 3
6 q 7 6 d0 Td0 0
Td0 Td0 76 DE q 7
6 7 6 76 7
4 DE_ 0fd 5 6 KA K5 0 KTA K6 T1 KATKvdc 7 4 DE0fd 5
6 T 7
DV_ dc 4 A A A A
5 DVdc
K7 0 K8 0 K9
Cdc Cdc Cdc ð4:95Þ
2 3
0 0 0 0
6 Kpme K
Mpde
Kpmb
M Mpdb 7
K 2 3
6 M 7 Dme
6 7
6 K K K K 76 Dd 7
6 qme qde qmB qdB 76 e 7
6 T0d0 T0d0 T0d0 T0d0 76 7
6 74 5
6 KA Kvme KA Kvde KA Kvmb KA Kvdb 7 Dmb
6 7
TA 5 Ddb
4 TA TA TA
Kdme Kdde Kdmb Kddb
Cdc Cdc Cdc Cdc
The linearized model is shown in Figs. 4.14 and 4.15, where
kp ¼ ½ Kpdc Kpme Kpde Kpmb Kpdb

kq ¼ ½ Kqdc Kqme Kqde Kqmb Kqdb
kv ¼ ½ Kvdc Kvme Kvde Kvmb Kvdb
ð4:96Þ
kd ¼ ½ Kdme Kdde Kdmb Kddb
uupfc1 ¼ ½ Dme Dde Dmb Ddb T
T
uupfc ¼ DVdc uTupfc1
4.3.2 Design of a UPFC Stabilizer Installed

in a Single-Machine Infinite-Bus Power System
4.3.2.1 Selection of Modulation Signal to Add the Damping Control

Signal of the UPFC Stabilizer by Using the Controllability
and Observability Index [6, 7]
Damping control signal of a UPFC stabilizer can be superimposed on a modulation

signal of the PWM control, Dme , Dde , Dmb , or Ddb . Selection of the modulation
signal to add the damping control signal ought to ensure the most effectiveness of
K1
1 Δω ω0 Δδ
- sM + D s
K4 K5
K2
kp uupfc kv
kq
1 KA
ΔE q ' Td0 's + K 3 - + 1 + sTA -
K6
Fig. 4.14 Extended Heffron–Phillips model of power system with a UPFC installed—part of the
power system
Fig. 4.15 Extended Heffron– Δδ

Phillips model of power
system with a UPFC installed
—part of the UPFC K7
1
ΔE q ' K8 + ΔVdc
sCdc − K 9
kd
uupfc1
uupfc1
UPFC stabilizer. It can be carried out by use of the modal controllability and
observability index introduced in Sect. 2.2.1.2.
Linearized model of the single-machine infinite-bus power system installed with
the UPFC of Eq. (4.95) can be written as
X
4
X_ ¼ AX þ bk Duk ð4:97Þ
k¼1
where
2 3
2 3 0 xo 0 0 0
Dd 6 K 7
6 1K MD K 2
0 Mpdc 7
6 7 6 M M 7
6 Dx 7 6 Kqdc 7
6 7 6 K0 4 K0 3 1
0 7
X¼6 0 7
6 DEq 7; A ¼ 6
6 Td0
0
Td0 T0d0 Td0 7 7;
6 0 7 6 K K 7
4 DEfd 5 6 A 5 0 TA K6
K T T 7
1 K A Kvdc 7
6 TA
DVdc 4 A A A
5
K7 0 K8 0 K9
Cdc Cdc Cdc
2 3 2 3 2 3 2 3
0 0 0 0
6 Kpme 7 6 Kpde 7 6 Kpmb 7 6 Kpdb 7
6 M 7 6 M 7 6 M 7 6 M 7
6 7 6 7 6 7 6 7
6 Kqme 7 6 Kqde 7 6 Kqmb 7 6 Kqdb 7
6 0 7 6 0 7 6 0 7 6 0 7
b1 ¼ 6 Td0 7; b2 ¼ 6 Td0 7; b3 ¼6 Td0 7; b4 ¼ 6 Td0 7;
6 7 6 7 6 7 6 7
6 KA Kvme 7 6 KA Kvde 7 6 KA Kvmb 7 6 KA Kvdb 7
6 TA 5 7 6 TA 7 6 TA 5 7 6 TA 7
4 4 5 4 4 5
Kdme Kdde Kdmb Kddb
Cdc Cdc Cdc Cdc
Du1 ¼ Dme ; Du2 ¼ Dde ; Du3 ¼ Dmb ; Du4 ¼ Ddb
If the transfer function and feedback signal of the UPFC stabilizer are Tupfck ðs) and
yk , respectively, state-space representation of the power system with the UPFC
stabilizer is
X
4
X_ ¼ AX þ bk Duk
k¼1
ð4:98Þ
yk ¼ ckT X
Duk ¼ Tupfck ðsÞyk
If ki is the electromechanical oscillation mode of the power system, from Eqs. (2.50
) and (2.54), it can have
Avi ¼ ki vi
ð4:99Þ
wTi A ¼ wTi ki
where vi and wTi are the right eigenvector and left eigenvector of state matrix, A,
with respect to mode ki , respectively. From the state transformation of Eqs. (2.53),
(2.55), (2.56), and (4.98), it can have
X
4 X
4
Z_ ¼ V1 AVZ þ V1 bk Duk ¼ K Z þ V1 bk Duk
k¼1 k¼1
ð4:100Þ
yk ¼ cTk VZ
where
2 3
k1 0 0 0
60 k2 0 07
6 7
V ¼ ½ v1 v2 . . . vn ; K¼6 .. 7
40 0 . 05
0 0 0 kn
According to Eq. (2.53),
V1 ¼ ½ w1 w2 . . . w M T ¼ W T ð4:101Þ
1 X 4
zj ¼ þ wTj bk Duk
s kj k¼1
X
yk ¼ cTk vj zj ; j ¼ 1; 2; . . . ð4:102Þ
j
The system represented by Eq. (4.102) is shown in Fig. 4.16.

From Fig. 4.16, it can be seen that wiT bk measures the influence of the UPFC
stabilizer on the oscillation mode and ckT vi measures how much the oscillation
mode is observed in the feedback signal of the stabilizer. Hence, it is defined that as
Fig. 4.16 Modal 1

decomposition of the system b k1 Ck1
to be installed with UPFC s − λ1
stabilizer
1
bk 2 Ck 2 +
s − λ2
1
b kn Ckn
s − λn
Δu k yk
Tupfck (s)
far as the oscillation mode ki is concerned, the controllability index and observ-
ability index of the UPFC stabilizer are, respectively,
bik ¼ wiT bk ; cik ¼ ckT vi ð4:103Þ
Their product is the residue
Rik ¼ bik cik ð4:104Þ
which measures the overall influence of the stabilizer on the oscillation mode.
Obviously, the controllability index, observability index, and residue are cal-
culated from the open-loop system model as far as the UPFC stabilizer is con-
cerned. Hence, it can be used to predict the effectiveness of the stabilizer to be
installed. The controllability index can be used to select the most effective modu-
lation signal to be superimposed with the damping control signal to enhance the
damping of system electromechanical oscillation mode.
4.3.2.2 Selection by Damping Torque Calculation
Electric torque contribution from a UPFC stabilizer can be calculated from the
forward path of UPFC damping control signal. The forward path can be obtained
from Figs. 4.14 and 4.15 with the blocks associated with K4 , K5 , and K7 ignored as
shown in Fig. 4.17,

when k ¼ 1; kpuk ¼ Kdme ;
kpuk ¼ ½ Kpdc Kpme ; kquk ¼ ½ Kqdc Kqme ; kvuk ¼ ½ Kvdc Kvme
when k ¼ 2; kpuk ¼ Kdde ;
kpuk ¼ ½ Kpdc Kpde ; kquk ¼ ½ Kqdc Kqde ; kvuk ¼ ½ Kvdc Kvde
when k ¼ 3; kpuk ¼ Kdmb ;
kpuk ¼ ½ Kpdc Kpmb ; kquk ¼ ½ Kqdc Kqmb ; kvuk ¼ ½ Kvdc Kvmb
when k ¼ 4; kpuk ¼ Kddb ;
kpuk ¼ ½ Kpdc Kpdb ; kquk ¼ ½ Kqdc Kqdb ; kvuk ¼ ½ Kvdc Kvdb
The forward path is described by the following state-space representation
X_ 1 ¼ A1 X1 þ b1k Duk
ð4:105Þ
DTuk ¼ cTupfc1 X1 þ kpuk Duk ; k ¼ 1; 2; 3; 4
Fig. 4.17 Forward path of ΔTuk

UPFC stabilizer K8 + K duk Δu k
- 1
sCdc − K 9
K2
⎡ ΔVdc ⎤
k puk ⎢ Δu ⎥
⎣ k⎦ k vuk
k quk
1 KA
ΔE q ' Td0 's + K 3 - + 1 + sTA -
K6
where
2 3
2 3 K
0 K0 3 1
0 qdc0
DEq 6 Td0 Td0 Td0 7
6 7 6 7
X1 ¼ 4 DE0fd 5; A1 ¼ 6 K
6 TA
A K6 K
T T 7
1 A Kvdc 7
4 A A 5
DVdc K8 K
Cdc 0 9
Cdc
T T
b11 ¼ ½ Kqme Kvme Kdme ; b12 ¼ ½ Kqde Kvde Kdde ;
b13 ¼ ½ Kqmb Kvmb Kdmb T ; b14 ¼ ½ Kqdb Kvdb Kddb T ;
cupfc1 ¼ ½ K2 0 Kpdc T
From Eq. (4.105), transfer function of the forward path can be obtained as
Fupfck ðsÞ ¼ cupfc1

T
ðsI A1 Þ1 b1k ; k ¼ 1; 2; 3; 4 ð4:106Þ
Hence, the electric torque contributed from the UPFC stabilizer is

DTuk ¼ Fupfck ðjxs ÞTupfck ðjxs Þyk ; k ¼ 1; 2; 3; 4 ð4:107Þ
where xs is the angular frequency of power oscillation. From Eq. (4.107), it can be
seen that the following index can also be used to predict the effectiveness of the
UPFC stabilizer to be installed. The index can be used to select the modulation
signal to add the damping control signal of the UPFC stabilizer

Fupfck ðjxs Þ; k ¼ 1; 2; 3; 4 ð4:108Þ
4.3.2.3 Robustness of the Damping Control to the Variations of Power

System Operating Conditions
Discussions so far have demonstrated that the effectiveness of a stabilizer may

change with variations of power system operating conditions (load conditions and
network configurations). Design of the stabilizer is carried out based on the lin-
earized model which is obtained at a selected operating condition. Hence, the
effectiveness of stabilizer is only guaranteed at the selected operating condition.
However, operating conditions of a power system are variable. In order to ensure
that the stabilizer is effective over a certain range of power system operating
conditions, when the stabilizer is designed its robustness to the variations of system
operating conditions should be considered. This is the topic to be discussed in this
section.
It is assumed that the set of system known operating conditions is X0 and the
stabilizer is to be designed to ensure its robustness over X0 , that is, to guarantee the
effectiveness of the stabilizer in damping power system oscillations when system
operating conditions change within X0 . If it is known that at an operating condition
rr 2 X0 , the stabilizer is least effective, rr 2 X0 can be selected as the operating
condition at which the stabilizer is designed. By doing so, the design can ensure the
effectiveness of stabilizer over the set of system operating conditions and hence
guarantees the robustness of the stabilizer to the variations of system operating
conditions within set X0 . Based on this principle, two examples to design a robust
stabilizer are introduced as follows.
1. Robust selection of modulation signal to add UPFC stabilizer
Selection of a modulation signal to add the damping control signal of a UPFC
stabilizer is in order to maximize its effectiveness. When the controllability index is
used, the criterion of selection is
Duselected ¼ Maxðbik ¼ uTi bk Þ; Duk 2 fDme ; Dmb ; Dde ; Ddb g ð4:109Þ

Duk
When the damping torque index is used, according to Eq. (4.108), the criterion of
selection is

Duselected ¼ Max Fupfck ðjxs Þ; k ¼ 1; 2; 3; 4 ; Duk 2 fDme ; Dmb ; Dde ; Ddb g
Duk
ð4:110Þ
To consider the robustness of stabilizer to the variations of system operating con-

ditions, the above criteria of Eqs. (4.109) and (4.110) should be calculated at a
selected operating condition of the power system when
rselected ¼ Minðbik ¼ uTi bk Þ; r 2 X0 ð4:111Þ

r
or

rselected ¼ Min Fupfck ðjxs Þ; k ¼ 1; 2; 3; 4 ; r 2 X0 ð4:112Þ
r
Equations (4.109)–(4.112) mean that the selection is made at an operating condition

of power system where the stabilizer is predicted to be least effective and the most
effective modulation signal is selected to add the UPFC stabilizer at the selected
operating condition.
2. Design of robust stabilizer by the phase compensation method
Denote the transfer function of a stabilizer installed in a single-machine infinite-bus
power system to be Tstab ðsÞ ¼ Kstab Tstabk ðsÞ, where Kstab is the gain of the stabilizer.
The feedback signal of the stabilizer is ystab , and its decomposition into the damping
torque and synchronizing torque is

x0 Ksstab
ystab ¼ Kdstab Dx þ Ksstab Dd ¼ Kdstab þ Dx ¼ Ky ðjxs ÞDx
jxs
ð4:113Þ
If the forward path of stabilizer to the electromechanical oscillation loop of gen-

erator is Fstab ðs), the damping torque provided by the stabilizer is
DTDstab ¼ Re½Fstab ðjxs ÞKy ðjxs ÞTstab ðjxs ÞDx ð4:114Þ
and the synchronizing torque is

xs
DTSstab ¼ Im½Fstab ðjxs ÞKy ðjxs ÞTstab ðjxs Þ Dd ð4:115Þ
x0
From Eq. (4.114), it can be seen that with the variations of system operating
conditions, change of the damping torque contributed by the stabilizer is mainly
determined by variations of Fstab ðjxs ÞKy ðjxs Þ. At an operating condition rj 2 X0 , if
it is denoted that
Fstab ðjxs ÞKy ðjxs Þ ¼ Hj \uj ; at rj 2 X0 ð4:116Þ
Tstab ðjxs Þ ¼ H\ / ð4:117Þ
From Eqs. (4.114) and (4.115), the damping torque and synchronizing torque
provided by the stabilizer at rj 2 X0 can be obtained as
DTdstab ¼ Hj H cosðuj /ÞDx ð4:118Þ
xs
DTsstab ¼ Hj H sinðuj /Þ Dd ð4:119Þ
x0
If it has
umax ¼ maxðuj Þ; rj 2 X0 ð4:120Þ
in order for the stabilizer to provide the power system with positive synchronizing
torque at all known operating conditions rj 2 X0 , design of the stabilizer must
ensure
/
uj ; rj 2 X0 ð4:120aÞ
Hence, simply it can choose

/ ¼ umax ð4:121Þ
Therefore, from Eqs. (4.118) and (4.121), the damping torque contributed by the
stabilizer can be obtained as
DTdstab ¼ Hj H cosðuj umax ÞDx ð4:122Þ
Equation (4.122) indicates that the effectiveness of stabilizer to be designed is

measured by Hj cosðuj umax Þ at rj 2 X0 . Hence, the operating condition at which
the stabilizer will be least effective can be selected by using the following criterion:
Hselected cosðuselected umax Þ ¼ Min½Hj cosðuj umax Þ; rj 2 X0 ð4:123Þ

j
At rselected 2 X0 determined by using the criterion of Eq. (4.123), the stabilizer can
be designed. The design will ensure that more damping torque will be provided by
the stabilizer when the power system operates at other operating conditions.
Therefore, the design guarantees the robustness of the stabilizer to the variations of
power system operating conditions.
4.4 Examples
4.4.1 An Example Single-Machine Infinite-Bus Power

System Installed with a BESS Stabilizer
4.4.1.1 Extended Heffron–Phillips Model of the Example Power

System Installed with a BESS
Configuration of an example single-machine infinite-bus power system installed

with a shunt-connected BESS is as same as that shown in Fig. 4.2. The model of
BESS is shown in Fig. 4.18.
rbess
Vc
VSC Cdc Vdc Vbess
BESS
Fig. 4.18 Model of BESS
Vbess Vdc
Idc2 ¼ ð4:124Þ
rbess
Parameters of the example system are as follows:

Generator xd ¼ 1:0; xq ¼ 0:6; x0d ¼ 0:3; M ¼ 8s:; D ¼ 0; T0d0 ¼ 5:044s:
The AVR KA ¼ 100; TA ¼ 0:01
Transmission line xts ¼ 0:3; xsb ¼ 0:3; xs ¼ 0:15
The BESS k ¼ 3; Cdc ¼ 1:0; Vdc0 ¼ 1:0; Vbess ¼ 1:0; rbess ¼ 0:01
Steady-state oper- Pt0 ¼ 0:5; Vt0 ¼ 1:0; Vb ¼ 1:0; Vs0 ¼ 1:0
ating point
Computation of initial values of various variables at the steady-state operation of

example power system is as same as that given in Sect. 3.3.1.2 as follows:
2sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 3

2
Vb 4 x P
Vb 5 ¼ 0:0337
sb t0
Qb0 ¼ V2s0
Xsb Vb
Pt0 jQb0
Isb0 ¼ ¼ 0:5 þ j0:0377
Vb
Vs0 ¼ jxsb Isb þ Vb ¼ 0:9887 þ j0:15 ¼ 1:0\8:6
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
V2t0 V2s0 ðPt0 xts Þ2 V2s0
Qs0 ¼ ¼ 0:0377
xts
Pt0 jQs0
Its0 ¼ ¼ 0:4887 þ j0:1123
Vs0
Is0 ¼ Its0 Isb0 ¼ 0:0113 þ j0:0746
Vt0 ¼ jxts Its0 þ Vs0 ¼ 0:955 þ j0:2966
EQ ¼ Vt0 þ jxq Its0 ¼ 0:8876 þ j0:5898 ¼ 1:07\33:6
4.4 Examples 165
Denote the real and imaginary part of a complex number by Re( ) and Im( ),
respectively, from the phasor diagram of Fig. 4.3 it can have
itsd0 ¼ ReðIts0 Þ sin d0 ImðIts0 Þ cos d0 ¼ 0:1769

itsq0 ¼ ReðIts0 Þ cos d0 þ ImðIts0 Þ sin d0 ¼ 0:4692
vtd0 ¼ ReðVt0 Þ sin d0 ImðVt0 Þ cos d0 ¼ 0:2815
vtq0 ¼ ReðVt0 Þ cos d0 þ ImðVt0 Þ sin d0 ¼ 0:9596
isd0 ¼ ReðIs0 Þ sin d0 ImðIs0 Þ cos d0 ¼ 0:0684
isq0 ¼ ReðIs0 Þ cos d0 þ ImðIs0 Þ sin d0 ¼ 0:0318
E0q0 ¼ EQ ðxd x0d Þitsd0 ¼ 1:0126

Vc0 ¼ Vs0 jxs Is0 ¼ 0:9999 þ j0:1517 ¼ 1:011\8:6
Vc0
m0 ¼ ¼ 0:3371
kVdc0
From Fig. 4.3, it can be seen that w0 and wx0 ¼ 8:6 is the phase of Vc0 in the d–
q coordinate and x–y coordinate, respectively. Hence,
p
w0 ¼ wx0 þ d0 ¼ 65:0250
2
From Eqs. (4.7) and (4.15), it is obtained that

c11 c12 0:6667 1:0000 d11 d12 0:9524 1:4286
¼ ; ¼
c21 c22 2:6667 0:6667 d21 d22 2:8571 0:9524
By using Eqs. (4.16)–(4.19), following coefficients in the Heffron–Phillips

model of Figs. 4.4 and 4.5 are obtained
K1 ¼ 0:3330; K2 ¼ 0:6702; K3 ¼ 2; K4 ¼ 0:1845; K5 ¼ 0:0290;

K6 ¼ 0:5483; K7 ¼ 1:9772; K8 ¼ 0:4067; K9 ¼ 0:0746
Kdm ¼ 0:2212; Kpm ¼ 0:5355; Kqm ¼ 1:8130; Kpw ¼ 0:7086;
Kqw ¼ 0:2847; Kdw ¼ 2:6858; Kpdc ¼ 0:1805; Kqdc ¼ 0:6112;
Kvm ¼ 0:8882; Kvw ¼ 0:0138; Kvdc ¼ 0:2994
1
DIdc2 ¼ DVdc ð4:125Þ
rbess
From Eqs. (4.19) and (4.125), state equation of the example power system is
obtained as
X_ ¼ AX þ Bu ð4:126Þ
where
2 3
Dd
6 7
6 Dx 7
6 7 D/
6 0 7
X ¼ 6 DEq 7; u ¼ ;
6 7 Dm
4 DE0fd 5
DVdc
2 3
0 xo 0 0 0
6 K 7
6 K 1
MD K 2
0 Mpdc 7
6 M M 7
6 Kqdc 7
6 K0 4 0 K3 1 7
A¼6 T T T 0
T 0
7
6 d0 d0 d0 d0 7
6 KA K5 K A K6 K 7
6 0 T T 1
T vdc 7
4 TA A A A 5
K7 0 K8 0 K9 C 1r
2 dc bess
3
0 314:16 0 0 0
6 7
6 0:1302 0 0:0838 0 0:0226 7
6 7
¼66 0:0930 0 0:3965 0:1983 0:1212 7;
7
6 7
4 428:242 0 5483:2 100 2994:2 5
0:7086 0 0:4067 0 100:0746
2 3
0 0 2 3
6 Kpw Kpm 7 0 0
6 M M 7 6 7
6 7 6 0:0886 0:0669 7
6 Kqw K 7 6 7
B¼6
6
0
Td0
qm 7 6
T0d0 7 ¼ 6 0:0564 0:3594 7
7
6 7 6 7
6 KA Kvw 7 4 138:495 8882 5
4 T KATKvm 5
A A
2:6858 0:2212
Kdw Kdm
4.4 Examples 167
Eigenvalues of state matrix are

k1 ; k2 ¼ 0:0591 j6:4978
k3 ¼ 13:0310
k4 ¼ 100:2902
k5 ¼ 87:2679
k1 and k2 are the pair of electromechanical oscillation mode of the example power
system.
4.4.1.2 Design of BESS Stabilizers by Use of the Phase

Compensation Method
To work out the transfer function of the forward path from a BESS stabilizer to the
electromechanical oscillation loop, the path related to Dd in the Heffron–Phillips
model of Figs. 4.4 and 4.5 can be ignored. Hence, from Fig. 4.4 with Dd being
ignored, it can have
1
DVdc ¼ ðK8 DE0q þ Kdm Dm þ Kdw D/Þ ð4:127Þ
s K9 þ C 1r
dc bess
Figure 4.19 shows the forward path from the BESS to the electromechanical
oscillation loop of generator. From Fig. 4.19, it can be obtained

1 KA
DE0q ¼ ðKvdc DVdc Kvw Dw Kvm Dm K6 DE0q Þ
sT0d0 þ K3 1 þ sTA

ð4:128Þ
Kqdc DVdc Kqw D/ Kqm Dm
DTet ¼ K2 DE0q þ Kpdc DVdc þ Kpw D/ þ Kpm Dm
-
ΔTet
[ ΔVdc Δψ Δm ]
K2
⎣⎡ K pdc K pψ K pm ⎦⎤ ⎣⎡ K vdc K vψ K vm ⎦⎤
⎣⎡ K qdc K qm K qψ ⎦⎤
ΔE q '
-
1 KA
K 3 + sTd0 ' + 1 + sTA -
K6
Fig. 4.19 Forward path of BESS

Deleting variable DVdc and DE0q in Eqs. (4.126) and (4.127), it can be obtained
that
DTes ¼ Fact ðsÞD/ þ Frct ðsÞDm ð4:129Þ
With xs ¼ Imðk1 or k2 Þ ¼ 6:4978, it can be computed that
Fact ðjxs Þ ¼ 0:7164 j0:0017 ¼ 0:7164\ 0:136

ð4:130Þ
Frct ðjxs Þ ¼ 0:2467 þ j0:4706 ¼ 0:5313\117:661
A simple way to calculate the transfer function of the forward path with s ¼ jxs
is to obtain the following state-space representation of the system of Eqs. (4.127)
and (4.128)
X_ r ¼ Ar Xr þ Br u
ð4:131Þ
DTet ¼ Cr Xr þ Dr u
2 3
DE0q
D/
where Xr ¼ 4 DE0fd 5; u ¼ , Ar and Br are obtained directly from A and
Dm
DVdc
B from Eq. (4.126) by deleting two state variables Dd and Dx, and the output
equation is obtained from the second equation in Eq. (4.128). That is
2 3
TK30 1 Kqdc 2 3
6 d0 Td0 0 Td0 0 7 0:3965 0:1983 0:1212
6 7 6 7
Ar ¼ 6 KA K6 1 KTvdc 7 ¼ 4 5483:2 100 2994:2 5;
4 TA TA A 5
K8 0 K9 C 1r 0:4067 0 100:0746
2 3 dc bess
K K 2 3
T pw0 T pm0 0:0564 0:3594
6 7
6 7 6 7
d0 d0
Br ¼ 6 KA Kvw KA Kvm 7 ¼ 4 138:495 8882 5;

4 TA TA 5
2:6858 0:2212
Kdw Kdm
2 3 2 3
K2 0:6702
6 7 6 7
Cr ¼ 4 0 5 ¼ 4 0 5; Dr ¼ ½ Kpw Kpm ¼ ½ 0:7086 0:5355
Kpdc 0:1805
It can have
DTet ¼ ½CTr ðsI Ar Þ1 Br þ Dr u ¼ Fact ðsÞD/ þ Frct ðsÞDm ð4:132Þ

4.4 Examples 169
Hence,
½ Fact ðjxs Þ Frct ðjxs Þ ¼ CTr ðjxs I Ar Þ1 Br þ Dr

¼ ½ 0:7164\ 0:136 0:5313\117:661
The above results are as same as those given in Eq. (4.130). Take the transfer
function of the BESS stabilizer to be
ð1 þ sT2 Þ
Tbess ðsÞ ¼ Kbess with T1 ¼ 0:38s:
ð1 þ sT1 Þ
Deviation of active power delivered along the transmission line, DPt , is taken as the
feedback signal of the stabilizer. Hence, the electric torque supplied by a BESS
active power stabilizer is
DTetactive ¼ Fact ðjxs ÞD/ ¼ Fact ðjxs ÞTbess ðjxs ÞDPt

¼ Fact ðjxs ÞTbess ðjxs ÞðD þ jxs MÞDx ð4:133Þ
The electric torque supplied by a BESS reactive power stabilizer is
DTetreactive ¼ Frct ðjxs ÞDm ¼ Frct ðjxs ÞTbess ðjxs ÞDPt

¼ Frct ðjxs ÞTbess ðjxs ÞðD þ jxs MÞDx ð4:134Þ
The BESS active and reactive power stabilizer can be designed by using the phase
compensation method to supply a pure damping torque DTetactive ¼ 9Dx and
DTetreactive ¼ 20Dx, respectively. The parameters of stabilizers obtained from the
design are as follows:
1. The BESS active power stabilizers: Kbess ¼ 15:9567; T2bess ¼ 0:3826s:;
2. The BESS reactive power stabilizers: Kbess ¼ 99:7980; T2bess ¼ 0:0151s:
Table 4.1 gives the computational results of oscillation mode of example power
system without and with the BESS stabilizers installed. They confirm the effec-
tiveness of the BESS stabilizer in damping the power oscillations.
Figure 4.20 shows the simulation results of the example power system without
and with the BESS stabilizers installed. In the simulation, a three-phase to-earth
Table 4.1 Oscillation mode of example power system when Pt0 = 0.5 p.u.
Without any stabilizer With active power stabilizer With reactive power stabilizer
0:0591 j6:4978 0:6541 j6:4621 0:5315 j5:1291
Rotor angle δ (degree)

with reactive stabilizer installed
with active stabilizer installed

without stabilizers
time(second)
Fig. 4.20 Simulation results of example power system when Pt0 ¼ 0:5 p.u.
short circuit occurred at 0.5 s of simulation for 100 ms. The results confirm the
effectiveness of the BESS stabilizers designed by use of the phase compensation
method to damp the power oscillation.
4.4.1.3 Robustness of the BESS Stabilizers to the Variations of Power

System Loading Conditions
Effectiveness of the BESS stabilizers is examined when Pt0 varies from 0.1 p.u. to
1.0 p.u. with Vt0 ¼ 1:0; Vb ¼ 1:0; Vs0 ¼ 1:0 being fixed. Figure 4.21 shows the
damping torque provided by the BESS stabilizers with variations of system load
conditions. From Fig. 4.21, it can be seen that provision of the damping torque by
the BESS active power stabilizer changes much less than that by the reactive power
stabilizer. Hence, the active power stabilizer is more robust to the variations of
system load conditions. Table 4.2 gives the computational results of electrome-
chanical oscillation mode. Figures 4.22 and 4.23 are the results of simulation of
example power system at different load conditions. Those results obviously confirm
that the BESS active power stabilizer is more robust than the reactive power sta-
bilizer to the variations of power system loading conditions as concluded by the
discussion in Sect. 4.1.2.2.
4.4 Examples 171
Damping torque supplied by stabilizer
Operating condition
Fig. 4.21 Damping torque provided by BESS stabilizers with variations of system loading
conditions
Table 4.2 Oscillation mode of example power system with variable loading conditions
Pt0 = 0.2 p.u. Pt0 = 0.9 p.u.
Without any stabilizer 0:0070 j6:3064 0:2505 j6:8670
With active power stabilizer 0:6685 j6:2676 0:5427 j6:8615
With reactive power stabilizer 0:2050 j5:7870 0:9353 j4:6027

without stabilizers time(second)
without stabilizers
time(second)
4.4.2 An Example Single-Machine Infinite-Bus Power

System Installed with a UPFC Stabilizer
4.4.2.1 Extended Heffron–Phillips Model of Example Power

System Installed with a UPFC
Configuration of an example single-machine infinite-bus power system installed

with a UPFC is as same as that shown in Fig. 4.12. Parameters of example system
are as follows:
Generator xd ¼ 1:0; xq ¼ 0:6; x0d ¼ 0:3; M ¼ 8s:; D ¼ 0; T0d0 ¼ 5:044s:
The AVR KA ¼ 100; TA ¼ 0:01s:
Transmission xtb ¼ 0:3; xte ¼ 0:15; xbt ¼ 0:15:
line
The UPFC ke ¼ 3; kb ¼ 3; Cdc ¼ 1:0; Vdc0 ¼ 1:0; xes ¼ 0:02; xbs ¼ 0:02:
Steady-state Pt0 ¼ 0:1; Vt0 ¼ 1:05; Vb ¼ 1:0:
operating point
This example is used to demonstrate the design and effectiveness of the UPFC
stabilizers. For the simplicity of presentation, normal control functions of the UPFC
are not included in the following demonstration. The procedure will be exactly as
4.4 Examples 173
same as what is presented below when the UPFC normal control functions are
considered. Hence, it is assumed that Ies ¼ 0; Vbt ¼ 0 which can give
Vbs ¼ jxbs Ibs ; Ves ¼ Vet ð4:135Þ
Total reactance of the transmission lines is xt ¼ xxtbtb ðþxxte teþþxxbtbtÞ ¼ 0:15. From

Vt0 Vb
Vb
jxt
¼ Pt0 þ jQb0 , the reactive power received at the infinite bus bar Qb0 can be
obtained as
2sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 3

2
Vb 4 xt Pt0
Qb0 ¼ V2t0 Vb 5 ¼ 0:3326 ð4:136Þ
xt Vb
Thus,
Pt0 jQb0
Itb0 þ Ite0 ¼ ¼ 0:1 j0:3326
Vb
Vt0 ¼ jxt ðItb0 þ Ite0 Þ þ Vb ¼ 1:0499 þ j0:0150 ¼ 1:05\0:82
xtb
Ite0 ¼ Ibs0 ¼ ðItb0 þ Ite0 Þ ¼ 0:05 j0:1663 ¼ 0:1737\ 73:27
xtb þ xte þ xbt
EQ ¼ Vt0 þ jxq ðItb0 þ Ite0 Þ ¼ 1:2495 þ j0:0750 ¼ 1:252\3:435
Ves0 ¼ Vet0 ¼ Vt0 jxte Ite0 ¼ 1:0249 þ j0:0075 ¼ 1:025\0:419 ¼ me0 ke Vdc0 \86:99
Vbs0 ¼ jxbs Ibs ¼ 0:0033 j0:001 ¼ 0:0034\16:86 ¼ mb0 kb Vdc0 \86:56
Hence, d0 ¼ 3:435 ; me0 ¼ 0:3417; mb0 ¼ 0:0012; de0 ¼ 0; db0 ¼ 2:8459 ,
vtd0 ¼ ReðVt0 Þ sin d0 ImðVt0 Þ cos d0 ¼ 0:0479;

vtq0 ¼ ReðVt0 Þ cos d0 þ ImðVt0 Þ sin d0 ¼ 1:0489;
ited0 ¼ ReðIte0 Þ sin d0 ImðIte0 Þ cos d0 ¼ 0:1690;
iteq0 ¼ ReðIte0 Þ cos d0 þ ImðIte0 Þ sin d0 ¼ 0:0399;
vtq0 Vb cos d0
itbd0 ¼ ¼ 0:1690;
xtb
Vb sin d0 vtd0
itbq0 ¼ ¼ 0:0399;
xtb
vetd0 ¼ ReðVet0 Þ sin d0 ImðVet0 Þ cos d0 ¼ 0:0539;
vetq0 ¼ ReðVet0 Þ cos d0 þ ImðVet0 Þ sin d0 ¼ 1:0236;
iesd0 ¼ 0; iesq0 ¼ 0;
ibsd0 ¼ ReðIbs0 Þ sin d0 ImðIbs0 Þ cos d0 ¼ 0:1690;
ibsq0 ¼ ReðIbs0 Þ cos d0 þ ImðIbs0 Þ sin d0 ¼ 0:0399;
E0q0 ¼ EQ ðxd x0d Þðited0 þ itbd0 Þ ¼ 1:1503
Various coefficient matrices in Eqs. (4.86) and (4.88) are computed to be

0:9805 0 0:4001 0:0367 1:4073 7:9475
Fq ¼ Fd ¼ ;
0:0134 0 0:2689 0:0408 0:1656 5:0655

1:1742 7:6144 0:9001 0:0174
Gq ¼
0:7903 5:1251 0:9515 0:0184

23:3120 0:4196 14:8739 0:001
Gd ¼
14:8800 0:2678 15:8655 0:0011
Hd ¼ ½ 0:8589 0 0:0262
Ld ¼ ½ 0:0768 0:4979 0:0103 0:0002
Hq ¼ ½ 0:0416 0:2641 0:4323
Lq ¼ ½ 1:2648 0:0228 0:1487 0
Following state equation of extended Heffron–Phillips model of Eq. (4.95) is

obtained
2 3 2 32 3
Dd_ 0 314:1593 0 0 0 Dd
6 7 6 76 7
6 Dx_ 7 6 0:2244 0 0:0173 0:0025 76 Dx 7
0
6 7 6 76 7
6 DE_ 7 6
0
6 q 7 ¼ 6 0:0192 0 0:5387 0:1983 0:2000 7 6 0 7
76 DEq 7
6 7
6 _0 7 6 76
2638:0 100 4330:5 54 DE0fd 5
7
4 DEfd 5 4 23:2301 0
DV_ dc 1:0061 0 0:0759 0 0:0189 DVdc
2 3
0 0 0 0 2 3
6 7 DmE
6 0:0074 0:1301 0:0012 0:00005 76
6 76 DdE 77
6 0:5851 0:0105 0:0688 0:000004 7 ð4:137Þ
6 764
7
6 7 Dm B5
4 12670 0:1463 1490:5 0:0127 5
DdB
0:0549 7:7987 0:0308 0:0172
Eigenvalues of state matrix are
k1 ¼ 94:4375
k2 ; k3 ¼ 0:0106 j8:4249
k4 ¼ 6:0714
k5 ¼ 0:0276
Obviously, k2 ; k3 ¼ 0:0106 j8:4249 are the pair of electromechanical

oscillation mode of example power system.
4.4 Examples 175
4.4.2.2 Robust Selection of a Modulation Signal to Add the Damping

Control Signal of the UPFC Stabilizer by Using the Damping
Torque Analysis and Residue Index
With Vt0 ¼ 1:05; Vb ¼ 1:0, Pt0 changes from 0.1 p.u. to 1.2 p.u. Deviation of
modulation signal to add a UPFC stabilizer is Duk ; k ¼ 1; 2; 3; 4 where
Du1 ¼ Dme ; Du2 ¼ Dde ; Du3 ¼ Dmb ; Du4 ¼ Ddb ð4:138Þ

According to the definition of Eq. (4.108) and from Fig. 4.17, Fupfck ðjxs Þ; k ¼
1; 2; 3; 4 is calculated as shown in Fig. 4.24 (jxs ¼ j8:4249).
From Fig. 4.24, it can be seen that

rselected ¼ fPt0 ¼ 0:1g ¼ MinðFupfck ðjxs Þ; k ¼ 1; 2; 3; 4; r 2 fPt0 ¼ 0:1 1:2gÞ
r
It means that the UPFC stabilizer should be designed at load condition Pt0 ¼ 0:1
when the stabilizer is least effective to damp the electromechanical oscillation
mode. It is also shown in Fig. 4.23 that the stabilizer can provide the most amount
of damping torque when it is added on the modulation signal Dde . That is

Duselected ¼ Dde ¼ Max½Fupfck ðjxs Þ; k ¼ 1; 2; 3; 4; Duk 2 fDme ; Dmb ; Dde ; Ddb g
Duk
Magnitude of Forward path
with δ e signal
with m e signal
with m b signal
with δ b signal
Operating condition

Fig. 4.24 Computational result of Fupfck ðjxs Þ; k ¼ 1; 2; 3; 4
Hence, to have a robust UPFC stabilizer, the stabilizer should be added on Dde
and the design carried out at operating condition Pt0 ¼ 0:1; Vt0 ¼ 1:05; Vb ¼ 1:0.
The robust modulation signal to add a UPFC stabilizer can also be selected by
using the residue index defined by Eq. (4.104). Take the feedback signal of sta-
bilizer to be the locally available signal, the deviation of line active power Pte .
Similar to Eq. (4.83), it has
Pte ¼ vtq iteq þ vtd ited ð4:139Þ
By using Eqs. (4.86) and (4.88), similar to the derivation of Eq. (4.91), lin-
earization of Eq. (4.139) can be obtained to be
DPte ¼ iteq0 Dvtq þ ited0 Dvtd þ vtq0 ðDiesq þ Dibsq Þ þ vtd0 ðDiesd þ Dibsd Þ
2 3
Dd 2 3
6 7 Dme
6 Dx 7 6 Dd 7
6 7 6 e7 ð4:140Þ
¼ ½ c1 0 c3 0 c5 6 0 7
6 DEq 7 þ ½ d1 d2 d3 d4 6 7
6 7 4 Dmb 5
4 DE0fd 5
Ddb
DVdc
where
½ c1 0 c3 0 c5 ¼ Itq0 Hq þ Itd0 Hd þ Vtq0 Fq1 þ Vtq0 Fq2

þ Vtd0 Fd1 þ Vtd0 Fd2 ¼ ½ 1:1616 0 0:0859 0 0:0212
½ d1 d2 d3 d4 ¼ Itq0 Lq þ Itd0 Ld þ Vtq0 Gq1 þ Vtq0 Gq2
þ Vtd0 Gd1 þ Vtd0 Gd2 ¼ ½ 0:0620 2:7016 0:0141 0:0011
Fq1 and Fq2 , Fd1 and Fd2 , Gq1 and Gq2 and Gd1 and Gd2 are the first and
second rows of Fq , Fd , Gq , and Gd , respectively.
At the selected operating condition rselected ¼ fPt0 ¼ 0:1; Vt0 ¼ 1:05;
Vb0 ¼ 1:0g, for the oscillation mode k2;3 ¼ 0:0106 j8:4249, the right and left
eigenvectors are obtained as
2 3 2 3
0:0091 þ j0:0206 0:0091 j0:0206
6 7 6 7
6 0:7702 þ j0:3380 7 6 0:7702 j0:3380 7
6 7 6 7
v2 ¼ 6 7 6 7
6 0:0013 þ j0:0008 7 and v3 ¼ 6 0:0013 j0:0008 7
6 7 6 7
4 0:0000 þ j0:0000 5 4 0:0000 j0:0000 5
0:0010 j0:0010 0:0010 þ j0:0010
2 3 2 3
0:0899 j0:2015 0:0899 þ j0:2015
6 7 6 7
6 0:0054 j0:0024 7 6 0:0054 þ j0:0024 7
6 7 6 7
2 ¼ 6
w 7
6 0:0021 j0:0224 7 and w 3 ¼ 6
6 0:0021 þ j0:0224 7
7
6 7 6 7
4 0:9747 5 4 0:9747 5
0:0243 þ j0:0107 0:0243 j0:0107
4.4 Examples 177
From Eqs. (4.102) and (4.103), the controllability of four input signals as far as
the electromechanical oscillation mode is concerned is calculated as
k ¼ 1; Du1 ¼ DmE ;
T2 b1 ¼ 2:4108 j2:3637;
b21 ¼ w b31 ¼ w
T3 b1 ¼ 2:4108 þ j2:3637
k ¼ 2; Du2 ¼ DdE ;

b22 ¼ T2 b2
w ¼ 10:7641 j5:1591; b32 ¼ w
T3 b2 ¼ 10:7641 þ j5:1591
k ¼ 3; Du3 ¼ Dmb ;

b23 ¼ T2 b3
w ¼ 0:2548 j0:2928; b33 ¼ w
T3 b3 ¼ 0:2548 þ j0:2928
k ¼ 4; Du4 ¼ Ddb ;
T2 b4 ¼ 0:0023 j0:0001;
b24 ¼ w b34 ¼ w
T3 b4 ¼ 0:0023 þ j0:0001
where
2 3 2 3 2 3 2 3
0 0 0 0
6 0:0074 7 6 0:1301 7 6 0:0012 7 6 0:0000 7
6 7 6 7 6 7 6 7
b1 ¼ 6 0:5851 7; b2 ¼ 6 0:0105 7; b3 ¼ 6 0:0688 7; b4 ¼ 6
6 7 6 7 6 7 7
6 0:0000 7;
4 12670 5 4 0:1463 5 4 1490:5 5 4 0:0127 5
0:0549 7:7986 0:0308 0:0172
The observability is given by
c2k ¼ cTk
v2 ¼ 0:0002 j0:0029 and c3k ¼ cTk v3 ¼ 0:0002 þ j0:0029
where
cTk ¼ ½ 1:1616 0 0:1387 0 0:0202 ðk ¼ 1; 2; 3; 4Þ
The residue index for four input signals fDmE ; DdE ; DmB ; DdB g of the UPFC
stabilizer, thus, is obtained as
k ¼ 1; Du1 ¼ DmE ;
R21 ¼
b21c21 ¼ 0:0073 j0:0065; R31 ¼ b31c31 ¼ 0:0073 þ j0:0065;
k ¼ 2; Du2 ¼ DdE ;
R22 ¼
b22c22 ¼ 0:0128 þ j0:0322; R32 ¼ b32c32 ¼ 0:0128 j0:0322;
k ¼ 3; Du3 ¼ Dmb ;
R23 ¼
b23c23 ¼ 0:0009 j0:0007; R33 ¼ b33c33 ¼ 0:0009 þ j0:0007;
k ¼ 4; Du4 ¼ Ddb ;
24c24 ¼ 0:0000 þ j0:0000; R34 ¼ b34c34 ¼ 0:0000 j0:0000
R24 ¼ b
Because R2k and R3k (k = 1, 2, 3, 4) are pairs of conjugate vectors, and the
amplitude of residue measures the overall influence of the UPFC stabilizer on the
oscillation mode. It can be obtained that
Residue
with m e signal
with δ e signal
with m b signal
with δ b signal
Pt0
Fig. 4.25 Computational result of jRki j; k ¼ 1; 2; 3; 4

R21 ¼ R31 ¼ 0:0097

R22 ¼ R32 ¼ 0:035

R23 ¼ R33 ¼ 0:0011

R24 ¼ R34 ¼ 0
The largest residue indicates the most effective feedback signal to design the UPFC
stabilizer. Hence,

uselected ¼ dE where Rik max ¼ Ri2 ¼ 0:035ði ¼ 2; 3Þ
The selection is as same as that obtained previously by using the damping torque
analysis. Variation of the residue with system load condition from 0.1 to 1.2 p.u. is
shown in Fig. 4.25. Obviously, Pt0 ¼ 0:1 should be selected, which again is as same
as that obtained previously.
4.4.2.3 Design of the UPFC Stabilizer by Using the Phase

Compensation Method
State Eq. (4.137) and output Eq. (4.140) are obtained at the selected operating
condition Pt0 ¼ 0:1; Vt0 ¼ 1:05; Vb ¼ 1:0 of the example power system.
With Dde being selected to add the damping control signal of the UPFC stabi-
lizer, it can be obtained that
4.4 Examples 179
k ¼ 0:2244; d ¼ 0; C1 ¼ 1:1616; C2 ¼ 0; C3 ¼ ½ 0:0859 0 0:0212 ;

dm ¼ 2:7016
A23 ¼ ½ 0:0173 0 0:0025
2 3 2 3 2 3
0:0192 0 0:5387 0:1983 0:2
6 7 6 7 6 7
A31 ¼ 4 23:2301 5; A32 ¼ 4 0 5; A33 ¼ 4 2638:0 100 4330:5 5
1:0061 0 0:0759 0 0:0189
B2 ¼ 0:1301
2 3
0:0105
6 7
B3 ¼ 4 0:1463 5
7:7986
Results of computation according to Eq. (4.61) are (jxs ¼ j8:4249)

jxs jxs
Kðjxs Þ ¼ k þ d þ AT23 ðjxs I A33 Þ1 A31 þ A32 ¼ 0:2259 þ j0:0006
x0 x0

x0 x
C1 þ C2 þ CT3 ðjxs I A33 Þ1
0
Ko ðjxs Þ ¼ A31 þ A32 ¼ 0:0663 j43:5996
jxs jxs
Kc ðjxs Þ ¼ AT23 ðjxs I A33 Þ1 B3 þ B2 ¼ 0:1418 þ j0:0047
Kil ðjxs Þ ¼ CT3 ðjxs I A33 Þ1 B3 þ dm ¼ 2:7599 þ j0:0155
Let the design of the UPFC stabilizer provide a pure damping torque
Dupfc Dx ¼ 1Dx. Take DPte as the feedback signal. Thus,
ð1 þ sT2 Þ
Dde ¼ Tupfc ðsÞDPte ¼ Kupfc DPte ð4:141Þ
ð1 þ sT1 Þ
where T1 ¼ 0:3 s.
From Eq. (4.64), the forward path of the stabilizer is calculated as
Fupfc ðjxs Þ ¼ Kc ðjxs ÞKo ðjxs Þ þ Dupfc Kil ðjxs Þ ¼ 2:9764 j6:1674
Thus, parameters of the UPFC stabilizer are obtained as
Kupfc ¼ 0:2689; T2 ¼ 0:1288:
Table 4.3 gives the computational results of the electromechanical oscillation

mode of example power system at three different load conditions. Figures 4.26,
4.27, and 4.28 are the results of nonlinear simulation. Those results confirm that the
UPFC stabilizer is effectively designed in damping the power oscillation and robust
to the variations of system load conditions.
Table 4.3 Oscillation mode of example power system with variable load conditions
Pt0 ¼ 0:1 p.u. Pt0 ¼ 0:8 p.u. Pt0 ¼ 1:2 p.u.
Without stabilizer 0:0106 j8:4249 0:0133 j8:9982 0:1457 j9:4250
With stabilizer 0:6880 j7:0257 1:2283 j7:5835 1:2930 j8:1842
δ (degree)
without stabilizer
with UPFC stabilizer

time(second)
Fig. 4.26 Simulation results of example power system when Pt0 = 0.1 p.u.
δ (degree)
without stabilizer
time(second)
4.4 Examples 181
δ (degree)

without stabilizer
time(second)
In order to demonstrate that the robustness of the UPFC stabilizer is due to the
selection of the load condition at Pt0 ¼ 0:1; Vt0 ¼ 1:05; Vb ¼ 1:0 where the
stabilizer is designed, the stabilizer is designed at another load condition
Pt0 ¼ 0:8; Vt0 ¼ 1:05; Vb ¼ 1:0. Then, its effectiveness is checked at different
load conditions by modal computation presented in Table 4.4. From the results, it
can be seen that though the UPFC stabilizer is designed properly at the selected
load condition Pt0 ¼ 0:8; Vt0 ¼ 1:05; Vb ¼ 1:0, where the power oscillation is
effectively suppressed (see the second column of Table 4.4), its effectiveness is
not ensured at Pt0 ¼ 0:1; Vt0 ¼ 1:05; Vb ¼ 1:0 (see the first column of
Table 4.4).
At the selected operating condition Pt0 ¼ 0:1; Vt0 ¼ 1:05; Vb ¼ 1:0, the UPFC
stabilizer can also be added to other modulation signals. Table 4.5 gives the results
of designing the UPFC stabilizers added on other two modulation signals and the
oscillation mode of example power system when stabilizers are installed. From
Table 4.5, it can be seen that the gain value of the UPFC stabilizers added on Dme
and Dmb is much greater than that of the stabilizer added on Dde in order to
Table 4.4 Oscillation mode of example power system with the stabilizer designed at
Pt0 ¼ 0:8; Vt0 ¼ 1:05; Vb ¼ 1:0
Pt0 ¼ 0:1 p.u. Pt0 ¼ 0:8 p.u. Pt0 ¼ 1:2 p.u.
Without stabilizer 0:0106 j8:4249 0:0133 j8:9982 0:0133 j8:9982
With stabilizer 0:4291 j6:8578 0:7517 j7:3203 0:7565 j7:5281
Table 4.5 Results of designing the UPFC stabilizers added on other two modulation signals
Modulation signal added with Parameters of the UPFC Oscillation mode with the
a UPFC stabilizer stabilizer stabilizer installed
Dme Kp ¼ 1:6654; Tp ¼ 0:0926 0:7030 j6:9400
Dmb Kp ¼ 15:3864; Tp ¼ 0:1069 0:7078 j7:0774
suppress the power oscillation effectively. This means that more control cost is
required. Hence, the UPFC stabilizer added on Dde is the most efficient damping
controller.
References
1. CIGRE TF 30-01-08 Report, Modelling of power electronics equipment (FACTS) in load flow
and stability programs (1999)
2. Wang HF (1999) Phillips-Heffron model of power systems installed with STATCOM and
applications. IEE Proc Part C 146(5):521–527
3. Du W (2009) Power system small signal oscillation stability as affected by static synchronous
compensator (STATCOM) and energy storage system (ESS). PhD thesis, University of Bath,
UK
4. Wang HF (2000) Static synchronous series compensation to damp power system oscillations.
Int J Power Syst Res 54(2)
5. Wang HF (1999) Applications of modelling UPFC into multi-machine power systems. IEE Proc
Part C (3)
6. Wang HF (1999) Damping function of unified power flow controller. IEE Proc Part C. (1)
7. Wang HF (2000) A unified model for the analysis of FACTS devices in damping power system
oscillations part III: unified power flow controller. IEEE Trans Power Delivery (3)
Chapter 5
A Multi-machine Power System Installed
with Power System Stabilizers
5.1 Mathematical Model of a Multi-machine Power

System Installed with Power System Stabilizers
5.1.1 A Two-Machine Power System Installed with Power

System Stabilizers
Figure 5.1 shows the configuration of a two-machine power system where two
synchronous generators are connected by a transmission line. This simple system is
used in this section to demonstrate the procedure of establishing the mathematical
model of a multi-machine power system. Later, it will be used for introducing basic
concepts and methods in the analysis of power system oscillations and design of
PSSs in the multi-machine power system. For the purpose of demonstration, a load is
connected at the terminal of two generators and a node along the transmission line,
respectively. The load is represented by equivalent impedance rLi þ jxLi ; i ¼ 1; 2; 3:
5.1.1.1 Network Equations
Connections of two synchronous generators in the two-machine power system are

given by following network equations expressed in a common x–y coordinate
Ig1 ¼ 1 g1 þ 1 3 Þ ¼ ðgL1 þ jbL1 þ jb13 ÞV 3

g1 jb13 V
V ðVg1 V
rL1 þ jxL1 jx13
ð5:1Þ
Ig2 ¼ 1 g2 þ 1 3 Þ ¼ ðgL2 þ jbL2 þ jb23 ÞV 3
g2 jb23 V
V ðVg2 V
rL2 þ jxL2 jx23

DOI 10.1007/978-1-4899-7696-3_5
184 5 A Multi-machine Power System Installed …
Vg1 V3 Vg 2 G2
G1 Ig1 x13 x 23 Ig 2
rL1 + jx L1 rL3 + jx L3 rL2 + jx L2
Fig. 5.1 A two-machine power system
1 3 þ 1 ðV 3 V g1 Þ þ 1 ðV 3 V g2 Þ
0¼ V
rL3 þ jxL3 jx13 jx23
¼ ðgL3 þ jbL3 þ jb13 þ jb23 ÞV g1 jb23 V
3 jb13 V g2 ð5:2Þ
where jb13 ¼ jx1 ; jb23 ¼ jx1 ; gLi þ jbLi ¼ r þ1jx ; i ¼ 1; 2; 3:

13 23 Li Li
Matrix form of above equations is
2 3 2 32 g1 3
Ig1 gL1 þ jbL1 þ jb13 0 jb13 V
4 Ig2 5 ¼ 4 0 gL2 þ jbL2 þ jb23 jb23 56
4 g2 7
V 5
0 jb13 jb23 gL3 þ jbL3 þ jb13 þ jb23 V3
ð5:3Þ
3 ¼ jb13 g1 jb23 g2
V V V ð5:4Þ
gL3 þ jbL3 þ jb13 þ jb23 gL3 þ jbL3 þ jb13 þ jb23
Substituting above equation into Eq. (5.1), it can be obtained that

2 3
" # jb jb
gL1 þ jbL1 þ jb13 þ g 13þ jb13
jb13 jb23 " #
Ig1
6 R gL3 þ jbR 7 Vg1
¼4
L3
5
Ig2 jb23 jb13 jb23 jb23 g2
V
gL3 þ jbR g L2 þ jb L2 þ jb 23 þ gL3 þ jbR
" #" # " #

yN11 yN12 V g1 g1
V
¼ ¼Y N

yN21 yN22 V g2 g2
V
ð5:5Þ
5.1 Mathematical Model of a Multi-machine … 185
Fig. 5.2 Phasor diagram in qi

x–y coordinate and di qi y
coordinate
E qi '
(x qi − x di ')Iqi
jx di ' Igi
V gi
δi
x
di
where bLR ¼ bL3 þ b13 þ b23 and Y N is the network admittance matrix with only
nodes of generator terminal left. Equation (5.2) can be rearranged as
2 3 2 32 3
igx1 gN11 bN11 gN12 bN12 vgx1
6 igy1 7 6 bN11 gN12 7 6 7
6 7 6 gN11 bN12 76 vgy1 7
4 igx2 5 ¼ 4 gN21 bN21 gN22 bN22 5 4 vgx2 5
ð5:6Þ
igy2 bN21 gN21 bN22 gN22 vgy2
where subscript x and y denotes x and y component of a variable in common x–y

coordinate, respectively. According to the phasor diagram of Fig. 2.9, in d–q
coordinate of the ith generator, di -qi , it can have (i ¼ 1; 2)
V 0 jx0 Igi þ ðxqi x0 Þiqi

gi ¼ E ð5:7Þ
qi di di
0 is
From Fig. 5.2, it can be seen that in the common x–y coordinate, phase of E qi
0
di and that of ðxiq xid Þiqi is di 90 . Hence, in x–y coordinate, Eq. (5.7) is
(i ¼ 1; 2)
gi ¼ E0 ejdi jx0 Igi þ ðxqi x0 Þiqi ejðdi 90 Þ

V ð5:8Þ
qi di di
In matrix form, Eq. (5.8) is

" # " #" # " #
g1
V ejd1 0 E0q1 jx0d1 0 Ig1
¼
g2
V 0 e jd2 0
Eq2 0 jxd2 Ig2
0
"
#
ðxq1 x0d1 Þiq1 ejðd1 90 Þ 0 iq1
þ 0 jðd2 90 Þ
ð5:9Þ
0 ðxq2 xd2 Þiq2 e iq2
Denote
gi ¼ E0 ejdi þ ðxqi x0 Þiqi ejðdi 90 Þ ;

E i ¼ 1; 2 ð5:10Þ
qi di
as an “internal voltage” of generator. Equation (5.9) becomes

0
g1
E g1
V jxd1 0 Ig1
g2 ¼ g2 þ 0 Ig2 ð5:11Þ
E V jx0d2
gi ; i ¼ 1; 2 can be
Equation (5.11) indicates that introduction of internal voltage E
seen as addition of an extra node behind the node of generator terminal as shown in
Fig. 5.3. The added extra “internal voltage” node is connected to the generator
terminal through x0di ; i ¼ 1; 2.
0 1
Ig1 1 jxd1 0 g1
E g1
E

¼ YN þ
¼Y ð5:12Þ
Ig2 0 jx0d2 g2
E Eg2
where matrix Y is the network admittance matrix with only internal voltage nodes
of generators left.
E g1 Vg1 V3 Vg 2 Eg 2
x d1 ' Ig1 x13 x 23 Ig 2 xd2 '
rL1 + jx L1 rL3 + jx L3 rL2 + jx L2
Fig. 5.3 Introduction of internal voltage nodes of generators

That is ði ¼ 1; 2Þ
X
2 X
2
Igi ¼ igxi þ jigyi ¼ gj ¼
yij E

yij ½E0qj ejdj þ ðxqj x0dj Þiqj ejðdj 90 Þ ð5:13Þ
j¼1 j¼1
where According to Fig. 5.2, in di -qi coordinate

yij ¼ yij ejaij is the element of Y.
Eq. (5.13) is (i ¼ 1; 2)
X
2

Igi ¼ idi þ jiqi ¼ ðixi þ jiyi Þejð90 di Þ
¼

yij ½E0qj ejdj þ ðxqj x0dj Þiqj ejðdj 90 Þ ejð90 di Þ
j¼1
X
2

¼ yij ½E0qj ejð90
þ dj di Þ
þ ðxqj x0dj Þiqj ejðdj di Þ
j¼1
X
2

¼ yij ½E0qj ejð90 þ dj di þ aij Þ
þ ðxqj x0dj Þiqj ejðdj di þ aij Þ
j¼1
X
2
¼ yij ½E0qj cos(90 þ dj di þ aij Þ þ jE0qj sin(90 þ dj di þ aij Þ
j¼1
þ ðxqj x0dj Þiqj cos(dj di þ aij Þ þ j(xqj x0dj Þiqj sin(dj di þ aij Þ
X2
¼ yij ½E0qj sin(dj di þ aij Þ þ jE0qj cos(dj di þ aij Þ
j¼1
þ ðxqj x0dj Þiqj cos(dj di þ aij Þ þ j(xqj x0dj Þiqj sin(dj di þ aij Þ
ð5:14Þ
From above equation, it can have ði ¼ 1; 2Þ
X
2
idi ¼ yij ½E0qj sinðdj di þ aij Þ þ ðxqj x0dj Þiqj cosðdj di þ aij Þ
j¼1
ð5:15Þ
X
2
iqi ¼ yij ½E0qj cos(dj di þ aij Þ þ ðxqj x0dj Þiqj sin(dj di þ aij Þ
j¼1
5.1.1.2 Linearized Model When the Classical Model of Synchronous

Generators Is Used
For the simplicity of discussion, firstly the simplest case is considered when a
synchronous generator is represented by the so-called classic model, i.e., the rotor
motion equation of generator. That is the model given by Eq. (2.41) with function
of the AVR ignored and E0q being constant. In di -qi coordinate, model of generator
is (i ¼ 1; 2)
d_ i ¼ xo ðxi 1Þ
1
x_ i ¼ ½Pmi Pi Di ðxi 1Þ
Mi
ð5:16Þ
Pi ¼ vdi idi þ vqi iqi
vdi ¼ xqi iqi
vqi ¼ E0qi x0di idi
Hence, linearized classical model of generator is (i ¼ 1; 2)
Dd_ i ¼ xo Dxi
1
Dx_ i ¼ ðDPi Di Dxi Þ ð5:17Þ
Mi
DPi ¼ ðvdi0 iqi0 x0di ÞDidi þ ðvqi0 þ idi0 xqi ÞDiqi
Linearization of Eq. (5.15) is (E0qi is a constant, i; j ¼ 1; 2)
Didi ¼ hdi ðDdj Ddi Þ

Diqi ¼ hqi ðDdj Ddi Þ ð5:18Þ
i 6¼ j
where hdi and hdi i ¼ 1; 2 are constants. Hence, from Eqs. (5.17) and (5.18), it can
have (i; j ¼ 1; 2)
DPi ¼ ki ðDdj Ddi Þ ð5:19Þ
where ki i = 1, 2 are constants. State equation of the two-machine power system is

2 3 2 0 0 xo 0
32 3 2 3
Dd_ 1 Dd1 Dd1
6 _ 7 6 0 0 0 xo 776 Dd2 7 6 Dd 7
6 Dd2 7 6 76 7 6 27
6 7¼6 k Mk1 MD1 6 7 ¼ A 46 7 ð5:20Þ
4 Dx_ 1 5 6 0 74 Dx 5
1
4 M1 1 1 5 1 4 Dx1 5
Dx_ 2 Mk2 k2 0 MD2 Dx2 Dx2
2 M2 2
Equation (5.20) is equivalent to the following equation

0 2 3 2 31
D1 k1 k1
0 M Dd1
@s2 I þ sxo 4 M1 5 þ xo 4 M1 1 5A
¼0 ð5:21Þ
0 D2 Mk2 k2 Dd2
M2 2 M2
Characteristic equation of the system is

2 D 1 þ x k1 k1
k þ kxo M xo M
1
oM
1 1 ¼0 ð5:22Þ
k D k
xo M2 k þ kxo M2 þ xo M2
2
2 2 2
Obviously, k ¼ 0 is one of the eigenvalues of the system. Because state matrix

A4 is a real matrix, if there is a power oscillation, other three eigenvalues must be a
pair of conjugate complex eigenvalues and a real eigenvalue. The pair of conjugate
complex eigenvalues is the electromechanical oscillation mode, because the only
dynamic of system is the rotor motion of generators. Therefore, in the two-machine
power system, there is only one electromechanical oscillation mode. In fact, if it is
denoted that Dd12 ¼ Dd1 Dd2 , the state equation of system of Eq. (5.20) becomes
2 3 2 0 xo 0
3
2 3 2 3
Dd_ 12 Dd12 Dd12
6 7 6 k1 MD1 7
4 Dx_ 1 5 ¼ 6
4 M1 1
0 74 Dx 5 ¼ A 4 Dx 5
5 1 3 1 ð5:23Þ
Dx_ 2 k2 0 D2
M Dx2 Dx2
M2 2
Obviously, the system can only have one pair of conjugate complex eigenvalues
and hence one electromechanical oscillation mode. The reduced-order state matrix
A3 does not have to have one zero eigenvalue when the relative angular position of
generators is used as the state variable.
From Eq. (5.19), it can be seen that variations of the active power delivered
along the transmission line connecting two generators are proportional to the rel-
ative angular position. For example, if Dd1 ¼ Dd2 , DPi ¼ 0 ði ¼ 1; 2Þ, there will be
no variations of active power even if the absolute angular positions of generators
vary. This means that the power oscillations are related only to the relative angular
positions of generators.
5.1.1.3 Heffron–Phillips Model
Heffron–Phillips model of the two-machine power system installed with the PSS
can be established based on the following 4th-order model of synchronous gener-
ators ði ¼ 1; 2Þ
d_ i ¼ xo ðxi 1Þ
1
x_ i ¼ ½Pmi Pi Di ðxi 1Þ
Mi
0 1 ð5:24Þ
E_ qi ¼ 0 ðEqi þ Efd0i þ E0fdi Þ
Tdoi
0 1 0 KAi
E_ fdi ¼ E þ ðVrefgi Vgi þ upssi Þ
TAi fdi TAi
where
vid ¼ xqi iqi

vqi ¼ E0qi x0di idi
Pi ¼ vdi idi þ vqi iqi ð5:25Þ
Eqi ¼ E0qi þ ðxdi x0di Þidi
Vgi ¼ v2di þ v2qi
Linearization of Eq. (5.15) is (i ¼ 1; 2)
X
2
Didi ¼ hdij DE0qj þ hdj ðDdj Ddi Þ
j¼1
X
2 ð5:26Þ
Diqi ¼ hqij DE0qj þ hqj ðDdj Ddi Þ
j¼1
i 6¼ j
By using Eq. (5.26), linearization of Eq. (5.25) can be obtained as
DP1 ¼ k11 ðDd2 Dd1 Þ þ k211 DE0q1 þ k212 DE0q2

DP2 ¼ k12 ðDd1 Dd2 Þ þ k221 DE0q1 þ k222 DE0q2
DEq1 ¼ k41 ðDd2 Dd1 Þ þ k311 DE00q1 þ k312 DE0q2
ð5:27Þ
DEq2 ¼ k42 ðDd1 Dd2 Þ þ k321 DE0q1 þ k322 DE0q2
DVg1 ¼ k51 ðDd2 Dd1 Þ þ k611 DE00q1 þ k612 DE0q2
DVg2 ¼ k52 ðDd1 Dd2 Þ þ k621 DE00q1 þ k622 DE0q2
By substituting Eq. (5.27) into the linearization of Eq. (5.24), it can have for G1
Dd_ 1 ¼ xo Dx1
1
Dx_ 1 ¼ ½k11 ðDd2 Dd1 Þ k211 DE0q1 k212 DE0q2 D1 Dx1
M1
0 1
DE_ q1 ¼ 0 ½k41 ðDd2 Dd1 Þ k311 DE0q1 k312 DE0q2 þ DE0fd1
Tdo1
0 1 KA1
DE_ fd1 ¼ DE0fd1 þ ½k51 ðDd2 Dd1 Þ k611 DE0q1 k612 DE0q2 þ Dupss1
TA1 TA1
ð5:28Þ
and for G2
Dd_ 2 ¼ xo Dx2
1
Dx_ 2 ¼ ½k12 ðDd1 Dd2 Þ k221 DE0q1 k222 DE0q2 D2 Dx2
M2
0 1
DE_ q2 ¼ 0 ½k42 ðDd1 Dd2 Þ k321 DE0q1 k322 DE0q2 þ DE0fd2
Tdo2
0 1 KA2
DE_ fd2 ¼ DE0 þ ½k52 ðDd1 Dd2 Þ k621 DE0q1 k622 DE0q2 þ Dupss2
TA2 fd2 TA2
ð5:29Þ
Equations (5.28) and (5.29) are the Heffron–Phillips model of the two-machine
power system installed with PSS as shown in Fig. 5.4.
State-space representation of the Heffron–Phillips model is
2 : 32 32 3 2 3
Dd 0 xo I 0 0 Dd 0
6 D x: 7 6
6 7 6 M1 K1 M1 D M1 K2 0 7 6 7 6 7
76 Dx0 7 þ 6 0 7Dupss
6 _0 7¼4 54 DEq 5 4 0 5
4 DE q 5 T1d0 K4 0 T1
d0 K3 T1
d0
_ 0
DEfd T 1
A K5 KA 0 T1
A K5 KA T1
A DE0fd T1
A KA
ð5:30Þ
where
" # 0
Dd1 Dx1 DE0q1 DEfd1
Dd ¼ ; Dx ¼ ; ¼DE0q 0 ; DE 0
¼ ;
Dd2 Dx2 DEq2 fd
DE0fd2

k11 k11 k211 k212 k311 k312
K1 ¼ ; K2 ¼ ; K3 ¼ ;
k12 k12 k221 k222 k321 k322

k41 k41 k51 k51 k611 k612
K4 ¼ ; K5 ¼ ; K6 ¼ ;
k42 k42 k52 k522 k621 k622
0
M1 0 D1 0 Tdo1 0
M¼ ;D¼ ; Td0 ¼ ;
0 M2 0 D2 0 T0do2

KA1 0 TA1 0 Dupss1
KA ¼ ; TA ¼ ; Dupss ¼
0 KA2 0 TA2 Dupss2
The model can be shown in Fig. 5.5 which is the matrix form of Fig. 5.4.
Δδ2
k1−1 k1−1
+
ΔE q 2 ' - 1 Δω1 ω0 Δδ1
k 2−12 Δδ2
- D1 + sM1 s
-
ΔE q 2 ' Δδ2
k 2−11
k 3−12 k 4−1 k 4−1 k 5−1 k 5−1
ΔE q1 '
- -+ +
1 K A1 + Δu pss1
+
k 3−11 + sTd01 ' ΔE fd1 ' 1 + sTA1 -
-
k 6−12
Δδ1 K 6−11
ΔE q 2 '
k1− 2 k1− 2
+
ΔE q1 ' - 1 Δω2 ω0 Δδ2
k 2− 21
- D 2 + sM 2 s Δδ1
-
ΔE q1 ' Δδ1
k 2− 22
k 3− 21 k 4− 2 k 4− 2 k 5− 2 k 5− 2
ΔE q 2 '
- -+ + -
1 + KA2 + Δu pss2
k 3− 22 + sTd02 ' ΔE fd 2 ' 1 + sTA 2 -
-
k 6− 22 k 6− 21
ΔE q1 '
Fig. 5.4 Heffron–Phillips model of two-machine power system installed with PSS
5.1.2 A Multi-machine Power System Installed with Power

System Stabilizers
5.1.2.1 Heffron–Phillips Model of an N-Machine Power System

Installed with Power System Stabilizers
In an N-machine power system installed with PSSs, the 4th-order model of syn-
chronous generator is given by Eqs. (5.24) and (5.25) for i ¼ 1; 2; . . .; N.
Equation (5.8) for i ¼ 1; 2; . . .; N can be written in the following matrix form
K1
-
ω0 I
(sM + D) −1
s
-
K4 K5
K2
- -
+ +
(K 3 + sTd0 ) −1 −1
(I + sTA ) K A Δu pss
K6
Fig. 5.5 Heffron–Phillips model of two-machine power system in matrix form

g ¼ ejd E0 jx0 Ig þ ðx x0 Þejðd90 Þ iq
V ð5:31Þ
q d q d
where

g ¼ V
V gN T ; Ig ¼ Ig1 Ig2 . . . IgN T ;
g2 . . . V
g1 V
T
E0q ¼ E0q1 E0q2 . . . E0q1 ; iq ¼ ½ iq1 iq2 . . . iq1 ;
T

ejd ¼ diagðejdi Þ; x0d ¼ diagðx0di Þ; ðxq x0d Þejðd90 Þ ¼ diag½ðxqi x0di Þejðdi 90 Þ
ð5:32Þ
In Eq. (5.32), diag(ci Þ denotes an Nth-order diagonal matrix with ci being the ith
diagonal element.
Let the network equation be
Ig ¼ Y g
NV ð5:33Þ
where Y N is the network admittance matrix with only the nodes of generator
terminal left. From Eqs. (5.31) and (5.33), it can have
h i
Ig ¼ Y
ejd E0 þ ðxq x0 Þejðd90 Þ iq ¼ Y
E g ð5:34Þ
q d
where Y ¼ ðY 1 þ jx0 Þ1 . The above equation gives the generator current in x–y
N d
coordinate as (i ¼ 1; 2; . . .; N)
X
N
Igi ¼
yij ½E0qj ejdj þ ðxqj x0dj Þiqj ejðdj 90 Þ ð5:35Þ
j¼1
where Hence, Eq. (5.15) can be obtained

yij ¼ yij ejaij is the elements of matrix Y.
for i ¼ 1; 2; . . .; N. Their linearization can be expressed in the following matrix
form
DId ¼ Fdd Dd þ Gdd DE0q þ Hdd DIq

ð5:36Þ
DIq ¼ Fqq Dd þ Gqq DE0q þ Hqq DIq
where
T
Dd ¼ ½ Dd1 Dd2 . . . DdN T ; DE0q ¼ DE0q1 DE0q2 ... DE0qN ;
DId ¼ ½ Did1 Did2 ... DidN T ; DIq ¼ ½ Diq1 Diq2 ... DiqN T
f ddij ¼ yij ½E0qj0 cosðdj0 di0 þ aij Þ þ ðxqj x0dj Þiqj0 sinðdj0 di0 þ aij Þ; i 6¼ j,
f qqij ¼ yij ½E0qj0 sinðdj0 di0 þ aij Þ ðxqj x0dj Þiqj0 cosðdj0 di0 þ aij Þ; i 6¼ j,
X
N X
N
f ddii ¼ f ddij ; f qqii ¼ f qqij ;
j¼1;j6¼i j¼1;j6¼i
gddij ¼ yij sinðdj0 di0 þ aij Þ; gqqij ¼ yij cosðdj0 di0 þ aij Þ;
hddij ¼ yij ðxqj x0dj Þ cosðdj0 di0 þ aij Þ; hqqij ¼ yij ðxqj x0dj Þ sinðdj0 di0 þ aij Þ
and f ddij ; f qqij ; gddij ; gqqij ; hddij and hqqij are the elements of matrix Fdd ; Fqq ; Gdd ;
Gqq ; Hdd and Hqq , respectively.
DId ¼ Fd Dd þ Gd DE0q
ð5:37Þ
DIq ¼ Fq Dd þ Gq DE0q
where
Fd ¼ Fdd þ Hdd Fq ; Gd ¼ Gdd þ Hdd Gq ;

1
Fq ¼ ðI Hqq Þ Fqq ; Gq ¼ ðI Hqq Þ1 Gqq
Linearization of Eqs. (5.24) and (5.25) is
Dd_ i ¼ xo Dxi
1
Dx_ i ¼ ðDPi Di Dxi Þ
Mi
0 1 ð5:38Þ
DE_ qi ¼ 0 ðDEqi þ DE0fdi Þ
Tdoi
0 1 KAi
DE_ fdi ¼ DE0 þ ðDVgi þ Dupssi Þ
TAi fdi TAi
Dvdi ¼ xqi Diqi

Dvqi ¼ DE0qi x0di Didi
DPi ¼ vdi0 Diid þ vqi0 Diqi þ idi0 Dvdi þ iqi0 Dvqi ð5:39Þ
DEqi ¼ DE0qi þ ðxdi x0di ÞDidi
vgdi0 vqi0
DVgi ¼ Dvgdi þ Dvgqi
Vgi0 Vgi0
In matrix form, the above equations are
Dd_ ¼ xo IDx
Dx_ ¼ M1 ðDP DDxÞ
0 ð5:40Þ
DE_ q ¼ T01 0
d0 ðDEq þ DEfd Þ
0
DE_ fd ¼ T1 0 1
A DEfd þ TA KA ðDVg þ Dupss Þ
DP ¼ Iq0 DE0q þ ðVd0 Iq0 X0d ÞDId þ ðVq0 þ Id0 Xq ÞDIq

DEq ¼ DE0q þ ðXd X0d ÞDId ð5:41Þ
DVg ¼ V01 0
g0 Vgq0 DEq V01 0
g0 Vgq0 Xd DId þ V01
g0 Vgd0 Xq DIq
where
T
Dx ¼ ½ Dx1 Dx2 ... DxN T ; DE0fd ¼ DE0fd1 DE0fd2 . . . DE0fdN ;
DP ¼ ½ DP1 DP2 ... DPN T ; DEq ¼ ½ DEq1 DEq2 ... DEqN T ;
T T
DVg ¼ ½ DVg1 DVg2 . . . DVgN ; Dupss ¼ ½ Dupss1 Dupss2 . . . DupssN ;
M ¼ diag(Mi Þ; D ¼ diag(Di Þ; T0d0 ¼ diag(T0d0i Þ; TA ¼ diag(TAi Þ; KA ¼ diag(KAi Þ;
Id0 ¼ diag(idi0 Þ; Iq0 ¼ diag(iqi0 Þ; Vd0 ¼ diag(vdi0 Þ; Vq0 ¼ diag(vqi0 Þ; Vg0 ¼ diag(Vgi0 Þ;
X0d ¼ diag(x0di Þ; Xq ¼ diag(xqi Þ; Xd ¼ diag(xdi Þ
DP¼K1 Dd þ K2 DE0q
DEq ¼K3 DE0q þ K4 Dd ð5:42Þ
DVg ¼K5 Dd þ K6 DE0q
By substituting Eq. (5.42) into Eq. (5.40), Heffron–Phillips model of the

N-machine power system with PSS installed can be obtained as
2 3 2 32 3 2 3
Dd_ 0 xo I 0 0 Dd 0
6 Dx_ 7 6
6 7 6 M1 K1 M1 D M1 K2 0 7 6 7 6 7
76 Dx0 7 þ 6 0 7Dupss
6 DE_ 0 7¼4 54 DEq 5 4 0 5
4 q 5 T1
d0 K4 0 T1
d0 K3 T1
d0
DE_
0
T1
A K5 KA 0 T1
A K5 KA T1
A DE0fd T1
A KA
fd
ð5:43Þ
It is in the exactly same format to that of two-machine power system given by

Eq. (5.30). Hence, the model can also be shown in Fig. 5.5.
5.1.2.2 Linearized Model When Full Mathematical Model

of Synchronous Generators Is Used
Linearized full dynamic mathematical model of the ith synchronous generator in the
N-machine power system is (i ¼ 1; 2; . . .; N)
Dw_ di ¼ Dvdi þ rai Didi þ x0 Dwqi þ wqi0 Dxi

Dw_ qi ¼ Dvqi þ rai Diqi x0 Dwdi wdi0 Dxi
Dw_ fi ¼ Dvfi rfi Difi ð5:44Þ
Dw_ Di ¼ rDi DiDi
Dw_ Qi ¼ rQi DiQi
2 3 2 31 2 3
Didi xdi xadi xadi Dwdi
6 7 6 7 6 7
4 Difi 5 ¼ 4 xadi xfi xadi 5 4 Dwfi 5
DiDi xadi xadi xDi DwDi ð5:45Þ
" # " #1 " #
Diqi xqi xaqi Dwqi
¼
DiQi xaqi xQi DwQi
Dd_ i ¼ x0 Dxi
1
Dx_ i ¼ ðDTei þ Di Dxi Þ ð5:46Þ
Mi
DTei ¼ wdi0 Diqi wqi0 Didi þ iqi0 Dwdi idi0 Dwqi
1 KAi
Dv_ 0fi ¼ Dv0 ðDVgi þ Dupssi Þ
TAi fi TAi
vgdi0 vgqi0 ð5:47Þ
DVgi ¼ Dvgdi þ Dvgqi
Vgi0 Vgi0
In matrix form, Eqs. (5.44) to (5.47) can be arranged as
X_ g ¼ Ag Xg þ Bg Dupss þ Bgv DVdq

ð5:48Þ
DIdq ¼ Cg Xg
where
T T
Dwd ¼ Dwd1 Dwd2 . . . DwdN ; Dwq ¼ Dwq1 Dwq2 . . . DwqN ;
Dwf ¼ ½ Dwf1 Dwf2 . . . DwfN T ; DwD ¼ ½ DwD1 DwD2 . . . DwDN T ;
T
DwQ ¼ DwQ1 DwQ2 . . . DwQN ; Dv0f ¼ ½ Dv0f1 Dv0f2 . . . Dv0fN T ;
h iT
Xg ¼ DdT DxT DwTd DwTq DwTf DwTD DwTQ Dv0T f
Dupss ¼ ½ Dupss1 Dupss2 ... DupssN T

DVdq ¼ ½ Dvgd1 Dvgq1 Dvgd2 Dvgq2 ... ... DvgdN DvgqN T
DIdq ¼ ½ Did1 Diq1 Did2 Diq2 ... ... DidN DiqN T
Conversion equations between x–y and di qi coordinate systems as shown in

Fig. 5.2 are (i ¼ 1; 2; . . .; N)

f di sin di cos di f xi f sin di cos di f di
¼ ; xi ¼ ð5:49Þ
f qi cos di sin di f yi f yi cos di sin di f qi
where f i is a phasor (i ¼ 1; 2; . . .; N). Linearization of above equations is

Didi sin di0 cos di0 Dixi ixi0 cos di0 þ iyi0 sin di0
¼ þ Ddi
Diqi cos di0 sin di0 Diyi ixi0 sin di0 þ iyi0 cos di0

Dvgdi sin di0 cos di0 Dvgxi vgxi0 cos di0 þ vgyi0 sin di0
¼ þ Ddi
Dvgqi cos di0 sin di0 Dvgyi vgxi0 sin di0 þ vgyi0 cos di0
ð5:50Þ
In matrix form, above equation can be written as
DIdq ¼ Tg0 DIxy þ BgI Xg

ð5:51Þ
DVdq ¼ Tg0 DVxy þ BgV Xg
where
DVxy ¼ ½ Dvgx1 Dvgy1 Dvgx2 Dvgy2 ... . . . DvgxN DvgyN T

DIxy ¼ ½ Dix1 Diy1 Dix2 Diy2 ... . . . DixN DiyN T
In the x–y coordinate, network equation of Eq. (5.33) can be written as

2 3 2 32 v 3
ix1 gN11 bN11 gN12 bN12 gN1N bN1N gx1
6 iy1 7 6 ... 6 vgy1 7
6 7 6 bN11 gN11 b gN12 b gN1N 7 6 7
6 7 N12 N1N 7 6 7
6 ix2 7 6 gN21 bN21 bN22 bN2N 767 6 v 7
6 7 66
gN22
...
gN2N
76 v 7
gx2
6 iy2 7 6 bN21 g gN2N 76 gy2 7
6 7¼ bN22 gN22 bN2N 7
6 . 7 6 76
N21
6 .. 7 6 . .. .. .. 76 .. 7
6 7 6 .. . . . 7
76 . 7 7
6 7 6
bNNN 54 vgxN 7
6
4 ixN 5 4 gNN1 bNN1 gNN2 bNN2 gNNN 5
...
iyN bNN1 gNN1 bNN2 gNN2 bNNN gNNN vgyN
ð5:52Þ
where N.
yNij ¼ gNij þ jbNij is the elements of network admittance matrix Y
DIxy ¼ YG DVxy ð5:53Þ
where
2 3
gN11 bN11 gN12 bN12 gN1N bN1N
6
6 b gN11 b gN12 b gN1N 7
6 N11 N12 N1N 7
6 gN21 bN21 gN22 bN22 gN2N bN2N 77
6 7
YG ¼ 6
6
bN21 gN21 bN22 gN22 bN2N gN2N 7
7
6 .. .. .. .. 7
6 . . 7
6 . . 7
4 gNN1 bNN1 gNN2 bNN2 gNNN bNNN 5

bNN1 gNN1 bNN2 gNN2 bNNN gNNN
X_ g ¼ Ag Xg þ Bg Dupss þ Bgv ðTg0 DVxy þ BgV Xg Þ

ð5:54Þ
Cg Xg ¼ Tg0 YG DVxy þ BgI Xg
Hence, state equation of the power system is
X_ g ¼ AXg þ Bg Dupss ð5:55Þ
where
A ¼ Ag þ Bgv ½Tg0 ðTg0 YG Þ1 ðCg BgI Þ þ BgV Xg
5.2 Modal Analysis and Control of Power System

Oscillations in a Multi-machine Power System
Installed with Power System Stabilizers
5.2.1 Eigensolution for the Analysis of Power System

Oscillations
5.2.1.1 Participation Factor, Correlation Ratio of Electromechanical

Loop, Modal Shape, and Eigensolution
Modal analysis is a method to study power system small-signal angular stability by

calculating various eigenparameters of state matrix of state-space representation of
power system linearized model of Eqs. (5.43) and (5.55).
In Sect. 2.2.1.1, following definition on eigenvalues ki , corresponding left and
right eigenvectors, vi and wTi , of state matrix A is given
Avi ¼ ki vi ;
ð5:56Þ
wTi A ¼ wTi ki ; i ¼ 1; 2; . . .; M
Also following two matrices are constructed by eigenvectors as

2 3
wT1
6 wT 7
6 27
V ¼ ½ v1 v2 . . . vn ; V1 ¼6 7
6 .. 7 ¼ W
T
ð5:57Þ
4 . 5
wTn
Equations (2.61) and (2.66) establish the following relationship between the state
variables xi ðtÞ; i ¼ 1; 2; . . .; M and modes of the system zi ðtÞ ¼ zi ð0Þeki t ; i ¼
1; 2; . . .; M to be
xk ðt) ¼ vk1 z1 ð0Þek1 t þ vk2 z2 ð0Þek2 t þ þ vkn zn ð0ÞekM t ð5:58Þ

zi ðt) ¼ w1i x1 ðt) þ w2i x2 ðt) þ þ wMi xM ðt) ð5:59Þ
where vki is the kth-row ith-column element of matrix V and wki the ith-row
kth-column element of matrix W.
From Eq. (5.58), it can be seen that the magnitude of vki measures how much the
ith mode zi ðt) ¼ zi ð0Þeki t ; i ¼ 1; 2; . . .; M, contributes to the kth state variable
xk ðt). Thus, jvki j is a kind of measurement of “observability” of the ith mode in the
kth state variable. Equation (5.59) indicates that the magnitude of jwki j measures the
influence of the kth state variable xk ðt) on the ith mode zi ðt) ¼ zi ð0Þeki t ; i ¼
1; 2; . . .; M of the system. It is a kind of measurement of “controllability” of the kth
state variable on the ith mode. Hence, jvki wki j measures how much the ith mode and
the kth state variable are connected. Its normalized value is defined to be the
participation factor
jvki wki j
pki ¼ PM ð5:60Þ
i¼1 jvki wki j
If it is assumed that xk ð0Þ ¼ 1; xj ð0Þ ¼ 0 (j 6¼ k, j ¼ 1; 2; . . .; n), from

Eq. (5.59), it can have
zi ð0Þ ¼ wki ; i ¼ 1; 2; . . .; M ð5:61Þ
By substituting Eq. (5.61) into Eq. (5.58) it can have
X
n
xk ðt) ¼ vki wki eki t ð5:62Þ
i¼1
Equation (5.62) indicates that the participation factor jvki wki j measures how
much the ith mode excited by the kth state variable participates the time response of
the kth state variable. That is why the index defined by Eq. (5.60) is named as the
participation factor.
As it is pointed out previously in Chap. 2, the electromechanical power oscil-
lations are closely related to the rotor motion equation of generators where state
variables are the deviation of rotor position Ddi and speed Dxi ; i ¼ 1; 2; . . .; M in
the state Eqs. (5.43) or (5.55). Hence for a pair of conjugate complex eigenvalue, k i ,
of state matrix, if the associated participation factors of state variables Ddi and
Dxi ; i ¼ 1; 2; . . .; M are much higher than those of other state variables, this means
i is the dominant oscillation mode responsible for the electromechanical oscilla-
k
tion. Hence, ki is an electromechanical oscillation mode. That is, if
5.2 Modal Analysis and Control of Power System … 201
!
p ðxk ¼ DdÞ þ pki ðxk ¼ DxÞ X
ri ¼ Pki 1; pki ¼ 1 ð5:63Þ
k pki ðxk 6¼ Dd and xk 6¼ DxÞ k
i is an electromechanical oscillation mode. ri defined by Eq. (5.63) is called the

k
correlation ratio of electromechanical loop, which measure how much closely the
eigenvalue k i is related to the rotor motion of generators (hence the electrome-
chanical oscillation). ri is often used to identify whether an eigenvalue k i is an
oscillation mode or not.
For the oscillation mode of interests, k i , let its corresponding (right) eigenvector
be vi ¼ ½ v1i v2i . . . vMi . Assume the rotor speed of generators, Dxj
T
(j ¼ 1; 2; . . .; N), to be the (N + 1)th to (2N)th state variables of the system of

Eq. (4.1), i.e.,
Dxj ¼ Dxj þ N ; j ¼ 1; 2; . . .; N ð5:64Þ

2 3 2 3
x1 ðt) v1i
6 .. 7 6 .. 7
6 . 7 6 . 7
6 7 6 7
6 Dx1 7 6 vN þ 1i 7
6 7 6 7
6 Dx2 7 6 vN þ 2i 7
6 7 6 7
6 .. 7 ¼ v1 z1 ð0Þe þ þ 6 .. 7zi ð0Þeki t þ
k1 t
ð5:65Þ
6 . 7 6 . 7
6 7 6 7
6 DxN 7 6 v2Ni 7
6 7 6 7
6 .. 7 6 .. 7
4 . 5 4 . 5
xM ðt) vMi
The (N + 1)th to (2N)th elements of vi , vki ¼ vki \uki ; k = N + 1, N +

2; . . .; 2N describe the mode shape of oscillation mode ki , because vki ¼ vki \u
ki
weighs the contribution of the oscillation mode to the rotor motion of involved
generators. By sketching the phasor diagram of vki ¼ vki \uki , relative positions of
rotor motion of involved generators can be found to determine how generators
participate the power oscillation. For example, let
i ¼ n þ jxi ; vN þ 1i ¼ jvN þ 1i j\0 ; vN þ 2i ¼ jvN þ 2i j\180

k ð5:66Þ
i
Equations (5.65) and (5.66) indicate that the rotor motion of the 1st and 2nd
generator, Dx1 and Dx2 , must have component jvN þ 1i jzi ð0Þeni cos xi t and
jvN þ 2i jzi ð0Þeni cosðxi t þ 180 Þ, respectively. Hence, as far as oscillation mode
i ¼ n þ jxi is concerned, the 1st and 2nd generators oscillate against each
k i
other. Thus, by looking at vki ¼ vki \uki ; k ¼ N þ 1; N þ 2;. . .; 2N, how groups
of generators take part in the oscillation associated with the mode can be
determined.
Procedure of modal analysis from the eigensolution of power system oscillations
normally is
1. Calculation of eigenvalues of state matrix;
2. Identification of electromechanical oscillation modes by checking the imaginary
part of eigenvalues within the range 2pf o ðf o ¼ 0:1 2 HzÞ and computing the
correlation ratio of electromechanical loop;
3. Examining how much each generator is involved in a particular oscillation mode
by calculating the modal participation index;
4. Determining how groups of generators oscillate against each other by computing
the modal shape.
5.2.1.2 Selective Reduced-Order Calculation of Electromechanical

Oscillation Modes
Commonly used numerical method of eigensolution is the QR decomposition. It is a

method to compute all eigenvalues of system state matrix. From above introduction
on the eigensolution for the analysis of power system oscillations, it can be seen
that in fact the majority of computational results of the QR method is redundant,
because only eigenvalues and associated eigenvectors related to the electrome-
chanical oscillation modes are needed. In addition, for a large-scale power system
with a high-dimensional state matrix, numerical difficulty of QR method may occur.
Hence, a lot of effort has been spent by many researchers to propose and develop
numerical methods for the efficient and effective eigensolution of large-scale power
system. The reduced-order modal analysis is one of the methods to carry out the
eigensolution from the reduced-order matrices rather than directly from the system
high-dimensional state matrix so as to avoid the numerical difficulty. The selective
modal analysis is another computational strategy to compute only the eigenvalues
of interests from system state matrix to save computational time and resources. In
this section, a selective reduced-order method of modal computation is introduced
by using Heffron–Phillips model.
Heffron–Phillips model of N-machine power system of Eq. (5.43) can be written
as (without PSS being considered)
sDd ¼ xo Dx
sDx ¼ M1 ðK1 Dd K2 DE0q DDxÞ
ð5:67Þ
sDE0q ¼ T01 0 0
d0 ðK3 DEq K4 Dd þ DEfd Þ
sDE0fd ¼ T1 0 1 0
A DEfd þ TA KA ðK5 Dd K6 DEq Þ
fs2 M þ sD þ xo K1 K2 ½ðI þ sTA ÞðK3 þ sT0d0 Þ þ KA K6 1

ð5:68Þ
½ðI þ sTA ÞK4 þ KA K5 gDd ¼ FðsÞDd ¼ 0
is an electromechanical oscillation mode, it must have

Obviously, if ki
i Þ ¼ 0
DET½Fðk ð5:69Þ
where DET½M ¼ 0 denotes the determinant of matrix M being equal to zero.

Usually, the oscillation modes of interests are poorly damped, close to the
imaginary axis of the complex plan. The angular oscillation frequency of elec-
tromechanical oscillation modes is close to the natural oscillation frequency, which
can be determined from the computation of eigenvalues of matrix xo M1 K1 . In
fact, in the N-machine power system, linearized 2N rotor motion equations of
generators are (with D being ignored)

Dd_ 0 xo I Dd
¼ ð5:70Þ
Dx_ K1 0 Dx
The characteristic equation of the system is

kI 0 xo I ¼ jkI Ar j ¼ 0 ð5:71Þ
K1 0
The equivalent form of Eq. (5.71) is
ðs2 M þ xo K1 ÞDd ¼ ðs2 I þ xo M1 K1 ÞDd ¼ 0 ð5:72Þ
Hence, 2N eigenvalues of system state matrix Ar are also solutions of the

following equation
2
s I þ xo M1 K1 ¼ 0 ð5:73Þ
Therefore, if N eigenvalues of matrix xo M1 K1 are gj ; j ¼ 1; 2;. . .; N, 2N

pffiffiffiffiffi
eigenvalues of matrix Ar are gj ; j ¼ 1; 2; . . .; N. It can be proved that all
eigenvalues of xo M1 K1 are real negative numbers except one zero eigenvalue.
Hence, 2N 2 eigenvalues of Ar are N − 1 pairs of the conjugate imaginary roots
pffiffiffiffiffi
j gj ; j ¼ 1; 2; . . .; N. They are in fact the N − 1 electromechanical oscillation
modes of the power system when only the rotor motion of generators is considered.
pffiffiffiffiffi
gj 6¼ 0; j ¼ 1; 2; . . .; N 1; are the N − 1 angular natural oscillation frequency
of electromechanical oscillation modes.
For a poorly damped oscillation mode of interests, k i ¼ n jxi , because it is
i
close to the imaginary axis of the complex plan and the angular oscillation
Fig. 5.6 Optimal direct

one-dimensional searching y
along the direction of
imaginary and real axis on
Contour of f (λ )
complex plane
λi
0 + j ηi
pffiffiffiffiffi
frequency xi is close to the angular natural oscillation frequency, gi , k i ð0Þ ¼
pffiffiffiffiffi i ¼ n jxi .
0 þ j gi should be a good initial guess of oscillation mode, k i
Therefore, following objective function can be established
fðkÞ ¼ jDET½FðkÞj ð5:74Þ
A direct searching method can be used to find the minimal optimum of above
pffiffiffiffiffi
objective function on the complex plan, starting from 0 þ j gi . That will be the
i ¼ n jxi , which satisfies Eq. (5.74). The direct
oscillation mode of interests, k i
searching can be implemented in the direction of imaginary and real axis iteratively
i ¼ n jxi is very close to 0 þ jpffiffiffiffi
as shown in Fig. 5.6. If k
ffi
gi , k i ¼ ni jxi will
i
be found by applying one-dimensional optimal searching just a couple of times in
each of two directions.
5.2.2 Design of Power System Stabilizers

in Multi-machine Power System
5.2.2.1 Coordinated Design of Multiple Power System Stabilizers
In an N-machine power system, it is assumed that there are L oscillation modes of

interests which need to be damped by installing PSSs in the power system. The first
task of designing PSSs is to determine the effective installing locations of PSSs.
Though nowadays it may be the case that every generator is equipped with a PSS, it
is still necessary to determine how the PSS is designed to effectively damp a
particular oscillation mode. This is an equivalent task to selecting the installing

locations of the PSS.
By computing the participation factors and/or the correlation ratio of elec-
tromechanical loop, it can be decided which generators are most strongly related to
a particular oscillation mode of interests. This is one of the most commonly used
methods to select the installing locations of PSSs to damp the particular oscillation
mode of interests. Without loss of generality, it can be assumed that L generators in
the N-machine power system are selected as the most effective installing locations
of PSSs to damp L oscillation modes of interests. The ith oscillation mode of
interests ki ¼ n jxi is selected to be damped by the PSS installed on the jth
i
generator in the power system, denoted as the jth PSS. In total, L PSSs are to be
designed to damp L oscillation modes. (It may be possible that more than L PSSs
are needed to damp L oscillation modes in some cases.)
A simple way to design L PSSs is to set parameters of L PSSs one by one in a
sequence. For example, method of pole assignment introduced in Sect. 2.2.2.3 can
be used to assign the ith oscillation mode k ¼
i ¼ n jxi to a target position k
i i
n i jx i by setting the parameters of the jth PSS. Afterwards, the (i + 1)th oscil-
lation mode is assigned by designing the (j + 1)th PSS, etc., until L oscillation
modes are assigned to the target positions by setting parameters of L PSSs.
However, this method of sequential design of L PSSs has a problem of “eigenvalue
drifting”, because when the (j + 1)th PSS is designed to assign the (i + 1)th
oscillation mode, the ith oscillation mode which has been assigned may drift away
from the target position. Hence at the end, L PSSs may not assign L oscillation
modes to their target positions at all.
To solve the problem of “eigenvalue drift” associated with the method of
sequential design of multiple PSSs, parameters of L PSSs can be set to move L
oscillation modes to the target positions simultaneously. This strategy of simulta-
neous tuning of parameters of all PSSs is also referred to as the coordinated design
of multiple PSSs. So far, there have been many methods of simultaneous tuning or
coordinated design of PSSs being proposed and developed. The majority of those
proposed methods adopt a certain objective function,
fðpÞ; p 2 P ð5:75Þ
where p is the vector of parameters of L PSSs and P is the parameter space of L

PSSs. A method of optimization can be used to find the optimum of objective
function, which is the solution of coordinated design of L PSSs.
So far, methods of optimization for the coordinated design of multiple PSSs can
be classified into two categories: (1) mathematical methods and (2) heuristic
methods. Mathematical methods include linear programming, nonlinear program-
ming and other mathematical algorithms of optimization. Examples of heuristic
methods are genetic algorithm, simulated annealing, and particle swarm algorithm.
Those methods have all been tried for optimally tuning parameters of multiple PSSs
to effectively damp power system oscillations. Application of mathematical methods
must have a computable objective function and sometimes is constrained by the

problem of local optimum, while heuristic methods can handle non-computable
objective function and are more effective to find the global optima.
In the following section, a mathematical method is introduced to tune parameters
of multiple PSSs installed in a multi-machine power system on the basis of
Heffron–Phillips model. In Sect. 7.4.1, a heuristic method to tune parameters of
multiple stabilizers added on a UPFC will be introduced.
5.2.2.2 Parameter Tuning Algorithm for the Design of Multiple Power

System Stabilizers
Let the objective function to simultaneously tune parameters of L PSSs to assign L

¼ n jx , i ¼ 1; 2; . . .; L be
i ¼ n jxi , to target positions k
oscillation modes, k i i i i
X
L

fðpÞ ¼ i ðpÞk
½k ðp Þ2 ð5:76Þ
i
i¼1
The minimal optimum p 2 P is when the objective function is equal to zero and
L oscillation modes are moved to the target positions. If the deepest decent method
in nonlinear programming is used, the optimal searching algorithm is
pðk þ 1Þ ¼ pðkÞ st rf½pðkÞ ð5:77Þ
where st is the optimal one-dimensional searching length and rf½pðkÞ is the

searching decent direction of objective function at the (k + 1)th step. A selective
modal algorithm to implement the deepest decent searching can be developed as
follows.
Heffron–Phillips model of N-machine power system of Eq. (5.43) can be written
as (with PSSs being considered)
sDd ¼ xo IDx
ð5:78Þ
sDE0q ¼ T01 0 0
sDE0fd ¼ T1 0 1 0
A DEfd þ TA KA ðK5 Dd K6 DEq þ Dupss Þ

½ðI þ sTA ÞK4 þ KA K5 gDd þ K2 ½ðI þ sTA ÞðK3 þ sT0d0 Þ ð5:79Þ
1
þ KA K6 KA Dupss ¼ 0
Let the transfer function of the jth PSS be Gj ðsÞ and the feedback signal Dxj .
Dupss can be denoted as
s
Dupss ¼ Gpss ðp; sÞDd ð5:80Þ
x0
where Gpss ðp; sÞ ¼ diagðGpssj Þ, Gpssj ¼ Gj ðs) when the jth generator is installed
with a PSS; otherwise, Gpssj ¼ 0. From Eqs. (5.79) and (5.80), it can have

s ð5:81Þ
½ðI þ sTA ÞK4 þ KA K5 KA Gpss ðp; sÞDdgDd ¼ Fðp; sÞDd ¼ 0
x0
Hence, it must have
DET½Fðp; ki Þ ¼ 0 ð5:82Þ
Equation (5.82) indicates that there should exist a nonzero vector vi and w
i
satisfying
Fðp; ki Þvi ¼ 0
ð5:83Þ
wTi Fðp; ki Þ ¼ 0
Let p ¼ ½ p1 p2 . . . ps T . A small increment of pk , Dpk , will cause that of ki

and
vi , Dki and Dvi . It can have
F½pðpk þ Dpk Þ; ki þ Dki ðvi þ Dvi Þ ¼ 0 ð5:84Þ
Because
@Fðp; ki Þ @Fðp; ki Þ
F½pðpk þ Dpk Þ; ki þ Dki ¼ Fðp; ki Þ þ Dki þ Dpk ð5:85Þ
@ki @pk
@ki Ti @Fðp;k
w iÞ
@ pk vi

¼ ð5:86Þ
@pk T
w @Fðp;k i Þ
vi
i @ki
@ki
rf½pðkÞ ¼ 2ðki k i Þ ð5:87Þ
@pðkÞ
h iT
@ki @ki @ki @ki
where @pðkÞ
¼ @ p1 @ p2 @ ps and @@k
pk (k ¼ 1; 2; . . .; s) is calculated by
i
using Eq. (5.86).

When the deepest decent algorithm of Eqs. (5.77) and (5.87) is implemented, it
needs to compute the oscillation mode of interests, ki , at each step of optimal
searching. Because it is an iterative algorithm, at the (k + 1) step, ki ðk þ 1Þ is very
close to ki ðkÞ and it should have DETF½pðk þ 1Þ; ki ðk þ 1Þ ¼ 0. Hence, set up the
following objective function,
f k þ 1 ðki Þ ¼ jDETF½p; ki j ð5:88Þ
The direct searching method introduced based on Eq. (5.74) can be used to find
ki ðk þ 1Þ on the complex plane, starting from ki ðkÞ.
In the multi-machine power system, objective of installation of multiple PSSs is
to ensure sufficient damping to be provided to the oscillation modes of interests.
Hence, pole assignment to the exact target positions, such as that to be achieved by
the objective function of Eq. (5.76), is usually not necessary. A more relaxed
objective function can be used for the coordinated design of multiple PSSs. For
example, following objective function
fðpÞ ¼ minðfi Þ
0:05 ð5:89Þ
where
ki ¼ ni jxi 2 fall oscillation modes to be damped by the PSS designg;

fi
fi ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ni þ x2i
2
is for the coordinated design of multiple PSSs to ensure the damping of all oscil-
lation modes to be greater than .05. For the objective function of Eq. (5.89), some
heuristic methods of optimization, such as the genetic algorithm, can be used.
Strategies of sequential setting and simultaneous tuning for the design of mul-
tiple PSSs can also be jointly used to achieve more efficient and practical design.
For example, it can be assumed that in the N-machine power system, the jth PSS
provides a pure damping torque, Dpssj Dxj (j ¼ 1; 2; . . .; L), to the electromechanical
oscillation loop of the jth generator. With this assumption, Eq. (5.81) becomes
fs2 M þ sðD þ Dpss Þ þ xo K1 K2 ½ðI þ sTA ÞðK3 þ sT0d0 Þ þ KA K6 1

ð5:90Þ
½ðI þ sTA ÞK4 þ KA K5 gDd ¼ Fðp; sÞDd ¼ 0
where
Dpss ¼ diag(dpssj Þ; dpssj ¼ Dpssj
when the jth generator is installed with a PSS; otherwise, dpssj ¼ 0 ðj ¼ 1; 2; . . .; L),
T
p ¼ Dpssj Dpssðj þ 1Þ . . . Dpssðj þ LÞ .
The method of coordinated design of PSSs introduced from Eqs. (5.76) to (5.88)
above can be used to set the damping torque provision from each PSS for the
multiple PSSs to damp all the oscillation modes of interests. Afterwards, each PSS
can be designed individually by using the phase compensation method to ensure
that the right amount of damping torque, Dpssj Dxj (j ¼ 1; 2; . . .; L), is supplied to
each generator.
5.2.3 Fixed Modes Associated with PSS Control
A PSS is a local controller, and hence, an N-machine power system with L PSSs
installed forms a decentralized control system. Fixed modes are an important
concept about the controllability and observability of a decentralized control sys-
tem, a natural generalization of the well-known concept of uncontrollable and
unobservable modes in centralized control systems. In this section, a mathematical
proof is presented to show that in the decentralized control of N-machine power
system by PSSs, any electromechanical oscillation mode is not a fixed mode. This
means that in the N-machine power system, any oscillation mode can be damped by
the appropriate design of multiple PSSs. In other words, in the parameter space of
coordinated design of multiple PSSs, the optimum always exists if the objective
function is appropriately set. Theoretical proof given below is based on the
Heffron–Phillips model of N-machine power system installed with multiple PSSs.
Let the realization of the jth PSS, Dupssj ¼ Gj ðsÞDxj , be
Dupssj ¼ Hj zj ðtÞ þ kj Dxj

ð5:91Þ
z_ j ðtÞ ¼ Fj zj ðtÞ þ Sj Dxj
Hence, it can have
Dupss ¼ HzðtÞ þ Ky
ð5:92Þ
z_ ðtÞ ¼ FzðtÞ þ Sy
where
T T
zðtÞ ¼ z1 ðtÞT z2 ðtÞT zL ðtÞT ; y ¼ ½ Dxj Dxj þ 1 Dxj þ L ;
H ¼ blockdiagðHj Þ; F ¼ blockdiagðFj Þ; S ¼ blockdiagðSj Þ; K = diag(kj Þ
and blockdiagðMj Þ denotes a block diagonal matrix with the diagonal matrix to be Mj .
From Eq. (5.92), state equation of Heffron–Phillips model of N-machine power
system with L PSSs installed of Eq. (5.43) can be written as
_ ¼ AxðtÞ þ BDupss
xðtÞ
ð5:93Þ
y ¼ CxðtÞ
where
2 3
0 xo I 0 0
6 7
6 M1 K1 M1 D M1 K2 0 7
A¼6
6 T1 K
7;
7
4 d0 4 0 T1
d0 K3 T1
d0 5
T1
A K5 KA 0 T1
A K5 KA T1
A
2 3
0
6 7
6 0 7
6
B¼6 7; C ¼ ½ 0 I 0 0
7
4 0 5
T1
A KA
The set of fixed modes of the decentralized control system of Eqs. (5.92) and
(5.93), denoted as KðA; B; C; KÞ, is defined as
KðA; B; C; KÞ ¼ \ 0 k½A þ BKC ð5:94Þ

K2K
where k½A þ BKC denotes the set of all eigenvalues of matrix A þ BKC, and K0 is
the following set
K0 ¼ fKjK ¼ diagðkj Þ; K 2 RNN g ð5:95Þ
Definition of the fixed modes given by Eqs. (5.94) and (5.95) in fact means that
the fixed modes are not affected by variations of any kj . This is the key in the
following proof.
2 3
0 xo I 0 0
6 M1 K1 M1 D M1 K2 0 7
A þ BKC ¼ 6
4 T1 K4
7
5 ð5:96Þ
d0 0 T1
d0 K3 T1
d0
T1
A K5 KA
1 1
TA KA K TA K5 KA TA 1
If ko is a fixed mode, it must be the eigenvalue of matrix A þ BKC. Hence,

2 32 3 2 3
0 xo I 0 0 v1 v1
6 M K1 1
M 1
D M 1
K 0 7 6 7 6 7
6 2 76 2 7 ¼ ko 6 v2 7 ð5:97Þ
v
4 T1 K4 0 1
Td0 K3 1 54
Td0 v3 5 4 v3 5
d0
1 1 1 1
TA K5 KA TA KA K TA K5 KA TA v4 v4
That is,
xo v2 ¼ ko v1
K1 v1 Dv2 K2 v3 ¼ ko Mv2
ð5:98Þ
K4 v1 K3 v3 þ v4 ¼ ko Td0 v3
K5 KA v1 KA Kv2 K6 KA v3 v4 ¼ ko TA v4

KA K
A4 k4o þ A3 k3o þ A2 k2o þ A1 k o þ A 0 v1 ¼ 0 ð5:99Þ
xo
where
A0 ¼ ðK3 þ KA K6 ÞK1
2 K1 K4 KA K5
1
A1 ¼ ðTA K3 þ Td0 ÞK1
2 K1 þ ðK þ KA K6 ÞK1
2 D TA K4
x0 3
1 1
A2 ¼ ðTA K3 þ Td0 ÞK1 1
2 D þ TA Td0 K2 K1 þ ðK þ KA K6 ÞK1
2 M
x0 x0 3
1 1
A3 ¼ TA Td0 K1
2 Dþ ðT K3 þ Td0 ÞK1
2 M
x0 x0 A
1
A4 ¼ TA Td0 K1
2 M
x0
v 6¼ 0; hence, v1 6¼ 0. It should have

KA K
DET A4 k4o þ A3 k3o þ A2 k2o þ A1 ko þ A0 ¼ 0 ð5:100Þ
xo
Denote

KA K
Fðko ; k1 ; k2; . . .kN Þ ¼ A4 k4o þ A3 k3o þ A2 k2o þ A1 ko þ A0
xo
2 3
f 11 ðko ; k1 Þ f 12 ðko Þ ... f 1N ðko Þ
6 f 21 ðko Þ f 22 ðko ; k2 Þ . . . f 2N ðko Þ 7 ð5:101Þ
6 7
¼6
6 .. .. .. ..
7
7
4 . . . . 5
f N1 ðko Þ f N2 ðko Þ ... f NN ðko ; kN Þ
Also denote
2 3
f 11 ðko ; k1 Þ f 12 ðko Þ ... f 1N ðko Þ
6 f 21 ðko Þ f 22 ðko ; k2 Þ . . . f 2N ðko Þ 7
6 7
Fk ðko ; k1 ; k2 ; . . .kN Þ ¼ 6 .. .. .. .. 7 ð5:102Þ
4 . . . . 5
f N1 ðko Þ f N2 ðko Þ ... f NN ðko ; kN Þ
Obviously, FN ðko ; k1 ; k2 ; . . .kN Þ ¼ Fðko ; k1 ; k2 ; . . .kN Þ. Let the companion

matrix of Fk ðko ; k1 ; k2; . . .kk Þ be F k ðko ; k1 ; k2; . . .kk Þ and denote
2 3
f k;11 f k;12 ... f k;1k
6 f k;21 f k;22 ...
f k;2k 7
6 7
Fk ðko ; k1 ; k2; . . .kk Þ ¼ 6 . .. .. .. 7 ð5:103Þ
4 .. . . . 5
f k;k1 f k;k2 ... f k;kk
f k;kk ¼ DET½Fk1 ðko ; k1 ; k2; . . .kk1 Þ ð5:104Þ
Denote
gk ðko ; k1 ; k2; kk Þ ¼ DET½Fk ðko ; k1 ; k2; kk Þ ð5:105Þ
When k ¼ N, from Eqs. (5.100) and (5.105) it can have
gN ðko ; k1 ; k2 ; . . .kN Þ ¼ DET½FN ðko ; k1 ; k2 ; . . .kN Þ

ð5:106Þ
¼ DET½Fðko ; k1 ; k2 ; . . .kN Þ ¼ 0
Because ko is a fixed mode,
@ko
¼ 0ðj ¼ 1; 2; . . .; NÞ ð5:107Þ
@kj

@gN @ko @g
þ N¼0 ð5:108Þ
@ko @kN @kN
@gN
¼0 ð5:109Þ
@kN
For any matrix MðxÞ,
dDET½MðxÞ dMðxÞT
¼ Trace½M ðxÞ ð5:110Þ
dx dx
where M ðxÞ is the companion matrix of MðxÞ; thus,
@gN @FN ðko ; k1 ; k2 ; . . .kN ÞT

¼ Trace½F N ðko ; k1 ; k2 ; . . .kN Þ ¼0 ð5:111Þ
@kN @kN
@FN ðko ; k1 ; k2 ; kN ÞT h i
@ f NN ðko ;kN Þ
¼ diag 0 0 0
@kN @ kN
ð5:112Þ
h i
ko
¼ diag 0 0 0 x0 KNA
Hence from Eqs. (5.103), (5.111), and (5.112), it can have
@gN ko
¼ KNA f N;NN ¼ 0 ð5:113Þ
@kN x0
KNA 6¼ 0 and if ko is an electromechanical oscillation mode, ko 6¼ 0. Hence from

Eqs. (5.113), (5.104), and (5.105), it can have
f N;NN ¼ DET½FN1 ðko ; k1 ; k2 ; . . .kN1 Þ ¼ gN1 ðko ; k1 ; k2 ; . . .kN1 Þ ¼ 0 ð5:114Þ
From Eq. (5.114), it should have
@gN1 @ko @g
þ N1 ¼ 0 ð5:115Þ
@ko @kN1 @kN1
From Eqs. (5.101), (5.107), and (5.110), it can be obtained

" #
@gN1 @FN ðko ; k1 ; k2 ; . . .kN1 ÞT
¼ Trace FN1 ðko ; k1 ; k2 ; . . .kN1 Þ
@kN1 @kN1
n h io
¼ Trace F N1 ðko ; k1 ; k2 ; . . .kN1 Þdiag 0 0 . . . kko0 KðN1ÞA
@o @o
¼ KðN1ÞA f ðN1Þ;ðN1ÞðN1Þ ¼ KðN1ÞA gN2 ð@o ; k1 ; k2 ; . . .kN2 Þ ¼ 0
@0 @0
ð5:116Þ
Again, KðN1ÞA 6¼ 0 and ko 6¼ 0, it can have
gN2 ðko ; k1 ; k2 ; . . .kN2 Þ ¼ DET½FN2 ðko ; k1 ; k2 ; . . .kN2 Þ ¼ 0 ð5:117Þ
Above procedure from Eqs. (5.106) to (5.117) can be carried on iteratively until
it is obtained that
@g1 ko
¼ K1A ¼ 0 ð5:118Þ
@k1 x0
It contradicts to that K1A 6¼ 0 and ko 6¼ 0. Hence, ko cannot be a fixed mode and

an electromechanical oscillation mode.
5.3 An Example Three-Machine Power System
5.3.1 Example Power System and Its Linearized

Heffron–Phillips Model
5.3.1.1 System Parameters and Operating Conditions
Configuration of a three-machine six-node example power system is shown in

Fig. 5.7 [1]. Parameters of generators and the AVRs are given in Table 5.1.
Parameters of transformers and lines are given in Table 5.2[1]. Initial conditions of
example power system for the load flow computation are
VG1 ¼ 1:04\0; VG2 ¼ 1:025\0; VG3 ¼ 1:025\0;

V1;2;3 ¼ 1:00\0; VA;B;C ¼ 1:00\0;
SG1 ¼ 0:716 þ j0:27; SG2 ¼ 1:63 þ j0:067; SG3 ¼ 0:85 j0:109
LoadA ¼ 1:25 þ j0:5; LoadB ¼ 0:9 þ j0:3; LoadC ¼ 1 þ j0:35:
Tables 5.3 and 5.4 are the results of the load flow computation.
5.3 An Example Three-Machine Power System 215
0.716+j0.27 BusC 1.63 + j0.067

Bus1 1.25 + j0.5 Bus3
G1 G3
1.04∠0
1.025∠0
BusB BusC
0.9 + j0.3 1 + j0.35
Bus2
0.85 − j0.109
G2
1.025∠0
Fig. 5.7 Configuration of a three-machine example power system [1]
Table 5.1 Parameters of G1 G2 G3

generators and the AVRs [1,
2] xd 0.146 0.8958 1.313
x0d 0.0608 0.1189 0.1813
T0d0 8.96 6 5.89
xq 0.0969 0.8645 1.258
x0q 0.0969 0.1969 0.25
M 47.2 12.8 6.02
D 0 0 0
KA 200 200 200
TA 0.02 0.02 0.02
5.3.1.2 Calculation of Initial Values of State Variables
From active and reactive power output of each generator, Pi and Qi ; i = 1,2,3, and
gi ; i = 1,2,3, output current of each generator can be calcu-
its terminal voltage, V
lated as
Table 5.2 Parameters of transformers and lines [1]

Node Node Resistance Reactance Susceptance
1 A 0.01 0.085 0.088
1 B 0.017 0.092 0.079
A 2 0.032 0.161 0.153
B 3 0.039 0.17 0.170
2 C 0.085 0.072 0.0745
C 3 0.0119 0.1008 0.1045
G1 1 0 0.0576 0
G2 2 0 0.0625 0
G3 3 0 0.0586 0
Table 5.3 Results of load flow computation—node voltage

Node Magnitude of voltage Phase of voltage (deg.)
G1 1.04 0.000
G2 1.025 0.163
G3 1.025 0.081
1 1.018 −0.039
A 0.977 −0.070
B 1.007 −0.065
C 1.012 0.012
2 1.021 0.065
3 1.030 0.034
Table 5.4 Results of load flow computation—active and reactive power

Starting Ending Active Reactive Active Reactive
node (SN) node (EN) power at power at SN power at power at EN
SN EN
4 5 0.412 0.411 −0.409 −0.468
4 6 0.306 0.036 −0.304 −0.108
5 8 −0.841 −0.117 0.865 0.084
6 9 −0.596 −0.061 0.609 −0.065
8 7 0.765 0.018 −0.760 −0.054
7 9 −0.240 −0.204 0.241 0.103
G1 4 0.718 0.396 −0.718 −0.360
G2 8 1.630 0.142 −1.630 0.017
G3 9 0.850 −0.072 −0.850 0.112
Igi ¼ Pi jQi ; i ¼ 1; 2; 3

V gi
Qi ; i ¼ 1; 2; 3, can be computed as
The imaginary voltage of each generator, E
Qi ¼ Vgi þ jxqiIgi ;
E i ¼ 1; 2; 3
Phase of E Qi ; i ¼ 1; 2; 3 is the angular position of each generator, di , i ¼ 1; 2; 3.

From di , i ¼ 1; 2; 3, computational results of output current and terminal voltage of
each generator can be transformed from the common x–y coordinate to the d–q
coordinate as (Eq. (5.49))

idi sindi cosdi vgxi vdi sindi cosdi vgxi
¼ ¼
iqi cosdi sindi vgyi vqi cosdi sindi vgyi
According to Eq. (5.25), it can have
E0qi ¼ vqi þ x0di idi

Eqi ¼ E0qi þ (xdi x0di Þidi
Since E0fdi0 ¼ 0, from the third equation in Eq. (5.24) it can have
Efd0i = Eqi0
Computational results of all the above are presented in Table 5.5.
5.3.1.3 Linearized Heffron–Phillips Model
From Table 5.2 and Fig. 5.7, following network admittance matrix can be
established.

12
L ¼ Y11
Y
Y
21
Y 22
Y
Table 5.5 Initial value of G1 G2 G3

state variables
di0 (rad.) 0.0621 1.0388 0.9199
xi0 1 1 1
E0qi0 1.0637 0.8119 0.7882
E0fqi0 1.0998 1.8302 1.4334
where
2 3
j17:3611 0 0
11
Y ¼4 0 j16 0 5;
0 0 j17:0648
2 3
j17:3611 0 0 0 0 0
12
Y ¼ 4 0 0 0 0 j16 0 5;
0 0 0 0 0 j17:0648
2 3
j17:3611 0 0
6 0 0 0 7
6 7
6 7
21 6 0 0 0 7
Y ¼6 7;
6 0 0 0 7
6 7
4 0 j16 0 5
0 0 j17:0648
2 3
3:3074 j39:3089 1:3652 þ j11:6041 1:9422 þ j10:5107
6 1:3652 þ j11:6041 3:8620 j18:0714 0 7
6 7
6 1:9422 þ j10:5107 4:1115 j16:1367 7
221 6 0 7
Y ¼6 7;
6 0 0 0 7
6 7
4 0 1:1876 þ j 5:9751 0 5
0 0 1:2820 þ j5:5882
2 3
0 0 0
6 0 1:1876 þ j5:9751 0 7
6 7
6 1:2820 þ j5:5882 7
222 6 0 0 7
Y ¼6 7;
6 3:7482 j23:6449 1:6171 þ j13:6980 1:1551 þ j9:7843 7
6 7
4 1:6171 þ j13:6980 2:8047 j35:4456 0 5
1:1551 þ j9:7843 0 2:4371 j32:1539
22 ¼ ½Y
Y 222
221 Y
By keeping three nodes of generator terminals via deleting all the other nodes in
the above network admittance matrix, it can have
N ¼ Y
Y 11 Y 1 Y
12 Y
22 21
2 3
1:0994 j4:7578 0:0954 þ j2:2197 0:0057 þ j2:2568
6 7
¼ 4 0:0954 þ j2:2197 0:7374 j5:1374 0:1249 þ j2:8146 5
0:0057 þ j2:2568 0:1249 þ j2:8146 0:7236 j5:0289
From Eq. (5.34), it can be obtained

¼ Y
Y 1 þ jx0
N d
2 3
0:8312 j3:0501 0:2806 þ j1:4829 0:2054 þ j1:2048
6 7
¼ 4 0:2806 þ j1:4829 0:4192 j2:7499 0:2119 þ j1:0795 5
0:2054 þ j1:2048 0:2119 þ j1:0795 0:2759 j2:3757
2 3
3:1614\1:3047 1:5092\1:3838 1:2221\1:4019
6 7
¼ 4 1:5092\1:3838 2:7816\1:4195 1:1001\1:3770 5
1:2221\1:4019 1:1001\1:3770 2:3917\1:4552
Thus according to Eq. (5.37), it can have

2 3 2 3
0:1450 0:1484 0:0034 2:8414 1:5907 1:2507
6 7 6 7
Fdd ¼ 4 1:4886 2:4386 0:9500 5; Fqq ¼ 4 0:6023 1:2064 0:6042 5;
1:1275 0:8120 1:9395 0:6478 0:8347 1:4825
2 3 2 3
3:0501 1:0626 0:9434 0:8312 1:0718 0:7770
6 7 6 7
Gdd ¼ 4 0:5975 2:7499 1:0467 5; Gqq ¼ 4 1:3859 0:4192 0:3384 5;
0:6326 1:0970 2:3757 1:0457 0:0823 0:2759
2 3 2 3
0:0300 0:7991 0:8366 0:1101 0:7923 1:0157
6 7 6 7
Hdd ¼ 4 0:0500 0:3126 0:3643 5; Hqq ¼ 4 0:0216 2:0503 1:1270 5:
0:0377 0:0614 0:2971 0:0228 0:8179 2:5579
From Eq. (5.38), following matrices are calculated

2 3 2 3
0:3252 0:0095 0:3157 2:1589 1:3566 0:8023
6 7 6 7
Fd ¼ 4 1:5477 2:4103 0:8626 5; Fq ¼ 4 0:2670 0:3235 0:0565 5;
1:1306 0:8935 2:0241 0:2296 0:1689 0:3985
2 3 2 3
2:2175 1:2518 1:1666 1:6110 0:8094 0:4956
6 7 6 7
Gd ¼ 4 0:1557 2:7758 0:9860 5; Gq ¼ 4 0:6318 0:1512 0:1474 5:
0:3995 1:1026 2:3982 0:4495 0:0527 0:1083
Thus according to Eqs. (5.41) and (5.42), it can have

0
K1 ¼ ðVd Iq Xd ) Fd þ ðVq þ Id Xq Þ Fq ;
0
K2 ¼ Iq þ ðVd Iq Xd ) Gd þ ðVq þ Id Xq Þ Gq ;
0
K3 ¼ I þ (Xd Xd ) Gd ;
0
K4 ¼ ðXd Xd ) Fd ;
0
K5 ¼ V1 1
gen Vq Xd Fd þ Vgen Vd Xq Fq ;
0
K6 ¼ V1 1 1
gen Vq Vgen Vd Xd Gd þ Vgen Vd Xq Gq
Coefficient matrices of Heffron–Phillips model of example power system as

given by Eq. (5.43) are obtained to be
2 3 2 3
2:3373 1:4640 0:8733 2:4574 0:9034 0:5628
6 7 6 7
K1 ¼ 4 1:5290 2:2160 0:6870 5; K2 ¼ 4 1:0247 3:0671 0:4060 5;
1:0599 0:8201 1:8800 0:3694 0:6458 2:3234
2 3 2 3
1:1889 0:1067 0:0994 0:0277 0:0008 0:0269
6 7 6 7
K3 ¼ 4 0:1209 3:1565 0:7660 5; K4 ¼ 4 1:2024 1:8725 0:6701 5;
0:4521 1:2478 3:7141 1:2795 1:0112 2:2907
2 3 2 3
0:0068 0:0076 0:0143 0:8732 0:0711 0:0678
6 7 6 7
K5 ¼ 4 0:0596 0:0315 0:0281 5; K6 ¼ 4 0:4315 0:5292 0:1730 5:
0:0780 0:0499 0:1278 0:4691 0:1829 0:4790
2 3 2 3
47:2 0 0 0 0 0
6 7 6 7
M¼4 0 12:8 0 5; D ¼ 4 0 0 0 5;
0 0 6:02 0 0 0
2 3 2 3
200 0 0 0:02 0 0
6 7 6 7
KA ¼ 4 0 200 0 5; TA ¼ 4 0 0:02 0 5;
0 0 200 0 0 0:02
2 3
8:96 0 0
0 6 7
Td0 ¼4 0 6:00 0 5:
0 0 5:89
5.3.2 Modal Analysis of Example Power System
5.3.2.1 Electromechanical Oscillation Modes of Example Power

System
Eigenvalues of state matrix of Heffron–Phillips model obtained in the above section

are calculated as
1;2 ¼ 25:0316 j28:2289

k
k3 ¼ 34:2648
4;5 ¼ 25:0343 j9:5156
k
6 ¼ 13:5354
k
7;8 ¼ 1:5051 j9:1860
k
9;10 ¼ 0:1736 j7:0590
k
11;12 ¼ 0:0000 j0:0000
k
Frequency of k 7;8 , and k

4;5 , k 9;10 is 9:5156 ¼ 1:5145 Hz, 9:1860 ¼ 1:4620 Hz, and
2p 2p
7:0590
2p ¼ 1:1235 Hz; respectively, falling within the range of power system
low-frequency oscillation 0.1–2 Hz. They can be identified to be the electrome-
chanical oscillation modes of the example power system. To confirm the identifi-
cation, participation factor and the correlation ratio of electromechanical loop can
be calculated for k 4 ¼ 25:0343 þ j9:5156, k 7 ¼ 1:5051 þ j9:1860, and k 9 ¼
0:1736 þ j7:0590 as follows.
Right and left eigenvectors corresponding to k 4 ¼ 25:0343 þ j9:5156 are
calculated as
2 3 2 3
0:6852 j17:3385 j0:0001
6 0:6595 þ j5:9033 7 6 0:0002 þ j0:0003 7
6 7 6 7
6 1:3447 þ j11:4352 7 6 0:0001 þ j0:0001 7
6 7 6 7
6 64:7503 þ j192:9710 7 6 0 7
6 7 6 7
6 31:8356 j61:9802 7 6 0 7
6 7 6 7
6 32:9147 130:9908 7 6 0 7
6
7 ¼ 6 7 6
7 ¼ 6 7
v
1:9362 396:4038 7; w 0:0020 þ j0:0008 7
6 7 6 7
6 2:5524 þ j88:6491 7 6 0:0044 j0:0017 7
6 7 6 7
6 2:5028 þ j50:7603 7 6 0:0024 j0:0010 7
6 7 6 7
6 0:5973 j1:5444 7 6 0:5202 j0:0006 7
6 7 6 7
4 0:18208 þ j0:5224 5 4 0:7505 þ j0:0000 5
0:12978 þ j0:2957 0:4076 þ j0:0030
By use of Eq. (5.60), participation factors corresponding to each of state vari-

ables are calculated as
p41 ¼ 0:0017; p42 ¼ 0:0022; p43 ¼ 0:0018:
Thus, the correlation ratio of electromechanical loop can be calculated by using

Eq. (5.63) as r4 ¼ 0:0040, which is smaller than 1. Hence, k 4 ¼ 25:0343 þ
j9:5156 is not an electromechanical oscillation mode.
Right and left eigenvectors corresponding to k 7 ¼ 1:5051 þ j9:1860 are cal-
culated as
2 3 2 3
1:0451 þ j2:2184 0:0007 j0:0011
6 2:2567 þ j5:5000 7 6 0:0073 j0:0198 7
6 7 6 7
6 3:3018 j7:7184 7 6 0:0252 þ j0:0502 7
6 7 6 7
6 68:1811 j46:9144 7 6 0 7
6 7 6 7
6 170:8669 j105:1733 7 6 0:0005 þ j0:0003 7
6 7 6 7
6 239:0480 þ j152:08774 7 6 0:0013 j0:0010 7
7 ¼ 6
v 7; w
¼ 6
6 0:1000 þ j2:5611 7 7 6 0:0005 j0:0001 7
7
6 7 6 7
6 2:449 þ j6:2889 7 6 0:0039 þ j0:0037 7
6 7 6 7
6 3:5529 j8:2378 7 6 0:0042 j0:0188 7
6 7 6 7
6 0:0009 þ j0:0057 7 6 0:0017 þ j0:0447 7
6 7 6 7
4 0:0042 þ j0:0224 5 4 0:1917 þ j0:1403 5
0:0067 j0:0301 0:9683

p71 ¼ 0:1427; p72 ¼ 0:3021; p73 ¼ 0:0618:

Eq. (5.63) as r7 ¼ 4:8776, which is greater than 1. Hence, k 7 ¼
1:5051 þ j9:1860 is an electromechanical oscillation mode.
9 ¼ 0:1736 þ j7:0590
Similarly, right and left eigenvectors corresponding to k
are calculated as
2 3 2 3
0:0830 þ j2:4518 0:0018 j0:0581
6 0:0192 j1:8299 7 6 0:0085 þ j0:1649 7
6 7 6 7
6 0:0638 j0:62186 7 6 0:0082 þ j0:0985 7
6 7 6 7
6 108:9599 j6:3716 7 6 0:0013 þ j0:0001 7
6 7 6 7
6 81:3708 þ j2:8540 7 6 0:0037 j0:0003 7
6 7 6 7
6 27:5890 þ j3:5176 7 6 0:0022 j0:0002 7
9 ¼ 6
v 7; w
¼ 6
6 0:6337 þ j0:5454 7 9 6 0:0015 þ j0:0016 7
7
6 7 6 7
6 0:9340 j0:2801 7 6 0:0066 j0:0195 7
6 7 6 7
6 0:2936 j0:0842 7 6 0:0030 j0:0134 7
6 7 6 7
6 0:0012 þ j0:0014 7 6 0:0973 þ j0:0918 7
6 7 6 7
4 0:0029 j0:0014 5 4 0:8031 5
0:0009 j0:0004 0:5441 j0:0031i

p91 ¼ 0:0031; p92 ¼ 0:1254; p93 ¼ 0:4717:

Eq. (5.63) as q9 ¼ 34:3291, which is greater than 1. Hence, k 9 ¼ 0:1736 þ
j7:0590 is also an electromechanical oscillation mode.
According to Eq. (5.66), mode shape of electromechanical oscillation mode can
be identified from the elements of right eigenvector corresponding to
Dxj ; j ¼ 1; 2; 3, which is the 4th, 5th, and 6th element of right eigenvector,
respectively. For k7 ¼ 1:5051 þ j9:1860, the elements of right eigenvector are
v74 ¼ 0:0000 þ j0:0000 ¼ 0\42:2017

75 ¼ 0:0005 þ j0:0003 ¼ 0:0006\29:5912
v
76 ¼ 0:0013 j0:0010 ¼ 0:0017\144:0796
v
The mode shape is illustrated in Fig. 5.8. It shows that the electromechanical
oscillation associated with this mode is that of G3 against G2.
-4
x 10
4
G2
2
-2
-4
-6
-8
G3
-10
-12 -10 -8 -6 -4 -2 0 2 4 6
-4
x 10
7 ¼ 1:5051 þ j9:1860
Fig. 5.8 Mode shape of k
x 10 -3
4
G1
0
G2 G3
-1
-2
-3
-4
-4 -3 -2 -1 0 1 2 3 4
-3
x 10
9 ¼ 0:1736 þ j7:0590
Fig. 5.9 Mode shape of k
9 ¼ 0:1736 þ j7.0590, the elements of right eigenvector are

Similarly, for k

v94 ¼ 0:0013 þ j0:0001 ¼ 0:0013\3:2273

v95 ¼ 0:0037 j0:0003 ¼ 0:0037\175:6366
96 ¼ 0:0022 j0:0002 ¼ 0:0022\173:8540
v
The mode shape is illustrated in Fig. 5.9. It indicates that the electromechanical
oscillation associated with this mode is that of G1 against G2 and G3.
5.3.2.2 Selective Reduced-Order Calculation of Electromechanical

Oscillation Modes of Example Power System
From the Heffron–Phillips model of example power system, the compact model of
system of Eq. (5.68) in the form of transfer function matrix is obtained as
Fðs) ¼ s2 M þ sD þ x0 K1 K2 ½ðI þ sTA ÞðK3 þ sT0d0 Þ þ KA K6 ]1

½ðI þ sTA ÞK4 þ KA Kk ]
The Hooke-Jeeves optimum searching method in nonlinear programming can be

used to find the solution of following objective function of Eq. (5.74) to calculate
electromechanical oscillation modes as follows.

¼ DEL[FðkÞ
f(kÞ ð5:74Þ
pffiffiffiffiffi
Step 1: Calculate eigenvalues of matrix x0 M1 K1 as gi ; i ¼ 1; 2. Thus, gi 6¼
0; j ¼ 1; 2 is the angular natural oscillation frequency of the electromechanical
oscillation modes. Choose k i ð0Þ ¼ 0 þ jpffiffiffiffi
ffi
gi on the complex plane as the starting
point of optimum searching.
Step 2: Choose an initial searching scale S0 [ 0, a minimum searching scale
SF [ 0, and an accelerating factor a [ 0. Search on the complex plane from k i ðk)

to ki ðk þ 1Þ. Three different patterns of searching on the complex plane from k i ðk)

to ki ðk þ 1Þ are illustrated in Fig. 5.10 where S is the searching step and a ¼ 2. In
Fig. 5.10 if fðk SÞ\fðk i Þ or fðk
jSÞ\fðki Þ, the searching is successful in
i i
horizontal and vertical direction. Otherwise, it is a failure. From Fig. 5.10, it can be
seen that (1) pattern 1—there is one successful searching in both horizontal and
pattern 1 pattern 2
pattern 3
complex plane
the starting point λ i(k) the ending point λ i(k+1)

S searching scale
the point where the searching is successful(S reduces)
the point where the searching fails(S does not reduce)

the serach is accelerated along the successful direction
Fig. 5.10 Illustration of searching patterns of the Hooke-Jeeves optimum searching method in one
step
vertical direction; (2) pattern 2—only one searching in the horizontal direction is
successful; (3) pattern 3—only one searching in the vertical direction is successful.
Step 3: If searching in Step 2 fails, reduce the searching scale S by half and then go
back to Step 2;
Step 4: If at ki (kF ), the searching scale becomes less than SF , stop the searching and

take ki (kF ) as the solution of objective function of Eq. (5.74), i.e., the ith oscillation
mode of the power system.
Figure 5.11 presents the flow chart of the Hooke-Jeeves optimum searching
method as explained above for the selective reduced-order calculation of the
electromechanical oscillation modes
The minimum searching scale SF [ 0 is the maximum distance of k i (kF ) to the

position where the real solution ki locates as shown in Fig. 5.12, where the last
searching scale is not greater than 2SF . Therefore, SF defines the accuracy of
optimum searching which is fully controllable. Of course, theoretically, the algo-
rithm can reach a solution which is as accurate as it is wished.
In the case that two or more starting points might converge to the same solution,
preventing techniques, such as the addition of a penalty or barrier function into the
objective function [3], can be applied in the direct searching.
The computational complexity (CC) [3] or computational cost (CC) [4] normally
is used to estimate the computational efficiency of an algorithm where one opera-
tion of an addition or multiplication is defined as one CC. The CC of multiplication
of two n-order full matrices (none of them is a diagonal matrix) is about n3 and the
inverse calculation of an n-order full matrix is n3 . Thus, the CC to form the
3
polynomial matrix model of Eq. (5.74) (i.e. forming all coefficient matrices in F(s))
is estimated as 2n3 =3 þ 32n3 . The CC in calculating the determinant of an n-order
complex matrix is about 4n3 =3. Thus, the CC of the Hooke-Jeeves optimum
searching in one step is about 16n3 =3. Therefore, the CC of the Hooke-Jeeves
optimum searching method is (kHS þ KHF )16n3 =3, where kHS is the times of suc-
cessful searching and KHF the times of searching failure. KHF can be estimated as
kHF ¼ logðS0 =SF Þ= log R, where R is the ratio of reducing the searching scale when
the searching in one step fails. If R = 2 S0 ¼ 0:01; SF ¼ 0:000001, we can have
kHF 13. Hence, the total CC of the reduced-order method is about
102n3 þ kHS 16n3 =3 þ oðn2 Þ.
Figures 5.13 and 5.14 show the trajectories of optimal searching starting from (0,
7.5128) and (0, 10.5647), respectively. Searching results of the electromechanical
oscillation modes are k 1 ¼ 0:1736 þ j7:0590 and k 2 ¼ 1:5051 þ j9:1860,
which are exactly as same as those obtained in the previous section where they are
denoted as k 9 ¼ 0:1736 þ j7:0590 and k 7 ¼ 1:5051 þ j9:1860.
Start
Choose i (0)=0+j i ,i=1,2 as the searching

Step 1 start points,where i are the eigenvalues of
matrix - 0 M -1K 1 and i 0.
Step 2 Initalization:i=k=m=1, d1 1 , d 2 j ,
set the value of S0 , SF and .
No Yes
f( i (k)+d mS)<f( i (k))?
No Yes
f( i (k)-d mS)<f( i (k))? i (k+1)= i (k)+d mS
i (k+1)= i (k) i (k+1)= i (k)-d mS
No Yes
m<2? m=m+1
No Yes
f( i (k+1))<f( i (k))?
i (k+1)= i (k)+a( i (k+1)- i (k)),

k=k+1,m=1
Step 3 S= S 2,
No Yes
i (k+1)= i (k), S<SF?
k=k+1,m=1
Step 4
i (k) is the solution
End
Fig. 5.11 Flow chart of Hooke-Jeeves optimum searching method

SF
*
i
i (k F )
point at each step contour of S
searching succeeds
searching fails
Fig. 5.12 Illustration on the searching to arrive at its final point
7.6
Imaginary part 48000000
of eigenvalue
7.5 40000000
Start Point
7.4 32000000
24000000
7.3
7.2 1600
0000
8000
000
7.1 Final Point
6.9
Real part of
eigenvalue
6.8
-0.25 -0.2 -0.15 -0.1 -0.05 0 0.05
1 calculation using direct searching algorithm

Fig. 5.13 Path of k
11 Imaginary part
of eigenvalue
10.8 900000000
10.6 70000000
Start Point
0
10.4
500000
000
10.2
10
3000
9.8 0000
0
9.6 100000
000
9.4 50000
000
9.2
Final Point Real part
of eigenvalue
9
-1.6 -1.4 -1.2 -1 -0.8 -0.6 -0.4 -0.2 0
2 calculation using direct searching algorithm

Fig. 5.14 Path of k
5.3.2.3 Coordinated Design of PSSs
Damping ratio of k1 is 0.0246 and that of k

2 is 0.1617. Now PSSs will be installed
in three machines to enhance the damping ratio of k1 to 0.1. Parameters of PSSs can
be set by use of the method introduced previously in Sect. 5.2.2 to move k 1 to

k1 ¼ 0:7094 þ j7:0590. Hence, following objective function is set up
2 2
f ðpÞ ¼ ½Re k1 ðpÞ Re k
ðp Þ þ ½Im k1 ðpÞ Im k
ðp Þ
1 1
where p ¼ ½ Dpss1 Dpss2 Dpss3 and Dpssi is the coefficient of damping torque
provided by the PSS installed in the ith generator.
Choose p0 ¼ ½ 0 0 0 with f ðp0 Þ ¼ 0:2872 to start the optimal searching of
the solution of the above objective function by use of the method introduced in
Sect. 5.2.2.2. In 52 iterations, the searching stops to have p52 ¼
½ 21:73 11:84 21:94 with f ðp52 Þ ¼ 9:2687 1015 , k 1 is moved to k c1 .
Figure 5.15 is the trajectories of Dpssi ; i ¼ 1; 2; 3 in respective to the iterative
searching. The optimal searching method used is the Hooke-Jeeves optimum
searching. Figure 5.16 is the trajectory of the objective function during the optimum
searching with respect to the movement of oscillation mode k 1 .
25
20
Designed damping
15
10
5
Dpss1
Dpss2
Dpss3
0
0 10 20 30 40 50 60
Iterations of direct search method
Fig. 5.15 Trajectories of Dpssi , i = 1, 2, 3 in respective to the iterative searching
Start Point
Objective function
Projection
Final Point
Imaginary part of eigenvalue Real part of eigenvalue
1
Fig. 5.16 Trajectory of the objective function in respective to the movement of k
Table 5.6 Parameter of the Kpss T2 T4

PSSs
G1 4.1272 0.096 0.096
G2 1.4864 0.095 0.095
G3 3.0945 0.1 0.1
From Fig. 5.5, the forward path matrix from the stabilizing signals of PSSs to the
electromechanical oscillation loops of generators is obtained as

Þ ¼ K2 KA ðK þ k
Fpss ðk TA Þ þ K6 KA 1
T0 ÞðI þ k
1 3 1 d0 1
2 3
4:0590\29:34 1:7719\153:70 1:0376\155:74
6 7
¼ 4 0:6433\113:76 6:1802\28:79 2:6963\136:09 5
1:1801\125:70 2:6184\140:40 5:3227\31:52
The forward path of the PSS installed in each generator as the diagonal element
of the above matrix is
Þ ¼ 4:0590\29:34
Fpss11 ðk 1
Þ ¼ 6:1802\28:79
Fpss22 ðk 1
Þ ¼ 5:3227\31:52
Fpss33 ðk 1
Let the transfer function of the PSS be
ð1 þ sT2i Þ ð1 þ sT4i Þ
Tpssi ðsÞ ¼ Kpssi ; T1i ¼ T3i ¼ 0:05
By use of the phase compensation method for the PSS to compensate the for-
Þ; i ¼ 1; 2; 3 to provide a pure damping torque
ward path of the ith generator Fii ðk

i
Dpssi , where Dpss1 Dpss2 Dpss3 ¼ p52 ¼ ½ 21:73 11:84 21:94 , parameters
of PSSs are set and listed in Table 5.6.
With PSSs installed, the closed-loop state equation is
sDX ¼ Apss DX
where

Apss11 Apss12 Apss13
Apss ¼
Apss21 Apss22 Apss23
2 3
0 0 0 314:1592 0 0
6 0 0 0 0 314:1592 0 7
6 7
6 314:1592 7
6 0 0 0 0 0 7
6 7
6 0:0495 0:0310 0:0185 0 0 0 7
6 7
Apss11 ¼6
6 0:1195 0:1731 0:0537 0 0 0 7
7
6 0:1761 0:3123 7
6 0:1362 0 0 0 7
6 7
6 0:0031 0:0001 0:0030 0 0 0 7
6 7
4 0:2004 0:3121 0:1117 0 0 0 5
0:2172 0:1717 0:3889 0 0 0
2 3
0 0 0 0 0 0
6 0 0 0 0 0 0 7
6 7
6 0 7
6 0 0 0 0 0 7
6 7
6 0:0521 0:0191 0:0119 0 0 0 7
6 7
Apss12 ¼6
6 0:0801 0:2396 0:0317 0 0 0 7 7
6 0:0614 0:1073 0:3860 0 7
6 0 0 7
6 7
6 0:1327 0:0119 0:0111 0:1116 0 0 7
6 7
4 0:0202 0:5261 0:1277 0 0:1667 0 5
0:0768 0:2119 0:6306 0 0 0:1698
2 3
0 0 0 0 0 0
60 0 0 0 0 07
6 7
60 07
6 0 0 0 0 7
6 7
60 0 0 0 0 07
6 7
Apss13 ¼6
60 0 0 0 0 077
60 07
6 0 0 0 0 7
6 7
60 0 0 0 0 07
6 7
40 0 0 0 0 05
0 0 0 0 0 0
2 3
67:5657 75:7691 143:3348 0 0 0
6 595:9666 314:8297 281:1369 0 0 0 7
6 7
6 779:5679 498:8968 1278:4646 0 7
6 0 0 7
6 7
6 0:0951 0:0596 0:0355 20 0 0 7
6 7
Apss21 ¼6
6 0:7703 0:4825 0:2878 162:048 0 0 7 7
6 0:2270 0:3289 0 7
6 0:1020 0 20 7
6 7
6 0:8193 1:1874 0:3681 0 72:2 0 7
6 7
4 0:3521 0:2724 0:6246 0 0 20 5
3:9227 3:0351 6:9579 0 0 222:8
2 3
8731:9338 710:95188 678:1403 50 0 0
6 4315:4400 5291:9761 1729:6500 0 50 0 7
6 7
6 4691:1745 1828:9008 4789:5838 50 7
6 0 0 7
6 7
6 0:1000 0:0367 0:0229 0 0 0 7
6 7
Apss22 6
¼ 6 0:8099 0:2977 0:1855 0 0 0 77
6 0:1521 0:4553 0 7
6 0:0603 0 0 7
6 7
6 0:5491 1:6435 0:2176 0 0 0 7
6 7
4 0:1227 0:2146 0:7719 0 0 0 5
1:3673 2:3902 8:5990 0 0 0
2 3
0 10000 0 0 0 0
6 0 0 0 10000 0 0 7
6 7
6 10000 7
6 0 0 0 0 0 7
6 7
6 20 0 0 0 0 0 7
6 7
Apss23 ¼6
6 77:648 20 0 0 0 0 7 7
6 20 0 7
6 0 0 0 0 7
6 7
6 0 0 34:2 20 0 0 7
6 7
4 0 0 0 0 20 0 5
0 0 0 0 111:4 20
Eigenvalues of state matrix are calculated as
1 ¼ 44:7474 þ j0; k
k 2;3 ¼ 23:7299 j27:9946; k
4 ¼ 33:1534 þ j0,
5;6 ¼ 19:9770 j11:9203; k
k 7;8 ¼ 12:9863 j13:6705;
9;10 ¼ 0:6094 j7:0235; k
k 11;12 ¼ 4:5207 j8:0856;
13 ¼ 0:2772 þ j0,k
k 14 ¼ 0 þ j0, k
15 ¼ 23:6389 þ j0,
k 17 ¼ 17:7485 þ j0,k

16 ¼ 12:1456 þ j0,k 18 ¼ 15:9316 þ j0
Obviously, k 1 is moved to k c1 ¼ 0:6094 þ j7:0235, slightly away from the

¼ 0:7094 þ j7:0590. In Sect. 6.4.1, it is examined why
target required position k1
this happens.
Simulation results of example power system without and with PSSs installed are
presented in Fig. 5.17. At 0.5 s of simulation, a 3-phase short-circuit occurs at Bus
1, and 0.1 s later the short-circuit fault is cleared. Fig. 5.17, confirms the effec-
tiveness of PSSs designed by use of the introduced method.
2- 1(deg)
With PSS
Without PSS
time(ms)
Fig. 5.17 Simulation results with a 3-phase short-circuit fault at Bus 1
References
1. Anderson PM, Fouad AA (1977) Power system control and stability. The Iowa State University
Press
2. Wang XF, Song YH, Irving M (2011) Modern power systems analysis. Springer, Berlin
3. Bazarra MS et al (1993) Non-linear programming. Wiley, Hoboken
4. Marchuk GI (1994) Numerical methods and applications. CRC Press Inc, Boca Raton
Chapter 6
Multi-machine Power System Installed
with Thyristor-Based FACTS Stabilizers

System Installed with Thyristor-Based FACTS
Stabilizers
6.1.1 Heffron–Phillips Model [1]

Installed with an SVC Stabilizer
Figure 6.1 shows an N-machine power system with nodes of internal voltage of
generators added. From Fig. 6.1, the following network equations can be
established.

0 Y11 Y12 Vm
¼ ð6:1Þ
Ig Y21 Y22 Eg
T
where Ig ¼ Ig1 Ig2 IgN ; E g ¼ E g2 E
g1 E m is the
gN T ; V
vector of voltage at all non-internal nodes of generators in the power system and Eg

is defined by Eq. (5.34). “Deleting” all non-internal nodes is to remove Vm in the
network equations of Eq. (6.1) to have
Ig ¼ ðY 1 Y
21 Y 22 ÞE
12 þ Y g ð6:2Þ
11
where only N internal nodes of generators are kept. Comparing Eqs. (6.2) and (5.34),
it can be seen that in fact
¼Y
Y 1 Y
21 Y 12 þ Y
22 ð6:3Þ
11

DOI 10.1007/978-1-4899-7696-3_6
236 6 Multi-machine Power System Installed …
Fig. 6.1 Network of an E g1 Eg 2 E gN

N-machine power system Ig1 Ig 2 IgN
including internal nodes of
generators
x d1 ' xd2 ' x dN '
V g1 V g2 V gN
Transmission network YN
Fig. 6.2 An N-machine E1 E2 EN

power system installed with Ig1 Ig 2 IgN
an SVC-based stabilizer
X d1 ' Xd 2 ' X dN '
V g1 Vg 2 V gN
Node 1
y11 jbsvc
Now consider an N-machine power system installed with an SVC stabilizer as

shown in Fig. 6.2. Without loss of generality of discussion, it can be assumed that
(1) the SVC stabilizer is installed at node 1 in the transmission network, and
(2) after other nodes are deleted in the network equations of the N-machine power
system, there are only N internal nodes of generators and node 1 left. Hence, the
network equations with those N + 1 nodes left are

0 11
Y 12
Y 1n
V
Ig ¼ 21 22 g ð6:4Þ
Y Y E
11 ¼ y11 þ jbsvc and jbsvc is the equivalent admittance of the SVC.

where Y
6.1 Mathematical Model of a Multi-Machine Power System … 237
Ig ¼ 1 21 Y
12 E
g þ Y g
22 E
Y
ðy11 þ jbsvc Þ
ð6:5Þ
1 21 Y
12 þ Y
22 Eg ¼ Y
Eg
¼ Y
ðy11 þ jbsvc Þ

From Eq. (6.5), it can be seen that the elements of network admittance matrix, Y,
can be denoted as
yij ¼ y0ij þ yij ðbsvc Þ ð6:6Þ
X
N
Igi ¼ idi þ jiqi ¼ yij ½E0qj ejð90

þ dj di Þ
j¼1
X
N

¼ y0ij ½E0qj ejð90 þ aij þ dj di Þ
þ ðxqj x0dj Þiqj ejðaij þ dj di Þ ð6:7Þ
j¼1
X
N

þ yij ðbsvc Þ½E0qj ejð90 þ dj di Þ
j¼1
0ij ¼ y0ij ejaij . From Eq. (6.7), it can be obtained that

where y
X
N
idi ¼ y0ij ½E0qj sinðaij dij Þ þ ðxqj x0dj Þiqj cosðaij dij Þ
j¼1
X
N
þ Re½yij ðbsvc Þ½E0qj sin dij þ ðxqj x0dj Þiqj cos dij
j¼1
X
N
Im½yij ðbsvc Þ½E0qj cos dij ðxqj x0dj Þiqj sin dij
j¼1
ð6:8Þ
X
N
iqi ¼ yij ½E0qj cosðaij dij Þ þ ðxqj x0dj Þiqj sinðaij dij Þ
j¼1
X
N
þ Re½yij ðbsvc Þ½E0qj cos dij ðxqj x0dj Þiqj sin dij
j¼1
X
N
þ Im[yij ðbsvc Þ½E0qj sin dij þ ðxqj x0dj Þiqj cos dij
j¼1
where Re½yij ðbsvc Þ and Im½yij ðbsvc Þ denote the real and imaginary parts of yij ðbsvc Þ,
respectively.
DId ¼ Fdd Dd þ Gdd DE0q þ Hdd DIq þ Ldd Dbsvc

ð6:9Þ
DIq ¼ Fqq Dd þ Gqq DE0q þ Hqq DIq þ Lqq Dbsvc
DId ¼ Fd Dd þ Gd DE0q þ Ld Dbsvc

ð6:10Þ
DIq ¼ Fq Dd þ Gq DE0q þ Lq Dbsvc
Substituting Eq. (6.10) into Eqs. (5.40) and (5.41), the Heffron–Phillips model of
the N-machine power system installed with the SVC stabilizer can be obtained as
Dd_ ¼ xo Dx
Dx_ ¼ M1 ðK1 Dd K2 DE0q DDx þ KP Dbsvc Þ
0 ð6:11Þ
DE_ q ¼ T01 0 0
d0 ðK3 DEq K4 Dd þ DEfd þ Kq Dbsvc Þ
0
DE_ fd ¼ T1 0 1 0
A DEfd þ TA KA ðK5 Dd K6 DEq þ KV Dbsvc Þ

Installed with a TCSC or TCPS Stabilizer
Without loss of generality of discussion, it can be assumed that a TCSC stabilizer is

installed between nodes 1 and 2 in an N-machine power system as shown in
Fig. 6.3. The following network admittance matrix can be formed with only the
internal nodes of generators, nodes 1 and 2 left,

0
Y 12
Y 12
V
Ig ¼ 11 22 g ð6:12Þ
Y21 Y E
where
2 3
y11 þ 1 y12 z 1
11 4 z12 jxt csc 12 jxt csc 5
Y ¼
y21 1 y22 þ z 1jx
z12 jxt csc 12 t csc

12
V ¼ V 1 V 2 T
Hence, the network admittance matrix with only the internal nodes of generators
left is
Fig. 6.3 An N-machine E g1 Eg 2 E gN

power system installed with a Ig1 Ig 2 IgN
TCSC-based stabilizer
x d1 ' xd2 ' x dN '
V g1 Vg 2 V gN
y12 (y 21 )
V1 V2
Node 1 Node 2
z12
x t csc
y11 y 22
¼Y 1 Y
21 Y
Y 11 12 þ Y22 ð6:13Þ
Obviously, the element of the network admittance matrix yij can be denoted as
yij ¼ y0ij þ yij ðxt csc Þ ð6:14Þ

X
N

þ dj di Þ
j¼1
X
N

j¼1
X
N

þ yij ðxt csc Þ½E0qj ejð90 þ dj di Þ
j¼1
Similar to Eqs. (6.7)–(6.10), by using Eq. (6.15), the Heffron–Phillips model of

the N-machine power system installed with the TCSC stabilizer can be obtained as
Dd_ ¼ xo Dx
Dx_ ¼ M1 ðK1 Dd K2 DE0q DDx þ KP Dxt csc Þ
0 ð6:16Þ
DE_ q ¼ T01 0 0
d0 ðK3 DEq K4 Dd þ DEfd þ Kq Dxt csc Þ
0
DE_ fd ¼ T1 0 1 0
A DEfd þ TA KA ðK5 Dd K6 DEq þ KV Dxt csc Þ
Fig. 6.4 An N-machine E g1 Eg 2 E gN

power system installed with a Ig1 Ig 2 IgN
TCPS-based stabilizer
x d1 ' xd2 ' x dN '
V g1 Vg 2 V gN
y12 (y 21 )
V1 V2
Node 1 Node 2
z12
k :1
y11 y 22
Similarly, for the N-machine power system with a TCPS stabilizer installed
between nodes 1 and 2 as shown in Fig. 6.4, where k ¼ kej/ , it can have

0 11
Y 12
Y 12
V
Ig ¼ 21 22 g ð6:17Þ
Y Y E
where
2 j/ 3
y11 þ 2
1 y12 e
11 ¼ 4 k z 12 kz 12
5
Y
y21 e
j/
y22 þ z112
kz12

12n
V ¼ V1 V 2 T
Hence, the network admittance matrix with only the internal nodes of generators
left can be obtained as
¼Y 1 Y
21 Y
Y 11 12 þ Y22 ð6:18Þ
and the element of the network admittance matrix yij can be denoted as
yij ¼ y0ij þ yij ð/Þ ð6:19Þ

Hence, for the system with the TCPS stabilizer installed, it can have
X
N

þ dj di Þ
j¼1
X
N

j¼1
X
N

þ yij ð/Þ½E0qj ejð90 þ dj di Þ
j¼1
By taking the similar procedure from Eqs. (6.7) to (6.10), from Eq. (6.20), the
Heffron–Phillips model of the N-machine power system installed with the TCPS
stabilizer can be obtained as
Dd_ ¼ xo Dx
Dx_ ¼ M1 ðK1 Dd K2 DE0q DDx þ KP D/Þ
0 ð6:21Þ
DE_ q ¼ T01 0 0
d0 ðK3 DEq K4 Dd þ DEfd þ Kq D/Þ
0
DE_ fd ¼ T1 0 1 0
A DEfd þ TA KA ðK5 Dd K6 DEq þ KV D/Þ
The Heffron–Phillips model of the N-machine power system installed with an

SVC stabilizer, TCSC stabilizer, or TCPS stabilizer can be expressed in a unified
form as shown in Fig. 6.5. In Fig. 6.5, Dufactss ¼ Dbsvc for the SVC stabilizer,
Dufactss ¼ Dxt csc for the TCSC stabilizer, and Dufactss ¼ D/ for the TCPS sta-
bilizer, respectively.
K1
-
ω0 I
(sM + D) −1
s
- + Δu facts −s
KP Kq KV
K4 K5
K2
+ -
- + +
−1
(K 3 + sTd0 ) (I + sTA ) −1 K A Δu pss
+
-
K6
Fig. 6.5 Unified Heffron–Phillips model of an N-machine power system installed with a
thyristor-based FACTS stabilizer
6.1.2 General Linearized Model of an N-Machine Power

System Installed with Multiple Thyristor-Based
FACTS Stabilizers
In this section, a more general procedure to establish the linearized model of an

N-machine power system installed with multiple thyristor-based FACTS stabilizers
is introduced. In the established model, the generator can be described by either the
reduced-order dynamic model which is used in the above section to establish the
Heffron–Phillips model, or the full linearized model given by Eqs. (5.44)–(5.47). In
addition, the normal control functions of thyristor-based FACTS controllers and the
full dynamics of stabilizers are also included in the established model.
6.1.2.1 Linearized Model of Generators
When full linearized mathematical model of generators is used, by substituting

Eq. (5.51) into Eq. (5.48), it can be obtained that
X_ g ¼ Agxy Xg þ Bg Dupss þ Bgxy DVxy

ð6:22Þ
DIxy ¼ Cgxy Xg þ Dgxy DVxy
where
Agxy ¼ Ag þ Bgv BgV

Bgxy ¼ Bgv Tg0
Cgxy ¼ T1
g0 ðCg BgI Þ
Dgxy ¼ 0
In the case that the reduced-order model of generators which is used for
establishing the Heffron–Phillips mode, from Eq. (2.31), it can have
vgdi E0 vgqi
iqi ¼ ; idi ¼ d 0 ð6:23Þ
xqi xd
Linearization of the above equation is

0
Dvgdi DEq Dvgqi
Diqi ¼ ; Didi ¼ ð6:24Þ
xqi x0q
From Eq. (6.24) and (5.39), it can be obtained that
DIdq ¼ E0 DE0iq þ Xdq1 DVdq

DP ¼ Vdq0 DIdq þ Idq0 DVdq
ð6:25Þ
DEq ¼ DE0q þ X0dd DIdq
DV ¼ V0dq DVdq
where
1
E0 ¼ feij g; eij ¼ ;
x0di
for i ¼ 1; 3; . . .; 2N 1; j ¼ 1; 2; . . .; N; otherwise eij ¼ 0
1
Xdq1 ¼ diagðxdqi Þ; xdqi ¼ 0 ; for i ¼ 1; 3; . . .; 2N 1;
xdi
1
xdqi ¼ ¼ 0; for i ¼ 2; 4; . . .2N
xqi
Vdq0 ¼ fVdq0ij g; Vdq0ij ¼ vgdi0 ; Vdq0ij þ 1 ¼ vgqi0 ;
for i ¼ 1; 2; . . .; N; j ¼ 1; 3; . . .; 2N 1; otherwise Vdq0ij ¼ 0
Idq0 ¼ fIdq0ij g; Idq0ij ¼ igdi0 ; Idq0ij þ 1 ¼ igqi0 ;
for i ¼ 1; 2; . . .; N; j ¼ 1; 3; . . .; 2N 1; otherwise Idq0ij ¼ 0
X0dd ¼ fx0ddij g; x0ddij ¼ ðxdi x0di Þ;
for i ¼ 1; 2; . . .; N; j ¼ 1; 3; . . .; 2N 1; otherwise x0ddij ¼ 0
vgdi0 vgqi0
V0dq ¼ fV0dqij g; V0dqij ¼ ; V0dqij þ 1 ¼ ;
Vgi0 Vgi0
for i ¼ 1; 2; . . .; N; j ¼ 1; 3; . . .; 2N 1; otherwise Vdq0ij ¼ 0
By substituting the above equation into (5.40), it can be obtained that
Dd_ ¼ xo Dx
Dx_ ¼ M1 ðDDx Vdq0 E0 DE0iq Þ M1 ðVdq0 Xdq1 þ Idq0 ÞDVdq
0
DE_ q ¼ T01 0 0 0 0 01 0
d0 ðDEq Xdd E0 DEiq þ DEfd Þ Td0 Xdd Xdq1 DVdq ð6:26Þ
0
DE_ fd ¼ T1 0 1 1
A DEfd TA KA V0dq DVdq þ TA KA Dupss
DIdq ¼ E0 DE0iq þ Xdq1 DVdq
which can be written as

2 3 2 32 3
Dd_ 0 xo I 0 0 Dd
6 7 6
6 Dx_ 7 6 0 M1 D M1 Vdq0 E0 0 7 6
76 Dx 7
7
6 0 7¼6 7 6 7
6 DE_ 7 4 0 T1 0
T1 54 DE0q 5
4 q 5 0 d0 ðI þ Xdd E0 Þ d0
0
DE_ fd 0 0 0 T1A DE0fd
2 3 2 3
0 0
6 M1 ðV X þ I Þ 7 6 0 7
6 dq0 dq1 dq0 7 6 7 ð6:27Þ
6 7DVdq þ 6 7Dupss
4 T01
d0 X 0
X
dd dq1
5 4 0 5
T1
A KA V0dq T1
A KA
2 3
Dd
6 Dx 7
6 7
DIdq ¼ ½ 0 0 E0 0 6 7 þ Xdq1 DVdq
4 DE0q 5
DE0fd
Obviously, the above equation can be written in the similar form to Eq. (5.48) as
X_ g ¼ Ag Xg þ Bg Dupss þ Bgv DVdq

ð6:28Þ
DIdq ¼ Cg Xg þ Dg DVdq
where
2 3 2 3
Dd 0 xo I 0 0
6 Dx 7 6 0 M1 D 1
M Vdq0 E0 7
6 7 6 0 7
Xg ¼ 6 7; Ag ¼ 6 7;
4 DE0q 5 40 0 T1 0
d0 ðI þ Xdd E0 Þ T1
d0
5
DE0 0 0 0 T1
2 fd 3 2 3 A
0 0
6 7 6 7
6 M1 ðVdq0 Xdq1 þ Idq0 Þ 7 6 0 7
6 7 6 7
Bgv ¼ 6 7; Bg ¼ 6 7;
6 T 01 0
X X 7 6 0 7
4 d0 dd dq1 5 4 5
T1
A KA V0dq T1
A KA
Cg ¼ ½ 0 0 E0 0 ; Dg ¼ Xdq1
By substituting Eq. (5.51) into the above equation, it can be obtained that
X_ g ¼ Agxy Xg þ Bg Dupss þ Bgxy DVxy

ð6:29Þ
DIxy ¼ Cgxy Xg þ Dgxy DVxy
where
Agxy ¼ Ag þ Bgv BgV

Bgxy ¼ Bgv Tg0
Cgxy ¼ T1
g0 ðCg BgI Þ
Dgxy ¼ T1
g0 Dg
The format of Eqs. (6.22) and (6.29) is exactly the same, which will be used as
the general linearized model of generators in the N-machine power system with no
PSSs installed.
Let the transfer function of the ith PSS be Gi ðsÞ and the feedback signal Dxi , that is
Dupssi ¼ Gi ðsÞDxi ð6:30Þ
If the state-space realization of Eq. (6.30) is
X_ pssi ¼ Apssi Xpssi þ Bpssi Dxi

ð6:31Þ
Dupssi ¼ Cpssi Xpssi þ Dpssi Dxi
State-space representation of L PSSs installed in the N-machine power system

can be written as
X_ pss ¼ Apss Xpss þ Bpss Dx

ð6:32Þ
Dupss ¼ Cpss Xpss þ Dpss Dx
where
2 3 2 3
Xpss1 Bpss1 0 ... 0
6 Xpss2 7 6 0 Bpss2 ... 0 7
6 7 6 7
Xpss ¼6 7 6
6 .. 7; Bpss ¼ 6 .. .. ..
7;
7
4 . 5 4 . . ... . 5
XpssL 0 0 ... BpssL
2 3 2 3
Apss1 0 ... 0 Cpss1 0 ... 0
6 0 Apss2 ... 0 7 6 0 Cpss2 ... 0 7
6 7 6 7
Apss ¼6
6 .. .. ..
7; Cpss
7 ¼6
6 .. .. ..
7;
7
4 . . ... . 5 4 . . ... . 5
0 0 ... ApssL 0 0 ... CpssL
2 3
Dpss1 0 ... 0
6 0 Dpss2 ... 0 7
6 7
Dpss ¼6
6 .. .. .. 7
7
4 . . ... . 5
0 0 ... DpssL
By substituting Eq. (6.32) into Eqs. (6.22) or (6.29), it can be obtained that
X_ gp ¼ Agp Xgp þ Bgp DVxy

ð6:33Þ
DIxy ¼ Cgp Xgp þ Dgp DVxy
where

Xg Agxy þ Bg DDpss Bg Cpss
Xgp ¼ ; Agp ¼ ;
Xpss BDpss Apss

Bgxy Dgp
Bgp ¼ ; Cgp ¼ ½ Cgxy 0 ; Dgp ¼ ;
0 0
2 3 2 3
0 0 0 0 0 0
6 7 6 7
DDpss ¼ 4 0 Dpss 0 5; BDpss ¼ 4 0 Bpss 0 5
0 0 0 0 0 0
Equation (6.33) is the linearized model of generators with the PSSs installed.
6.1.2.2 Linearized Model of an SVC Installed in the N-Machine Power

System
For a thyristor-controlled reactor and fixed capacitor (TCR-FC) type of SVC with
an SVC stabilizer as shown in Fig. 3.1, from Eq. (3.1), it can be obtained that for
the ith SVC installed in the N-machine power system
1 2ai sin 2ai
bsvci ¼ ð6:34Þ
xsvcci 2pxsvcli
Without loss of generality, denote the installing location of the SVC as the ith
node in the power system. It can be assumed that the transfer function of voltage
controller and stabilizer in Fig. 3.1 is Tvsvci ðsÞ and Tssvci ðsÞ, respectively, and the
feedback signal of the SVC stabilizer is the integral signal of the active power, Pij ,
delivered through the node where the SVC is installed as shown in Fig. 6.6. From
Figs. 3.1 and 6.6, it can have
Fig. 6.6 An SVC installed on Vi Vj

the ith node in the N-machine Pij zij Iij
power system
Ii jbsvci
ai ¼ avi þ asi ð6:35Þ
where
avi ¼ Tvsvci ðsÞðVi Viref Þ
1 ð6:36Þ
asi ¼ Tssvci ðsÞ ðPij Pijref Þ
s
where Viref and Pijref are the reference signal of the SVC voltage and stabilizing
control, respectively.
Linearization of Eqs. (6.34) and (6.36) is
1 cos 2ai0
Dbsvci ¼ ðDavi þ Dasi Þ ð6:37Þ
pxsvcli
Davi ¼ Tvsvci ðsÞDVi

1 ð6:38Þ
Dasi ¼ Tssvci ðsÞ DPij
s
where
vix0 viy0 h i Dv h i
viy0 viy0
DVi ¼ Dvix þ Dviy ¼ vVix0 ix
¼ vVix0 Vi0 DVixy
Vi0 Vi0 i0 Vi0 Dviy i0

i V
V j
Pij ¼ Reð Þ ¼ Re½ðgij þ jbij ÞðV
V V
iV Þ
jV
zij i i i
ð6:39Þ
¼ gij ðv2xi þ v2yi Þ þ bij ðvxi vyi vyi vxj Þ
where * denotes the conjugate of a phasor. Linearization of the above equation

gives
DPij ¼ 2gij ðvxi0 Dvxi þ vyi0 Dvyi Þ þ bij ðvyj0 Dvxi vxj0 Dvyi Þ
þ bij ðvxi0 Dvyj vyi0 Dvxj Þ
¼ ð2gij vxi0 þ bij vyj0 ÞDvxi þ ð2gij vyi0 bij vxj0 ÞDvyi
þ bij ðvxi0 Dvyj vyj0 Dvxj Þ

Dvxi
¼ 2g v
ij xi0 þ b v
ij yj0 2g v
ij yi0 b v
ij xj0
Dvyi

Dvxj
þ ½ bij vxi0 bij vyi0
Dvyj

¼ 2gij vxi0 þ bij vyj0 2gij vyi0 bij vxj0 DVxyi
þ ½ bij vxi0 bij vyi0 DVxyj
¼ psvci DVxyi þ p0svcj DVxyj ð6:40Þ
Hence, from Eqs. (6.38) and (6.40), it can have

h i
v
Davi ¼ Tvsvci ðsÞ vVxi0 Vyi0 DVxyi
i0 i0
1 ð6:41Þ
Dasi ¼ Tssvci ðsÞ ðpsvci DVxyi þ p0svcj DVxyj Þ
s
Let the state-space realization of the voltage controller and stabilizer of the SVC
given by Eq. (6.41) be
X_ VSVCi ¼ AVSVCi XVSVCi þ BVSVCi DVxyi

ð6:42Þ
Davi ¼ CVSVCi XVSVCi þ DVSVCi DVxyi
X_ SSVCi ¼ ASSVCi XSSVCi þ BSSVCi DVxyi þ BSSVCj DVxyj

ð6:43Þ
Dasi ¼ CSSVCi XSSVCi þ DSSVCi DVxyi þ DSSVCj DVxyj
According to Fig. 6.6, the injected current from the SVC into the ith node is
Ii ¼ ixi þ jiyi ¼ jbsvci V

i ¼ jbsvci ðvxi þ jvyi Þ ð6:44Þ
ixi ¼ bsvci vyi ; iyi ¼ bsvci vxi ð6:45Þ

Dixi vyi0 bsvci0 0 Dvxi
¼ Dbsvci þ
Diyi vxi0 0 bsvci0 Dvyi

1 cos 2ai0 vyi0 bsvci0 0
¼ ðDavi þ Dasi Þ þ DVxyi
pXsvcli vxi0 0 bsvci0
ð6:46Þ
When only the SVC voltage control function is considered, by substituting

Eq. (6.42) into (6.46), it can have

Dixi vyi0 bsvci0 0 Dvxi
¼ Dbsvci þ
Diyi vxi0 0 bsvci0 Dvyi

1 cos 2ai0 vyi0
¼ ðCVSVCi XVSVCi þ DVSVCi DVxyi þ DaSi Þ
pxsvcli vxi0

bsvci0 0
þ DVxyi ð6:47Þ
0 bsvci0
That is
DISVCi ¼ CSVCi XVSVCi þ DSVCi DVxyi þ bSSVCi Dasi ð6:48Þ
where

Dixi 1 cos 2ai0 vyi0
DISVCi ¼ ; CSVCi ¼ CVSVCi ;
Diyi pxsvcli vxi0

1 cos 2ai0 vyi0 1 cos 2ai0 vyi0
DSVCi ¼ DVSVCi ; bSSVCi ¼
pxsvcli vxi0 pxsvcli vxi0
Equations (6.42) and (6.48) are the linearized model of the SVC without the
stabilizer. Feedback signal of the SVC stabilizer is DPij . Hence, output equation is
Eq. (6.40).
When the SVC stabilizer is considered, by substituting Eq. (6.43) into
Eq. (6.47), it can be obtained that

Dixi 1 cos 2ai0 vyi0
¼ ðCVSVCi XVSVCi þ CSSVCi XSSVCi Þ
Diyi pxsvcli vxi0

1 cos 2ai0 vyi0
ðDVSVCi DVxyi þ DSSVCi DVxyi þ DSSVCj DVxyj Þ
pxsvcli vxi0

bsvci0 0
þ DVxyi
0 bsvci0
ð6:49Þ
Put the state equation of the SVC voltage controller of Eq. (6.42) and stabilizer
of Eq. (6.43) together
X_ SVCi ¼ ASVCi XSVCi þ BSVCi DVxyi þ BSVCj DVxyj ð6:50Þ
where

XVSVCi AVSVCi 0
XSVCi ¼ ; ASVCi ¼ ;
XSSVCi 0 ASSVCi

BVSVCi 0
BSVCi ¼ ; BSVCj ¼
BSSVCi BSSVCj
DISVCi ¼ CSVCi XSVCi þ DSVCi DVxyi þ DSVCj DVxyj ð6:51Þ

where

Dixi
DISVCi ¼
Diyi

vyi0 vyi0
CSVCi ¼ 1cos 2ai0
pxsvcli CVSVCi 1cos
CSSVCi 2ai0
pxsvcli
vxi0 vxi0

1 cos 2ai0 vyi0 bsvci0 0
DSVCi ¼ ðDVSVCi þ DSSVCi Þ þ
pxsvcli vxi0 0 bsvci0

1 cos 2ai0 vyi0
DSVCj ¼ DSSVCj
pxsvcli vxi0
Equations (6.50) and (6.51) are the linearized model of the SVC, including
voltage controller and stabilizer, installed at the ith node in the N-machine power
system.
6.1.2.3 Linearized Model of a TCSC Installed in the N-Machine Power

System
At the location in the N-machine power system where a TCSC is installed, two
extra nodes, denoted as i and j, at the terminals of the TCSC can be created as
shown in Fig. 6.7. From Fig. 6.7, it can be seen that the function of the TCSC is
electrically equivalent to the injection of current into two nodes,

Ii ¼ Vj Vi ; Ij ¼ Vi Vj ð6:52Þ
jxtcsci jxtcsci
Vi Vj
Pij
Vi Vj
Ii Ij Ii Ij
jx tcsci
Fig. 6.7 Creation of two extra nodes at the location where the TCSC is installed
That is
Ii ¼ ixi þ jiyi ¼ j ðvxj þ jvyj vxi jvyi Þ

xtcsci
ð6:53Þ
Ij ¼ ixj þ jiyj ¼ j ðvxi þ jvyi vxj jvyj Þ
xtcsci
1 1
ixi ¼ ðvyj vyi Þ; iyi ¼ ðvxi vxj Þ;
xtcsci xt csci
ð6:54Þ
1 1
ixj ¼ ðvyi vyj Þ; iyj ¼ ðvxj vxi Þ
xtcsci xtcsci
The normal control function of the TCSC is the load flow regulation. Let the
transfer function of the TCSC load flow regulator be Tltcsci ðsÞ. It can be assumed
that the transfer function of the TCSC stabilizer and feedback signal is Tstcsci ðsÞ and
the integral of deviation of the active power delivered along the transmission line,
Pij , where the TCSC is installed. Control signal of the TCSC stabilizer can be
superimposed on the TCSC load flow controller, that is
1
xtcsci ¼ ½Tltcsci ðsÞ þ Tstcsci ðsÞðPij Pijref Þ ð6:55Þ
s
where Pijref is the reference signal of the TCSC load flow and stabilizing control.
1
Dxtcsci ¼ ½Tltcsci ðsÞ þ Tstcsci ðsÞDPij ð6:56Þ
s
i V
V j
Pij ¼ Reð Þ ¼ Re½ 1 ðV
V V
iV Þ
jV
i i i
jxtcsci jxtcsci
ð6:57Þ
1
¼ ðvyi vxj vxi vyj Þ
xtcsci
1
DPij ¼ ðvyi0 vxj0 vxi0 vyj0 ÞDvtcsci
x2tcsci0
1 1 ð6:58Þ
þ ðvxj0 Dvyi vyj0 Dvxi Þ þ ðvyj0 Dvxj vxi0 Dvyj Þ
xtcsci0 xtcsci0
¼ atcsc0 Dxtcsci þ atcsci DVxyi þ atcscj DVxyj
Fig. 6.8 TCSC with a PI load

K ptcsc
flow controller superimposed
by a stabilizing signal
ΔPij
+ + Δx tcsci
K itcsc Δα stcsci
s Δx ltcsci
where
1
atcsc0 ¼ ðvyi0 vxj0 vxi0 vyj0 Þ
x2tcsci0
h i
atcsci ¼ xtcsci0
1 1
vyj0 xtcsci0 vxj0
h i
atcscj ¼ xtcsci0
1
vyj0 xtcsci0
1
vxi0
For the case that the TCSC stabilizer is not installed, Eq. (6.56) becomes
Dxtcsci ¼ Tltcsci ðsÞDPij þ Dastcsci ð6:59Þ
where Dastcsci is the stabilizing signal. Without loss of generality, it can be assumed
that the TCSC load flow controller adopts a proportional and integral (PI) control
law as shown in Fig. 6.8. From Fig. 6.8, it can have
Dxtcsci ¼ Dastcsci þ Kptcsc DPij þ Dxltcsci

ð6:60Þ
Dx_ ltcsci ¼ Kitcsc DPij
1 1
Dxtcsci ¼ Dastcsci þ Dxltcsci
1 þ Kptcsc atcsc0 1 þ Kptcsc atcsc0
ð6:61Þ
Kptcsc
þ ðatcsci DVxyi þ atcscj DVxyj Þ
1 þ Kptcsc atcsc0
Kitcsc atcsc0 Kitcsc atcsc0

Dx_ ltcsci ¼ Dxltcsci Dastcsci
1 þ Kptcsc atcsc0 1 þ Kptcsc atcsc0
ð6:62Þ
Kptcsc Kitcsc atcsc0
þ ðKitcsc Þðatcsci DVxyi þ atcscj DVxyj Þ
1 þ Kptcsc atcsc0
Dx_ ltcsci ¼ Altcsci Dxltcsci þ BLTCSCi DVxyi þ BLTCSCj DVxyj þ bltcsci Dastcsci ð6:63Þ

1 1
Dixi ¼ ðvyj0 vyi0 ÞDxtcsci þ
ðDvyj Dvyi Þ;
x2tcsci0 xtcsci0
1 1
Diyi ¼ 2 ðvxi0 vxj0 ÞDxtcsci þ ðDvxi Dvxj Þ;
xtcsci0 xtcsci0
ð6:64Þ
1 1
Dixj ¼ 2 ðvyi0 vyj0 ÞDxtcsci þ ðDvyi Dvyj Þ;
xtcsci0 xtcsci0
1 1
Diyj ¼ 2 ðvxj0 vxi0 ÞDxtcsci þ ðDvxj Dvxi Þ
xtcsci0 xtcsci0
That is
2 3
2 3 x21 ðvyj0 vyi0 Þ
Dixi 6 tcsci0
7
6 Di 7 6 1
ðvxi0 vxj0 Þ 7
6 yi 7 6 x 2 7
6 7¼6 7Dxtcsci
tcsci0
4 Dixj 5 6 7
6 x2tcsci0 ðvyi0 vyj0 Þ 7
1
4 5
Diyj x21 ðvxj0 vxi0 Þ ð6:65Þ
2 32 3
tcsci0
0 1 0 1 Dvxi
1 6 6 1 0 1 0 7 6 7
76 Dvyi 7
þ 6 76 7
xtcsci0 4 0 1 0 1 54 Dvxj 5
1 0 1 0 Dvyj
By using Eq. (6.61), the above equation becomes

DITCSCij ¼ CTCSCij Dxltcsci þ bTCSCij Dastcsci þ DTCSCij DVxyij ð6:66Þ
where
2 3 2 3
Dixi Dvxi
6 Diyi 7 6 Dvyi 7
6 7 6 7
DITCSCij ¼6 7; DVxyij ¼ 6 7;
4 Dixj 5 4 Dvxj 5
Diyj Dvyj
2 3
x21 ðvyj0 vyi0 Þ
6 tcsci0
7
6 21 ðvxi0 vxj0 Þ 7
1 6 xtcsci0 7
bTCSCij ¼ CTCSCij ¼ 6 7;
1 þ Kptcsc atcsc0 6 7
6 x2tcsci0 ðvyi0 vyj0 Þ 7
1
4 5
x21 ðvxj0 vxi0 Þ
2 3
tcsci0
0 1 0 1

Kptcsc atcsci 0 1 6 6 1 0 1 0 77
DTCSCij ¼ þ 6 7 ð6:67Þ
1 þ Kptcsc atcsc0 0 atcscj x tcsci0 4 0 1 0 1 5
1 0 1 0
Equations (6.63) and (6.66) are the linearized model of the TCSC without the TCSC
stabilizer installed. Output equation can be obtained from Eqs. (6.58) and (6.61) as
DPij ¼ dtcsci Dastcsci þ ctcsci Dxltcsci þ ptcsci DVxyi þ p0tcscj DVxyj ð6:68Þ
where
atcsc0 atcsc0
dtcsci ¼ ; ctcsci ¼ ;
1 þ Kptcsc atcsc0 1 þ Kptcsc atcsc 0

atcsc0 Kptcsc 0 atcsc0 Kptcsc
ptcsci ¼ 1 atcsci ; ptcscj ¼ 1 atcscj
1 þ Kpt csc atcsc0 1 þ Kptcsc atcsc0
When the TCSC stabilizer is installed as shown in Eq. (6.55), denote

Ttcsci ðsÞ ¼ Tltcsci ðsÞ þ s1 Tstcsci ðsÞ. From Eqs. (6.56) and (6.58), it can have
Ttcsci ðsÞ
Dxtcsci ¼ ðatcsci DVxyi þ atcscj DVxyj Þ
1 þ atcsc0 Ttcsci ðsÞ ð6:69Þ
¼ T0t csc i ðsÞðatcsci DVxyi þ atcscj DVxyj Þ
Let the state-space realization of Eq. (6.69) be
X_ TCSCi ¼ ATCSCi XTCSCi þ BTCSCi DVxyi þ BTCSCj DVxyj ð6:70Þ
Dxtcsci ¼ CTCSCi XTCSCi þ DTCSCi DVxyi þ DTCSCj DVxyj ð6:71Þ
By substituting Eq. (6.71) into Eq. (6.65), it can be obtained that
DITCSCij ¼ CTCSCij XTCSCi þ DTCSCij DVxyij ð6:72Þ
where
2 3
2 3 x21 ðvyj0 vyi0 Þ
Dixi 6 tcsci0
7
6 Di 7 6 21 ðvxi0 vxj0 Þ 7
6 7 6 x 7
7; CTCSCij ¼ 6 7
yi
DITCSCij ¼6 6 1 ðv v Þ 7CTCSCi ;
tcsci0
4 Dixj 5 6 x2tcsci0 yi0 yj0 7

4 5
Diyj x21 ðvxj0 vxi0 Þ
2 3 2 3
tcsci0
Dixi 0 1 0 1
6 Di 7 6 1
6 yi 7 D TCSCi 0 1 6 0 1 0 7
7
DVxyij ¼6 7; DTCSCij ¼ þ 6 7
4 Dixj 5 0 DTCSCj xtcsci0 4 0 1 0 1 5
Diyj 1 0 1 0
Equations (6.70) and (6.72) are the linearized model of the TCSC with stabilizer
installed between ith and jth nodes in the N-machine power system.
6.1.2.4 Linearized Model of a TCPS Installed in the N-Machine Power

System
Without loss of generality, it can be assumed that a TCPS is installed between

nodes i and j in an N-machine power system as shown in Fig. 6.9. From Fig. 6.9, it
can have
0 Iij
V
i ¼ I0 ¼ e ð6:73Þ
i j/i
V ij
0 j/i
I0 ¼ Vi Vj ¼ Vi e Vi þ Vi Vj
ij
zij zij
ð6:74Þ
j/i i V
j þ V
j V
j ej/i
Iij ¼ Iij ej/i ¼ Vi Vj e ¼
V
zij zij
Hence, installation of the TCPS is electrically equivalent to the addition of a

current source at nodes i and j, respectively, as shown in Fig. 6.9, where
j/i
Ij ¼ Vi e Vi
zij
ð6:75Þ
j/i j
V
Ii ¼ Vj e
zij
Denote z1ij ¼ gij þ jbij . From Eq. (6.75), it can be obtained that
ixi ¼ ðgij vxj þ bij vyj Þðcos/i 1Þ þ ðgij vyj þ bij vxj Þsin/i
iyi ¼ ðgij vxj þ bij vyj Þsin/i þ ðgij vyj þ bij vxj Þðcos/i 1Þ
ð6:76Þ
ixj ¼ ðgij vxi þ bij vyi Þðcos/i 1Þ ðgij vyi þ bij vxi Þsin/i
iyj ¼ ðgij vxi þ bij vyi Þsin/i þ ðgij vyi þ bij vxi Þðcos/i 1Þ
Vi Vj
z iijj
Pij
Vi Vj
I ij I ij ' z iijj
Vi '
k :1
Ii Ij
k =1
Fig. 6.9 Installation of a TCPS between node i and j in an N-machine power system
Dixi ¼ ½ðgij vxj0 þ bij vyj0 Þ sin /i0 þ ðgij vyj0 þ bij vxj0 Þ cos /i0 D/i
þ ½gij ðcos /i0 1Þ þ bij sin /i0 Dvxj þ ½bij ðcos /i0 1Þ þ gij sin /i0 Dvyj
Diyi ¼ ½ðgij vxj0 þ bij vyj0 Þ cos /i0 ðgij vyj0 þ bij vxj0 Þ sin /i0 þ D/i
þ ½gij sin /i0 þ bij ðcos /i0 1ÞDvxj þ ½bij sin /i0 þ gij ðcos /i0 1ÞDvyj
Dijx ¼ ½ðgij vxi0 þ bij vyi0 Þ sin /i0 ðgij vyi0 þ bij vxi0 Þ cos /i0 D/i
þ ½gij ðcos /i0 1Þ bij sin /i0 Dvxi þ ½bij ðcos /i0 1Þ gij sin /i0 Dvyi
Dijy ¼ ½ðgij vxi0 þ bij vyi0 Þ cos /i0 ðgij vyi0 þ bij vxi0 Þ sin /i0 D/i
þ ½gij sin /i0 þ bij ðcos /i0 1ÞDvxi þ ½bij sin /i0 þ gij ðcos /i0 1ÞDvyi
That is
DITCPSij ¼ bTCPSi D/i þ DTCPSi DVxyij ð6:77Þ
where
2 3 2 3
Dixi Dvxi
6 Di 7 6 Dv 7
6 yi 7 6 yi 7
DITCPSij ¼6 7; DVxyij ¼ 6 7;
4 Dixj 5 4 Dvxj 5
Diyj Dvyj
2 3
ðgij vxj0 þ bij vyj0 Þ sin /i0 þ ðgij vyj0 þ bij vxj0 Þ cos /i0
6 ðg vxj0 þ bij vyj0 Þ cos / ðg vyj0 þ bij vxj0 Þ sin / 7
6 ij i0 ij i0 7
bTCPSi ¼6 7;
4 ðgij vxi0 þ bij vyi0 Þ sin /i0 ðgij vyi0 þ bij vxi0 Þ cos /i0 5
ðgij vxi0 þ bij vyi0 Þ cos /i0 ðgij vyi0 þ bij vxi0 Þ sin /i0

0 DTCSCij
DTCPSi ¼
DTCSCii 0
" #
gij ðcos /i0 1Þ þ bij sin /i0 bij ðcos /i0 1Þ þ gij sin /i0
DiTCSCj ¼
gij sin /i0 þ bij ðcos /i0 1Þ bij sin /i0 þ gij ðcos /i0 1Þ
" #
gij ðcos /i0 1Þ bij sin /i0 bij ðcos /i0 1Þ gij sin /i0
DiTCSCi ¼
gij sin /i0 þ bij ðcos /i0 1Þ bij sin /i0 þ gij ðcos /i0 1Þ
Typical application of a TCPS is to regulate power flow. Let the TCPS load flow
controller be implemented by a PI control law and a stabilizing control signal,
Dastcpsi , be added as shown in Fig. 6.10. It can have
Fig. 6.10 A PI TCPS load

K ptcps
flow controller superimposed
with a stabilizing signal
ΔPij
+ + Δφi
K itcps Δα stcpsi
s Δx ltcpsi
D/i ¼ Dastcpsi þ Kpt c ps DPij þ Dxltcpsi

ð6:78Þ
Dx_ ltcpsi ¼ Kit c ps DPij
j/i
Vi Vj e
i Iij ¼ V
Sij ¼ V i z ij
j ej/i Þ ð6:79Þ
¼ ðgij þ jbij ÞðVi Vi V
2
¼ ðgij þ jbij ÞV2i ðgij þ jbij Þðvxi jvyi Þðvxj þ jvyj Þðcos /i j sin /i Þ
Hence,
Pij ¼ ReðSij Þ ¼ gij ðv2xi þ v2yi Þ

þ gij ½ðvxi vxj þ vyi vyj Þ cos /i þ ðvxi vyj vyi vxj Þ sin /i ð6:80Þ
bij ½ðvxi vxj þ vyi vyj Þ sin /i þ ðvxi vyj vyi vxj Þ cos /i
DPij ¼ at c ps0 D/i þ atcpsi DVxyi þ atcpsj DVxyj ð6:81Þ
where
at c ps0 ¼ gij ½ðvxi0 vxj0 þ vyi0 vyj0 Þ sin /i0 ðvxi0 vyj0 vyi0 vxj0 Þ cos /i0
þ bij ½ðvxi0 vxj0 þ vyi0 vyj0 Þ cos /i0 þ ðvxi0 vyj0 vyi0 vxj0 Þ sin /i0
" #T
2gij vxi0 þ gij ðvxj0 cos /i0 þ vyj0 sin /i0 Þ bij ðvxj0 sin /i0 þ vyj0 cos /i0 Þ
atcpsi ¼
2gij vyi0 þ gij ðvyj0 cos /i0 vxj0 sin /i0 Þ þ bij ðvyj0 sin /i0 þ vxj0 cos /i0 Þ
" #T
gij ðvxi0 cos /i0 vyi0 sin /i0 Þ þ bij ðvxi0 sin /i0 þ vyi0 cos /i0 Þ
atcpsj ¼
gij ðvyi0 cos /i0 þ vxi0 sin /i0 Þ bij ðvyi0 sin /i0 þ vxi0 cos /i0 Þ
1
D/i ¼ Dastcpsi
1 þ atcps0 Kptcps
Kptcps 1
þ ðatcpsi DVxyi þ atcpsj DVxyj Þ þ Dxltcpsi
1 þ atcps0 Kptcps 1 þ atcps0 Kptcps
ð6:82Þ
atcps0 Kitcps atcps0 Kitcps

Dx_ ltcpsi ¼ Dxltcpsi þ Dastcpsi
1 þ atcps0 Kptcps 1 þ atcps0 Kptcps
ð6:83Þ
atcps0 Kitcps Kptcps
þ ðKptcps ðatcpsi DVxyi þ atcpsj DVxyj Þ
1 þ atcps0 Kitcps
Dx_ ltcpsi ¼ Altcpsi Dxltcpsi þ BLTCPSi DVxyi þ BLTCPSj DVxyj þ bltcpsi Dastcpsi ð6:84Þ
DITCPSij ¼ CTCPSij xltcpsi þ bTCPSij Dastcpsi þ DTCPSij DVxyij ð6:85Þ
where
1
bTCPSij ¼ CTCPSij ¼ bTCPSi
1 þ at c ps0 Kpt c ps
ð6:86Þ
Kpt c ps atcpsi 0
DTCPSij ¼ þ DTCPSi
1 þ at c ps0 Kpt c ps 0 atcpsj
Equations (6.84) and (6.85) are the linearized model of the TCPS without the
TCPS stabilizer installed. From Eqs. (6.81) and (6.82), the output equation can be
obtained as
at c ps0 at c ps0
DPij ¼ Dastcpsi þ Dxltcpsi
1 þ at c ps0 Kpt c ps 1 þ at c ps0 Kpt c ps
1 ð6:87Þ
þ ðatcpsi DVxyi þ atcpsj DVxyj Þ
1 þ at c ps0 Kpt c ps
¼ dtcpsi Dastcpsi þ ctcpsi Dxltcpsi þ ptcpsi DVxyi þ p0tcpsj DVxyj
Let the transfer function of the TCPS load flow controller be Tltcpsi ðsÞ. It is
assumed that the transfer function of the TCPS stabilizer and feedback signal is
Tltcpsi ðsÞ and the integral of deviation of the active power delivered along the
transmission line, DPij , where the TCSC is installed. When the TCPS stabilizer is
installed, from Fig. 6.10, it can have

1
D/i ¼ Tltcpsi ðsÞ þ Tstcpsi ðsÞ DPij ¼ Ttcpsi ðsÞDPij
s
Ttcpsi ðsÞ
D/i ¼ ðatcpsi DVxyi þ atcpsj DVxyj Þ
1 þ at c ps0 Ttcpsi ðsÞ ð6:88Þ
¼ T0tcpsi ðsÞðatcpsi DVxyi þ atcpsj DVxyj Þ
Let the state-space realization of Eq. (6.88) be
X_ TCPSi ¼ ATCPSi XTCPSi þ BTCPSi DVxyi þ BTCPSj DVxyj ð6:89Þ
D/i ¼ CTCPSi XTCPSi þ DTCPSi DVxyi þ DTCPSj DVxyj ð6:90Þ
By substituting Eq. (6.90) into Eq. (6.77), it can be obtained that
DITCSCij ¼ CTCSCij XTCSCi þ DTCSCij DVxyij ð6:91Þ
where
CTCSCij ¼ bTCPSi CTCPSi ;

bTCPSi DTCPSi 0
DTCSCij ¼ þ DTCPSi
0 DTCPSj
Equations (6.89) and (6.91) are the linearized model of the TCPS with the
stabilizer installed between the ith and jth nodes in the N-machine power system.
6.1.2.5 General Linearized Model of the N-Machine Power System

Without the PSSs and Thyristor-Based FACTS Stabilizers
Installed
Similar to Eq. (5.52), network equations of M nodes in the N-machine power

system installed with thyristor-based FACTS stabilizers can be written as
2 3 2 32 3
I1 Y11 Y12 ... Y1M V1
6 I2 7 6 Y21 Y22 ... Y21 76 V2 7
6 7 6 76 7
6 .. 7¼6 .. .. .. .. 76 .. 7 ð6:92Þ
4 . 5 4 . . . . 54 . 5
IM YM1 YM1 . . . YMM VM
where

ixi v g bij
Ii ¼ ; Vi ¼ xi ; Yij ¼ ii ; i; j ¼ 1; 2; . . .; M ð6:93Þ
iyi vyi bji gjj
Let the number of the SVCs, TCSCs, and TCPSs installed in the system be Lsvc ,
Ltcsc and Ltcps , respectively. Denote
T
DISVC ¼ DITSVC1 DITSVC2 . . . DITSVCLsvc ;
h iT
DITCSC ¼ DITTCSC1j DITTCSC2j . . . DITTCSCLtcsc j ;
h iT
DITCPS ¼ DITTCPS1j DITTCPS2j . . . DITTCPSLtcps j ;
T ð6:94Þ
DVSVC ¼ DVTSVC1 DVTSVC2 . . . DVTSVCLsvc ;
h iT
DVTCSC ¼ DVTTCSC1j DVTTCSC2j . . . DVTTCSCLtcsc j ;
h iT
DVTCPS ¼ DVTTCPS1j DVTTCPS2j . . . DVTTCPSLtcps j
When the SVCs, TCSCs, and TCPSs are not equipped with stabilizers,
Eqs. (6.42) and (6.48) are the linearized model of the ith SVC, Eqs. (6.63) and
(6.66) are that of the ith TCSC, and Eqs. (6.84) and (6.85) are that of the ith TCPS.
Hence, by using the definition of Eq. (6.94), the open-loop (without thyristor-based
FACTS stabilizers), state-space representation of Lsvc SVCs, Lt csc TCSCs, and Ltcps
TCPSs can be obtained, respectively, from Eqs. (6.42) and (6.48), Eqs. (6.63) and
(6.66), and Eqs. (6.84) and (6.85) as
X_ VSVC ¼ AVSVC XVSVC þ BVSVC DVSVC

ð6:95Þ
DISVC ¼ CSVC XVSVC þ DSVC DVSVC þ bSSVC DaS
X_ LTCSC ¼ ALTCSC XLTCSC þ BTCSC DVTCSC þ bLTCSC DaSTCSC

ð6:96Þ
DITCSC ¼ CTCSC XLTCSC þ DTCSC DVTCSC þ bTCSC DaSTCSC
X_ LTCPS ¼ ALTCPS XLTCPS þ BTCPS DVTCPS þ bLTCPS DaSTCPS

ð6:97Þ
DITCPS ¼ CTCPS XLTCPS þ DTCPS DVTCPS þ bTCPS DaSTCPS
where
2 3 2 3 2 3
XVSVC1 Dxlt csc 1 Dxlt c ps1
6 X 7 6 Dx 7 6 7
6 VSVC2 7 6 lt csc 2 7 6 Dxlt c ps1 7
6 7
XVSVC ¼ 6
6 .. 7; XLTCSC ¼ 6
7 6 .. 7; XLTCPS ¼ 6
7 .. 7;
4 . 5 4 . 5 6 . 7
4 5
XVSVCLsvc Dxlt csc 1 Lt csc xltcpsLtcps
2 3 2 3 2 3
Das1 Dast csc 1 Dastcps1
6 Da 7 6 Da 7 6 7
6 s2 7 6 st csc 2 7 6 Dastcps2 7
6 7
DaS ¼ 6 7
6 .. 7; DaSTCSC ¼ 6
6 .. 7; DaSTCPS ¼ 6
7 .. 7
4 . 5 4 . 5 6 . 7
4 5
DasLsvc Dast csc Ltcsc DastcpsLtcps
Without loss of generality and for the simplicity of expression, let the order of
arranging the network equations of Eq. (6.92) be (1) N nodes of generator termi-
nals; (2) Lsvc nodes where Lsvc SVCs are installed; (3) Lt csc nodes where Lt csc
TCSCs are installed; (4) Ltcps nodes where Ltcps TCPSs are installed; and (5) other
nodes in the system. Linearized network equations of Eq. (6.92) can be written as
2 3 2 32 3
DIxy Ygg Ygs Ygc Ygp Ygo DVxy
6 DISVC 7 6 Ysg Yso 7 6 7
6 7 6 Yss Ysc Ysp 76 DVSVC 7
6 DITCSC 7 ¼ 6 Ycg Ycs Ycc Ycp Yco 7 6 DVTCSC 7 ð6:98Þ
6 7 6 76 7
4 DITCPS 5 4 Ypg Yps Ypc Ypp Ypo 54 DVTCPS 5
0 Yog Yos Yoc Yop Yoo DVO
where VO is the vector of the voltage at other nodes in the N-machine power
system.
The output equation of open-loop system of the SVC stabilizer is Eq. (6.40), that
of the TCSC stabilizer is Eq. (6.68), and that of the TCPS stabilizer is Eq. (6.87).
According to the notations in Eqs. (6.94)–(6.98), the output equation of open-loop
system of Lsvc SVC stabilizers, Lt csc TCSC stabilizers, and Ltcps TCPS stabilizers
can be written, respectively, as
ySVC ¼ pSVC DVSVC þ pSVCO DVO

yTCSC ¼ dTCSC DaSTCSC þ cTCSC DXLTCSC þ pTCSC DVTCSC ð6:99Þ
yTCPS ¼ dTCPS DaSTCPS þ cTCPS DXLTCPS þ pTCPS DVTCPS
where
T T
ySVC ¼ ½ DP1j DP1j DPLsvc j ; yTCSC ¼ ½ DP1j DP1j DPLt csc j ;
T
yTCPS ¼ DP1j DP1j DPLcps j ; dTCSC ¼ diagðdt csc i Þ;
cTCSC ¼ diagðct csc i Þ; dTCPS ¼ diagðdtcpsi Þ; cTCSC ¼ diagðctcpsi Þ;
2 0 3
psvc1
6 0 7
2 3 6 p0svc2 7
psvc1 6 7
0 6 .. 7
6 psvc2 7 6 . 7
6 7 6 7
pSVC ¼ 6
6 . 7; pSVCO ¼ 6
7 6 p0svcLsvc 77;
4 . . 5 6 7
0 6 0 7
6 0 7
psvcLsvc 6 7
4 5
..
.
2 3
ptcsc1
6 p0tcsc1 7
6 0 7
6 7
6 ptcsc2 7
6 7
6 p0tcsc2 7
pTCSC ¼6
6
7
7
6 .. 7
6 . 7
6 7
6 7
4 0 ptcscLtcsc 5
p0tcscLtcsc
2 3
ptcps1
6 7
6 p0tcps1 0 7
6 7
6 ptcps2 7
6 7
6 7
6 p0tcsc2 7
pTCPS ¼6 7
6 .. 7
6 . 7
6 7
6 7
6 0 ptcpsLtcps 7
4 5
p0tcpsLtcps
Substituting the second equation in Eqs. (6.95), (6.96), (6.97), and Eq. (6.29)
into Eq. (6.98), it can be obtained that

Cxy X þ Ba Du
0
22 3 2 33
Ygg Dgxy Ygs Ygc Ygo Ygo
66 7 6 Y 77
66 Ysg Yss DSVC Ysc Ysp 7 6 so 7 7
66 7 6 7 7 ½DV
6
¼6 4 Ycg Ycs Ycc DTCSC Ycp 5 4 Yco 5 7
7 DV
6 7 O
4 Ypg Yps Ypc Ypp DTCPS Ypo 5
½ og
Y Y os Y oc Yop Yoo

Y0M Y1o ½DV
¼
Y2o Yoo DVO
ð6:100Þ
where
2 3 2 3 2 3
Xg DVxy Cgxy
6X 7 6 DV 7 6 0 7
6 VSVC 7 6 SVC 7 6 CSVC 7
X¼6 7; DV¼6 7; Cxy ¼ 6 7;
4 XLTCSC 5 4 DVTCSC 5 4 CTCSC 5
0
XLTCPS DVTCPS CTCPS
2 3 2 3
Dupss 0
6 Da 7 6 0 7
6 S 7 6 bSSVC 7
Du ¼ 6 7; B a ¼ 6 7
4 DaSTCSC 5 4 bTCSC 5
0
DaSTCPS bTCPS
DVO ¼ Y1
oo Y2o DV
0 ð6:101Þ
Cxy X þ Ba Du ¼ ðYM Y1o Y1
oo Y2o ÞDV ¼ YM DV
By arranging the state equation (the first equation) in Eqs. (6.95), (6.96), (6.97),
and (6.29) together according to the notation in Eq. (6.101), it can have
X_ ¼ Axy X þ Bp Du þ Bxy DV ð6:102Þ
where
2 3
Agxy
6 0 7
6 AVSVC 7
Axy ¼ 6 7;
4 ALTCSC 5
0
ALTCPS
2 3 2 3
Bg Bgxy
6 0 7 6 0 7
6 0 7 6 BVSVC 7
Bp ¼ 6 7; Bxy ¼ 6 7
4 bLTCSC 5 4 BTCSC 5
0 0
bLTCPS BTCPS
Deleting DV in Eq. (6.102) by substituting the second equation of Eq. (6.101)

into Eq. (6.102), it can be obtained that
X_ ¼ AX þ BDu ð6:103Þ
where A ¼ Axy þ Bxy Y1 1

M Cxy ; B ¼ Bp þ Bxy YM Ba . Equation (6.103) is the state
equation of open-loop system without the PSS and thyristor-based FACTS stabi-
lizers installed.
Let the feedback signal of a PSS be the rotor speed of generator where it is
installed. Hence, the output equation of PSS can be written as
yPSS ¼ Dx ¼ ½ 0 I 0 Xg ¼ cPSS Xg ð6:104Þ
By using the notation of Eq. (6.100), from Eq. (6.99), the first equation in
Eq. (6.101) and (6.104), it can have
y ¼ C0 X þ D0 Du þ Pxy DV ð6:105Þ
where
2 3 2 3 2 3
yPSS cPSS 0
6y 7 6 0 7 6 0 7
6 SVC 7 0 6 0 7 0 6 0 7
y¼6 7; C ¼ 6 7; D ¼ 6 7;
4 yTCSC 5 4 cTCSC 5 4 dTCSC 5
0 0
yTCPS cTCPS dTCPS
2 3 2 3
0 0
6 0 7 6 p 1 7
6 pSVC 7 6 SVCO Yoo Y2o 7
Pxy ¼6 7þ6 7
4 pTCSC 5 4 0 5
0
pTCPS 0
Deleting DV in Eq. (6.105) by substituting the second equation of Eq. (6.101)

into Eq. (6.105), the output equation of the system about the PSS and
thyristor-based FACTS stabilizers can be obtained as
Dy ¼ CX þ DDu ð6:106Þ
where C ¼ C0 þ Pxy Y1 0 1

M Cxy ; D ¼ D þ Pxy YM Ba Du:
6.1.2.6 General Linearized Model of the N-Machine Power System

with the PSSs and Thyristor-Based FACTS Stabilizers
Installed
Equations (6.50) and (6.51) are the linearized model of an SVC with stabilizer
installed. Equations (6.70) and (6.72) are that of a TCSC with stabilizer installed.
Equations (6.89) and (6.91) are that of a TCPS with stabilizer installed. By using
the notation in Eqs. (6.94)–(6.97), the linearized model of Lsvc SVCs, Lt csc TCSCs,
and Ltcps TCPSs with stabilizers installed can be obtained from Eqs. (6.50) and
(6.51), Eqs. (6.70) and (6.72), and Eqs. (6.89) and (6.91), respectively, as
X_ SVC ¼ ASVC XSVC þ BSVC DVSVC

ð6:107Þ
DISVC ¼ CSVC XSVC þ DSVC DVSVC þ DSVCO DVO
X_ TCSC ¼ ATCSC XTCSC þ BTCSC DVTCSC

ð6:108Þ
DITCSC ¼ CTCSC XTCSC þ DTCSC DVTCSC
X_ TCPS ¼ ATCPS XTCPS þ BTCPS DVTCPS

ð6:109Þ
DITCPS ¼ CTCPS XTCPS þ DTCPS DVTCPS
where
2 3 2 3 2 3
XSVC1 XTCSC1 XTCPS1
6 XSVC2 7 6 XTCSC2 7 6 XTCPS2 7
6 7 6 7 6 7
XSVC ¼ 6 .. 7; XTCSC ¼ 6 .. 7; XTCPS ¼ 6 .. 7
4 . 5 4 . 5 4 . 5
XSVCLsvc XTCSCLtcsc XTCPSLtcps
By substituting the second equation in Eqs. (6.107), (6.108), (6.109), and

Eq. (6.33) into Eq. (6.98), it can be obtained that
" 0 #
Cp X Yp Y1po ½DV
¼
0 Y2o Yoo DVO
22 3 2 33
Ygg Dgp Ygs Ygc Ygo Ygo
66 7 6 Y D 77
6 6 Ysg Yss DSVC Ysc Ysp 7 6 so SVCO 7 7
66 7 6 77 ½DV
¼6
6 4 Ycg Ycs Ycc DTCSC Ycp 5 4 Yco 577 DV
6 7 O
4 Ypg Yps Ypc Ypp DTCPS Ypo 5
½ Yog Yos Yoc Yop Yoo
ð6:110Þ
where
2 3 2 3
Xg Cgp
6 XSVC 7 0
6 7 6 CSVC 7
X¼4 ;C ¼4 5
XTCSC 5 p CTCSC
0
XTCPS CTCPS

1
DVO ¼ Yoo Y2o DV
0
1
ð6:111Þ
Cp X ¼ ðYp Y1po Yoo Y2o ÞDV ¼ Yp DV
By arranging the state equation (the first equation) in Eqs. (6.107), (6.108),
(6.109), and (6.33) together, it can have
X_ ¼ Ap X þ B0p DV ð6:112Þ
where
2 3
Agp
6 0 7
6 ASVC 7
Ap ¼ 6 7;
4 ATCSC 5
0
ATCPS
2 3
Bgp
6 0 7
6 BSVC 7
B0p ¼ 6 7
4 BTCSC 5
0
BTCPS
X_ ¼ Aclose X ð6:113Þ
where Aclose ¼ Ap þ B0p Y1

p Cp . Equation (6.113) is the state equation of the
closed-loop system with the PSSs and thyristor-based FACTS stabilizers installed
in the N-machine power system.
6.2 Analysis and Damping Control of Thyristor-Based

FACTS Stabilizers Installed in a Multi-machine Power
System
6.2.1 Damping Torque Analysis in a Multi-machine Power

System
6.2.1.1 Damping Torque Contribution from a Stabilizer Derived

from the Heffron–Phillips Model
Equation (5.43) is the Heffron–Phillips model of the N-machine power system with
the PSSs installed. As this is a linear system satisfying the principle of superim-
position, it only needs to consider the function of one PSS in the discussion. Now, if
6.2 Analysis and Damping Control … 267
K1
-
ω0 I
(sM + D) −1
s
- TPSS
K4 K5
K2
- -
+ Δu pssk
−1 −1 +
(K 3 + sTd0 ) (I + sTA ) K A Fk
-
K6
Fig. 6.11 Heffron–Phillips model of a multi-machine power system installed with a PSS
only the kth PSS is considered, the Heffron–Phillips model of the power system can
be obtained from Eq. (5.43) as
sDd ¼ xo IDx
ð6:114Þ
sDE0q ¼ T01 0 0
sDE0fd ¼ T1 0 1 0
A DEfd þ TA KA ðK5 Dd K6 DEq þ Fk Dupssk Þ
where Fi ¼ ½ 0 ... 0 1 0 . . . 0 T and Dupssk is the output stabilizing

"
the kth element
signal from the kth PSS. From Eq. (6.114), the Heffron–Phillips model with the kth
PSS installed is shown in Fig. 6.11.
Similar to the procedure of damping torque analysis for a PSS installed in a
single-machine infinite-bus power system as described in Chap. 2, from Fig. 6.11,
the transfer function matrix from the PSS stabilizing signal to the electromechanical
oscillation loops of all generators can be obtained as
DTPSS
¼ M1 K2 ½ðI þ sTA ÞðK3 þ sTd0 Þ þ KA K6 1 KA Fk ¼ FPSS ðsÞ ð6:115Þ
Dupssk
Let the transfer function and feedback signal of the PSS be Tpssk ðsÞ and Dyk ,
respectively, that is
Dupssk ¼ Tpssk ðsÞDyk ð6:116Þ
The feedback signal Dyk can always be reconstructed by using Dxj ; j ¼

1; 2; . . .; N (see Sect. 6.2.1.2), that is
Dyk ¼ cj ðsÞDxj ; j ¼ 1; 2; . . .; N ð6:117Þ
If the low-frequency power oscillation of interests is associated with the ith

i ¼ n þ jxi , the electric torque contribution
oscillation mode of the power system, k i
of the PSS at the complex angular oscillation frequency k i ¼ n þ jxi can be
i
obtained from Eqs. (6.115) to (6.117) as
2 3
i ÞT
pss1 ðk pssk ðk i Þc ðk

F 1 i ÞDx1
6 7
6 7
DTPSS ¼F i )T
PSS ðk i ÞDy ¼ 6 Fpss2 ðki ÞTpssk ðki Þc2 ðki ÞDx2 7
pssk ðk ð6:118Þ
k 6 .. 7
4 . 5
i ÞT
pssN ðk
F pssk ðki Þc ðki ÞDxN
N
where F i Þ; j ¼ 1; 2; . . .N is the jth element of F

pssj ðk i Þ. Hence, the damping
PSS ðk
torque contribution from the PSS to the electromechanical oscillation loop of each
generator is
i ÞT
~pssj ðk
DTdampj ¼ Ddampj Dxj ¼ Re½F i Þc ðk
pssk ðk
j i ÞDxj ð6:119Þ
From the unified Heffron–Phillips model of the N-machine power system

installed with a thyristor-based FACTS stabilizer of Fig. 6.5, the transfer function
matrix from the FACTS stabilizing signal to the electromechanical oscillation loops
of all generators can be obtained as
DTFACTS
¼ FFACTS ðs)
Dufactss
¼ M1 KP M1 K2 ½ðI þ sTA ÞðK3 þ sTd0 Þ þ KA K6 1 ½ðI þ sTA ÞKq þ KA KV
ð6:120Þ
Similarly, let the transfer function and feedback signal of the thyristor-based
FACTS stabilizer be Tfacts ðsÞ and Dyk , respectively, that is
Dufactss ¼ Tfacts ðs)Dyk ð6:121Þ
The electric torque contribution of the thyristor-based FACTS stabilizer at the

complex angular oscillation frequency k i ¼ n þ jxi can be obtained from
i
Eqs. (6.117), (6.120), and (6.121) as
2 3
i ÞT
facts1 ðk
F ~ facts ðki Þ i ÞDx1
c1 ðk
6F i ÞT
facts2 ðk facts ðki Þ ÞDx2 7
c2 ðk
DTFACTS ¼F i ÞT
FACTS ðk i ÞDy ¼ 6
facts ðk 6 .
i 7
7 ð6:122Þ
k
4 .. 5
i ÞT
factsN ðk
F facts ðki Þ
c ðki ÞDxN
N
where F i Þ; j ¼ 1; 2; . . .; N is the jth element of F

factsj ðk i Þ. Hence, the
FACTS ðk
damping torque contribution from the thyristor-based FACTS stabilizer to the
electromechanical oscillation loop of each generator is
i ÞT
factsj ðk
DTdampj ¼ Ddampj Dxj ¼ Re½F i Þc ðk
facts ðk
j i ÞDxj ð6:123Þ
6.2.1.2 Damping Torque Contribution from a Stabilizer Derived

from the General Linearized Model of an N-Machine
Power System
The open-loop general linearized model of the N-machine power system is the state
equation of Eq. (6.103), that is
X_ ¼ AX þ BDu ð6:124Þ
where
2 3 2 3 2 3
Xg Dupss Dupss1
6X 7 6 Da 7 6 Dupss2 7
6 VSVC 7 6 7 6 7
7; Dupss ¼ 6 7;
S
X¼6 7; Du ¼ 6 6 . 7
4 XLTCSC 5 4 DaSTCSC 5 4 .. 5
XLTCPS DaSTCPS DupssN
2 3 2 3 2 3
Das1 Dastcsc1 Dastcps1
6 Da 7 6 Da 7 6 7
6 s2 7 6 stcsc2 7 6 Dastcps2 7
6 7 6 7 6 7
DaS ¼ 6 . 7; DaSTCSC ¼ 6 .. 7; DaSTCPS ¼ 6 .. 7
4 .. 5 4 . 5 6
4 . 7
5
DasLsvc DastcscLtcsc DastcpsLtcps
Equation (6.124) can be rearranged into the following form:

2 3 2 32 Dd 3 2 3
Dd_ x0 I 0
6 Dx_ 7 4
0 0
6 7 X 6 7
4 5 ¼ A21 A22 A23 54 Dx 5 þ 4 B2k 5Duk ð6:125Þ
DZ_ A31 A32 A33 DZ k
B3k
where Duk is the output stabilizing signal of the kth stabilizer installed in the power
system, which can be a PSS or a thyristor-based FACTS stabilizer, that is
Duk ¼ Dupssj ; j ¼ 1; 2; . . .; N; Duk ¼ Dasj ; j ¼ 1; 2; . . .; Lsvc ;

Duk ¼ Dastcscj ; j = 1,2,. . .; Ltcsc ; Duk ¼ Dastcpsj ; j ¼ 1; 2; . . .; Ltcps
A 21
A 22
Δ Δ
1 ω0 I
+
s s
Δu k
B 2k B3k A 32 A 31
A 23
1
Δz +
s
A 33
Fig. 6.12 General linearized model of an N-machine power system
The general linearized model of the N-machine power system of Eq. (6.125) is
shown in Fig. 6.12. From the third row of Eq. (6.125), it can be obtained that
!
1
X
DZ ¼ ðsI A33 Þ A31 Dd þ A32 Dx þ B3k Duk ð6:126Þ
k
Substituting the above equation into Eq. (6.125), it can have
Dd_ ¼ x0 Dx
Dx_ ¼ ½A21 þ A23 ðsI A33 Þ1 A31 Dd þ ½A22 þ A23 ðsI A33 Þ1 A32 Dx
X
þ ½B2k þ A23 ðsI A33 Þ1 B3k Duk ð6:127Þ
k
X
¼ A21 ðsÞDd þ A22 ðsÞDx þ Bk ðs)Duk
k
Hence, the general linearized model of Fig. 6.12 is shown by its compact form in
Fig. 6.13.
Let the transfer function and feedback signal of the kth stabilizer be Tstk ðsÞ and
Dyk , respectively, that is
A 21 (s)
A 22 (s)
Δ Δ
1 ω0 I
+
s s
Δu k
Σ B k (s)
Fig. 6.13 Compact form of general linearized model of an N-machine power system
Duk ¼ Tstk ðsÞDyk ð6:128Þ
From Fig. 6.13, Eqs. (6.117), and (6.128), the electric torque contribution from
the kth stabilizer can be obtained as
2 3
i ÞT
k1 ðk i Þc ðk
stk ðk
B 1 i ÞDx1
6 7
6 7
i ÞT
k ðk
DTST ¼ B i ÞDy ¼ 6 Bk2 ðki ÞTstk ðki Þc2 ðki ÞDx2 7
stk ðk k 6 .. 7
4 . 5

BkN ðki ÞTstk ðki Þc ðki ÞDxNN
i Þ; j ¼ 1; 2; . . .; N is the jth element of Bk ðk

where Bkj ðk i Þ. Hence, the damping
torque contribution from the kth stabilizer to the electromechanical oscillation loop
of each generator is (for j ¼ 1; 2; . . .; N)
i ÞT
kj ðk
DTdampj ¼ Ddampj Dxj ¼ Re½B i Þc ðk
stk ðk
j i ÞDxj ð6:129Þ
For the kth stabilizer, the general linearized model of the multi-machine power
system of Eq. (6.125) can also be rearranged as (for j ¼ 1; 2; . . .; N)
2 3 2 32 Dd 3 2 0 3
Dd_ j 0 x0 0 j
6 _ 7 4 6 7 6 7
4 D x j5 ¼ k j dj Aj 54 Dxj 5 þ 4 bjk 5Duk ð6:130Þ
DZ_ j Aj1 Aj2 Aj3 DZj Bjk

hx i
DZj ¼ ðsI Aj3 Þ1
0
Aj1 þ Aj2 þ Bjk Tstk ðsÞcj ðsÞ Dxj ð6:131Þ
s
The output equation of the system about the feedback signal of the stabilizer can
be written as
2 3
Ddj
6 7
Dyk ¼ c1kj c2kj C3kj 4 Dxj 5 þ djk Duk ð6:132Þ
DZj
By using Eqs. (6.117), (6.128), and (6.131), Eq. (6.131) becomes

x
1 x0
0
Dyk ¼ c1kj þ c2kj Dxj þ C3kj sI Aj3 Aj1 þ Aj2 Dxj
s s ð6:133Þ
½C3kj ðsI Aj3 Þ1 Bjk þ djk Tstk ðsÞcj ðsÞDxj
Dyk x0 x
c1kj þ c2kj þ C3kj ðsI Aj3 Þ1
0
cj ðsÞ ¼ ¼ Aj1 þ Aj2
Dxj s s ð6:134Þ
½C3kj ðsI Aj3 Þ1 Bjk þ djk Tstk ðsÞcj ðsÞ
Hence, finally, it can have
ðx0 c1kj þ c2kj Þ þ C3kj ðsI Aj3 Þ1 ðxs0 Aj1 þ Aj2 Þ
cj ðsÞ ¼ s ð6:135Þ
½C3kj ðsI Aj3 Þ1 Bjk þ djk Tstk ðsÞ
6.2.2 Selection of Installing Location and Feedback Signal

of a Stabilizer in a Multi-machine Power System
6.2.2.1 Selection of Installing Location of a PSS by Participation

Factor and Sensitivity Index
Selection of installing location and feedback signal of a stabilizer is considered

before a stabilizer is installed in a multi-machine power system. At this stage, both
the structure and parameters of the stabilizer are not decided. The selection is based
on the open-loop linearized model of the power system to predict the effectiveness
of the stabilizer. The participation factor, the sensitivity index, the residue, and the
damping torque analysis are among the most popular techniques for the selection.
Consider the general linearized open-loop system of Eqs. (6.130) and (6.132).
Denote the right and left eigenvectors of the state matrix with respect to the mode k i
to be (see Sect. 2.2.1 and 5.2.1.1)
2 3
v1i
vi ¼ 4 v2i 5; w Ti ¼ w 1i 2i
w T3i
w ð6:136Þ
v3i
The following participation factor (see Eq. (5.60)
jw 2i v2i j
pji ¼ T ð6:137Þ
i vi
w
measures how much the ith oscillation mode and the variation of rotor speed of the
jth generator are connected. If the connection is the highest among all generators as
measured by the above-defined participation factor, the jth generator is chosen to be
the best installing location of a PSS to suppress the power system oscillation
associated with the oscillation mode ki . The general procedure of the selection is to
arrange the state matrix of the system according to Eq. (6.125), that is
2 3
0 x0 I 0
A ¼ 4 A21 A22 A23 5 ð6:138Þ
A31 A32 A33
Denote the right and left eigenvectors of the above state matrix with respect to
i to be
the mode k
2 3
v1i T
vi ¼ 4 v2i 5; w
Ti ¼ w
1i T2i
w T3i
w ð6:139Þ
v3i
The participation factors of all generators can be calculated as

vj2i uj2i

pji ¼ T ðfor j ¼ 1; 2; . . .; NÞ ð6:140Þ
v2i
w2i
The installing location of the PSS is the generator with the highest participation
factor.
The sensitivity index for the selection of installing location of the PSS is
computed by adding an increment of artificial damping in the electromechanical
oscillation loop of each generator in the power system, Ddj Dxj (for j ¼ 1; 2; . . .; N).
If the maximum improvement, Dk i , is
i , for the oscillation mode of interests, k
achieved by the increment on the jth generator, it is selected as the best installing
location of the PSS. The physical explanation of the sensitivity index is based on
the understanding that the PSS is equivalent to an addition of damping torque in the
electromechanical oscillation loop of generators. Hence, if the addition of damping
torque on the jth generator achieves the maximum improvement of the oscillation
mode of interests, it should be the most effective installing location of the PSS as far
as the improvement of the oscillation mode of interests is concerned. The sensitivity
index in fact is the sensitivity of the oscillation mode of interests to the coefficient
of the added extra damping torque in the electromechanical oscillation mode, Ddj .
That is

ij ¼ Dki ¼ @ ki
S ð6:141Þ
Ddj @dj

2 32 3
0 0 0 v1i
ij ¼ @ ki ¼ w
S Ti
@A
vi ¼ w 1i 2i
w T3i 4 0
w 1 0 54 v2i 5 ¼ w
2i v2i ð6:142Þ
@dj @dj v3i
0 0 0
Equation (6.142) shows that in fact, the participation factor is the normalized
sensitivity index. Both methods are equivalent. Application of the sensitivity index
often only considers the real part of the oscillation mode of interests, as the
objective of installing stabilizer is to improve the damping of the oscillation mode.
Let the oscillation mode be k i ¼ n þ jxi . The sensitivity index used is
i
Dni
Sijn ¼ ð6:143Þ
Ddj
Though Eqs. (6.119) and (6.129) show that a PSS supplies damping torque to the
electromechanical oscillation loop of all generators in the power system, usually the
amount of damping torque contribution to the generator where it is installed is much
greater than that of other generators. Hence, participation factor and sensitivity
index, though they only tell how much a generator is connected to the oscillation
mode of interests, can be used to select the installing location of the PSS. In the case
of a thyristor-based FACTS stabilizer, however, they cannot be applied directly.
This is because how the thyristor-based FACTS stabilizer distributes its damping
torque contribution among all generators also plays a very important role in
determining how much it affects the oscillation modes of interest.
6.2.2.2 Selection of Installing Locations and Feedback Signals

of a Stabilizer by Damping Torque Analysis [2]
Equations (6.119) and (6.129) demonstrate that a stabilizer (the PSS or

thyristor-based FACTS stabilizer) installed in a power system contributes a
damping torque, Ddampj Dxj , to the electromechanical oscillation loop of every
generator in the system. By using the sensitivity index defined in Eqs. (6.141) or
(6.143), the addition of damping torque, Ddampj Dxj (for j ¼ 1; 2; . . .; N), on each
generator affects the oscillation mode of interests. Hence, the total improvement of
the oscillation mode of interests ki by the stabilizer can be obtained as
X
N X
N
i ¼
Dk ij ; or Dni ¼
Ddampj S Ddampj Sijn ð6:144Þ
j¼1 j¼1
Generator 1
Ddamp1 Si1
Generator 2
Ddamp2 Si2
The
oscillation
A
mode of
stabilizer
interests
DdampN SiN
Generator N
Damping torque contribution Damping torque contribution is

to each generator converted to the improvement
of the oscillation mode of
interests
Fig. 6.14 The oscillation mode of interests as affected by a stabilizer
where Ddampj is given by Eq. (6.119) for a PSS and Eq. (6.123) for a thyristor-based
FACTS stabilizer based on the Heffron–Phillips model as well as by Eq. (6.129) for
the PSS or thyristor-based FACTS stabilizer based on the general linearized model
of the N-machine power system.
Equation (6.144) is shown in Fig. 6.14. It presents a clear and full picture on how
the stabilizer (the PSS or thyristor-based FACTS stabilizer) contributes the damping
torque to every generator and then how the contribution of the damping torque is
converted to the influence on the oscillation mode of interests. Damping torque
analysis in the multi-machine power system of Eqs. (6.119), (6.123), and (6.129)
gives the left-hand half picture of Fig. 6.14 on how the stabilizer distributes the
damping torque to all generators. Participation factor or sensitivity index gives
equivalently the right-hand half picture of Fig. 6.14 on how the oscillation mode of
interests is connected to generators.
Based on Fig. 6.14 or Eq. (6.144), the following indices of the damping torque
analysis (DTA) can be proposed for the selection of installing location and feedback
signal of the stabilizer:
X
N
DTA ¼ F i Þc ðk
pssj ðk
j i ÞSij ðfor a PSS using HeffronPhillips modelÞ ð6:145Þ
j¼1
X
N
DTA ¼ F i Þc ðk
fsctsj ðk
j i ÞSij ð6:146Þ
j¼1
(for a thyristor-based FACTS stabilizer using Heffron–Phillips model)
X
N
DTA ¼ B i Þc ðk
kj ðk
j i ÞSij ð6:147Þ
j¼1
(for a PSS or thyristor-based FACTS stabilizer using general linearized model)

where cj ðsÞ ¼ ðxs0 c1kj þ c2kj Þ þ C3kj ðsI Aj3 Þ1 ðxs0 Aj1 þ Aj2 Þ which is given by
Eq. (6.135).
In the above DTA indices, often S ij is replaced by Sijn so as to focus the
selection on the influence of the stabilizer on the damping of the oscillation mode of
interests.
6.2.2.3 Residue Index and Its Connection with the Damping Torque
Analysis [3]
Consider the kth stabilizer to be installed in a multi-machine. The general linearized

model of the system is (see Eq. (6.125))
2 3 2 32 Dd 3 2 3
Dd_ 0 x0 I 0 0
6 7 76 7 6
6 Dx_ 7 ¼ 6 A22 A23 56 7 7
4 5 4 A21 4 Dx 5 þ 4 B2k 5Duk
DZ_ A31 A32 A33 DZ B3k
ð6:148Þ
2 3
Dd

T 6 7
Dyk ¼ CT1k CT2k C3k 4 Dx 5
DZ
where Dyk is the feedback signal of the stabilizer. If the right and left eigenvectors
i , are vi and w
of the state matrix with respect to the oscillation mode of interests, k Ti ,
respectively, the controllability, observability, and residue index of the stabilizer are
(see Sect. 2.2.1)
2 3
0
Ti 4 B2k 5; cik ¼ CT1k
bik ¼ w CT2k ik ¼ bikcik
CT3k vi ; R ð6:149Þ
B3k
Figure 6.15 shows that cik measures how much the oscillation mode k i is
observed in Dyk and bik measures how much the stabilizing signal Duk affects k i .

Hence, the residue index Rik measures how much the stabilizer is connected with
i . It can be used as an index for the selection of
the oscillation mode of interests k
the installing location and feedback signal of the stabilizer.
Fig. 6.15 Configuration of

1
the system in modal bik cik
decomposition s − λi
Channels associated with other

eigenvalues of the system
+
Δu k Δy k
stabilizer
In fact, the damping torque analysis and the residue introduced above for the
selection of installing location and feedback signal of the stabilizer are equivalent.
This equivalence can be proved as follows.
2 3
2 3 z1 ð0Þek1 t
Dd 6 z2 ð0Þek2 t 7
4 Dx 5 ¼ V6
6 ..
7
7 ð6:150Þ
4 . 5
DZ
zn ð0Þekn t
Hence,
X
N
v2ij zi ð0Þ
Dxj ðsÞ ¼ i ð6:151Þ
i¼1 sk
v2ij is the jth element of vi2 in Eq. (6.139). While

where
2 3
2 3 z1 ð0Þek1 t
Dd 6 k2 t 7
6 z2 ð0Þe 7
Dyk ¼ CT1k CT2k CT3k 4 Dx 5 ¼ CT1k CT2k CT3k V6 .. 7
4 . 5
DZ
zn ð0Þekn t
Hence,
X
N
cik zi ð0Þ
Dyk ðsÞ ¼ i ð6:152Þ
i¼1 sk

PN c ik zi ð0Þ i Þ PN c ik zið0Þ
Dyk ðsÞ i¼1 sk i ðs k i¼1 ski
¼ ¼ ð6:153Þ
Dxj ðsÞ PN v 2ij zi ð0Þ ðs k i Þ PN v 2ij zi ð0Þ
i¼1 ski i¼1 ski
i and using Eq. (6.117), it can be obtained

In the above equation, by letting s ¼ k
that
i Þ cik zi ð0Þ
Dyk ðk cik
i Þ ¼
cj ðk
i Þ ¼ v2ij zi ð0Þ ¼ v2ij
Dxj ðk
ð6:154Þ
On the other hand, from Eqs. (6.139) and (6.148), it can have
2 3
0 x0 I 0 T
T1i
w T2i
w T3i 4 A21
w A22 i w
A23 5 ¼ k 1i T2i
w T3i
w ð6:155Þ
A31 A32 A33
w T3i A33 ¼ w
T2i A23 þ w i
T3i k ð6:156Þ
Solution of Eq. (6.156) is
T3i ¼ w
w i I A33 Þ1
T2i A23 ðk ð6:157Þ
Hence, from Eqs. (6.148), (6.149), and (6.157), it can have

2
3 3 2
0 0
6 7 T 6 7
Ti 4 B2k 5 ¼ w
bik ¼ w 1i T2i w
w T3i 4 B2k 5 ¼ w T2i B2k þ w
T3i B3k
B3k B3k ð6:158Þ
X
¼w T i I A33 Þ B3k ¼ w
2i ½B2k þ A23 ðk 1 i Þ ¼
T2i Bk ðk i Þ
2ij Bkj ðk
w
j

2 3
0 x0 I 0
@ 4 A21 A22 þ diagðdj Þ A23 5
i
@k A31 A32 A33
Sij ¼ i
¼w T
vi ¼ w
2ij v2ij ð6:159Þ
@dj @dj
Hence, from Eqs. (6.154), (6.158), and (6.159), the residue index can be
obtained as
X X
ik ¼
bikcik ¼ i Þc ðk
kj ðk
2ij B ij B i Þc ðk
kj ðk
R w j i Þv2ij ¼ S j iÞ ð6:160Þ
j j
This proves that the DTA index of Eq. (6.147) for the selection of installing
ik j.
location and feedback signal is equivalent to the residue index jR
6.2.3 Selection of Robust Installing Locations and Feedback

Signals of a Stabilizer by an Eigensolution-Free
Method
6.2.3.1 Robust Selection of Installing Locations and Feedback Signals

of a Stabilizer [4]
A good design of a stabilizer must ensure the maximum effectiveness of the sta-
bilizer. In addition, the robustness of the stabilizer to the variations of power system
operation conditions is also an important factor to be considered in the design. In
Sect. 4.3.2.3, the issue of robust design of the stabilizer is introduced in the
selection of control functions of a UPFC to add the stabilizing control. In this
section, it is discussed how the robustness of the stabilizer is considered at the stage
of selecting the installing location and feedback signal.
Let the set of the candidate installing locations and feedback signals of the
stabilizer in a multi-machine power system be UðuÞ and that of power system
operating conditions XðlÞ. The effectiveness of the stabilizer is the function of φ
and μ, C(φ, μ). If only the maximum effectiveness of the stabilizer to damp a power
oscillation is considered, the criterion of selecting the installing location and
feedback signal of the stabilizer is
Max½Cðu; l0 Þ; l0 2 XðlÞ ð6:161Þ

u
In the discussion in previous sections, only the above criterion is considered.

In order to consider the robustness of the stabilizer to the variations of power
system operating conditions, the following two criteria should be added.
Max½min Cðu; lÞ ð6:162Þ

u l
Min½max Cðu; lÞ min Cðu; lÞ ð6:163Þ

u l l
1. The criterion of Eq. (6.162) means that the selection of installing location and
feedback signal is made at the operating condition where the stabilizer is least
effective. This selection guarantees that when the stabilizer operates at other
operating conditions, it is more effective. Hence, the selection is robust to the

variations of system operating conditions.
2. If the damping contribution from the stabilizer changes significantly, on the one
hand, with the variations of power system operating conditions, the control
could be overstrong at some operating conditions, which would pose much
unwanted influence on other modes in the power system. Hence, the criterion of
Eq. (6.163) requires that the change of damping contribution from the stabilizer
is minimum with the variations of power system operating conditions. However,
this criterion should be applied jointly with that of Eq. (6.162), since failing to
meet the requirement of the effectiveness obviously is not a proper selection.
Obviously, in order to apply the criteria of Eqs. (6.162) and (6.163), the effec-
tiveness of the stabilizer has to be examined over the range of all system operating
conditions. If a method dependent on the eigensolution of system state matrix, such
as the residue index, is used, it could mean a heavy computational burden. In the
following section, an equivalent residue index is derived which can be used for the
eigensolution-free selection of robust installing location and feedback signal of the
stabilizer.
6.2.3.2 An Equivalent Residue Index and Eigensolution-Free Selection

of Robust Installing Location and Feedback Signal
of the Stabilizer
Consider the kth stabilizer to be installed in the N-machine power system. The
linearized model of the system is given by Eqs. (6.130) and (6.132) as follows:
2 3 2 32 Dd 3 2 3
Dd_ j 0 x0 j0 0
6 7 76 7 6
6 Dx_ j 7 ¼ 6 Aj 56 7 7
4 5 4 kj dj 4 Dxj 5 þ 4 bjk 5Duk
DZ_ j Aj1 Aj2 Aj3 DZj Bjk
2 3 ð6:164Þ
Ddj
6 7
Dyk ¼ c1kj c2kj C3kj 6 7
4 Dxj 5
DZj
Let the right and left eigenvectors of the state matrix, respectively, be
2 3
v1i
vi ¼ 4 v2i 5; w Ti ¼ w 1i 2i
w T3i
w ð6:165Þ
v3i
According to the definition of the right and left eigenvectors, it can have
2 3
0 x0 0
6 7
1i
w 2i
w T3i 4 kj
w dj Aj 5 ¼ k 1i
i w 2i
w T3i
w
Aj1 Aj2 Aj3
2 32 3 2 3 ð6:166Þ
0 x0 0 v1i v1i
6 76 7 6 7
4 kj dj Aj 54 v2i 5 ¼ ki 4 v2i 5
Aj1 Aj2 Aj3 v3i v3i
Ti3 ¼ Aj ðk
w i I Aj3 Þ1 w
i2
x0
vi1 ¼ vi2
ki ð6:167Þ

i I Aj3 Þ1 Aj1 x0 þ Aj2 vi2
vi3 ¼ ðk i
k
From Eq. (6.167), controllability and observability index can be obtained,

respectively, as
2 3
0
6 7
bik ¼ w 1i 2i
w T3i 4 bjk 5
w
Bjk
i I Aj3 Þ1 Bjk w
¼ ½bjk þ Aj ðk 2i ¼ K i Þw
bi ðk i2
2 3
v1i
6 7
cik ¼ c1kj c2kj C3kj 4 v2i 5
v3i

x0 1 x0 i Þv2i
ci ðk
¼ c1kj þ c2kj þ C3kj ðki I Aj3 Þ ðAj1 þ Aj2 Þ v2i ¼ K
ki ki
ð6:168Þ
Hence, the equivalent residue index is
ik ¼ bikcik ¼ K
R i ÞK
bi ðk i Þvi2 w
ci ðk i2 ð6:169Þ
In the selection of installing location and feedback signal of the stabilizer to be

installed in the multi-machine power system, the residue is calculated for com-
parison among various candidate locations and feedback signals. For example, if
there are two candidate installing locations or feedback signals, A and B, and
ikA j [ jR
jR ikB j, then A is considered to be better than B to be the installing location

R
or the feedback signal of the stabilizer. Therefore, it is the ratio, , not the
ikA
RikB
values of the residue, that determines the selection.
On the other hand, in system state equation of Eq. (6.164), if the installing
location or the feedback signal of the stabilizer is different, the control and the
output vectors are not the same, but the open-loop state matrix is unchanged,
provided that the operating point of the system remains the same. That is,
vi2A ¼ vi2B ; w
i2A ¼ w
i2B ð6:170Þ
Therefore, from Eqs. (6.169) and (6.170), it can be obtained that

ikA j K
jR i ÞK
biA ðk i Þvi2A w
ciA ðk i2A

ikB j ¼ K
jR i ÞK
biB ðk i Þvi2B w
ciB ðk i2B
ð6:171Þ
K i ÞK
biA ðk i Þjvi2A w
ciA ðk i2A j K i ÞK
biA ðk i Þ
ciA ðk
¼ ¼
i ÞK
biB ðk
K i Þjvi2B w
ciB ðk i2B j K i ÞK
biB ðk i Þ
ciB ðk
Equation (6.171) shows that jK bi ðki ÞK

ci ðki Þj can replace jR ik j to be the index to
measure the effectiveness of the stabilizer so as to select the installing location and
feedback signal. Since in most cases, the oscillation
mode of interest is lightly
damped, i.e. ki jxi , so K ÞK
bi ðk Þ K
ci ðk ci ðjx Þ, the index can be
bi ðjx ÞK
i i i i
calculated without knowing the eigensolution of the open-loop state matrix.
Therefore, the selection can be made eigensolution-free with the following criteria:

bi ðjxi ÞK
Max Min½jK ci ðjxi Þj; for all l 2 XðlÞ ; for all u 2 UðuÞ
u l

bi ðjxi ÞK
ci ðjxi Þj Min½jK bi ðjxi ÞK
ci ðjxi Þj; ð6:172Þ
Min Max½jK
u l l
for all l 2 XðlÞg; for all u 2 UðuÞ
6.2.4 Stabilizer Design in a Multi-machine Power System

Considering Robustness and Interaction of Stabilizers
6.2.4.1 Selection of an Operating Condition for the Robust

Coordinated Design of Multiple Stabilizers in a Multi-machine
Power System
Section 4.3.2.3 discusses the selection of an operating condition for the design of a
robust stabilizer in a single-machine infinite-bus power system by using the phase
compensation method. The selection strategy presented in Sect. 4.3.2.3 is to choose
an operating condition where it is predicted that the stabilizer to be installed will be

least effective. Hence, once the stabilizer is designed at the selected operating
condition to be effective in damping power system oscillation, its effectiveness is
ensured when the system operates at other operating conditions.
In a multi-machine power system installed with multiple stabilizers, such a
strategy to select an operating condition for the design of robust stabilizers can be
similarly applied by using the damping torque analysis as follows.
Consider the kth stabilizer installed in the system to improve the damping of the
i ¼ n jxi , and use the following sensitivity index given by
ith oscillation mode, k i
Eq. (6.143)
X
N
Dni ¼ Ddampj Sijn ð6:173Þ
j¼1
X
N X
N
Dni ¼ Ddampj Sijn ¼ i Þc ðk
kj ðk
Re½B j i ÞSijn ð6:174Þ
j¼1 j¼1
which gives the prediction of the amount of damping provided by the kth stabilizer
to the ith oscillation mode, that is, the effectiveness of the stabilizer. Therefore, if it
is assumed that the set of all known candidate operating conditions of the
multi-machine power system is X0 , the following index can be used to predict the
effectiveness of the kth stabilizer to damp the ith oscillation mode at an operating
condition:
X
N
Wik ðmÞ ¼ i Þc ðk
kj ðk
Re½B j i ÞSijn ; lik ðmÞ 2 X0 ð6:175Þ
j¼1
If Wik ðmt Þ ¼ minðWk Þ; lik ðmÞ 2 X0 , it is determined that the kth stabilizer is
m
least effective in damping the ith oscillation mode at the operating condition
lik ðmt Þ 2 X0 . Hence, lik ðmt Þ 2 X0 can be chosen as the operating condition to
design the kth stabilizer. An effective design of the stabilizer at lik ðmt Þ 2 X0 can
ensure the effectiveness of the stabilizer at all operating conditions lik ðmt Þ 2 X0 .
Obviously, this selection above may result in different operating conditions for
different stabilizers and oscillation modes. So, it may be difficult or impossible to
select a common operating condition at which all stabilizers can be designed
simultaneously in coordination. In other words, if they are designed in coordination
at a common operating condition simultaneously, their robustness may not be
ensured due to the difference in the changing pattern of their effectiveness to the
variations of system operating conditions. This problem can be solved by designing
the multiple stabilizers at two stages. The first stage of the design is to set the phase
of stabilizers one by one in a sequence according to the selection of robust oper-
ating conditions of stabilizers. At the second stage, gains of all stabilizers are tuned
simultaneously in coordination to avoid the problem of eigenvalue drift.
For simplicity of expression, the following conventional form of the transfer
function of stabilizers is used:
sTwk 1 ð1 þ sTk2 Þð1 þ sTk4 Þ

Tstk ðsÞ ¼ Kk ¼ Kk Tk ðsÞ ð6:176Þ
1 þ sTwk 1 þ sTk ð1 þ sTk1 Þð1 þ sTk3 Þ
It is assumed that the kth stabilizer is designed mainly to improve the damping of
the ith oscillation mode in the power system, k i ¼ n jxi . Without loss of gen-
i
erality, it is assumed that at the operating condition of the kth stabilizer selected by
the use of Eq. (6.175), Si1n
Si2n

SiNn . A compensation angle /k is
chosen to satisfy
90 [ /k ukj [ 0 ; j ¼ 1; 2; . . .; N ð6:177Þ
where ukj is the phase of B i Þc ðk

kj ðk
j i Þ. If such a compensation angle cannot be
obtained, the selection of the compensation angle /k can be made among the first
(N − 1) machines to satisfy
90 [ /k ukj [ 0 ; j ¼ 1; 2; . . .; N 1 ð6:178Þ
This procedure can be continued until such a compensation angle is obtained.

Then, Tk ðsÞ is set to have a phase angle /k , that is, T i Þ ¼ jTk j\ / .
k ðk k
According to Fig. 6.14, as far as the setting of the transfer function of the stabilizer is
concerned, the damping contribution by the kth stabilizer to k i can be expressed as
X
N X
N
Dni ðkÞ ¼ Ddampj Sijn ¼ i Þc ðk
kj ðk
Re½B j i ÞSijn Re½DTstk ðki Þ
j¼1 j¼1
ð6:179Þ
X
N
¼ Sijn Hkj Tk cos bkj DKk ; j ¼ 1; 2; . . .; N
j¼1
where B i Þ
kj ðk i ÞT
cj ð k i Þ ¼ Hkj Tk \b ; j ¼ 1; 2; . . .; N and b ¼ / þ u .
stk ðk kj kj k kj
The procedure above in selecting the compensation angle /k ensures that
0 [ bkj [ 90 is tenable to the machines which are more sensitive to k i so that
they are provided with a positive synchronizing torque. Equation (6.179) shows the
damping contribution to the ith oscillation mode by the kth stabilizer set at the
selected operating condition. If there are total L stabilizers considered, the total
damping contributed from L stabilizers to the ith oscillation mode can be obtained
from Eq. (6.179) as
X
L L X
X N
Dni ¼ Dni ðkÞ ¼ Sijn Hkj Tk cos bkj DKk ; j ¼ 1; 2; . . .; N ð6:180Þ
k¼1 k¼1 j¼1
Let the target damping of the oscillation modes by the coordinated design of L
stabilizers be ni ; i ¼ 1; 2; . . .; L. An objective function can be formed as
X
L
JðKÞ ¼ Qi ½ni ðKÞ ni 2 ð6:181Þ
i¼1
where K ¼ ½ K1 K2 . . . KL T and Qi is the weight for the assignment of the ith

oscillation mode. The simultaneous tuning of gains of L stabilizers can be carried
out by using a method of optimization to find the solution of the objective function
defined by Eq. (6.181). This simultaneous tuning is conducted at the selected
operating condition of each stabilizer. At the selected operating condition of the first
stabilizer, the initial values of all gains are simply taken to be zero. From the
selected operating condition of the second stabilizer on, the gain values set by the
previous tuning are taken to be the initial values to start the current tuning. Also in
the tuning process, the following constraint is checked:

jni ðKÞj
ni ; i ¼ 1; 2; . . .; L ð6:182Þ
If Eq. (6.182) is satisfied for any kl ; l ¼ 1; 2; . . .; L, which is supposed to be

damped by the kth stabilizer, Ql ¼ 0 is set in Eq. (6.181). This arrangement frees
the eigenvalue drift towards the “good” direction, the left half complex plane over
the target value. Therefore, a unique final solution of gain tuning will be much
easier to be found.
6.2.4.2 Non-negatively Interactive Design of a Stabilizer

in a Multi-machine Power System [5]
For the design of multiple stabilizers (the PSSs or FACTS-based stabilizers) in a

multi-machine power system, the techniques of sequential setting require the sta-
bilizers to be designed one by one in a sequence. However, the stabilizer set later
may move the oscillation modes damped by the stabilizers designed previously.
This problem of “eigenvalue drift” caused by the interactions among the multiple
stabilizers has motivated the development of the coordinated tuning of multiple
stabilizers. By coordinated tuning, the parameters of all interactive stabilizers are set
simultaneously to achieve an accurate assignment of all oscillation modes of
interest in the power system so that the interactions among the stabilizers are fully
taken account of.
However, in the design of multiple stabilizers, only an approximate assignment
of oscillation modes is needed as long as the damping of the oscillation modes is
improved to meet a certain target value of damping. If the interactions among the
stabilizers are positive, in fact, extra benefits of damping provision can be obtained.
Hence, the positive interactions should be encouraged instead of being strictly
constrained by the coordinated tuning. While what ought to be avoided is the
negative interactions among multiple interactive stabilizers, which may result in the
provision of negative damping to some oscillation modes, at the same time that
positive damping is contributed by the design of the stabilizer. Furthermore, in
practice, it is not always the case that when a new stabilizer is put into a power
system, all old interactive stabilizers are adjustable so that the coordinated tuning
can be implemented. As long as the design of this new stabilizer can ensure no
provision of negative damping to all interactive oscillation modes in the power
system, it can be set without the coordinated tuning with other old stabilizers. In this
section, it is discussed how such a non-negatively interactive stabilizer can be
designed.
From Eq. (6.179), the damping provided by a stabilizer (the kth stabilizer) in a
multi-machine power system can be denoted as
X
N
Dni ðkÞ ¼ i Þc ðk
kj ðk
Re½B j i ÞSijn Re½DTstk ðki Þ
j¼1
( ) ð6:183Þ
X
N
¼ Re ½Sijn H i ÞDT
kij ðk i Þ
stk ðk ki DT
¼ Re½C i Þ
stk ðk
j¼1
P i Þ ¼ jC i Þ. If the target oscillation mode for

where C ki ¼ N Sijn Hkij ðk ki j\aki ðk
j¼1
the installation 0 ¼ n jx0
k
of the kth stabilizer is 0 and
0 Þ ¼ K k T
stk ðk
T 0 Þ\b ðk
k ðk 0 Þ, from Eq. (6.183) the damping improvement of the
k0
target oscillation mode by the stabilizer can be obtained as

k0 DT
Dn0 ¼ Re½C 0 Þ ¼ jC
stk ðk k0 jD½Kk T 0 Þ cos½b ðk
k ðk
k0 0 Þ þ ak0 ðk0 Þ ð6:184Þ
i ¼ n jxi ; i 6¼ 0, in the power system, the

For other oscillation modes, k i
damping increment of the ith oscillation mode due to the installation of the kth
stabilizer is

ki DT
Dni ¼ Re½C i Þ ¼ jC
stk ðk ki jD½Kk T i Þ cos½b ðk
k ðk
ki i Þ þ aki ðki Þ ð6:185Þ
To ensure a positive damping provision by the stabilizer to the target oscillation

mode and at the same time, the imposition of a non-negatively interaction upon the
ith oscillation mode, it should have
0 Þ þ ak0 ðk
cos½bk0 ðk 0 Þ [ 0; cos½b ðk

ki i Þ þ aki ðki Þ
0 ð6:186Þ
Therefore, a successful design of non-negatively interactive stabilizer is equiv-

i Þ, onto the first quadrant in the complex plane,
ki ðk
alent to shifting all the vectors, C
where k i ¼ n jxi ; i ¼ 0; 1; . . . are oscillation modes associated with the inter-
i
active stabilizers in the power system. From Eq. (6.184) to (6.186), obviously, it
can be seen that the following two factors decide whether the shifting may be
successful:
1. the difference between the angles ak0 ðk0 Þ and aki ðk
i Þ;
2. the setting of the phase angle of the stabilizer.
The most successful shifting for C i Þ is onto the positive real axis so that
ki ðk

cos½bki ðki Þ þ aki ðki Þ ¼ 1; i ¼ 0; 1; . . ..
In the case that aki ðk i Þ [ ak0 ðk0 Þ, to achieve the maximum efficiency of the
stabilizer to damp its target oscillation mode, according to Eq. (6.184), the phase of
the stabilizer could be set to be bk0 ðk 0 Þ ¼ ak0 ðk 0 Þ so that C 0 Þ is shifted onto
k0 ðk

the real axis with cos½bk0 ðk0 Þ þ ak0 ðk0 Þ ¼ 1. As shown in Fig. 6.16a, C 0 Þ may
k0 ðk
also be moved into the first quadrant. This results in a design of an effective and
non-negatively interactive stabilizer.
However, if aki ðk i Þ\ak0 ðk 0 Þ as shown in Fig. 6.16b, the priority of shifting may
have to be given to C 0 Þ so as to move it to the first quadrant and in the same
k0 ðk

time to shift Ck0 ðk0 Þ into the first quadrant instead of onto the real axis. Thus, the
design may not be most efficient as far as the damping of the target oscillation
mode, k 0 , is concerned. However, the stabilizer is positively interactive with k i .

In both cases above, if the angle between Ck0 ðk0 Þ and Cki ðki Þ is large and even
ak0 ðk0 Þ aki ðki Þ [ 90 as shown in Fig. 6.16c, there is the danger that when
0 Þ is shifted into the first quadrant, C
k0 ðk
C i Þ may not be able to be moved into
ki ðk
Fig. 6.16 Non-negatively (a) y

interactive design of a
stabilizer.
i Þ [ ak0 ðk
a aki ðk 0 Þ, Cki ∠α ki (λ i ) + βki (λ i )
i Þ\ak0 ðk
b aki ðk 0 Þ,
Ck 0 ∠α k 0 (λ 0 ) + βk 0 (λ 0 ) = 0
0 Þ aki ðk
c ak0 ðk i Þ [ 90
α k 0 (λ 0 ) x
Cki (λ i )
C k 0 (λ 0 )
(b) y
Ck 0 ∠α k 0 (λ 0 ) + βk 0 (λ 0 )
Cki ∠α ki (λ i ) + βki (λ i ) = 0
x
α ki (λ i )
C k 0 (λ 0 )
Cki (λ i )
(c) y
Cki ∠α ki (λ i ) + βki (λ i ) C ki (λ i )
Ck 0 ∠α k 0 (λ 0 ) + βk 0 (λ 0 ) = 0
x
α k 0 (λ 0 )
C k 0 (λ 0 )
the first quadrant in the same time which could result in a negative damping
provision to the ith oscillation mode. This means a negative interaction between the
stabilizer being designed and the stabilizer which damps the ith oscillation mode.
6.3 An Example Two-Area Four-Machine Power System
6.3.1 Linearized Model
6.3.1.1 System Parameters and Operating Conditions
Configuration of a four-machine eleven-node example power system is shown in

Fig. 6.17.
Line parameters are given in Table 6.1. Parameters of generators and the AVRs
are given in Table 6.2.
5 7 8 9
G1 1 6 10 11 3 G3
L7 L9
C7 C9
2 4
G2 G4
Fig. 6.17 Configuration of a four-machine example power system
Table 6.1 Network Node Node Resistance Reactance Susceptance

parameters
1 5 0.0000 0.0167 0.0000
2 6 0.0000 0.0167 0.0000
3 11 0.0000 0.0167 0.0000
4 10 0.0000 0.0167 0.0000
5 6 0.0025 0.0250 0.04375
6 7 0.0010 0.0100 0.0175
7 8 0.0110 0.1100 0.1925
7 8 0.0110 0.1100 0.1925
8 9 0.0110 0.1100 0.1925
8 9 0.0110 0.1100 0.1925
9 10 0.0010 0.0100 0.0175
10 11 0.0025 0.0250 0.04375
Table 6.2 Generator parameters

G1 G2 G3 G4
TJ 13.0 13.0 12.35 12.35
Xd 1.8000 1.8000 1.8000 1.8000
Xq 1.7000 1.7000 1.7000 1.7000
X0d 0.3000 0.3000 0.3000 0.3000
T0d0 8.000 8.000 8.000 8.000
D 0 0 0 0
M 6.5 6.5 6.175 6.175
KA 50 50 50 50
TA 0.1 0.1 0.1 0.1
Table 6.3 Results of load flow computation—node voltage and node injection power
Node Magnitude of Phase of voltage (deg.) Active Reactive
voltage power power
G1 1.0300 0.5362 7.0000 2.7299
G2 1.0100 0.3616 7.0000 4.4625
G3 1.0300 0 7.3319 3.1059
G4 1.0100 −0.1910 7.0000 5.1643
5 0.9923 0.4218 0.0000 0.0000
6 0.9434 0.2388 0.0000 0.0000
7 0.8989 0.0792 −9.6700 −1.0000
8 0.8636 −0.2000 0.0000 0.0000
9 0.8799 −0.4834 −17.6700 −1.0000
10 0.9320 −0.3153 0.0000 0.0000
11 0.9869 −0.1205 0.0000 0.0000
Initial conditions of the example power system for the load flow computation are
VG1 ¼ 1:03; VG2 ¼ 1:01; VG3 ¼ 1:03\0; VG4 ¼ 1:03;

PG1 ¼ 7:0; PG2 ¼ 7:0; PG4 ¼ 7:0;
Load L7 ¼ 9:67 þ j1:00; Load L9 ¼ 13:67 þ j1:00:
Results of load flow computation are given in Tables 6.3 and 6.4. Based on the
results of load flow, initial values of state variables can be calculated as follows.
6.3 An Example Two-Area Four-Machine Power System 291
Table 6.4 Results of load flow computation—line power and power loss
Starting node Ending node Power at SN Power at EN Power loss
(SE) (EN)
1 5 7.00 + j2.73 −7.00 − j1.84 0.00 + j0.88
2 6 7.00 + j4.46 −7.00 − j3.33 0.00 + j1.12
3 11 7.33 + j3.10 −7.33 − j2.11 0.00 + j0.99
4 10 7.00 + j5.16 −7.00 − j3.93 0.00 + j1.23
5 6 7.00 + j1.82 −6.86 − j0.57 0.13 + j1.25
6 7 13.87 + j3.88 −13.63 − j1.58 0.23 + j2.30
7 8 1.98 + j0.21 −1.93 + j0.04 0.05 + j0.25
7 8 1.98 + j0.21 −1.93 + j0.04 0.05 + j0.25
8 9 1.93 − j0.19 −1.87 + j0.44 0.05 + j0.25
8 9 1.93 − j0.19 −1.87 + j0.44 0.05 + j0.25
9 10 −13.93 − j2.04 14.18 + j4.57 0.25 + j2.53
10 11 −7.18 − j0.67 7.33 + j2.08 0.15 + j1.42
From the output active and reactive power of each generator, pi þ jQi , and its
terminal voltage,Vgi0 ; i ¼ 1; 2; 3; 4 ; output current of each generator can be cal-
culated as
Pi jQi
Igi0 ¼ ; i ¼ 1; 2; 3; 4
Vgi0
It gives
Ig10 ¼ 7:0811 þ j1:1871

Ig20 ¼ 7:7126 j1:2230
Ig30 ¼ 7:0287 j2:1249
Ig40 ¼ 6:2746 j4:1969
The imaginary voltage of each generator can be computed as
Q10 ¼ V
E g10 þ jXq1Ig10 ¼ 1:9976\70:7
Q20 ¼ V
E g20 þ jXq2Ig20 ¼ 2:2619\56:1
Q30 ¼ V
E g30 þ jXq3Ig30 ¼ 2:0896\40:0
Q40 ¼ V
E g40 þ jXq4Ig40 ¼ 2:3702\22:6
The phase of imaginary voltage is the angular position of each generator,

di0 ; i ¼ 1; 2; 3; 4. From di0 ; i ¼ 1; 2; 3; 4, computational results of output current and
terminal voltage of each generator can be transformed from the common x–y
coordinate to the d–q coordinate as (Eq. (5.49))
Table 6.5 Initial values of G1 G2 G3 G4

state variables
δi0 (rad.) 1.2341 0.9788 0.6990 0.3942
ωi0 1 1 1 1
E0qi0 1.0025 1.0775 1.0181 1.1116
E0fdi0 2.0688 2.3456 2.1662 2.4601

idi sin di cos di vgxi vdi sin di cos di vgxi
¼ ; ¼
iqi cos di sin di vgyi vqi cos di sin di vgyi
Since E0fdi0 ¼ 0, from the third equation of Eq. (5.24), it can have
Efdi0 ¼ Eqi0
E0q10 ¼ Vgq10 þ X0d1 Igd10 ¼1:0025; Efd10 ¼ E0q10 þ ðXd1 X0d1 ÞIgd10 ¼ 2:0688
Computational results of all the above are presented in Table 6.5. Initial values
of SVC are given as
Xsvcc ¼ 0:02; Xsvcl ¼ 0:01; Bsvc0 ¼ 0
According to Eq. (6.34), it can have a0 ¼ 90 .
6.3.1.2 Linearized Heffron–Phillips Model
From Tables 6.1 and 6.3, the following network equations can be obtained:

Ig
Y 12
Y g
E
¼ 11 22 m
0 Y21 Y V
where
2 3
j59:9880 0 0 0
6 0 7
11 ¼ 6 7
0 j59:9880 0
Y 6 7;
4 0 0 j59:9880 0 5
0 0 0 j59:9880
2 3
j59:9880 0 0 0 0 0 0
6 0 7
12 6 0 j59:9880 0 0 0 0 7
Y ¼6 7;
4 0 0 0 0 0 0 j59:9880 5
0 0 0 0 0 j59:9880 0
2 3
j59:9880 0 0 0
6 0 7
6 0 j59:9880 0 7
6 7
6 0 0 0 0 7
6 7
21 6 7
Y ¼6 0 0 0 0 7;
6 7
6 0 0 0 0 7
6 7
6 7
4 0 0 0 j59:9880 5
0 0 j59:9880 0

22
Y
¼ Y221 Y222 Y223 ,
2 3
3:9604 j99:5701 3:9604 þ j39:6040 0
6 3:9604 þ j39:6040 13:8614 j198:5712 9:90010 þ j99:0100 7
6 7
6 7
6 0 9:9010 þ j99:0100 23:6692 j118:0481 7
6 7
221 6 7
Y ¼6 0 0 1:8002 þ j18:0020 7;
6 7
6 0 0 0 7
6 7
6 7
4 0 0 0 5
0 0 0
2 3
0 0
6 7
6 0 0 7
6 7
6 1:8002 þ j18:0020 0 7
6 7
222 6 7
Y ¼ 6 3:6004 j35:6186 1:8002 þ j18:0020 7;
6 7
6 1:8002 þ j18:0020 34:5237 j118:1020 7
6 7
6 7
4 0 9:9001 þ j99:0100 5
0 0
2 3
0 0
6 7
6 0 0 7
6 7
6 0 0 7
6 7
223 6 7
Y ¼6 0 0 7;
6 7
6 9:9010 þ j99:0100 0 7
6 7
6 7
4 13:8614 j198:5712 3:9604 þ j39:6040 5
3:9604 þ j39:6040 3:9604 j99:57010
By keeping only three nodes of generator terminals via deleting all the other
nodes in the above network admittance matrix, it can have
N ¼ Y
Y 11 Y 1 Y
12 Y
22 21
2 3
1:5150 j17:7819 1:1172 þ j15:2884 0:1867 þ j0:4071 0:5275 þ j0:9893
6 1:1172 þ j15:2884 5:0840 0:5275 þ 1:4666 þ j2:3934 7
6 j21:9474 j0:9893 7
¼6 7
4 0:1867 þ j0:4071 0:5275 þ j0:9893 1:9728 j18:0523 2:2208 þ j14:5440 5
0:5275 þ j0:9893 1:4666 þ j 2:3934 2:2208 þ j14:5440 7:7304 j23:9732
Thus,
1
¼ Y
Y 1 þ jx0
N d
2 3
1:2870 9:3383i 1:4623 þ 5:9214i 0:4321 þ 0:4042i 0:6681 þ 0:5845i
6 1:4623 þ 5:9214i 2:4883 11:1930i 0:6681 þ 0:5845i 1:0309 þ 0:8430i 7
6 7
¼6 7
4 0:4321 þ 0:4042i 0:6681 þ 0:5845i 1:7113 98981i 2:0705 þ 5:0608i 5
0:6681 þ 0:5845i 1:0309 þ 0:8430i 2:0705 þ 5:0608i 3:3572 12:5138i

2 3
6:9805 5:4977 0:6487 0:8340
6 3:2377 1:0716 1:4891 7
6 5:7984 7
Fdd ¼6 7;
4 0:4782 0:8486 6:8161 5:4893 5
0:4958 1:0508 3:4812 5:0277
2 3
3:2724 3:9725 0:1877 0:5123
6 6:6150 0:4642 7
6 6:1187 0:0321 7
Fqq ¼6 7;
4 0:4880 0:5283 3:6934 2:6771 5
0:9808 1:2240 5:8082 8:0130
2 3
9:3383 5:3603 0:1274 0:1073
6 6:0988 0:3772 0:1341 7
6 11:1930 7
Gdd ¼6 7;
4 0:5680 0:7462 9:8981 4:2062 5
0:8876 1:2719 5:4489 12:5138
2 3
1:2870 2:9102 0:5778 0:8811
6 0:0804 1:3249 7
6 2:4883 0:8035 7
Gqq ¼6 7;
4 0:1656 0:4807 1:7113 3:4938 5
0:0108 0:3945 0:4563 3:3572
2 3
0:2002 0:4527 0:0899 0:1371
6 0:0125 0:3871 0:1250 0:2061 7
6 7
Hdd ¼6 7;
4 0:0258 0:0748 0:2662 0:5435 5
0:0017 0:0614 0:0710 0:5222
2 3
1:4526 0:8338 0:0198 0:0167
6 0:9487 1:7411 0:0587 0:0209 7
6 7
Hqq ¼ 6 7
4 0:0884 0:1161 1:5397 0:6543 5
0:1381 0:1979 0:8476 1:9466
Thus, according to Eq. (5.38), the following matrices are calculated:

2 3
5:9711 4:9200 0:7327 0:3184
6 7
6 4:2784 6:4231 1:2632 0:8814 7
6 7
Fd ¼ 6 7;
6 1:0557 1:0252 6:1938 4:1129 7
4 5
0:9626 1:1694 4:2732 6:4053
2 3
0:5778 0:9782 0:1172 0:2833
6 7
6 2:2267 1:8849 0:0617 0:2801 7
6 7
Fq ¼ 6 7;
6 0:4224 0:2615 1:0323 0:3484 7
4 5
0:5768 0:4099 1:6646 2:6513
2 3
9:5556 4:2382 0:2888 0:9817
6 7
6 6:0083 11:8998 0:0148 0:7474 7
6 7
Gd ¼ 6 7;
6 0:4687 0:2367 10:3994 2:6752 7
4 5
0:8284 0:9169 5:1327 13:6144
2 3
0:5850 1:6999 0:3921 0:6178
6 7
6 0:1761 1:5087 0:4498 0:7509 7
6 7
Gq ¼ 6 7;
6 0:1130 0:4304 0:8213 1:8852 7
4 5
0:0754 0:4386 0:4397 1:7610

K1 ¼ Vtd0 Iq0 X0d Fd þ Vtd0 Id0 Xq ;

K2 ¼ Iq0 þ Vtd0 Id0 Xq Gd Vtd0 Id0 Xq Gq ;

K3 ¼ I þ Xd X0d Gd ;

K4 ¼ Xd X0d Fd ;
K5 ¼ Vt01 Vq0 X0d Fd þ V1
t0 Vd0 Xq Fq ;
K6 ¼ Vt01 Vq0 Vt01 Vq0 X0d Gd þ V1

t0 Vd0 Xq Gq
Hence, coefficient matrices of the Heffron–Phillips model of the example power

system as given by Eq. (5.43) are obtained as
2 3
4:4089 4:6359 0:1653 0:3923
6 7:0963 0:4686 0:2093 7
6 7:3556 7
K1 ¼ 6 7;
4 1:4589 1:1060 5:5378 2:9728 5
1:8094 1:5087 5:9085 9:2266
2 3
9:8813 1:0858 0:9408 1:7693
6 2:4940 2:0583 7
6 12:2358 1:0102 7
K2 ¼ 6 7;
4 0:0196 0:7701 10:9008 2:4791 5
0:2018 0:6184 1:3160 13:3819
2 3
2:5926 0:7064 0:0481 0:1636
6 1:0014 2:9833 0:0025 0:1246 7
6 7
K3 ¼ 6 7;
4 0:0781 0:0395 2:7332 0:4459 5
0:1381 0:1528 0:8555 3:2691
2 3
0:9952 0:8200 0:1221 0:0531
6 0:7131 0:2105 0:1469 7
6 1:0705 7
K4 ¼ 6 7;
4 0:1760 0:1709 1:0323 0:6855 5
0:1604 0:1949 0:7122 1:0675
2 3
0:0824 0:0069 0:0329 0:0425
6 0:1271 0:0315 0:0411 0:0546 7
6 7
K5 ¼ 6 7;
4 0:0244 0:0056 0:0326 0:0626 5
0:0334 0:0103 0:0549 0:0986
2 3
0:5932 0:3146 0:0402 0:0499
6 0:1826 0:6569 0:0496 0:0618 7
6 7
K6 ¼ 6 7:
4 0:0257 0:0583 0:5999 0:2974 5
0:0309 0:0712 0:1885 0:6390
2 3 2 3
6:5 0 0 0 0 0 0
0
6 0 6:5 7 6 07
6 0 0 7 60 0 0 7
M¼6 7; D ¼ 6 7;
4 0 0 6:175 0 5 40 0 05
0
0 0 0 6:175 0 0 0
0
2 3 2 3
50 0 0 0 0:1 0 0 0
6 0 50 0 7 6 0 0:1 0 7
6 0 7 6 0 7
KA ¼ 6 7; T A ¼ 6 7;
4 0 0 50 0 5 4 0 0 0:1 0 5
0 0 0 50 0 0 0 0:1
2 3
8 0 0 0
60 8 0 07
6 7
T0do ¼ 6 7
40 0 8 05
0 0 0 8
6.3.1.3 Linearized Heffron–Phillips Model of the System with an SVC

Stabilizer to Be Installed
Consider that the SVC stabilizer is installed at node 9 in the example power system.
From Fig. 6.2 and Eq. (6.4), it can have

Ig 11
Y 12
Y g
E
¼ 21 22 9
0 Y Y V
Thus,
Ig ¼ 1 12 Y
21 E
g þ Y g ¼ ½
11 E 1 12 Y
21 þ Y
11 E
g ¼ Y
E g
Y Y
ð
y22 þ jbsvc Þ ðy22 þ jbsvc Þ
where

11 ¼ Y
Y 111 112 ;
Y
2 3
j59:9880 0
6 8:9964 j89:9640 7 j59:9880
111 ¼ 6
Y 6
7
7
4 33:5910 j254:3110 9:5608 j92:1942 5
45:8239 j640:5978 10:0729 j231:8408
2 3
0 0
6 7
112 6 0 0 7
Y ¼6 7;
4 1:2539 j20:7674 0:0170 þ j7:8663 5
0:0170 þ j7:8663 1:2218 j40:3316
2 3
j59:9880
6 8:9931 þ j149:9192 7
12 ¼ 6
Y 6
7
7;
4 60:1463 þ j359:6609 5
96:3963 þ j907:8644
2 3T
498:2591 þ j1004:9460
6 169:1906 þ j372:5732 7
21 ¼ 6
Y 6
7
7 ;
4 1:2703 þ j12:9860 5
1:2278 þ j32:6445

y22 ¼ 757:0770 1400:3458

2 3
7:0361 5:5021 0:6541 0:8800
6 7
6 3:4671 6:0160 1:0512 1:4976 7
6 7
Fdd ¼6 7;
6 0:5079 0:8661 6:8227 5:4488 7
4 5
0:6231 1:1592 3:9462 5:7286
2 3
3:2904 3:9105 0:1630 0:4571
6 7
6 6:2778 5:8578 0:0239 0:3961 7
6 7
Fqq ¼6 7;
6 0:4221 0:4362 3:4441 2:5858 7
4 5
0:8044 0:9453 4:9335 6:6832
2 3
9:3383 5:4253 0:1673 0:0150
6 7
6 6:0990 11:1930 0:2855 7
0:4145
6 7
Gdd ¼6 7;
6 0:5551 0:7229 9:8981 4:4351 7
4 5
0:8776 1:2181 5:4050 12:5138
2 3
1:2870 2:7871 0:5675 0:8875
6 7
6 0:0588 2:4883 0:7849 1:3007 7
6 7
Gqq ¼6 7;
6 0:2047 0:5151 1:7113 3:1982 7
4 5
0:1328 0:5381 0:8275 3:3572
2 3
0:2002 0:4335 0:0883 0:1381
6 0:0091 0:2023 7
6 0:3871 0:1221 7
Hdd ¼6 7;
4 0:0318 0:0801 0:2662 0:4975 5
0:0207 0:0837 0:1287 0:5222
2 3
1:4526 0:8439 0:0260 0:0023
6 0:9487 1:7411 0:0645 7
6 0:0444 7
Hqq ¼6 7;
4 0:0864 0:1125 1:5397 0:6899 5
0:1365 0:1895 0:8408 1:9466
2 3 2 3
0:1043 0:0788
6 0:2946 7 6 0:1519 7
6 7 6 7
Ldd ¼6 7; Lqq ¼ 6 7
4 0:6485 5 4 0:1730 5
1:6297 0:0638

2 3
6:1526 4:9855 0:7232 0:4439
6 7
6 4:4148 6:6100 1:0000 7
1:1952
Fd ¼ 6 6 0:9953
7;
4 0:9652 6:4009 4:4404 7 5
1:0862 1:2197 4:5385 6:8444
2 3
0:6169 0:9820 0:0895 0:2756
6 7
6 2:0932 1:7871 0:0404 0:2658 7
Fq ¼ 6 6 0:3695
7;
4 0:2106 0:9843 0:4041 7 5
0:4845 0:3115 1:3867 2:1827
2 3
9:5940 4:3354 0:2644 0:8437
6 7
6 5:9571 11:9432 0:0136 0:5853 7
6
Gd ¼ 6 7;
7
4 0:4126 0:1860 10:4629 3:0182 5
0:7581 0:7453 4:9475 13:7236
2 3
0:6082 1:6579 0:3971 0:6468
6 7
6 0:2375 1:5002 0:4534 0:7684 7
6
Gq ¼ 6 7;
7
4 0:1473 0:4578 0:8635 1:7847 5
0:1306 0:4865 0:5748 1:7280
2 3 2 3
0:1662 0:0604
6 7 6 7
6 0:3477 7 6 0:0793 7
6 7
Ld ¼ 6 7
6 0:7077 7; Lq ¼ 6 7
4 5 6 7
4 0:0886 5
1:6776 0:0548
Substituting into Eqs. (5.40) and (5.41), it can have

K1 ¼ Vtd0 Iq0 X0d Fd þ Vtq0 Id0 Xq Fq ;

K2 ¼ Iq0 þ Vtd0 Iq0 X0d Gd þ Vtq0 Id0 Xq Gq ;

Kp ¼ Vtd0 Iqo X0d Ld þ Vtd0 Ido X0q Lq ;

K3 ¼ I þ Xd X0d Gd ;

K4 ¼ Xd X0d Fd ;

Kq ¼ Xd X0d Ld ;
K5 ¼ V1 0 1
t0 Vq0 Xd Fd þ Vt0 Vd0 Xq Fq ;
K6 ¼ V1 1 0 1
t0 Vq0 Vt0 Vq0 Xd Gd þ Vt0 Vd0 Xq Gq ;
Kv ¼ V1 0 1
t0 Vq0 Xd Ld Vt0 Vd0 Xq Lq
2 3
4:6390 4:6990 0:2294 0:2894
6 7
6 6:7233 7:1872 0:0586 7
0:5225
6 7
K1 ¼ 6 7;
6 1:2956 0:9680 5:6138 3:3503 7
4 5
1:5697 1:2865 5:2531 8:1092
2 3
10:1342 0:8111 0:9219 1:7323
6 7
6 2:5271 12:5659 0:9623 1:9405 7
6 7
K2 ¼ 6 7;
6 0:0495 0:7864 11:4295 1:7433 7
4 5
0:1357 0:6002 1:4532 14:2532
2 3
2:5990 0:7226 0:0441 0:1406
6 7
6 0:9929 2:9905 0:0023 0:0976 7
6 7
K3 ¼ 6 7;
6 0:0688 0:0310 2:7438 0:5030 7
4 5
0:1263 0:1242 0:8246 3:2873
2 3
1:0254 0:8309 0:1205 0:0740
6 7
6 0:7358 1:1017 0:1992 0:1667 7
6 7
K4 ¼ 6 7;
6 0:1659 0:1609 1:0668 0:7401 7
4 5
0:1810 0:2033 0:7564 1:1407
2 3
0:0776 0:0029 0:0293 0:0454
6 7
6 0:1258 0:0326 0:0362 0:0571 7
6 7
K5 ¼ 6 7;
6 0:0231 0:0035 0:0300 0:0566 7
4 5
0:0306 0:0063 0:0511 0:0881
2 3
0:5876 0:3150 0:0428 0:0593
6 0:1844 0:0735 7
6 0:6494 0:0528 7
K6 ¼ 6 7;
4 0:0290 0:0633 0:5884 0:3030 5
0:0352 0:0778 0:1962 0:6259
2 3 2 3 2 3
0:2106 0:0277 0:0033
6 7 6 7 6 7
6 0:3466 7 6 0:0580 7 6 0:0000 7
6 7 6 7 6 7
KP ¼ 6 7; Kq ¼ 6 7; Kv ¼ 6 7:
6 0:5811 7 6 0:1179 7 6 0:0059 7
4 5 4 5 4 5
1:0042 0:2796 0:0364
6.3.2 Selection of Installing Locations of Stabilizers
6.3.2.1 Selection of the PSS Installing Locations
From the linearized Heffron–Phillips model of the N-machine power system,

eigenvalues of the state matrix are calculated as
1;2 ¼ 5:1604 j6:0402

k
3;4 ¼ 5:3163 j4:9927
k
5;6 ¼ 0:1979 j6:1696
k
7;8 ¼ 0:1707 j6:3459
k
9;10 ¼ 0:1641 j3:2658
k
11;12 ¼ 5:8428 j0:0805
k
13;14 ¼ 4:1998 j0:0862
k
15;16 ¼ 0 j0
k
According to Eqs. (5.60) and (5.63), the correlation ratio of electromechanical

loop can be calculated to identify the electromechanical oscillation modes of the
example power system as
r1;2 ¼ 0:0080
r3;4 ¼ 0:0503
9
r5;6 ¼ 23:5268 >=
r7;8 ¼ 27:3818 electromechanical modes
>
;
r9;10 ¼ 17:0246
r11;12 ¼ 0:1747
r13;14 ¼ 0:1658
Hence, three electromechanical oscillation modes of the example power system

are
5;6 ¼ 0:1979 j6:1696

k
7;8 ¼ 0:1707 j6:3459
k
9;10 ¼ 0:1641 j3:2658
k
Damp ratio of inter-area oscillation mode k 9;10 is −0.0437. Right and left

eigenvectors corresponding to k9 ¼ 0:1641 þ j3:2658 are calculated as
2 3 2 3
0:0352 j0:0797 0:7994 þ j0:5947
6 0:0312 j0:0629 7 6 0:7288 þ j0:4218 7
6 7 6 7
6 0:1850 þ j0:0850 7 6 0:7969 j0:6005 7
6 7 6 7
6 0:1683 þ j0:0753 7 6 0:7313 j0:4160 7
6 7 6 7
6 0:0007 þ j0:0003 7 6 73:0984 j88:6036 7
6 7 6 7
6 0:0006 þ j0:0002 7 6 52:7805 j81:4813 7
6 7 6 7
6 0:0008 j0:0016 7 6 73:7524 þ j88:2869 7
6 7 6 7
6 0:0007 j0:0014 7 6 52:1265 þ j81:7980 7
v9 ¼ 6
6 0:0087 j0:0165 7
7; w
9 ¼ 6
6 0:5131 þ j1:5515 7
7
6 7 6 7
6 0:0050 j0:0249 7 6 0:3709 þ j1:8021 7
6 7 6 7
6 0:0052 j0:0066 7 6 0:5572 j1:6701 7
6 7 6 7
6 0:0043 j0:0093 7 6 0:4024 j1:9651 7
6 7 6 7
6 0:5060 þ j0:1325 7 6 0:0002 þ j0:0191 7
6 7 6 7
6 0:7359 7 6 0:0023 þ j0:0214 7
6 7 6 7
4 0:1030 þ j0:1762 5 4 0:0002 j0:0206 5
0:1964 þ j0:1248 0:0026 j0:0233
1. Selection of installing locations based on the participation factor

From Eq. (6.140), the participation factor of each generator can be calculated to
examine how much each generator is involved in the oscillation modes. The results
are given in Table 6.6.
9 ¼ 0:1641 þ j3:2658, it can be seen that generator 3 is the best installing
For k
location.
2. Selection of installing locations by damping torque analysis
The linearized Heffron–Phillips model for the PSSs to be installed is given in the
above section.
sDd ¼ xo IDx
sDE0 ¼ T01 ðK3 DE0 K4 Dd þ DE0 Þ
q d0 q fd
sDE0fd ¼ T1 0 1 0
A DEfd þ TA KA ðK5 Dd K6 DEq þ Fk Dupssk Þ
Table 6.6 Participation factor of each generator

Generator Electromechanical oscillation modes
−0.1979 ± j6.1696 −0.1707 ± j6.3459 0.1641 ± j3.2658
1 0.2269 0.0000 0.0868
2 0.2843 0.0031 0.0591
3 0.0017 0.2053 0.2032
4 0.0010 0.3062 0.1551
From Eq. (6.145), the DTA can be used to select installing location of the PSSs.
If the PSS is installed in generator 3, F3 ¼ ½ 0 0 1 0 T . From Eq. (6.115), it
can have with s ¼ k 9 ¼ 0:1641 þ j3:2658.
9 Þ ¼ DTPSS ¼ M1 K2 ½ðI þ sTA ÞðK3 þ sTd0 Þ þ KA K6 1 KA F3

FPSS ðk
Dupss3
2 3
0:0365 j0:0099
6 0:0401 j0:0102 7
6 7
¼6 7
4 0:1147 j0:0109 5
0:1129 j0:0107
9 , the elements of right eigenvector corresponding to Dxj ; j ¼ 1; 2; 3; 4 are

For k
8
>
> v95 ¼ 0:0007 þ j0:0003
<
v96 ¼ 0:0006 þ j0:0002
>
> v ¼ 0:0008 j0:0016
: 97
v98 ¼ 0:0007 j0:0014
Let the feedback signal be Dx3 , and thus,
Dx3 ¼ Dy ¼ cT3 X
where the 7th element of C3 is 1 and all other elements are zero. From Eq. (6.154),
it can have
8
>
cT3 v9 v97
>
> c1 k 9 ¼ ¼ ¼ 1:7522 j1:5478
>
> v95 v95
>
>
>
>
>

9 ¼ c3 v9 v97
T
>
>
< c2 k ¼ ¼ 2:2552 j1:8217
v96 v96
>

9 ¼ 3 v9 v97
T
>
> c3 k
c
>
> ¼ ¼1
>
> v97 v97
>
>
>

> 9 ¼ 3 v9 v97
T
> c
: c4 k ¼ ¼ 1:1042 j0:0115
v98 v98
From Eq. (6.154), sensitive index Sij can be calculated as

8
> 291 v291 ¼ 0:0755
S91 ¼ w j0:0427
>
>
<S ¼ w 292 v292 ¼ 0:0493 j0:0327
92
ð6:187Þ
>
> S ¼ 293 v293 ¼ 0:1985 þ
w j0:0433
>
:
93
294 v294 ¼ 0:1544 þ
S94 ¼ w j0:0150
Generator 1
Ddamp1 = −0.0092 − j0.0074 S91 = 0.0755 − j0.0427
Generator 2
S92 = 0.0493 − j0.0327
Ddamp2 = −0.0124 − j0.0144

Generator 3 λ 9,10
PSS
Ddamp3 = 0.0881 − j0.0931 S93 = 0.1985 + j0.0433
Generator 4
Ddamp4 = −0.0107 + j0.0526 S94 = 0.1544 + j0.0150
Fig. 6.18 The PSS to be installed on G3 to affect the inter-area oscillation mode
According to Fig. 6.14 and Eq. (6.145), the DTA can be calculated as
X
4

DTA3 ¼ F 9 c k
pssj k 9 S
9j ¼ 0:0576
j
j¼1
where
2 3
0:0092 j0:0074
6 0:0124 j0:0144 7
9 Þc ðk
pssj ðk 6 7
F j 9Þ ¼ 6 7
4 0:0881 j0:0931 5
0:0107 þ j0:0526
Figure 6.18 shows how the PSS to be installed on generator 3 will affect the
inter-area oscillation mode. When the PSS is installed on generators 1, 2, or 4, the
DTA can be calculated similarly as
DTA1 ¼ 0:0150; DTA2 ¼ 0:0114; DTA4 ¼ 0:0354
Obviously, generator 3 is the best installing location.
6.3.2.2 Selection of Installing Location of the SVC Stabilizer
From the Heffron–Phillips model of the example power system installed with the
SVC, eigenvalues of the state matrix are calculated as
1;2 ¼ 0:1960 j6:1676 )

k
electromechanical modes
3;4 ¼ 0:1607 j6:3093
k
5;6 ¼ 5:5111 j5:6757
k
7;8 ¼ 0:2433 j3:1216 electromechanical mode
k
9;10 ¼ 5:9764 j4:4089
k
11 ¼ 5:8189
k
12 ¼ 5:5025
k
13 ¼ 4:9368
k
14 ¼ 4:1875
k
15 ¼ 2:4164
k
16 ¼ 0
k
17 ¼ 0
k
An SVC stabilizer is to be installed to improve the damping of the inter-area

7 ¼ 0:2433 þ j3:1216. Right and left eigenvectors corresponding
oscillation mode k

to k7 ¼ 0:2433 þ j3:1216 are calculated as
2 3 2 3
0:0624 þ j0:0911 0:7677 þ j0:5166
6 0:0563 j0:0725 7 6 0:6952 þ j0:3532 7
6 7 6 7
6 0:1749 þ j0:0796 7 6 0:7436 j0:4640 7
6 7 6 7
6 0:1611 þ j0:0693 7 6 0:7193 j0:4058 7
6 7 6 7
6 0:0008 þ j0:0005 7 6 69:2025 j87:3225 7
6 7 6 7
6 0:0006 þ j0:0004 7 6 48:8981 j80:1477 7
6 7 6 7
6 0:0008 j0:0014 7 6 62:6622 þ j84:9249 7
6 7 6 7
6 0:0007 0:0013i 7 6 55:4385 þ j82:5452 7
6 7 6 7
7 ¼ 6
v 7 7 ¼ 6 0:5691 þ j1:5609 7
6 0:0097 j0:0170 7; w 6 7
6 0:0064 j0:0265 7 6 0:4629 þ j1:9346 7
6 7 6 7
6 0:0029 j0:0014 7 6 0:1767 j0:8646 7
6 7 6 7
6 0:0032 j0:0021 7 6 0:1625 j0:8590 7
6 7 6 7
6 0:5220 þ j0:1297 7 6 0:0010 þ j0:0194 7
6 7 6 7
6 0:7683 þ j0:0000 7 6 0:0014 þ j0:0232 7
6 7 6 7
6 0:0338 þ j0:1256 7 6 0:0010 j0:0103 7
6 7 6 7
4 0:0012 þ j0:1189 5 4 0:0011 j0:0101 5
0:0005 þ j0:0008 3:5217 þ j14:0586

Davi ¼ Tvsvci ðsÞDVi
Dbsvci ¼ 1cos
pxsvcli ðDavi þ Dasi Þ
2ai0
Let the transfer function of the SVC voltage controller be

Kvi
Tvsvci ðsÞ ¼ Kvp þ
s
where Kvp ¼ 1; Kvi ¼ 8. Let XVSVCi ¼ DV s . It can have

i

X_ vsvci ¼ DVi
Davi ¼ Kvi Xvsvci þ Kvp DVi

8 h i
> v viy0
< Avsvci ¼ 0; Bvsvci ¼ Vix0i0 Vi0
h i
>
: cvsvci ¼ Kvi ; Dvsvci ¼ vix0 viy0
Vi0 Vi0
Hence, from Eq. (6.48), it can be obtained that

1 cos 2ai0 vyi0 26:0361
bssvci ¼ ¼
pxsvcli vxi0 49:5981
Substituting the above into Eq. (6.103), it can have

2 3
0
6 0 7
6 7
6 0 7
6 7
6 0 7
6 7
6 0:0215 7
6 7
6 0:0284 7
6 7
6 7
2 3 6 0:0818 7
0 6 0:0905 7
6 7
B ¼ 4 B2k 5 ¼ 6
6 0:0520 7
7
B3k 6 0:0627 7
6 7
6 0:0882 7
6 7
6 0:0372 7
6 7
6 3:0862 7
6 7
6 4:3028 7
6 7
6 143:3026 7
6 7
4 178:5977 5
0:5828
The feedback signal of the SVC stabilizer is taken to be the active power, Pij ,
delivered through the node where the SVC is to be installed. Linearization of
Vi Vj Vi Vj
Pij ¼ X sinðui uj Þ ¼ X sinuij gives
Vi Vj Vi0 Vj0 Vi0 Vj0

DPij ¼ sinuij0 DVi þ sinuij0 DVj þ cosuij0 Dui
X X X
Vi0 Vj0
cosuij0 Duj
X
where X is the line reactance between nodes i and j. In the common x–y coordinate,
DVk and Duk ; k ¼ i; j can be written as
Vxk0 Vyk0
DVk ¼ DVxk þ DVyk
Vk0 Vk0
Vyk0 Vxk0
Duk ¼ 2
DVx þ 2 DVyk
Vk0 Vk0
Hence,
Vi0 Vj0 Vxi0 Vi0 Vj0 Vyi0

DPij ¼ sin uij0 DVxi þ sin uij0 DVyi
X Vi0 X Vi0
Vi0 Vj0 Vxj0 Vi0 Vj0 Vyj0
þ sin uij0 DVxj þ sin uij0 DVyj
X Vj0 X Vj0
Vi0 Vj0 Vyi0 Vi0 Vj0 Vxi0
þ cos uij0 2 DVxi þ cos uij0 2 DVyi
X Vi0 X Vi0
Vi0 Vj0 Vyj0 Vi0 Vj0 Vxj0
cos uij0 DVxj cos uij0 2 DVyj
X V2j0 X Vj0
which can be written as
DPij ¼ a11 DVxi þ a12 DVyi þ a21 DVxj þ a22 DVyj ð6:188Þ
where
8 Vi0 Vj0 Vxi0 Vi0 Vj0 Vyi0
>
> a11 ¼ sin uij0 cos uij0 2
>
>
>
>
X Vi0 X Vi0
>
>
>
> V V V V V V
> i0 j0 yi0 i0 j0 xi0
< a12 ¼ X sin uij0 Vi0 þ X cos uij0 V2
>
i0
>
> Vi0 Vj0 Vxj0 Vi0 Vj0 Vyj0
>
> a21 ¼ sin uij0 þ cos uij0 2
>
> X Vj0 X Vj0
>
>
>
>
>
> Vi0 Vj0 Vyj0 Vi0 Vj0 Vxj0
: a22 ¼ sin uij0 cos uij0 2
X Vj0 X Vj0
DV ¼ Y1
M Cxy DX
Hence, the voltage at any node can always be reconstructed by state variables,
that is
DVxi ¼ YixT DX
DVyi ¼ YiyT DX
ð6:189Þ
DVxj ¼ YjxT DX
DVyj ¼ YjyT DX
2 3
Dd
Because Dy ¼ DPij ¼ cTk 4 Dx 5 ¼ cTk DX; from Eqs. (6.188) and (6.189) above,
DZ
it can have
ck ¼ a11 Yix þ a12 Yiy þ a21 Yjx þ a22 Yjy ð6:190Þ
For the example power system, when the SVC stabilizer is to be installed at node
9, the feedback signal is taken to be P89, the active power delivered through node 8
to node 9. From the results of power flow calculation, it can have
V80 ¼ 0:8636; V90 ¼ 0:8799; u80 ¼ 0:2001; u90 ¼ 0:4834
Thus,
a11 ¼ 0:3999; a12 ¼ 0:8355; a21 ¼ 0:2098; a22 ¼ 0:8856
According to Eq. (6.101), for the example power system, it has

YM1 ¼ YM1 1 YM1 2
2 3
7:8484 46:0583 0:9779 15:1586 0 0
6 24:9097 4:9233 15:1586 0:9779 0 0 7
6 7
6 0:9779 15:1586 15:5048 45:9152 0 0 7
6 7
6 15:1586 0:9779 33:8809 6:0720 0 0 7
6 7
6 0 0 0 0 13:3470 35:9030 7
Y1 ¼6
6
7
M1
6 0 0 0 0 40:9433 10:8393 7
7
6 0 0 0 0 0:0156 7:8445 7
6 7
6 0 0 0 0 7:8445 0:0156 7
6 7
4 0:1448 2:0653 0:6714 5:1397 1:2690 12:9499 5
2:0653 0:1448 5:1397 0:6714 12:9499 1:2690
2 3
0 0 0:1448 2:0653
6 0 0 2:0653 0:1448 7
6 7
6 0 0 0:6714 5:1397 7
6 7
6 0 0 5:1397 0:6714 7
6 7
6 0:0156 7:8445 1:2690 12:9499 7
¼6 7
1
YM2 6 7:8445
6 0:0156 12:9499 1:2690 77
6 10:1321 49:4841 1:2300 32:5540 7
6 7
6 66:5814 7:7021 32:5540 1:2300 7
6 7
4 1:2300 32:5540 3:9996 65:4289 5
32:5540 1:2300 98:4466 48:9889
Cxy ¼ ½ Cxy1 Cxy2

2 3
18:5288 0 28:8999 0 0 0 0 0 0 0
6 5:8704 0 8:0494 0 0 0 0 0 0 07
6 7
6 0 0 0 0 18:4483 0 25:8688 0 0 07
6 7
6 0 0 0 0 3:2065 0 15:1922 0 0 07
6 7
6 0 0 0 0 0 0 0 0 18:5570 07
Cxy1 ¼6
6
7
6 0 0 0 0 0 0 0 0 5:6022 077
6 0 0 0 0 0 0 0 0 0 07
6 7
6 0 0 0 0 0 0 0 0 0 07
6 7
4 0 0 0 0 0 0 0 0 0 05
0 0 0 0 0 0 0 0 0 0
2 3
0 0 0 0 0 0 0
6 0 0 0 0 0 0 0 7
6 7
6 0 0 0 0 0 0 0 7
6 7
6 0 0 0 0 0 0 0 7
6 7
6 18:9260 0 0 0 0 0 0 7
Cxy2 6
¼6 7
7
6 23:2767 0 0 0 0 0 0 7
6 0 0 16:9781 0 11:7722 0 0 7
6 7
6 0 0 7:6725 0 27:5938 0 0 7
6 7
4 0 0 0 0 0 0 201:7248 5
0 0 0 0 0 0 398:2669
Thus,
1
Y1
M Cxy ¼ YM Cxy 1 Y1
M Cxy 2 Y1
M Cxy 3
2 3
0:4930 0 0:2838 0 0:1810 0 0:3392 0
6 0:5948
6 0 0:6333 0 0:2330 0 0:0626 077
6 0:2527
6 0 0:1235 0 0:2656 0 0:5422 077
6 0:3017
6 0 0:1663 0 0:5832 0 0:3967 077
6 0:0098
6 0 0:0107 0 0:0048 0 0:0249 077
6 0:0505
6 0 0:0233 0 0:0571 0 0:0103 077
6 0:0006
6 0 0:0192 0 0:0074 0 0:0343 077
6 0:0664
6 0 0:0341 0 0:0765 0 0:0197 077
6 0:0201
6 0 0:0466 0 0:0382 0 0:0701 077
6 0:0878
6 0 0:0505 0 0:1034 0 0:0355 077
6 0:3899 0 0:2505 0 0:1775 0 0:3858 077
Y1
M Cxy 1 ¼6
6 0:5133
6 0 0:4797 0 0:3233 0 0:1282 077
6 0:2474
6 0 0:1774 0 0:1587 0 0:4654 077
6 0:3753
6 0 0:2541 0 0:4580 0 0:2195 077
6 0:1828
6 0 0:1857 0 0:0946 0 0:4404 077
6 0:3638
6 0 0:2132 0 0:4303 0 0:1542 077
6 0:0820
6 0 0:1175 0 0:0282 0 0:2581 077
6 0:2283
6 0 0:1331 0 0:2697 0 0:0956 077
6 0:0090
6 0 0:0326 0 0:0220 0 0:0522 077
6 0:0763
6 0 0:0422 0 0:0892 0 0:0278 077
4 0:0030 0 0:0190 0 0:0051 0 0:0354 05
0:0613 0 0:0314 0 0:0706 0 0:0179 0
2 3
0:0341 0 0:0519 0 0:0538 0
6 0:0111 0 0:0061 0 0:0081 07
6 7
6 0:0378 0 0:0564 0 0:0591 07
6 7
6 0:0184 0 0:0075 0 0:0145 07
6 7
6 0:0050 0 0:5385 0 0:0196 07
6 7
6 0:6827 0 0:4218 0 0:2397 07
6 7
6 0:0268 0 0:1369 0 0:1462 07
6 7
6 0:2835 0 0:1461 0 0:5675 07
6 7
6 0:1330 0 0:1905 0 0:2048 07
6 7
6 0:2698 0 0:1561 0 0:3211 07
6 7
6 0:0393 0 0:0574 0 0:0610 07
Y1
M Cxy2 ¼6
6 0:0200
7
6 0 0:0035 0 0:0176 077
6 0:0484 0 0:0660 0 0:0732 07
6 7
6 0:0325 0 0:0003 0 0:0309 07
6 7
6 0:0593 0 0:0756 0 0:0875 07
6 7
6 0:0453 0 0:0047 0 0:0451 07
6 7
6 0:0973 0 0:1345 0 0:1479 07
6 7
6 0:1591 0 0:0811 0 0:1849 07
6 7
6 0:0806 07
6 0 0:2184 0 0:1665 7
6 0:3305 07
6 0 0:1886 0 0:3922 7
4 0:0376 0 0:4166 0 0:0745 05
0:5445 0 0:3213 0 0:3041 0
2 3
0:0701 0 0:5669
6 0:0180 0 0:8487 7
6 7
6 0:0756 0 0:5768 7
6 7
6 0:0251 0 1:2875 7
6 7
6 0:1712 0 2:2984 7
6 7
6 0:0275 0 4:5241 7
6 7
6 0:5319 0 1:8674 7
6 7
6 0:2074 0 5:6191 7
6 7
6 0:2520 0 1:6762 7
6 7
6 0:1031 0 6:9322 7
6 7
6 0:0764 0 0:5434 7
Y1
M Cxy3 ¼6
6 0:0190
7
6 0 1:1876 7
7
6 0:0857 0 0:4571 7
6 7
6 0:0214 0 1:6917 7
6 7
6 0:0957 0 0:3264 7
6 7
6 0:0208 0 2:1281 7
6 7
6 0:1758 0 1:0071 7
6 7
6 0:0414 0 4:5797 7
6 7
6 0:3350 1:8476 7
6 0 7
6 0:1216 6:1937 7
6 0 7
4 0:2343 0 2:1590 5
0:0690 0 5:2016
Thus,
2 3
0:0650 0:2336 0:0246 0:0886
6 0 0 0 0 7
6 7
6 0:1252 0:1235 0:0471 0:0467 7
6 7
6 0 0 0 0 7
6 7
6 0:0082 0:2696 0:0427 0:1022 7
6 7
6 0 0 0 0 7
6 7
6 0:2609 0:0767 0:0685 0:0288 7
6 7
6 0 0 0 0 7
6 7
½ Y8x Y8y Y9x Y9y ¼ 6
6 0:0972 0:1612 0:1345 0:2724 7
7
6 0 0 0 0 7
6 7
6 0:1353 0:0838 0:1935 0:1609 7
6 7
6 0 0 0 0 7
6 7
6 0:1476 0:1818 0:2072 0:3158 7
6 7
6 0 0 0 0 7
6 7
6 0:1755 0:0396 0:2542 0:1013 7
6 7
4 0 0 0 0 5
0:9722 4:6129 1:6229 6:9396
Finally, according to Eq. (6.190), it is obtained that

2 3
0:0856
6 0 7
6 7
6 0:1020 7
6 7
6 0 7
6 7
6 0:1225 7
6 7
6 0 7
6 7
6 0:1285 7
6 7
6 0 7
6 7
ck ¼ 6 7
6 0:0959 7
6 0 7
6 7
6 0:0589 7
6 7
6 0 7
6 7
6 0:1122 7
6 7
6 0 7
6 7
6 0:0398 7
6 7
4 0 5
2:3398
1. Selection of installing location by using damp torque analysis

2 3
0:0740 þ j0:0070
6 0:0102 þ j0:0092 7
Bk ¼ B2k þ A23 ðsI A33 Þ1 B3k ¼6
4 0:0205 þ
7
j0:0287 5
0:0206 þ j0:0374
From Eq. (6.154), it can be calculated that

8
>
cTk v7
>
> c1 k7 ¼ ¼ 25:8592 j63:8552
>
> v75
>
>
>
>
>

7 ¼ k v7
T
>
>
c
< c2 k ¼ 26:4325 j78:5172
v76
>

7 ¼ k v7
T
>
>
c
>
> c k ¼ 6:2615 þ j39:1056
> 3
> v77
>
>
>

> 7 ¼ k v7
T
> c
: c4 k ¼ 7:7533 þ j42:6894
v78
Thus, the following coefficients in Eq. (6.147) are obtained:

2 3
0:0500 j0:0515
6 0:0474 j0:0663 7
i Þc ðk
kj ðk 6 7
B j iÞ ¼ 4
0:1877 þ j0:1421 5
0:2250 þ j0:1462
From Eq. (6.142), the sensitive index Sij can be calculated as

8
> T291 v291 ¼ 0:0950 j0:0377
S91 ¼ w
>
>
< T291 v292 ¼ 0:0648 j0:0305
S92 ¼ w
> S93 ¼ w
> T291 v293 ¼ 0:1670 þ j0:0220
>
:
T291 v294 ¼ 0:1440 þ j0:0156
S94 ¼ w
DTA9 ¼ 0:3585
Figure 6.19 illustrates how the SVC stabilizer to be installed at node 9 con-
tributes the damping to the inter-area oscillation mode.
When the installing location of SVC stabilizer is at node 7 and node 8, the
following results can be obtained similarly:
DTA7 ¼ 0:2316; DTA8 ¼ 0:0478
Obviously, the best installing location for the SVC stabilizer is at node 9.
Generator 1
Ddamp1 = 0.0500 − j0.0515 S 71 = 0.0950 − j0.0377
Generator 2
Ddamp2 = 0.0474 − j0.0663 S 72 = 0.0648 − j0.0305

SVC
Generator 3 λ 7,8
Stabilizer
Ddamp3 = −0.1877 + j0.1421 S 73 = 0.1670 + j0.0220
Generator 4
Ddamp4 = −0.2250 + j0.1462 S 74 = 0.1440 + j0.0156
Fig. 6.19 The SVC stabilizer to be installed at node 9 contributing damping to the inter-area
oscillation mode
2. Selection of installing location based on the residue index

9 and W
Substituting Bk ; ck ; V T into Eq. (6.149), it can have
9
ik ¼
R bikcik ¼ ð5:5927 j0:9465Þð0:0498 j0:0389Þ ¼ 0:3153 þ j0:1705
ik j ¼ 0:3585, being equal to DTA9 ¼ 0:3585.

thus, jR
Residue index of node 7 and node 8 is calculated as
Node 7:
ik ¼ b
R ikcik ¼ ð1:3986 j0:8958Þð0:0854 þ j0:1103Þ ¼ 0:2182 0:0777i

jRik j ¼ 0:2316
Node 8:
ik ¼
R bikcik ¼ ð0:4780 þ j0:1553Þð0:0791 þ j0:0528Þ ¼ 0:0296 j0:0375

jRik j ¼ 0:0478
They are equal to DTA7 ¼ 0:2316 and DTA8 ¼ 0:0478, respectively.
6.3.2.3 Robust Installing Locations and Feedback Signals of the SVC

Stabilizer
Let six candidate installing locations and feedback signals of the SVC stabilizer in
the example power system be
8
>
> u: node 7 DP67
> 1
>
>
> u : node 7 DP78
< 2
u3 : node 8 DP78
>
> u4 : node 8 DP89
>
>
>
> u: node 9 DP89
: 5
u6 : node 9 DP910
Consider XðlÞ the set of operating conditions of the example power system as
given in Table 6.7.
Obviously, in order to apply the criteria of Eqs. (6.162) and (6.163), the effec-
tiveness of the stabilizer has to be examined at ten given system operating condi-
tions. Table 6.8 presents the results of residue calculation.
Table 6.7 System operating MW Load at node 7 Load at node 9

conditions
l1 927 + j100 1767 + j100
l2 937 + j100 1767 + j100
l3 947 + j100 1767 + j100
l4 957 + j100 1767 + j100
l5 967 + j100 1767 + j100
l6 977 + j100 1767 + j100
l7 987 + j100 1767 + j100
l8 997 + j100 1767 + j100
l9 1007 + j100 1767 + j100
l10 1017 + j100 1767 + j100
Table 6.8 Residue (Cðu; lÞ) u1 u2 u3 u4 u5 u6

l1 0.2263 0.2240 0.0585 0.0534 0.3485 0.0311
l2 0.2283 0.2260 0.0559 0.0512 0.3509 0.0314
l3 0.2302 0.2279 0.0532 0.0489 0.3534 0.0317
l4 0.2319 0.2298 0.0505 0.0465 0.3559 0.0319
l5 0.2336 0.2316 0.0478 0.0442 0.3585 0.0322
l6 0.2353 0.2334 0.0451 0.0418 0.3610 0.0325
l7 0.2369 0.2351 0.0424 0.0394 0.3637 0.0327
l8 0.2384 0.2368 0.0397 0.0370 0.3663 0.0330
l9 0.2399 0.2384 0.0371 0.0346 0.3689 0.0333
l10 0.2413 0.2400 0.0344 0.0322 0.3716 0.0335
The following two indices are used to examine the effectiveness of the stabilizer.
8
>
> 1: min Cðu; lÞ
< l
max Cðu; lÞ min Cðu; lÞ
>
> l l
: 2:
Cðu; lÞ

where Cðu; lÞ is the mean value of the residue (Cðu; lÞ) at all operating condi-
tions. Two indices above are applied for u1 u6 , and the results of selection are
given in Table 6.9.
In order to consider the robustness of the stabilizer to the variations of power
system operating conditions, the criteria (6.161), (6.162), and (6.163) should be
taken into account. That is
Table 6.9 Results of indices u1 u2 u3 u4 u5 u6

for u1 u6
Index 1 0.2263 0.2240 0.0478 0.0442 0.3485 0.0311
Index 2 0.0640 0.0689 0.5187 0.4939 0.0642 0.0742
8
< Max½min Cð/; lÞ
/ l
: Min½max Cð/; lÞ min Cð/; lÞ
/ l l
Obviously, from Table 6.9, the final selection should be u5 . That is, the
installing location is at node 9, and the feedback signal is DP89 .
6.3.2.4 Design of Robust SVC Stabilizer
According to the results obtained in the previous section, the SVC stabilizer is
installed at node 9 and the integral of P89 is used as the feedback signal. The
stabilizer is designed at operating condition l1 , where it is estimated to be least
effective. At l1 , the inter-area oscillation mode is k7 ¼ 0:2473 þ j3:0340. Right and
left eigenvectors corresponding to k 7 ¼ 0:2473 þ j3:0340 are calculated as
2 3 2 3
0:0512 þ j0:0923 0:7809 j0:5431
6 0:0466 þ j0:0745 7 6 0:7129 j0:3833 7
6 7 6 7
6 0:1739 j0:0788 7 6 0:7625 þ j0:4873 7
6 7 6 7
6 0:1611 j0:0682 7 6 0:7314 þ j0:4390 7
6 7 6 7
6 0:0008 j0:0003 7 6 75:8300 þ j91:8156 7
6 7 6 7
6 0:0006 j0:0003 7 6 55:1927 þ j84:9901 7
6 7 6 7
6 0:0007 þ j0:0013 7 6 68:6860 j90:0933 7
6 7 6 7
6 0:0006 þ j0:0012 7 6 62:3367 j86:7124 7
6 7 6 7

9 ¼ 6 0:0102 þ j0:0178 7; W 6 7
V 6 7 9 ¼ 6 0:6701 j1:6183 7
6 0:0069 þ j0:0278 7 6 0:5915 j2:0830 7
6 7 6 7
6 0:0033 þ j0:0024 7 6 0:2084 þ j0:8511 7
6 7 6 7
6 0:0038 þ j0:0026 7 6 0:2230 þ j0:8620 7
6 7 6 7
6 0:5193 j0:1312 7 6 0:0022 j0:0204 7
6 7 6 7
6 0:7712 þ j0:0000 7 6 0:0002 j0:0253 7
6 7 6 7
6 0:0006 j0:1244 7 6 0:0005 þ j0:0102 7
6 7 6 7
4 0:0208 j0:1235 5 4 0:0003 þ j0:0104 5
0:0006 j0:0010 4:2900 j15:2077
Taking the same procedure of computation, it can have

2 3 2 3T
0 0:0773
6 0 7 6 0 7
6 7 6 7
6 0 7 6 0:1042 7
6 7 6 7
6 0 7 6 0 7
6 7 6 7
6 0:0199 7 6 0:1159 7
6 7 6 7
6 0:0270 7 6 0 7
6 7 6 7
6 0:0793 7 6 0:1367 7
6 7 6 7
6 0:0914 7 6 0 7
6 7 6 7
Bk ¼ 6
6 0:0520 7; Ck ¼ 6 0:0881 7
7 6 7
6 0:0643 7 6 0 7
6 7 6 7
6 0:0821 7 6 0:0488 7
6 7 6 7
6 0:0384 7 6 0 7
6 7 6 7
6 2:5208 7 6 0:1051 7
6 7 6 7
6 2:6311 7 6 0 7
6 7 6 7
6 143:597 7 6 0:0307 7
6 7 6 7
4 176:7661 5 4 0 5
0:5767 2:3522
9 and W
Substituting Bk Ck V T into Eq. (6.149), the residue is calculated as
9
ik ¼
R bikcik ¼ ð5:8380 þ j1:1365Þð0:0448 þ j0:0378Þ ¼ 0:3485\150:8661
Let the transfer function of the SVC stabilizer be
sTw 1 ð1 + sT2 Þ ð1 + sT4 Þ

Tsvc ðsÞ = Ksvc = Ksvc Tk ðsÞ
1 + sTw 1 + sT ð1 + sT1 Þ ð1 + sT3 Þ
where T1 ¼ 0:79s; T3 ¼ 0:79s; T ¼ 0:01s; Tw ¼ 10s.

Firstly, T2 and T4 are set such that the SVC stabilizer to compensate the phase of
the residue index to ensure the SVC stabilizer to move the oscillation mode hori-
zontally towards the left on the complex plane. The result is
T2 ¼ 0:2371s,T4 ¼ 0:2703s
It gives

7 ¼ 1\ 150:8661
Tk k
To determine the gain value of the SVC stabilizer, Ksvc , a direct searching
7 to k
method in nonlinear programming can be used to move k by minimizing the
7
following objective function:

7 ðK) Re k
f ðKÞ ¼ Re k ðK Þ
7

ðK Þ ¼ 0:45. The steps of direct searching are as follows:
where Re k 7
Step 1: Set the initial searching scale S ¼ 2 and minimum searching scale
Sm ¼ 0:0001. Choose K0 ¼ 50 as the starting point of the searching. The oscilla-
tion mode is calculated as k 7 ¼ 0:2031 j4:6030. Thus, the initial value of
objective function is fð0Þ ¼ 0:0610.

Step 2: Search from Kð0Þ to Kð0Þ þ S, if f K0 þ S \f K0 let Kð0Þ ( K0 þ S and

execute Step 2 again; if not, search from Kð0Þ to Kð0Þ S, if f K0 þ S \f K0 let
Kð0Þ ( K0 S and execute Step 2 again; if not, Step 2 fails and go to Step 3.
Step 3: If searching in Step 2 fails, reduce the searching scale S by half and then go
back to Step
2.
Step 4: If f K0 \Sm , stop searching and take K0 as the solution K .
For
theSVC stabilizer, in 36 steps, the searching stops to have K = 112.5000
with f K0 ¼ 0. Thus, the design of the SVC stabilizer is completed. Parameters of
SVC stabilizer are set to be
Ksvc ¼ 112:5000; T2 ¼ 0:2371 s; T4 ¼ 0:2704 s
With the SVC stabilizer installed, eigenvalues of the state matrix are calculated as
1;2 ¼ 49:7492 j85:8280

k
)
3;4 ¼ 0:1615 j6:3364
k
electromechanical mode
5;6 ¼ 0:1840 j6:1485
k
7;8 ¼ 5:7637 j5:5700
k
9;10 ¼ 0:4500 j4:5671
k electromechanical mode
11;12 ¼ 4:3308 j3:9562
k
13;14 ¼ 0:2099 j22:0032
k

k15 ¼ 6:3962

k16 ¼ 3:4632
17;18 ¼ 4:0904 j0:2451
k
19 ¼ 4:3536
k
20 ¼ 5:6946
k
21 ¼ 0
k
22 ¼ 0
k
Fig. 6.20 Simulation result δ1 − δ3 /degree

with and without the SVC
stabilizer
t/s
Table 6.10 Oscillation mode Operating condition Eigenvalue Damp ratio

when the SVC stabilizer is
installed and the system is l1 −0.4500 ± j4.5671 0.0983
operated at varied operating l2 −0.4510 ± j4.5600 0.0984
conditions l3 −0.4517 ± j4.5553 0.0987
l4 −0.4526 ± j4.5512 0.0990
l5 −0.4576 ± j4.5344 0.1004
l6 −0.4633 ± j4.5269 0.1018
l7 −0.4706 ± j4.5197 0.1036
l8 −0.4838 ± j4.4630 0.1078
l9 −0.4870 ± j4.4068 0.1098
l10 −0.4901 ± j4.3511 0.1119
It can be observed that the mode of interest is moved to 0:4500 j4:5671

which is well damped.
The simulation result of the example power system at operating condition l1
without and with the SVC stabilizer installed is presented in Fig. 6.20. At 0.5 s of
simulation, a three-phase short-circuit fault occurred at node 8 and was cleared in
0.1 s. From the simulation result, it can be seen that the low-frequency oscillation is
well damped.
To examine the robustness of the SVC stabilizer, the oscillation mode of interest
is calculated at other operating conditions l2 l10 with the SVC stabilizer
installed, and the results are given in Table 6.10. From Table 6.10, it can be
observed that when the SVC stabilizer is designed at the selected operating con-
dition l1 which is predicted to be least effective, its effectiveness is ensured when
the system operates at other operating conditions. Hence, the design ensures the
robustness of the stabilizer.
6.4 Example Three-Machine Power System
6.4.1 Dynamic Interactions Among PSSs Installed

in Example Three-Machine Power System
In Sect. 5.3, an example three-machine power system is presented. The demon-

stration of the example is finished with the coordinated design of three PSSs
installed in three synchronous generators in Sect. 5.3.2.3, where the forward path of
the PSSs before they are designed and installed is calculated as

Þ ¼ K2 KA ðK þ k
Fpss ðk TA Þ þ K6 KA 1
T0 ÞðI þ k
1 3 1 d0 1
2 3
4:0590\29:34 1:7719\153:70 1:0376\155:74
6 7
¼ 4 0:6433\113:76 6:1802\28:79 2:6963\136:09 5
1:1801\125:70 2:6184\140:40 5:3227\31:52
where k ¼ 0:7094 þ j7:0590 is the target oscillation mode for the coordinated
1
design of the PSSs, i.e. the position of the oscillation mode k 1 ¼
0:1736 þ j7:0590 to be assigned to by the coordinated design of the PSSs. With
the PSSs being designed by the use of the method introduced in Sect. 5.2.2, the
. The mode with the
oscillation mode is not moved to the exact target position, k1
pss1 ¼ 0:6094 þ j7:0235. The eigenvalue drift
PSSs installed is in fact moved to k
from the target position k to k
pss1 ¼ 0:6094 þ j7:0235 is due to the dynamic
1
interactions between the PSSs as to be explained as follows.
Consider the example three-machine power system with the PSSs installed on
each of the synchronous generators. Transfer function of the PSSs is
Tpssi ðsÞ ¼ Kpssi ; T1i ¼ T3i ¼ 0:05; i ¼ 1; 2; 3
Parameters of the PSSs are obtained as given in Table 5.6 in Sect. 5.3.2.3. The
forward path of the PSSs with kpss1 ¼ 0:6094 þ j7:0235 can be obtained to be

pss1 Þ ¼ K2 KA ðK þ k
Fpss ðk pss1 TA Þ þ K6 KA 1
pss1 T0 ÞðI þ k
3 d0
2 3
4:2427\29:72 1:8428\153:34 1:0765\155:48
6 7
¼ 4 0:7022\115:88 6:4424\29:37 2:8752\135:23 5
1:2780\125:43 2:7740\139:57 5:5693\32:24
According to Eq. (6.115), damping torque provided by each PSS to each of the
three generators can be calculated as
6.4 Example Three-Machine Power System 321
DTPSSi ¼ Re½F pss1 ÞT

PSS ðk pss1 ÞDxi ¼ Re½Fpss ðk
pssi ðk pss1 ÞFi T pss1 ÞDxi ; i
pssi ðk
¼ 1; 2; 3
where
2 3 2 3 2 3
1 0 0
F1 ¼ 4 0 5; F1 ¼ 4 1 5; F1 ¼ 4 0 5
0 0 1
The following results can be obtained according to the above equations:

2 3 2 3
22:7134Dx1 3:5277Dx2
6 7 6 7
DTPSS1 ¼ 4 3:0879Dx1 5; DTPSS2 ¼ 4 12:3402Dx2 5;
6:1895Dx1 5:2042Dx2
2 3
4:4036Dx3
6 7
DTPSS3 ¼ 4 11:5351Dx3 5
22:9518Dx3
The above results give the damping torque contribution from the ith PSS to the
jth generator in the form of Dpssij Dxi ; i; j ¼ 1; 2; 3; i 6¼ j. This needs to be con-
verted to the form Dpssij Dxj ; i; j ¼ 1; 2; 3; i 6¼ j. The conversion is expressed
generally by Eq. (6.117). Without the PSS installed, the conversion coefficient
i Þ can be calculated as explained from Eqs. (6.128) to (6.135) or from
cj ð k
Eqs. (6.150) to (6.154). With the PSSs installed, the conversion can be made as
explained as follows.
With the PSSs installed, the closed-loop state equation is
sDX ¼ Apss DX
which is given at the end of Sect. 5.3.2.3. Hence,
pss1 DXð
k kpss1 Þ ¼ Apss DXð
kpss1 Þ
pss1 to be
Denote the right eigenvector of the state matrix corresponding to k

vpss1 ; it should have
pss1 Þ vi ðk
i ðk
DX pss1 Þ
¼

DXj ðkpss1 Þ vj ðkpss1 Þ
where DX pss1 Þ and DX

i ðk pss1 Þ is the ith and jth state variables (element) of
j ðk
pss1 Þ, and
k
DXð pss1 Þ and vj ðk
vi ð k pss1 Þ are the ith and jth elements of vpss1 ,
respectively.
From the closed-loop state matrix obtained in Sect. 5.3.2.3, the right eigenvector
pss1 ¼ 0:6094 þ j7:0235; vpss1 is calculated as
corresponding to k
vpss1 ¼ ½ vpss11 vpss12

vpss13 T
where
2 3 2 3
0:0159 j0:0289 0:0049 þ j0:0047
6 0:0554 + j0:0831 7 6 0:0042 j0:0198 7
6 7 6 7
6 7 6 7
6 0:0163 þ j0:0429 7 6 0:0107 j0:0087 7

vpss11 ¼6 7
6 ð6:1622 þ j4:1263Þ 104 7; vpss12 ¼6 7
6 0:3141 þ j0:2922 7;
6 7 6 7
6 7 6 7
4 ð1:7503 j1:3991Þ 103 5 4 0:7337 þ j0 5
ð9:2666 j4:4722Þ 104 0:3582 j0:3680
2 3
0:0005 þ j0:0006
6 0:0016 þ j0:0035 7
6 7
6 7
6 0:0015 j0:0020 7

vpss13 6
¼6 7
7
6 0:0015 j0:0039 7
6 7
4 0:0009 j0:0008 5
0:0021 j0:0035
DX is defined in Sect. 5.3.2.3 to be

T
DX ¼ DX1 DX2 DX 3
where
DX1 ¼ ½ Dd1 Dd2 Dd3 Dx1 Dx2 Dx3 T

T
DX2 ¼ DE0q1 DE0q2 DE0q3 DE0fd1 DE0fd2 DE0fd3
DX3 ¼ ½ DXpss1 DUpss1 DXpss2 DUpss2 DXpss3 DUpss3 T
Thus, from the last three elements of vpss11 , the following results can be obtained
as
4
Dx1 ¼ pss1 ÞDx2 ¼ ð6:1622 þ j4:1263Þ 10 Dx2 ¼ ð0:3298 þ j0:0279ÞDx2 ;
c12 ðk
ð1:7503 j1:3991Þ 103
4
Dx1 ¼ pss1 ÞDx3 ¼ ð6:1622 þ j4:1263Þ 10
c12 ðk Dx3 ¼ ð0:7137 j0:1009ÞDx3 ;
ð9:2666 j4:4722Þ 104
3
Dx2 ¼ pss1 ÞDx3 ¼ ð1:7503 j1:3991Þ 10 Dx3 ¼ ð2:1230 þ j0:4853ÞDx3
c23 ðk
ð9:2666 j4:4722Þ 104
Hence, with Dpssij Dxi ; i; j ¼ 1; 2; 3; i 6¼ j being converted to Dpssij Dxj ; i; j ¼

1; 2; 3; i 6¼ j, the damping torque contributed by each PSS to each of the generators
should be calculated as
2 3
pss1 ÞT
Re½Fpssi1 ðk pss1 Þc ðk
pssi ðk
i1 pss1 ÞDx1
6 pss1 ÞT pss1 ÞDx2 7
pss1 Þc ðk
pssi ðk
DTPSSi ¼ 4 Re½Fpssi2 ðk i2 5; i ¼ 1; 2; 3
pss1 ÞT
Re½Fpssi1 ðk pss1 Þc ðk
pssi ðk
i3 pss1 ÞDx3
By using the results obtained and equation given above, the damping torque
contributed by each PSS to each of the generators is calculated as
2 3 2 3
22:7134Dx1 11:0506Dx1
6 7 6 7
DTPSS1 ¼ 4 0:8905Dx2 5; DTPSS2 ¼ 4 12:3402Dx2 5;
5:9790Dx3 14:7900Dx3
2 3
4:9442Dx1
6 7
DTPSS1 ¼ 4 3:4809Dx2 5
22:9518Dx3
The above results are summarized in Table 6.11.

In Sect. 5.3.2.3, the required damping torque provided by the PSSs is
Dpss1 Dpss2 Dpss3 ¼ p52 ¼ ½ 21:73 11:84 21:94 to move the oscillation
mode from k 1 ¼ 0:1736 þ j7:0590 to k ¼ 0:7094 þ j7:0590 by the design of
1
the PSSs. With the PSSs installed, the oscillation mode is only moved to
pss1 ¼ 0:6094 þ j7:0235. The reason is clearly indicated by the results in
k
Table 6.11. The actual damping torque provided by the PSSs is
½ Dpss1 Dpss2 Dpss3 ¼ ½ 38:71 9:75 14:14 , not the required damping torque

Dpss1 Dpss2 Dpss3 ¼ ½ 21:73 11:84 21:94 for the PSSs design, because
each PSS provides damping torque to all three generators as is shown in Table 6.11.
1. The amount of damping torque provided by each PSS to the generator where it is
installed
is ½ 22:7134
12:3402 22:9518 , very close to
Dpss1 Dpss2 Dpss3 ¼ ½ 21:73 11:84 21:94 , the required damping torque
by the design of the PSS. The difference is caused by the difference between

k
pss1 ¼ 0:6094 þ j7:0235 and k1 ¼ 0:7094 þ j7:0590.
Table 6.11 The damping torque provided by each PSS

Damping torque contribution to By PSS1 By PSS2 By PSS3 Total Dpssi
G1 22.7134 11.0506 4.9442 38.7082
G2 0.8905 12.3402 −3.4809 9.7498
G3 5.9790 −14.7900 22.9518 14.1407
2. Each of the three PSSs provides positive damping torque to the first generator.
Thus, the first generator obtains more damping torque (38:7082Dx1 ) than
required by the design.
3. PSS installed on the second generator and third generator contributes negative
damping torque to the third and second generators, respectively, reducing the
actual amount of damping torque obtained by the second and third generators.
6.4.2 Design of Non-negatively Interactive PSSs Installed

in the Example Power System
In Sect. 6.2.4.2, a scheme of non-negatively interactive design of the PSSs is

introduced. In this section, the scheme is used to examine whether the coordinated
design of three PSSs to be installed in the three-machine power system can be
designed to be non-negatively interactive or not as follows.
The state matrix of the example three-machine power system is constructed in
Sect. 5.3.1.3 by the following coefficient matrices according to Eq. (5.43) when the
PSSs are not installed
2 3 2 3
2:3373 1:4640 0:8733 2:4574 0:9034 0:5628
6 7 6 7
K1 ¼ 4 1:5290 2:2160 0:6870 5; K2 ¼ 4 1:0247 3:0671 0:4060 5;
1:0599 0:8201 1:8800 0:3694 0:6458 2:3234
2 3 2 3
1:1889 0:1067 0:0994 0:0277 0:0008 0:0269
6 7 6 7
K3 ¼ 4 0:1209 3:1565 0:7660 5; K4 ¼ 4 1:2024 1:8725 0:6701 5;
0:4521 1:2478
3:7141 1:2795 1:0112 2:2907
2 3 2 3
0:0068 0:0076 0:0143 0:8732 0:0711 0:0678
6 7 6 7
K5 ¼ 4 0:0596 0:0315 0:0281 5; K6 ¼ 4 0:4315 0:5292 0:1730 5;
00780 0:0499 0:1278 0:4691 0:1829 0:4790
2 3 2 3
47:2 0 0 0 0 0
6 7 6 7
M¼4 0 12:8 0 5; D ¼ 4 0 0 0 5;
0 0 6:02 0 0 0
2 3 2 3
200 0 0 0:02 0 0
6 7 6 7
KA ¼ 4 0 200 0 5; T A ¼ 4 0 0:02 0 5;
0 0 200 0 0 0:02
2 3
8:96 0 0
6 7
T0d0 ¼ 4 0 6:00 0 5
0 0 5:89
In Sect. 5.3.2.3, it is found that when the damping torque provided

by the PSSs
installed on the three generators is Dpss1 Dpss2 Dpss3 ¼ p52 ¼ ½ 21:73
11:8421:94, the electromechanical oscillation mode of interests is moved to
¼ 0:7094 þ j7:0590. Hence, the state matrix of the example three-machine
k 1
power system with the PSSs installed (parameters of the PSSs have not been set yet)
can be constructed still according to Eq. (5.43) with coefficient matrix D being
changed to
2 3 2 3
Dpss1 0 0 21:73 0 0
D¼ 4 0 Dpss2 0 5¼4 0 11:84 0 5
0 0 Dpss3 0 0 11:84
Denote the state equation with the PSSs installed (their parameters have not been
set yet) to be
sDX ~ pss DX
~ ¼A ~
~ ¼ ½ DX1
where DX DX2 T
DX1 ¼ ½ Dd1 Dd2 Dd3 Dx1 Dx2 Dx3 T

T
DX2 ¼ DE0q1 DE0q2 DE0q3 DE0fd1 DE0fd2 DE0fd3
Obviously, k ¼ 0:7094 þ j7:0590 is the eigenvalue of state matrix A ~ pss . Let

1
~ pss corresponding to k
and v0 ðk

v0pss1 be the right eigenvector of state matrix A
1 j 1 Þ be
the (j + 3) element of v0pss1 ; j ¼ 1; 2; 3. It can have

ÞDxj ¼ vi ðk1 Þ Dxj
Dxi ¼ cij ðk 1 Þ
vj ðk 1
The damping torque provided by the ith PSS to the jth generator is
Þc ðk
DTpssij ¼ Re½Fpssij ðk
1 ij 1 ÞTpssi ðk1 ÞDxj ; i; j ¼ 1; 2; 3
Þ
f pssij ¼ Fpssij ðk
1 cij ðk1 Þ; i; j ¼ 1; 2; 3 is the forward path from the ith PSS to the
electromechanical oscillation loop of the jth generator. A forward path matrix Fð k Þ
1
can be formed with f pssij ¼ Fpssij ðk Þc ðk
Þ; i; j ¼ 1; 2; 3 being the elements. For the
1 ij 1
example three-machine power system, v0pss1 is calculated as
0
v0pss1
¼ v0pss1b
vpss1a v0pss1c T
0pss1a
v ¼ ½ 0:0080 j0:0556 0:0111 + j0:1722 0:0188 þ j0:0752 0:0013 j0:0001 T
0pss1b
v ¼ ½ 0:0038 j0:0006 0:0017 + j0:0003 0:0014 + j 0:0014 0:0086 j0:0222 T
v0pss1c ¼ ½ 0:0076 j0:0021 0:0958 + j0:0822 0:9324 + j0 0:16467 j0:2189 T
Thus, Þ is calculated as
cij ðk 1
0
c12 ¼ v0 pss14
v pss15 ¼ ð0:0013 j0:0001Þ=ð0:0038 j0:0006Þ ¼ 0:3187 + j0:0671;
0
0
c13 ¼ v pss14 v pss16 ¼ ð0:0013 j0:0001Þ=ð0:0017 + j0:0003Þ ¼ 0:7211 j0:0733;
c21 ¼ 1= c12 ¼1=ð0:3187 + j0:0671Þ ¼ 3:0048 j0:6331;
0
c23 ¼ v 0 pss15
v pss16 ¼ ð0:0038 j0:0006Þ=ð0:0017 + j0:0003Þ ¼ 2:1204 + j0:6768;
c31 ¼ 1= c13 ¼ 1=ð0:7211 j0:0733Þ ¼ 1:3726 + j0:1395;
c32 ¼ 1= c23 ¼ 1=ð2:1204 + j0:6768Þ ¼ 0:4280 j0:1366
k
Fð Þ is calculated as
1
2 3
4:059\29:341 5:441\38:198 1:4316\18:46
k
Fð Þ ¼ 4 0:2095\54:347 6:1802\28:783 1:2114\153:79 5
1
0:8554\60:103 5:8279\122:7 5:3227\31:527
Þc ðk
Three vectors f pssij ¼ Fpssij ðk
1 ij 1 Þ; j ¼ 1; 2; 3 for i = 1,2,3 are plotted in
Figs. 6.21, 6.22, and 6.23, respectively.
From Fig. 6.21, it can be seen that if the compensating phase of PSS1 is chosen
to be h10 ¼ 29:341 , it can have T Þ ¼ Tpss1 \h10 ¼ Tpss1 \29:341 . The
pss1 ðk
1
electric torque provided by PSS1 to each of the three generators is
Electric torque
provided to G3
-1 1 2 3
forward 0 Electric torque

path to G2 θ10 provided to G1
Electric torque
provided to G2
forward path
to G3
-1
Direction of positive
synchronizing torque forward path to G1
-2
Fig. 6.21 Forward path of PSS1 to each generator

Þc ðk
DTPSS11 ¼ Fpssi11 ðk
1 i1 1 ÞTpss1 ðk1 ÞDx1 ¼ 4:059\29:341 Tpss1 \h10 Dx1 ¼ 4:059Tpss1 Dx1

Þc ðk

DTPSS12 ¼ Fpssi12 ðk 1 i1 1 ÞTpss1 ðk1 ÞDx2 ¼ 0:2095\54:347 Tpss1 \h10 Dx2
¼ 0:2095Tpss1 cosð54:347 þ 29:341 ÞDx2
Þc ðk
DTPSS13 ¼ Fpssi13 ðk ÞT ÞDx3 ¼ 0:8554\60:103 Tpss1 \h10 Dx3
pss1 ðk
1 i1 1 1
¼ 0:8554Tpss1 cosð60:103 þ 29:341 ÞDx3
which are all positive. The synchronizing torque provided by PSS1 to each of the
generators is
Þc ðk

DTs11 ¼ jImfFpssi11 ðk 1 i1 1 ÞTpss1 ðk1 ÞgDx1
¼ j4:059Tpss1 sinð29:341 þ h10 ÞDx1 ¼ 0
DTs12 ¼ j0:2095Tpss1 sinð54:347 þ 29:341 ÞDx1
jx
j0:2095Tpss1 sinð54:347 þ 29:341 Þ s Dd2
x0
x
¼ 0:2095Tpss1 sinð54:347 þ 29:341 Þ s Dd2
x0
x
DTs13 ¼ 0:8554Tpss1 sinð60:103 þ 29:341 Þ s Dd3
x0
It shows that PSS1 provides positive synchronizing torque to the 2nd and 3rd
generators, as the projection of the vector of electric torque provided by PSS1 to G2
and G3 is on the direction of positive synchronizing torque.
From Fig. 6.22, it can be seen that if the compensating phase of PSS2 is chosen
to be h20 , PSS2 will provide positive damping and synchronizing torque to G1
and G2, but negative damping torque to G3. There is no suitable compensating
forward path to G3
4
Electric torque
provided to G3
Electric torque
-2 2 4 provided to G2 6
0
θ20
Electric torque
provided to G1
-2
forward path to G2
forward path to G1
-4 Direction of positive
synchronizing torque

forward path to G2
Electric torque
provided to G1 Electric torque
-1 1 3 provided to G2
0 θ30
Electric torque
provided to G2
forward path to G1
-1
-2
Direction of positive forward path to G3

synchronizing torque
phase to be selected to ensure the PSS to provide positive damping torque to three
generators. Similarly, from Fig. 6.23, it can also be seen that there is no suitable
compensating phase available for designing PSS3. Hence, in the example
three-machine power system presented in Sect. 5.3, it is inevitable that PSS2 and
PSS3 supply negative to other generators.
References
1. Wang HF, Swift FJ (1998) A unified model for the analysis of FACTS devices in damping
power system oscillations part II: multi-machine power systems. IEEE Trans Power Deliv (4)
2. Wang HF, Swift FJ (1997) The indexes for selecting the best locations of PSS or FACTS-based
stabilizers in multi-machine power systems: a comparison study. IEE Proc Part C (2)
3. Wang HF, Swift FJ (1997) The connection between modal analysis and electric torque analysis
in studying the oscillation stability of multi-machine power systems. Int J Electr Power Energy
Syst 19(5):321–330
4. Wang HF (1999) Selection of robust installing locations and feedback signals of FACTS-based
stabilizers in multi-machine power systems, IEEE Trans Power Syst (2)
5. Wang HF (1999) Design of non-negatively interactive FACTS-based stabilizers in
multi-machine power systems. Int J Power Syst Res 50 (3)
Chapter 7
Multi-machine Power Systems Installed
with VSC-Based Stabilizers

System Installed with VSC-Based Stabilizers
7.1.1 Mathematical Model of a Multi-machine Power

System Installed with a Shunt VSC-Based Stabilizer

Installed with a Shunt VSC-Based Stabilizer
Without loss of generality, it can be assumed that a shunt VSC is installed at a busbar
between nodes 1 and 2 in an N-machine power system, as shown in Fig. 7.1. The
shunt VSC can be connected to an energy storage system (ESS), a HVDC, or a static
renewable power station, such as a photovoltaic (PV) or fuel cell (FC) power station.
It can operate alone as a VSC-based FACTS device, STATCOM (see, Sect. 4.1.1.1).
From Fig. 7.1, it can be obtained by the following:
s ¼ jxsIs þ V
V c ¼ jxs ðI1s þ Is2 Þ þ V
c ð7:1Þ
Hence,
1 ¼ jx1sI1s þ V
V s ¼ jx1sI1s þ jxs ðI1s þ Is2 Þ þ V
c ¼ jðx1s þ xs ÞI1s þ jxsIs2 þ V
c
2 ¼ jxs2Is2 þ V
V s ¼ jxs2Is2 þ jxs ðI1s þ Is2 Þ þ V
c ¼ jxsI1s þ jðxs2 þ xs ÞIs2 þ V
c
ð7:2Þ
In matrix form, Eq. (7.2) is given as follows:

jðx1s þ xs Þ jxs I1s
V
V
¼ 1 c ð7:3Þ
jxs jðxs2 þ xs Þ Is2 V2 Vc

DOI 10.1007/978-1-4899-7696-3_7
330 7 Multi-machine Power Systems Installed with VSC-Based Stabilizers
Fig. 7.1 A multi-machine A multi-machine power system

power system installed with a
shunt VSC
V1 Vs V2
x 1s x s2
I1s Is2
Node 1 Node 2
Is
xs
VC
Shunt VSC
That is,
1
I1s jðx1s þ xs Þ jxs 1
V
V
Is2 ¼ c ð7:4Þ
jxs jðxs2 þ xs Þ V2 Vc
For the N-machine power system, firstly the set of network equations with N
generator’s internal nodes, nodes 1 and 2 left (without node s where the shunt VSC
is installed) can be established as follows:
2 3 2 32 3
0 y11 y12 13
Y 1
V
6 7 6 7 6
23 54 V2 7
4 0 5 ¼ 4 y21 y22 Y 5 ð7:5Þ
Ig Y 31 Y 32
Y33
Eg
Secondly, the network equations of (7.5) can be modified with node s added to be
0
0 y11 0 V1 I1s Y13
¼ þ þ E
0 0 y022 V 2 Is2 23 g
Y
ð7:6Þ
V1
Ig ¼ Y 31 Y
32
þ Y33 Eg
2
V
where y011 and

y022 are obtained by excluding x12 ¼ x1s þ xs2 from y11 and y22 in
Eq. (7.5), respectively.
Finally from Eqs. (7.4) and (7.6), it can have the following:

0 y011 0
V1
¼
0 0 y22 V0 2
1
jðx1s þ xs Þ jxs V1 Vc Y13
þ þ E
jxs jðxs2 þ xs Þ
V2
Vc 23 g
Y
7.1 Mathematical Model of a Multi-machine Power System … 331
That is.
0 1
V1 jðx1s þ xs Þ jxs y11 0
¼ Iþ
2
V jxs jðxs2 þ xs Þ 0 y022

Vc jðx1s þ xs Þ jxs Y13
E
c
V jxs jðxs2 þ xs Þ Y 23 g

jðx1s þ xs Þ jxs
1 Vc
¼Y
Y13
E ð7:7Þ
S
Vc jxs jðxs2 þ xs Þ Y 23 g
Substituting Eq. (7.7) into (7.6), it can be obtained that

1 1
V
Ig ¼ C c þ Y E g ¼ Y 31 Y 32 Ys c
V
1

1 jðx1s þ xs Þ jxs Y13
þ Y 33 Y 31 Y
32 Y g
E ð7:8Þ
s
jxs jðxs2 þ xs Þ 23
Y
From Eqs. (7.4) and (7.7), it can have the following:

1
I1s jðx1s þ xs Þ jxs 1 Vc 1 Y 13
¼
YS I YS
E ð7:9Þ
Is2 jxs jðxs2 þ xs Þ Vc Y23 g
Hence from Fig. 7.1 and Eq. (7.9), it can have the following:
X
I N
Is ¼ isx þ jisy ¼ I1s þ Is2 ¼ ½ 1 c þ
1 1s ¼ ys V gj
ysj E ð7:10Þ
Is2 j¼1
where E g . According to Eq. (5.10), E

gj is the jth element of E gj is given as follows:
gj ¼ E0 ejdj þ ðxqj x0 Þiqj eiðdj 90 Þ

E ð7:11Þ
qj dj
X
N
s ¼ jxsIs þ V
V c ¼ ðjxs þ ys ÞV
c þ gj
ysj E ð7:12Þ
j¼1
In x–y coordinate of the power system, it can have the following:
c ¼ mkVdc ðcos c þ j sin cÞ ¼ mkVdc \c

V ð7:13Þ
c and x-axis. Active power received by the capacitor

where c is the angle between V
of the VSC from the power system is given as:
Vdc Idc1 ¼ isx vcx þ isy vcy ¼ isx mkVdc cos c þ isy mkVdc sin c
Hence,
Idc1 ¼ isx mk cos c þ isy mk sin c ð7:14Þ
In this section, for the simplicity of discussion, effect of dynamic and control
functions associated with Idc2 (see, Fig. 4.1) is not considered, and the VSC adopts
the PWM (pulse width modulation) algorithm. Hence, the dynamic equation on the
DC side of the VSC is given as follows [1]:
1 1
V_ dc ¼ Idc1 ¼ ðisx mk cos c þ isy mk sin cÞ ð7:15Þ
Cdc Cdc
AC and DC voltage control functions implemented via the DC/AC converter are
as follows:
m ¼ m0 þ Tm ðsÞðVs Vsref Þ þ uvscpss
ð7:16Þ
/ ¼ /0 þ Tdc ðsÞðVdc Vdcref Þ
s and V
where uvscpss is the stabilizing control signal, and / is the angle between V c
vsy
ac tan1 ¼ /þc ð7:17Þ
vsx
Tm ðs) and Tdc ðs) are the transfer functions of AC and DC voltage control
function of the VSC, respectively.
Denote the elements of matrix C and Y
in Eq. (7.8) as
l m
¼ dci e;
C ¼ yij
Y ð7:18Þ
Similar to Eqs. (6.6) and (6.7), from Eqs. (7.8) and (7.18) it can have the
following:

!
X N h i
c þ yij Eqj e j þ xqj xdj iqj e ð j Þ

0 0 d 90
¼ e ð90 di Þ ci V
j jd j
j¼1
X
N h i
¼ ejð90

di Þ c þ
ci V

yij E0qj ejð90 þ dj di Þ þ xqj x0dj iqj ejðdj di Þ
j¼1
X
N h i
yij E0qj ejð90 þ aij þ dj di Þ þ xqj x0dj iqj ejðaij þ dj di Þ

¼ ci mkVdc ejð90 di þ c þ bi Þ
þ
j¼1
ð7:19Þ
where ci ¼ ci ejbi ;

yij ¼ yij ejaij . From Eq. (7.19), it can be obtained that
idi ¼ ci mkVdc sinðc þ bi di Þ

X N h i
þ yij E0qj sinðaij dij Þ þ xqj x0dj iqj cosðaij dij Þ
j¼1
ð7:20Þ
iqi ¼ ci mkVdc cosðc þ bi di Þ
XN h i
þ yij E0qj cosðaij dij Þ þ xqj x0dj iqj sinðaij dij Þ
j¼1
Linearization of Eq. (7.20) can be written in the matrix form to be
DId ¼ Fdd Dd þ Gdd DE0q þ Hdd DIq þ Ldddc DVdc þ Lddm Dm þ Lddg Dc
DIq ¼ Fqq Dd þ Gqq DE0q þ Hqq DIq þ Lqqdc DVdc þ Lqqm Dm þ Lqqg Dc
ð7:21Þ
That can give the following:
DId ¼ Fd Dd þ Gd DE0q þ Lddc DVdc þ Ldm Dm þ Ldg Dc

ð7:22Þ
DIq ¼ Fq Dd þ Gq DE0q þ Lqdc DVdc þ Lqm Dm þ Lqg Dc
Substituting Eq. (7.22) into Eqs. (5.40) and (5.41), it can have the following:
Dd_ ¼ xo Dx
Dx_ ¼ M1 ðK1 Dd K2 DE0q DDx þ Kpdc DVdc þ Kpm Dm þ Kpg DcÞ
0
DE_ q ¼ T01 0 0
d0 ðK3 DEq K4 Dd þ DEfd þ Kqdc DVdc þ Kqm Dm þ Kqg DcÞ
0
DE_ fd ¼ T1 0 1 0
A DEfd þ TA KA ðK5 Dd K6 DEq þ Kvdc DVdc þ Kvm Dm þ Kvg DcÞ
ð7:23Þ
Linearization of Eqs. (7.10) and (7.11) is given as follows:
Disx ¼ axdc DVdc þ axm Dm þ axg Dc þ axd Dd þ axe DE0q

ð7:24Þ
Disy ¼ aydc DVdc þ aym Dm þ ayg Dc þ ayd Dd þ aye DE0q
DV_ dc ¼ K7 Dd þ K8 DE0q þ K9 DVdc þ K10 Dm þ K11 Dc ð7:25Þ

Linearization of Eqs. (7.11) and (7.12) can give the following:
Dvsx ¼ bxdc DVdc þ bxm Dm þ bxg Dc þ bxd Dd þ bxe DE0q

Dvsy ¼ bydc DVdc þ bym Dm þ byg Dc þ byd Dd þ bye DE0q
Hence, linearization of Eq. (7.16) is given as
Dm ¼ Tm ðsÞDVs þ Duvscpss
¼ Tm ðsÞðbdc DVdc þ bm Dm þ bg Dc þ bd Dd þ be DE0q Þ þ Duvscpss ð7:26Þ
D/ ¼ Tdc ðsÞDVdc
That is,
Dm ¼ Tm ðsÞDVs þ Duvscpss
Tm ðsÞ
¼ ðbdc DVdc þ bg Dc þ bd Dd þ be DE0q Þ þ Duvscpss ð7:27Þ
1 bm Tm ðsÞ
D/ ¼ Tdc ðsÞDVdc
Dc ¼ Kgd Dd þ Kgq DE0q þ Kgdc DVdc þ Kgm Dm þ Kgf D/ ð7:28Þ
By substituting Eq. (7.28) into Eqs. (7.23), (7.25), and (7.27), Heffron–Phillips
model of the power system installed with the shunt VSC-based stabilizer can be
obtained as
Dd_ ¼ xo Dx
Dx_ ¼ M1 ðK01 Dd K02 DE0q DDx þ K0pdc DVdc þ K0pm Dm þ K0pg D/Þ
0
DE_ q ¼ T01 0 0 0 0 0 0 0
d0 ðK3 DEq K4 Dd þ DEfd þ Kqdc DVdc þ Kqm Dm þ Kqg D/Þ
0
DE_ fd ¼ T1 0 1 0 0 0 0 0 0
A DEfd þ TA KA ðK5 Dd K6 DEq þ Kvdc DVdc þ Kvm Dm þ Kvg D/Þ
DV_ dc ¼ K07 Dd þ K08 DE0q þ K09 DVdc þ K010 Dm þ K011 D/

Tm ðsÞ
Dm ¼ ðb0 DVdc þ b0g D/ þ b0d Dd þ b0e DE0q Þ þ Duvscpss
1 bm Tm ðsÞ dc
D/ ¼ Tdc ðsÞDVdc
ð7:29Þ
The model is shown in Figs. 7.2 and 7.3 [2, 3].
+ b′dc
b′e Eq '
Tm (s)
1 − b m Tm (s) K ′8
b′d Δm 1
+ K′10 + ΔVdc
s − K′9
Δu vsc − pss
K ′7 K′11 Tdc (s)
Δφ
b′g
Fig. 7.2 Heffron–Phillips model—part of shunt VSC
K ′1
-
ω0 I
(sM + D) −1
- + s
K ′4 K ′5
ΔVdc
[K ′p-dc K ′p-g K ′p-m ] [Δφ] [K ′v-dc K ′v-g K ′v-m ]
K ′2 Δm
[K ′q-dc K ′q-g K ′q-m ]
+ -
- +
(sTd0' + K ′3 ) −1 −1
(I + sTA ) K A
Eq ' + Efd'
-
K ′6
Fig. 7.3 Heffron–Phillips model—part of power system

7.1.1.2 General Linearized Model of an N-Machine Power System

Installed with a Shunt VSC-Based Stabilizer

Dvsx
Let DVsxy ¼ . Linearization of Eq. (7.17) can be obtained as
Dvsy
Dc ¼ D/ þ aTs DVsxy ð7:30Þ
vsx þ jvsy ¼ jxs ðisx þ jisy Þ þ mkVdc cos c þ jmkVdc cos c ð7:31Þ
By using Eq. (7.30), linearization of Eq. (7.31) can give the following:
Disx ¼ cxdc DVdc þ cxm Dm þ cxf D/ þ aTsx DVsxy

ð7:32Þ
Disy ¼ cydc DVdc þ cym Dm þ cyf D/ þ aTsy DVsxy
Hence, by using Eq. (7.32), linearization of Eq. (7.15) can be obtained as
DV_ dc ¼ cddc DVdc þ cdm Dm þ cdf D/ þ aTsdc DVsxy ð7:33Þ
Because
h i
vsy0
DVs ¼ vVsx0 Vs0 DVsxy ; ð7:34Þ
s0
the state-space realization of the AC controller of the shunt VSC (the first equation
in Eq. (7.26)) can be written as
X_ AC ¼ AAC XAC þ BAC DVsxy

ð7:35Þ
Dm ¼ CAC XAC þ DAC DVsxy þ Duvscpss
Let the state-space realization of the DC controller of the shunt VSC (the second
equation in Eq. (7.26)) be
X_ DC ¼ ADC XDC þ BDC DVdc

ð7:36Þ
D/ ¼ CDC XDC þ DDC DVdc
From Eqs. (7.33), (7.35), and (7.36), it can have the following:
DV_ dc ¼ ðcddc þ cdf DDC ÞDVdc þ cdm CAC XAC þ cdf CDC XDC
ð7:37Þ
þ ðcdm DAC þ aTsdc ÞDVsxy þ cdm Duvscpss
From Eqs. (7.35)–(7.37), the overall state equation of the shunt VSC dynamics
and control functions can be written as
X_ VSC ¼ AVSC XAC þ BVSC DVsxy þ Bvscpss Duvscpss ð7:38Þ
where
2 3 2 3
DVdc cddc þ cdf DDC cdm CAC cdf CDC
6 7 6 7
XAC ¼ 4 XAC 5; AVSC ¼4 0 AAC 0 5;
XDC BDC 0 ADC
2 3 2 3
cdm DAC þ aTsdc cdm
6 7 6 7
BVSC ¼4 BAC 5; Bvscpss ¼4 0 5
0 0
By using Eqs. (7.35) and (7.36), Eq. (7.32) becomes as
Disx ¼ ðcxdc þ cxf DDC ÞDVdc þ cxm CAC XAC þ cxf CDC XDC
þ ðcxm DAC þ aTsx ÞDVsxy þ cxm Duvscpss
ð7:39Þ
Disy ¼ ðcydc þ cyf DDC ÞDVdc þ cym CAC XAC þ cyf CDC XDC
þ ðcym DAC þ aTsx ÞDVsxy þ cym Duvscpss

Disx
Let DIsxy ¼ . From Eq. (7.39), it can have the following:
Disy
DIsxy ¼ CVSC XVSC þ DVSC DVsxy þ Dvscpss Duvscpss ð7:40Þ
where

cxdc þ cxf DDC cxm CAC cxf CDC
CVSC ¼ ;
cydc þ cyf DDC cym CAC cyf CDC
" #
cxm DAC þ aTsx cxm
DVSC ¼ ; Dvscpss ¼
cym DAC þ aTsy cym
If it is assumed that the feedback signal of the shunt VSC-based stabilizer is the
active power, Ps2 , delivered along the transmission line where the shunt VSC is
installed, similar to Eq. (6.40), it can have the following:
DPs2 ¼ psvsc DVsxy þ p2vsc DVxy2 ð7:41Þ


Dvx2 2 . Similar to Eq. (6.98), let the lin-
where DVxy2 ¼ and vx2 þ jvy2 ¼ V
Dvy2
earized network equation of the N-machine power system with the shunt
VSC-based stabilizer installed be (only with N nodes of generators, node s and 2 of
system of Fig. 7.1 left) the following:
2 3 2 32 3
DIxy Ygg Ygs Yg2 DVxy
4 DIsxy 5 ¼ 4 Ysg Yss Ys2 54 DVsxy 5 ð7:42Þ
0 Y2g Y2s Y22 DVxy2
By substituting Eq. (6.29) and (7.40) into Eq. (7.42), it can have the following:
2 3 2 32 3
Cgxy Xg Ygg Dgxy Ygs Yg2 DVxy
4 CVSC XVSC þ Dvscpss Duvscpss 5 ¼ 4 Ysg Yss DVSC Ys2 54 DVsxy 5
0 Y2g Y2s Y22 DVxy2
ð7:43Þ

DVxy
DV2xy ¼ Y1
22 ½ Y2g Y2s
DVsxy

Cgxy 0 Xg 0
þ Duvscpss
0 CVSC XVSC Dvscpss

Ygg Dgxy Ygs Yg2 1 DVxy
¼ Y22 ½ Y2g Y2s
Ysg Yss DVSC Ys2 DVsxy

DVxy
¼ Ygsxy
DVsxy
ð7:44Þ
By arranging the state equation of the generators of Eq. (6.29) and that of the
shunt VSC of Eq. (7.38) together, it can have the following:

X_ g Agxy 0 Xg Bgxy 0 DVxy 0
¼ þ þ Duvscpss
X_ VSC 0 AVSC XVSC 0 BVSC DVsxy Bvscpss
ð7:45Þ
By substituting Eq. (7.44) into (7.45), the open-loop state equation of the system
without the shunt VSC-based stabilizer installed can be obtained as
X_ ¼ AX þ BDuvscpss ð7:46Þ
where

Xg 0 Bgxy 0 1
0
X¼ ; B¼ þ Y
XVSC Bvscpss 0 BVSC gsxy Dvscpss

Agxy 0 Bgxy 0 1 Cgxy 0
A¼ þ Y
0 AVSC 0 BVSC gsxy 0 CVSC

DVxy
DPs2 ¼ ½ 0 psvsc p2vsc Y1
22 ½ Y2g Y2s
DVsxy

¼ ½0 psvsc p2vsc Y1
22 ½ Y2g Y2s Y1
gsxy

Cgxy 0 Xg 0
þ Duvscpss ð7:47Þ
0 CVSC XVSC Dvscpss
Hence, the output equation of the system is given as

Dy ¼ CX þ DDuvscpss ð7:48Þ
where

Cgxy 0
C ¼ ð½ 0 psvsc p2vsc Y1
22 ½ Y2g Y2s ÞY1
gsxy
0 CVSC

0
D ¼ ð½ 0 psvsc p2vsc Y1
22 ½ Y2g Y2s ÞY1
gsxy
Dvscpss

System Installed with a UPFC-Based Stabilizer

Installed with a UPFC-Based Stabilizer [4, 5]
Without loss of generality, it can be assumed that a UPFC is installed at a busbar

between nodes 1 and 2 in an N-machine power system, as shown in Fig. 7.4.
From Fig. 7.4, it can be obtained (see, Sect. 4.3.1.1) as
Et ¼ jxEIE þ V
V E; Bt ¼ jxBIE2 þ V
V B ð7:49Þ
Gn G2 G1
Ign I g2 I g1
Vgn Vg2 Vg1
Yt
V1 VEt VBt V2
I1E I E2
xB
x1E x E2
IE
VEt x E
Cdc
mE δE mB δB
UPFC
Fig. 7.4 A multi-machine power system installed with a UPFC
and
V 1 ¼ jx1EI1E þ V Et
Et ¼ jxE2IE2 þ V
V Bt þ V
2 ð7:50Þ
IE ¼ I1E IE2
Substituting Eq. (7.49) into (7.50), it can have the following:

I1E 1 jðxE þ xE2 þ xB Þ jxE 1
V 1 jðxE2 þ xB Þ jxE E
V
IE2 ¼ 2 þ B
xR jxE jðx1E þ xE Þ V xR jxE jðx1E þ xE Þ V
ð7:51Þ
where xR ¼ ðx1E þ xE ÞðxE þ xE2 þ xB Þ x2E . Similar to Eq. (7.6), it should have the
following:
0
0 y11 0 V1 I1E Y13
¼ þ þ E
0 0
0 y22 V2
IE2 23 g
Y
ð7:52Þ
V
Ig ¼ Y 31 Y
32 1
þY g
33 E
2
V
where y011 and y

022 are obtained by excluding x12 ¼ x1E þ xE2 from y11 and y22 in
Eq. (7.5). Substituting Eq. (7.51) into (7.52), it can be obtained that

0 y011 0
1 jðxE þ xE2 þ xB Þ jxE V1
¼ þ
0 0
0 y22 x R jxE jðx1E þ xE Þ 2
V
ð7:53Þ
1 jðxE2 þ xB Þ jxE VE Y13
þ þ E
xR jxE jðx1E þ xE Þ V B 23 g
Y
The above equation gives the following:

1
V 1 jðxE2 þ xB Þ jxE E
V 13
Y
1
¼ Y þ ð7:54Þ

V2 E
xR jxE jðx1E þ xE Þ B
V 23 Eg
Y
0
E ¼ y 11 0 jðxE þ xE2 þ xB Þ jxE
where Y þ 1
0 y022 xR jxE jðx1E þ xE Þ
EV
Ig ¼ C E þ C BVB þ Y E g

1 1 jðxE2 þ xB Þ jxE VE
¼ Y 31 Y 32 Y
E
xR jxE jðx1E þ xE Þ V B ð7:55Þ

1 Y13
þ Y 33 Y 31 Y 32 Y g
E
E 23
Y

I1E 1 jðxE þ xE2 þ xB Þ jxE
1
¼ Y þI
IE2 xR jxE jðx1E þ xE Þ E

1 jðxE2 þ xB Þ jxE VE
ð7:56Þ
xR jxE jðx1E þ xE Þ V B

1 jðxE þ xE2 þ xB Þ jxE
1 Y13
Y E
xR jxE jðx1E þ xE Þ E Y 23 g
Hence from Fig. 7.4 and Eq. (7.56), it can have the following:
X N
IE ¼ iEx þ jiEy ¼ I1E IE2 ¼ ½ 1 1 I1E ¼ yE V E þ B þ
yB V gj
yEj E
IE2
j¼1
XN
IE1 ¼ iE1x þ jiE1y ¼ ½ 1 0 I1E ¼ yE1 V E þ yB1 V
B þ gj
yE1j E
IE2
j¼1
XN
IE2 ¼ iE2x þ jiE2y ¼ ½ 0 1 I1E ¼ yE2 V E þ yB2 V
B þ gj
yE2j E ð7:57Þ
IE2
j¼1
where E g , which is given by Eq. (7.11). From Eqs. (7.49)

gj is the jth element of E
and (7.57), it can have the following:
X
N
Et ¼ ðjxE yE þ 1ÞV
V B þ jxE
E þ jxE yB V gj
yEj E
j¼1
ð7:58Þ
X
N
Bt ¼ jxB yE2 V
V E þ ðjxB yB2 þ 1ÞV
B þ jxB gj
yE2j E
j¼1
In x–y coordinate of the power system, let the following be denoted
E ¼ mE kE Vdc ðcos cE þ j sin cE Þ ¼ mE kE Vdc \cE

V
ð7:59Þ

VB ¼ mB kB Vdc ðcos cB þ j sin cB Þ ¼ mB kB Vdc \cB
Linearization of Eqs. (7.11) and (7.12) can give the following:
DvEtx ¼ bEtxdc DVdc þ bEtxmE DmE þ bEtxgE DcE

þ bEtxmB DmB þ bEtxgB DcB þ bEtxd Dd þ bEtxe DE0q
DvEty ¼ bEtydc DVdc þ bEtymE DmE þ bEtygE DcE
þ bEtymB DmB þ bEtygB DcB þ bEtyd Dd þ bEtye DE0q
ð7:60Þ
DvBtx ¼ bBtxdc DVdc þ bBtxmE DmE þ bBtxgE DcE
þ bBtxmB DmB þ bBtxgB DcB þ bBtxd Dd þ bBtxe DE0q
DvBty ¼ bBtydc DVdc þ bBtymE DmE þ bBtygE DcE
þ bBtymB DmB þ bBtygB DcB þ bBtyd Dd þ bBtye DE0q
Et and V
In x–y coordinate of the power system, denote the angle between V E to
be dE and that between V Bt and V
B to be dB ; that is,
vEty
ac tan1 ¼ dE þ cE
vEtx
vBty ð7:61Þ
ac tan1 ¼ dB þ cB
vBtx
From Eq. (7.60) and linearization of Eq. (7.61), it can be obtained that
DcE ¼ KgEd Dd þ KgEq DE0q þ KgEdc DVdc

þ KgEmE DmE þ KgEdE DdE þ KgEmB DmB þ KgEdB DdB
ð7:62Þ
DcB ¼ KgBd Dd þ KgBq DE0q þ KgBdc DVdc
þ KgBmE DmE þ KgBdE DdE þ KgBmB DmB þ KgBdB DdB
E, C
Denote the elements of matrix, C B , and Y
in Eq. (7.55) as
l m
E ¼ dcEi e;
C B ¼ dcBi e;
C ¼ yij
Y ð7:63Þ

X
N
¼ ejð90

di Þ E þ cBi V
ðcEi V B þ yij ½E0qj ejdj þ ðxqj x0dj Þiqj ejðdj 90 Þ Þ

j¼1
X
N
¼ ejð90

di Þ E þ cBi V
ðcEi V BÞ þ yij ½E0qj ejð90

þ dj di Þ
j¼1

di þ cE þ bEi Þ di þ cB þ bBi Þ
¼ cEi mE kE Vdc ejð90 þ cBi mB kB Vdc ejð90
X
N

þ yij ½E0qj ejð90 þ aij þ dj di Þ
þ ðxqj x0dj Þiqj ejðaij þ dj di Þ
j¼1
ð7:64Þ
where cEi ¼ cEi ejbEi ; cBi ¼ cBi ejbBi ; yij ¼ yij ejaij . From Eq. (7.64) it can be obtained
that
idi ¼ cEi mE kE Vdc sinðcE þ bEi di Þ cBi mB kB Vdc sinðcB þ bBi di Þ

X
N
þ yij ½E0qj sinðaij dij Þ þ ðxqj x0dj Þiqj cosðaij dij Þ
j¼1
ð7:65Þ
iqi ¼ cEi mE kE Vdc cosðcE þ bEi di Þ þ cBi mB kB Vdc cosðcB þ bBi di Þ
X
N
þ yij ½E0qj cosðaij dij Þ þ ðxqj x0dj Þiqj sinðaij dij Þ
j¼1
Linearization of Eq. (7.65) is given as
DId ¼ Fdd Dd þ Gdd DE0q þ Hdd DIq þ Ldddc DVdc

þ LddmE DmE þ LddgE DcE þ LddmB DmB þ LddgB DcB
ð7:66Þ
DIq ¼ Fqq Dd þ Gqq DE0q þ Hqq DIq þ Lqqdc DVdc
þ LqqmE DmE þ LqqgE DcE þ LqqmB DmB þ LqqgB DcB
By using Eq. (7.62), from above equation it can have the following:
DId ¼ Fd Dd þ Gd DE0q þ Lddc DVdc

þ LdmE DmE þ LddE DdE þ LdmB DmB þ LddB DdB
ð7:67Þ
DIq ¼ Fq Dd þ Gq DE0q þ Lqdc DVdc
þ LqmE DmE þ LqdE DdE þ LqmB DmB þ LqdB DdB
Substituting Eq. (7.67) into Eqs. (5.40) and (5.41), it can be obtained that
Dd_ ¼ xo Dx
Dx_ ¼ M1 ðK1 Dd K2 DE0q DDx þ Kpdc DVdc
þ KpmE DmE þ KpdE DdE þ KpmB DmB þ KpdB DdB Þ
0
DE_ q ¼ T01 0 0
d0 ðK3 DEq K4 Dd þ DEfd þ Kqdc DVdc ð7:68Þ
þ KqmE DmE þ KqdE DdE þ KqmB DmB þ KqdB DdB Þ
0
DE_ fd ¼ T1 0 1 0
A DEfd þ TA KA ðK5 Dd K6 DEq þ Kvdc DVdc
þ KvmE DmE þ KvdE DdE þ KvmB DmB þ KvdB DdB Þ
Dynamic equation of the VSCs of the UPFC on the DC side is given by the
following equation (see Eq. (4.73)):

_Vdc ¼ 1 mE kE VEt sin dE þ mB kB VBt sin dB ð7:69Þ
Cdc xE xB
By using Eqs. (7.60) and (7.62), linearization of Eq. (7.69) can be obtained as
DV_ dc ¼ K7 Dd þ K8 DE0q þ K9 DVdc

ð7:70Þ
þ K10E DmE þ K11E DdE þ K10B DmB þ K11B DdB
Without loss of generality and for the purpose of demonstration, it is assumed

that the shunt VSC of the UPFC implements the AC and DC voltage control
functions, the series VSC realizes the line active and reactive power regulation, and
the damping control is added on the AC voltage control function as follows:
mE ¼ mE0 þ Tac ðsÞðVEt VEtref Þ þ uUPFCpss

dE ¼ dE0 þ Tdc ðsÞðVdc Vdcref Þ
ð7:71Þ
mB ¼ mB0 þ Tap ðsÞðPEt PEtref Þ
dB ¼ dB0 þ Trp ðsÞðQEt QEtref Þ
where uUPFCpss is the damping control signal, Tac ðs), Tdc ðs),Tap ðs), and Trp ðs) is
the transfer function of the AC voltage, DC voltage, active, and reactive power
controller of the UPFC, respectively. Linearization of Eq. (7.71) is given as
DmE ¼ Tac ðsÞDVEt þ DuUPFCpss

DdE ¼ Tdc ðsÞDVdc
ð7:72Þ
DmB ¼ Tap ðsÞDPEt
DdB ¼ Trp ðsÞDQEt
By using Eqs. (7.60) and (7.62), it can be obtained that
DVEt ¼ bEtdc DVdc þ bEtmE DmE þ bEtdE DdE

ð7:73Þ
þ bEtmB DmB þ bEtdB DdB þ bEtd Dd þ bEte DE0q

" #

X
N

PEt þ jQEt ¼ VEt I ¼ ðjxE yE þ 1ÞVE þ jxE yB VB þ jxE
yEj Egj
E1
j¼1
" # ð7:74Þ
X
N

yE1 V
þ yB1 V þ
yE1j E
E B gj
j¼1
By using Eq. (7.62), linearization of the above equation can be obtained as
DPEt ¼ bpdc DVdc þ bpmE DmE þ bpdE DdE

þ bpmB DmB þ bpdB DdB þ bpd Dd þ bpe DE0q
ð7:75Þ
DQEt ¼ bqdc DVdc þ bqmE DmE þ bqdE DdE
þ bqmB DmB þ bqdB DdB þ bqd Dd þ bqe DE0q
Substituting Eqs. (7.73) and (7.75) into Eq. (7.72) gives the following:
Tac ðsÞ
DmE ¼ ðbEtdc DVdc þ bEtdE DdE þ bEtmB DmB
1 bEtmE Tac ðsÞ
1
þ bEtdB DdB þ bEtd Dd þ bEte DE0q Þ þ Dupss
1 bEtmE Tac ðsÞ
DdE ¼ Tdc ðsÞDVdc
Tap ðsÞ ð7:76Þ
DmB ¼ ðbpdc DVdc þ bpmE DmE
1 bpmB Tap ðsÞ
þ bpdE DdE þ bpdB DdB þ bpd Dd þ bpe DE0q Þ
Trp ðsÞ
DdB ¼ ðbqdc DVdc þ bqmE DmE
1 bqdB Trp ðsÞ
þ bqdE DdE þ bqmB DmB þ bqd Dd þ bqe DE0q Þ
Equation (7.68), (7.70), and (7.76) are the Heffron–Phillips model of the power
system installed with the UPFC.
7.1.2.2 General Linearized Model of an N-Machine Power System

Installed with a UPFC-Based Stabilizer
Let

DvEtx DvBtx Dvx2
DVEtxy ¼ ; DVBtxy ¼ ; DVE2xy ¼ ;
DvEty DvBty Dvy2

DiEx DiE2x
DIExy ¼ ; DIE2xy ¼
DiEy DiE2y
Et ¼ jxEIE þ V
V E; V
Et ¼ jðxE2 þ xB ÞIE2 þ V
B þ V
2
that is,
IE ¼ 1 ðV Et V EÞ
jxE
ð7:77Þ
IE2 ¼ 1 Et V
B V
2Þ
ðV
jðxE2 þ xB Þ
Bt ¼ xB Et V
B V
2Þ þ V
B
V ðV ð7:78Þ
ðxE2 þ xB Þ
By using Eq. (7.59), linearization of Eqs. (7.77) and (7.78) becomes the
following:
DIExy ¼ aEdc DVdc þ aEm DmE þ aEd DdE þ aTEEt DVEtxy

DIE2xy ¼ aE2dc DVdc þ aE2m DmB þ aE2d DdB ð7:79Þ
þ aTE2Et DVEtxy þ aTE22 DVE2xy
DVBtxy ¼ bBtdc DVdc þ bBtm DmB þ bBtd DdB

ð7:80Þ
þ bTBtEt DVEtxy þ bTBt2 DVE2xy
By using Eqs. (7.79) and (7.80), linearizing Eq. (4.73) gives the following:
DV_ dc ¼ cdc DVdc þ cmE DmE þ cdE DdE þ cmB DmB

ð7:81Þ
þ cdB DdB þ cTEt DVEtxy þ cT2 DVE2xy
From Eq. (7.74), it can have the following:
EtI ¼ V
PEt þ jQEt ¼ V Et ðIE þ IE2 Þ ð7:82Þ
E1
Hence by using Eq. (7.79), linearization of Eq. (7.82) can give the following:
DPEt ¼ cpdc DVdc þ cpmE DmE þ cpdE DdE þ cpmB DmB

þ cpdB DdB þ cTpEt DVEtxy þ cTp2 DVExy2
ð7:83Þ
DQEt ¼ cqdc DVdc þ cqmE DmE þ cqdE DdE þ cqmB DmB
þ cqdB DdB þ cTqEt DVEtxy þ cTq2 DVExy2
Because
h i
vEty0
DVEt ¼ vVEtx0 VEt0 DVEtxy ; ð7:84Þ
Et0
the state-space realization of four UPFC controllers of Eq. (7.72) can be written as
follows:
X_ UPFCAC ¼ AUPFCAC XUPFCAC þ BUPFCAC DVEtxy

ð7:85Þ
DmE ¼ CUPFCAC XUPFCAC þ DUPFCAC DVEtxy þ Dupss
X_ UPFCDC ¼ AUPFCDC XUPFCDC þ BUPFCDC DVdc

ð7:86Þ
DdE ¼ CUPFCDC XUPFCDC þ DUPFCDC DVdc
X_ UPFCP ¼ AUPFCP XUPFCP þ BUPFCP DPEt

ð7:87Þ
DmB ¼ CUPFCP XUPFCP þ DUPFCP DPEt
X_ UPFCQ ¼ AUPFCQ XUPFCQ þ BUPFCQ DQEt

ð7:88Þ
DdB ¼ CUPFCQ XUPFCAC þ DUPFCQ DQEt
Denote
T
XUPFC ¼ DVdc XTUPFCAC XTUPFCDC XTUPFCP XTUPFCQ ð7:89Þ
By using Eqs. (7.83), (7.85), and (7.86), Eqs. (7.87) and (7.88) can be written as
X_ UPFCP ¼ AUPFCPP XUPFC þ BUPFCPEt DVEtxy þ BUPFCP2 DVE2xy þ BUPFCPPSS Dupss

DmB ¼ CUPFCPP XUPFC þ DUPFCPEt DVEtxy þ DUPFCP2 DVE2xy þ DUPFCPPSS Dupss
ð7:90Þ
X_ UPFCQ ¼ AUPFCQQ XUPFC þ BUPFCQEt DVEtxy þ BUPFCQ2 DVE2xy þ BUPFCQPSS Dupss

DdB ¼ CUPFCQQ XUPFC þ DUPFCQEt DVEtxy þ DUPFCQ2 DVE2xy þ DUPFCQPSS Dupss
ð7:91Þ
By substituting the second equation of Eqs. (7.85), (7.86), (7.90), and (7.91) into
Eq. (7.81), it can have the following:
DV_ dc ¼ cDCUPFC XUPFC þ cTDCEt DVEtxy þ cTDC2 DVE2xy þ cDCPSS Dupss ð7:92Þ
From the first equation of Eqs. (7.85), (7.86), and (7.90)–(7.92), the overall state
equation of the UPFC dynamic and control functions can be written as
X_ UPFC ¼ AUPFC XUPFC þ BUPFCEt DVEtxy þ BUPFC2 DVE2xy þ BUPFCpss Dupss

ð7:93Þ
Let
h iT h iT
DVUPFC ¼ DVTEtxy DVTE2xy ; DIUPFC ¼ DITExy DITE2xy
X_ UPFC ¼ AUPFC XUPFC þ BUPFC DVUPFC þ BUPFCpss Dupss

ð7:94Þ
DIUPFC ¼ CUPFC XUPFC þ DUPFC DVUPFC þ DUPFCpss Dupss
Similar to Eq. (6.98) and (7.42), let the linearized network equation of the
N-machine power system with the UPFC-based stabilizer installed be
2 3 2 32 3
DIxy Ygg YgUPFC Yg2 DVxy
4 DIUPFC 5 ¼ 4 YUPFCg YUPFC YUPFCO 54 DVUPFC 5 ð7:95Þ
0 YOg YOUPFC YO DVO
Similar to the derivation from Eqs. (7.43)–(7.46), from Eqs. (7.94) and (7.95) the
open-loop, state equation of the system without the UPFC stabilizer installed can be
obtained. The output equation can be obtained by using Eqs. (7.83), (7.85), (7.86),
(7.90), and (7.91).
7.2 Design of a Shunt VSC-Based Stabilizer … 349
7.2 Design of a Shunt VSC-Based Stabilizer

by Localized Phase Compensation Method
to Suppress Inter-area Line Power Oscillations
In Sect. 2.3.1.3, the phase compensation method is introduced for the design of a
PSS on the basis of the Heffron–Phillips model. In Sect. 4.2.2, it is demonstrated
that when a general linearized model is used, a SSSC stabilizer can be designed by
the use of the phase compensation method. Those are examples of applying the
phase compensation method in a single-machine infinite-bus power system. They
are the cases of suppressing a local-mode power oscillation along a transmission
line, where the power oscillation is closely related to the rotor motion of the
synchronous generator which is connected at one end of the transmission line.
Acceleration and deceleration of the generator absorbs and releases the active
power in responding to the power oscillation dynamically. Hence if a stabilizer (the
PSS or FACTS-based stabilizer) is designed to supply positive damping torque to
damp the oscillation of the rotor motion of the generator, it thus can effectively
suppress the local-mode power oscillation.
As it is introduced in Chaps. 5 and 6, a stabilizer in a multi-machine power
system contributes the damping torque to the electromechanical oscillation loop of
every generator. Hence, the application of the phase compensation method involves
the dynamic interactions of all generators, as it is shown in Sect. 6.2.4.2. The main
reason is that the phase compensation method has been introduced on the basis of
the damping torque analysis. In addition, like the modal analysis, the damping
torque analysis (hence the phase compensation method) needs the global infor-
mation of the multi-machine power system to establish the linearized model of the
whole power system, which in practice may not always be readily available and is
difficult to be validated when the system is large and complex.
This section introduces a different phase compensation method for the design of
a VSC-based stabilizer to suppress an inter-area line power oscillation. It is
established straight on the damping analysis of the inter-area mode line power
oscillation along the transmission line where the VSC-based stabilizer locates,
rather than relying on the concept of the damping torque analysis. Its application
considers the direct contribution from the stabilizer to the damping of the inter-area
line power oscillation and does not involve in explicit consideration of interactive
dynamics of generators. Moreover, it does not need to obtain and validate the
complete power system information for applying the phase compensation method to
design the VSC-based stabilizer, because the design only requires the locally
available parameters of the transmission lines and the VSC-based unit. Hence, it is
named as the localized phase compensation method.
7.2.1 Localized Small-Signal Model of a VSC-Based Unit

in a Multi-machine Power System
7.2.1.1 Line Power Small-Signal Oscillation [6]
Figure 7.5 shows a VSC-based unit installed in a multi-machine power system. The
unit has a DC/DC converter behind the DC capacitor, Cdc , which can be connected
to an energy storage system. It can also be connected to a power generation plant,
such as a fuel cell (FC) or photovoltaic (PV) power plant. Without the DC/DC
converter (Idc2 ¼ 0), the unit is a STATCOM. From Eq. (7.13), it can have the
following:
c ¼ mkc Vdc \c ¼ mkc Vdc \u þ /

V ð7:96Þ
where the modulation phase, /, is the phase difference between V c ¼ Vc \c and

Vs ¼ Vs \u at node s, i.e. / ¼ c u. From Fig. 7.5, it can have the following:

S V
V C

Vs ¼ jxsB IsB þ VB ¼ jXsB IAs B
þV
jxs ð7:97Þ
xsB xsB
¼ jxsBIAs Vs þ B
Vc þ V
xs xs

s ¼ jxsB IAs þ
V
xsB c þ VB
V ð7:98Þ
1 þ xxsBs xs ð1 þ xxsBs Þ 1 þ xxsBs
Fig. 7.5 A VSC-based unit A multi-machine power system

in a multi-machine power
system x As Node s x sB
VA Vs VB
IAs I sB
Node A Ic xs Node B
Vc
m
DC/AC VSC
φ
I dc1
Cdc Idc2
DC/DC
dc
converter
power storage or generation unit

Fig. 7.6 Phasor diagram in Y

the x–y coordinate VB
Va
δΑΒ bVB
δΑΒ
δ ′ aVc
γ VA
X
Hence, it can be obtained that
s ¼ jðxAs þ xs xsB ÞIAs

A ¼ jxAsIAs þ V
V
xs þ xsB
xsB xs ð7:99Þ
þ Vc þ B ¼ jxRIAs þ V
V a
xs þ xsB xs þ xsB
where
xs xsB a ¼ xsB xs B ¼ aV
c þ bV
B
xR ¼ ðxAs þ Þ; V Vc þ V
xs þ xsB xs þ xsB xs þ xsB
From Eq. (7.99) and the phasor diagram of Fig. 7.6, the active power delivered
along the transmission line from node A to s in Fig. 7.5 can be obtained as
Va VA VA
PA ¼ sin d0 ¼ ðbVB sin dAB þ akmVdc sin cÞ
xR xR
ð7:100Þ
VA VA
¼ bVB sin dAB þ akmVdc sinðu þ /Þ
xR xR
Hence, the small-signal variation of line power is given as
DPA ¼ hAB DdAB þ hA DVA þ hB DVB þ hu Du

ð7:101Þ
þ hdc DVdc þ hm Dm þ hf D/
where
@PA @PA @PA @PA
hAB ¼ ; hA ¼ ; hB ¼ ; hu ¼ ;
@dAB @VA @VB @u
@PA VA0
hdc ¼ ¼ akm0 sinðu0 þ /0 ÞDVdc
@Vdc xR
@PA VA0
hm ¼ ¼ akVdc0 sinðu0 þ /0 Þ
@m xR
@PA VA0
hf ¼ ¼ akm0 Vdc0 cosðu0 þ /0 Þ
@/ xR
Equation (7.101) indicates that the small-signal variation of line power is directly
affected by the variation of line terminal variables, DdAB , DVA , DVB , Du and the
internal variables of the VSC-based unit, DVdc , Dm, D/:
7.2.1.2 Control Function Model of the VSC
The VSC is controlled by the DC/AC conversion to regulate active and reactive
power exchange between the VSC-based unit and the rest of the power system. Two
most popular control algorithms are the pulse width modulation (PWM) and the
pulse amplitude modulation (PAM). Figure 7.7 shows the configuration of a voltage
control type of the PWM, where m and / are the modulation ratio and phase of the
PWM, respectively.
From Fig. 7.7, the linearized model of the PWM VSC control functions can be
written as

Kac
Dm ¼ Kpac þ DVs þ Dupss
s
ð7:102Þ
Kdc
D/ ¼ Kpdc þ DVdc þ Dupss
s
where Kpac , Kac , Kpdc , and Kdc is the gain of the AC and DC voltage proportional
and integral (PI) controllers. Obviously, if the PWM VSC controllers are not PI
ac voltage controller dc voltage controller

Vs u pss Vdc u pss
− m − φ
+ +
K pac + K pdc +
Vs-ref Vdc-ref
K ac x ac K dc x dc
s s
Fig. 7.7 Configuration of PWM VSC controllers (voltage control type)

controllers, then small-signal model of the VSC control can always be expressed by
the following general state-space representation as
sXPWM ¼ APWM XPWM þ bPWMs DVs þ bPWMdc DVdc

Dm ¼ CAC XPWM þ dAC DVs þ Dupss ð7:103Þ
D/ ¼ CDC XPWM þ dDC DVdc þ Dupss
where XPWM ¼ ½ xac xdc T is the vector of state variables of the PWM PI con-
trollers as shown in Eq. (7.103) and in Fig. 7.7.
Figure 7.8 shows the configuration of the voltage control type of the PAM VSC
control. From Fig. 7.8, the small-signal model of the PAM VSC control can be
written as

Kpam Kac
D/ ¼ Kppam þ ðDVs þ Dupss Þ DVdc ð7:104Þ
s s
where Kp , Kppam , and Kpam is the gain of corresponding proportional or integral

controllers in Fig. 7.8. The state-space representation of Eq. (7.104) can be written as
sXPAM ¼ APAM XPAM þ bPAM DVs þ bpampss Dupss

ð7:105Þ
D/ ¼ CPAM XPAM þ dDC DVdc
T
where XPAM ¼ ½ xpam1 xpam2 the vector of state variables of the PAM PI con-
trollers as shown in Eq. (7.104) and in Fig. 7.8.
The VSC controllers can also be the current control type, where the limit of
current can be easily applied for the protection of the VSC-based unit. Figure 7.9
shows the configuration of the current control type of the VSC PWM controllers.
As the difference in the control type (voltage or current) does not affect the fol-
lowing discussion, the voltage control type of the VSC controllers is used as the
representative in this section to demonstrate how the localized small-signal model
of the VSC-based unit can be established.
Fig. 7.8 Configuration of Vs Vdc

PAM VSC controller (voltage
control type) − − φ
+ Kp + K pam +
Vs-ref s x pam1
+
u pss K pam x pam2

s
Vs I sq0 u pss-m
− + m
+ +
K ac (s) K rp (s) +
Vsref −
I sq
Vdc I sd0 u pss-f
− + φ
+ +
K dc (s) K ap (s) +
Vdcref −
Isd
Fig. 7.9 Configuration of PWM VSC controllers (current control type)
7.2.1.3 Dynamic Equation of the VSC
c ¼ jxsIs þ V
From Fig. 7.5, it can have V s . Hence,
mkVdc sin / Vs mkVdc cos /

isd ¼ ; isq ¼ ð7:106Þ
xs xs
where subscripts d and q denote the d and q components of the corresponding

variable, respectively, in the d–q coordinate of the VSC-based unit, which is shown
in Fig. 7.10. Linearization of the above equations is given as
Disd ¼ b1d DVdc þ b2d Dm þ b3d D/

ð7:107Þ
Disq ¼ b1q DVdc þ b2q Dm þ b3q D/ þ b4q DVs
The active power exchange between the VSC-based unit and the rest of the
power system is (see Eq. (7.96) given as vcd ¼ mkVdc cos /; vcq ¼ mkVdc sin /)
Vdc Idc1 ¼ isd vcd þ isq vcq ¼ isd mkVdc cos / þ isq mkVdc sin /
Fig. 7.10 Phasor diagram in y

the d–q coordinate of q
VSC-based unit Vc
Vs d
φ
ϕ
x
Hence
Idc1 ¼ isd mk cos / þ isq mk sin / ð7:108Þ
Thus, linearization of Eq. (7.108) can be obtained from Eq. (7.107) as
DIdc1 ¼ bd1 DVdc þ bd2 Dm þ bd3 D/ þ bd4 DVs ð7:109Þ
The small-signal model of a power storage or generation unit can always be

expressed by the following state-space representation:
sXP ¼ AP XP þ bP DVdc þ bvsc Duvsc

ð7:110Þ
DIdc2 ¼ CP XP þ dP DVdc
where XP and Duvsc are the state variable vector and input of the power storage or
generation unit, respectively. In Sect. 7.3, a FC (fuel cell) power plant is used as an
example to demonstrate the procedure to derive the small-signal model of
Eq. (7.110). A similar procedure to what is demonstrated in Sect. 7.3 can be applied
to derive the model of Eq. (7.110) for the energy storage system.
Dynamic equation of the VSC is given as
1
V_ dc ¼ ðIdc1 þ Idc2 Þ ð7:111Þ
Cdc
Hence by using Eqs. (7.109)–(7.111), linearization of Eq. (7.111) can be

obtained as
sDVdc ¼ b1 DVdc þ b2 Dm þ b3 D/ þ b4 DVs þ bdc XP ð7:112Þ
7.2.1.4 Localized Small-Signal Model of the VSC-Based Unit
From Fig. 7.5 and Eq. (7.99), it can have the following:
A ¼ jxR ðiAsx þ jiAsy Þ þ V

V a
ð7:113Þ
¼ xR iAsy þ Va cos d0 þ jxR iAsx þ jVa sin d0
where subscripts x and y denote the x and y components of the corresponding

variable, respectively, in the x–y coordinate of the power system as shown in
Fig. 7.6. In x–y coordinate, Eq. (7.113) is given as
Va sin d0 Va cos d0 VA
iAsx ¼ ; iAsy ¼ ð7:114Þ
xR xR
because
A ¼ jxAsIAs þ V
V s
ð7:115Þ
¼ jxAs ðiAsx þ jiAsy Þ XAs iAsY þ vsx þ jvsy
it can be obtained from Eqs. (7.114) and (7.115) that
Va cos d0 VA
vsx ¼ VA þ xAs iAsy ¼ VA þ xAs
xR

xAs x R x
Va cos d0 þ
As
¼ VA
xR xR

xAs xR xAs ð7:116Þ
¼ ðbVB cos dAB þ akmVdc cos cÞ þ VA
xR xR
Va sin d0
vsy ¼ xAs IAsx ¼ xAs
xR
xAs
¼ ðbVB sin dAB þ akmVdc sin cÞ
xR
Linearization of Eq. (7.116) gives the following:
Dvsx ¼ a01x DVdc þ a02x Dm þ a03x Dc þ a04x DdAB þ a05x DVA þ a06x DVB
ð7:117Þ
Dvsy ¼ a01y DVdc þ a02y Dm þ a03y Dc þ a04y DdAB þ a05y DVA þ a06y DVB
s and V
However, the modulation phase, /, is the angle between V c , as given in
Eq. (7.96), that is
vsy
c ¼ / þ u ¼ / þ arctan1 ð7:118Þ
vsx
from Eqs. (7.117) and (7.118), it can have the following:
Dc ¼ D/ þ a01 DVdc þ a02 Dm þ a03 Dc þ a04 DdAB þ a05 DVA þ a06 DVB ð7:119Þ
Dc ¼ ag D/ þ ag1 DVdc þ ag2 Dm þ ag4 DdAB þ ag5 DVA þ ag6 DVB ð7:120Þ
By substituting Eq. (7.120) into (7.117), it can be obtained that
Dvsx ¼ a1x DVdc þ a2x Dm þ a3x D/ þ a4x DdAB þ a5x DVA þ a6x DVB
ð7:121Þ
Dvsy ¼ a1y DVdc þ a2y Dm þ a3y D/ þ a4y DdAB þ a5y DVA þ a6y DVB
s ¼ vsx þ jvsy . Hence, the linearization

In x–y coordinate of the power system, V
of Vs can be obtained from Eq. (7.121) as
vsx0 vsy0
DVs ¼ Dvsx þ Dvsy ¼ a1f DVdc þ a2f Dm
Vs0 Vs0 ð7:122Þ
þ a3f D/ þ a4f DdAB þ a5f DVA þ a6f DVB
Finally from Eqs. (7.101), (7.102), (7.104), (7.110), (7.112), and (7.122), the
localized small-signal model of the VSC-based unit can be shown by the block
diagram in Fig. 7.11 when the VSC adopts the PWM control algorithm, and
Fig. 7.12 when the PAM is adopted by the VSC. To distinguish the stabilizer added
on the AC and DC voltage control loops of the PWM VSC (see Fig. 7.7) as well as
on the control function of the PAM VSC (see Fig. 7.8), in Figs. 7.10 and 7.11, the
stabilizing signal is denoted as Dupss ðmÞ, Dupss ð/Þ, and Dupss ðpamÞ, respectively.
In the following section, the corresponding stabilizer is named as the PWM reactive
power stabilizer, the PWM active power stabilizer, and the PAM stabilizer.
Obviously, the establishment of the localized small-signal model of the VSC-based
unit only requires the locally available data and parameters associated with the
transmission lines and the VSC-based unit. The global information of the whole
power system is not needed.
h AB
Δm
Δu pss (m) + hm hA
K ac
K pac + a 2f b2 hB
s
ΔδAB
a 4f
ΔVA Δϕ
a 5f + hϕ +
ΔPA
ΔVB
a 6f
a1f
b4
a 3f
ΔVdc
Δu pss (φ) Δφ 1
+ b3 + h dc
s − b1
bp hf
bdc
(sI - A P ) −1 +
Xp
b vsc
Δu vsc
K dc
K pdc +
s
Fig. 7.11 Block diagram of localized small-signal model of VSC-based unit adopting the PWM
control
hB
hA
h AB
ΔδAB
a 4f Δϕ hϕ
ΔPA
ΔVA
a 5f + +
ΔVB
a 6f a 3f
a1f
- K pam
+ Kp + K ppam + hf
s s Δφ
Δu pss (pam) -
b4 b3 h dc
Xp bdc +
1
(sI - A P ) −1
s − b1
b vsc + bp
Δu vsc ΔVdc
Fig. 7.12 Block diagram of localized small-signal model of VSC-based unit adopting the PAM
control
7.2.2 Design of VSC-Based Stabilizer by Localized Phase

Compensation Method
In the Heffron–Phillips model of a single-machine system installed with a PSS of

Fig. 2.10, the electric torque to the electromechanical oscillation loop is contributed
to the deviation of rotor angular position Dd and PSS stabilizing signal Dupss . The
principle of superimposition of linear systems is given as
DT ¼ DTðDdÞ þ DTðDupss Þ
ð7:123Þ
DTðDdÞ ¼ Fdelta ðsÞDd; DTðDupss Þ ¼ Fpss ðsÞDupss
where Fdelta ðs) and Fpss ðs) are the transfer functions of the forward path from Dd and
the stabilizing signal of the PSS to DT, the electric torque contribution to the
electromechanical oscillation loop, respectively. The damping torque analysis is the
decomposition of the electric torque on Dd Dx coordinate into a synchronizing

torque (proportional to Dd) and damping torque (proportional to Dx),
DTðDdÞ ¼ Fdelta ðjxs ÞDd ¼ Cdelta Dd þ Ddelta Dx

ð7:124Þ
DTðDupss Þ ¼ Fpss ðjxs ÞDupss ¼ Cpss Dd þ Dpss Dx
Since only the damping torque contribution affects the damping of the rotor
motion, the phase compensation method is to ensure that the direct contribution
from the PSS is only the damping torque; that is, DTðDupss Þ ¼ Dpss Dx. In
Eq. (7.124), DTðDdÞ ¼ Cdelta Dd þ Ddelta Dx represents the effect of the variable (i.e.
DdÞ external to the PSS on the damping of power oscillation. It is not considered
when the PSS is designed (though it is a part of dynamic of the power system).
As it has been explained in Sect. 2.3.2.2, in fact, the damping torque analysis
converts the problem of suppression of the local-mode line power oscillation (i.e.
DPt ) into that of the damping of the rotor motion (i.e. DdÞ. That is why the damping
torque analysis in a multi-machine power system involves all generators. However,
if the damping of the local-mode line power oscillation, DPt , is considered directly,
from Fig. 2.10, Eqs. (7.123) and (7.124), it can have the following:
DPt ¼ DPt ðDdÞ þ DPt ðDupss Þ

DPt ðDdÞ ¼ K1 Dd þ Cdelta Dd þ Ddelta Dx ð7:125Þ
DPt ðDupss Þ ¼ Cpss Dd þ Dpss Dx
According to the damping torque analysis, only the component in DPt propor-
tional to Dx determines the damping of line power oscillation. The phase com-
pensation method is to ensure that the PSS only contributes the damping
component. The localized phase compensation method introduced below is based
on the establishment of Eq. (7.125) in a multi-machine power system.
According to the principle of superimposition of linear systems, from Figs. 7.11
and 7.11, it can have the following:
DPA ¼ DPA ðDEÞ þ DPA ðDupss Þ

DPA ðDEÞ ¼ ½FAB ðsÞ FA ðsÞ FB ðsÞ Fu ðsÞ Fvsc ðsÞDE ð7:126Þ
DPA ðDupss Þ ¼ Fpss ðsÞDupss
where DE ¼ ½DdAB DVA DVB Du Duvsc T ; Dupss is Dupss ðmÞ, Dupss ð/Þ or
Dupss ðpamÞ; FAB ðsÞ, FA ðsÞ, FB ðsÞ,Fu ðsÞ; Fvsc ðsÞ, and Fpss ðsÞ is the transfer function
of the forward path from DdAB , DVA , DVB , Du, Duvsc , and Dupss to DPA in
Figs. 7.11 and 7.12, respectively.
T
In fact, by denoting XVSC ¼ DVdc XTPWM XTP , from Eqs. (7.101), (7.103),
(7.105), (7.110), (7.112), and (7.126), it can be obtained that
sX_ VSC ¼ AVSC XVSC þ BVSC DE þ BPSS Dupss

ð7:127Þ
DPA ¼ CVSC XVSC þ DVSC DE þ dPSS Dupss
Hence comparing Eqs. (7.126) and (7.127), it can have
DPA ðDEÞ ¼ ½CSVC ðsI ASVC Þ1 BSVC þ DSVC DE ð7:128Þ
DPA ðDupss Þ ¼ ½CSVC ðsI ASVC Þ1 BPSS þ dpss Dupss ð7:129Þ
That is,
½FAB ðsÞ FA ðsÞ FB ðsÞ Fu ðsÞ Fvsc ðsÞ ¼ ½CSVC ðsI ASVC Þ1 BSVC þ DSVC
ð7:130Þ
Fpss ðsÞ ¼ ½CSVC ðsI ASVC Þ1 BPSS þ dpss Dupss ð7:131Þ
In Eq. (7.129), DPA ðDupss Þ is the direct contribution from a VSC-based stabilizer
to the line power variation (oscillation), DPA . Obviously, Eqs. (7.128) and (7.129)
are the extensions of Eq. (7.125) in the multi-machine power system.
Main Conclusion: If the stabilizer is designed to ensure DPA ðDupss Þ to be pro-
portional to the time derivative of DPA , that is
DPA ðDupss Þ ¼ Dpss DP_ A ðDpss [ 0Þ ð7:132Þ
the stabilizer will supply positive damping to suppress the inter-area line power
oscillation.
The above main conclusion can be established by using the graphical explana-
tion (similar to the introduction in Sect. 2.3.2.2) based on the linearized P d curve
and equal-area criterion as follows.
Figure 7.13 shows the linearized PA dAB curve, where ðdAB0 ; PA0 Þ is the
operating point of the power system at steady state. It is assumed that the
small-signal oscillation of DPA starts from point “a” in Fig. 7.13 with the operating
point moving down. In dAB Dd_ AB coordinate, DPA ðDEÞ can always be decom-
posed to be
DPA ðDEÞ ¼ Cother DdAB þ Dother Dd_ AB ð7:133Þ
where Cother and Dother are two constants. Without affecting the establishment of the
main conclusion, it is assumed that Dother [ 0. When there is no stabilizing control
(DPA ðDupss Þ ¼ 0), DPA ¼ DPA ðDEÞ ¼ Cother DdAB þ Dother Dd_ AB which is shown
by the dashed curve in Fig. 7.13. This is because when the operating point moves
Fig. 7.13 Graphical

.
explanation of the main ΔPA(ΔE)=CotherΔδAB+DotherΔδAB a
conclusion Direction
. of
ΔPA<0 ΔPA(ΔE)=CotherΔδAB
A1
PA0 g d e
A2 .
c ΔPA=ΔPA(ΔE)+DpssΔPA
δAB2' δΑΒ2 δΑΒ0 δΑΒ1

δAB
down, Dd_ AB \0: Hence, Dother Dd_ AB \0 is added on the line DPA ðDEÞ ¼
Cother DdAB . When the operating point arrives at point “f”, it stops moving
(Dother Dd_ AB ¼ 0Þ. Hence, it should be on the line DPA ¼ DPA ðDEÞ ¼ Cother DdAB .
According to the equal-area criterion, the area “ade” is equal to that of “dgf”.
When the stabilizing control is added and is set according to Eq. (7.132),
Dpss DP_ A \0 (DP_ A \0) will be added on the dashed curve DPA ðDEÞ ¼
Cother DdAB þ Dother Dd_ AB as shown in Eq. (7.126). Hence, when the operating point
moves down, it should move below the dashed curve along the highlighted tra-
jectory in Fig. 7.13. When the operating point stops on the line
DPA ðDEÞ ¼ Cother DdAB , the area A1 is equal to area A2 at point “c”, where DP_ A ¼
0 and Dother Dd_ AB ¼ 0: Obviously, dAB1 dAB0 [ dAB0 dAB2 , indicating that
extra positive damping is provided by the stabilizing control to the power oscil-
lation. A similar analysis can be carried out to examine the case when the operating
point moves up from point “c”. Thus, the above main conclusion is established.
Hence, if DPA is used as the feedback signal of the VSC-based stabilizer and the
transfer function of stabilizer is constructed by a lead-lag, then it can have the
following:
sTw 1 þ sT2 1 þ sT4

Dupss ¼ Kpss DPA ð7:134Þ
1 þ sTw 1 þ sT1 1 þ sT3
According to Eqs. (7.129), (7.132), and (7.134), parameters of the stabilizer can
be set at the angular oscillation frequency, xs , to satisfy the following equation

Fpss ðsÞKpss ¼ Dpss s ð7:135Þ
with s ¼ jxs . That is, the stabilizer is set with a phase to compensate the phase of
Fpss ðsÞ
s such that a pure positive damping is provided to the line power oscillation.
7.2.3 Robustness of an ESS-Based Stabilizer to Variation

of Line-Loading Conditions
7.2.3.1 Localized Small-Signal Model of an Energy Storage System [7]
In Fig. 7.5, the shunt-connected VSC-based unit can be an energy storage system
(ESS). At the steady-state operation of the power system, there is no exchange of
active power between the ESS and the rest of power system, i.e. /0 ¼ 0. Hence
from Eq. (7.106), it can have isd0 ¼ 0 and
mkVdc sin / k sin /0

Disd ¼ D ¼ ðVdc0 Dm þ m0 DVdc Þ
xs xs
m0 kVdc0 cos /0 m0 kVdc0 Vc0
þ D/ ¼ D/ ¼ D/ ð7:136Þ
xs xs xs
Vs0 Vc0
isq0 ¼
xs
From the above equation, linearization of Eq. (7.108) can be obtained as
DIdc1 ¼ isd0 ðk cos /0 Dm m0 k sin /0 D/Þ

þ isq0 ðk sin /0 Dm þ m0 k cos /0 D/Þ
þ m0 k cos /0 Disd þ m0 k sin /0 Disq ð7:137Þ
Vc0
¼ isq0 m0 kD/ þ m0 k D/
xs
Hence, linearization of Eq. (7.111) for the ESS is given as
DIdc2
sDVdc ¼ kdc D/ þ ð7:138Þ
Cdc
where kdc is a constant.

Denote the transfer function of the AC and DC voltage controllers of the VSC
adopting the PWM algorithm (see Fig. 7.7 and Eq. (7.102) where the controllers
adopt the PI control law) generally to be Kac ðsÞ and Kdc ðsÞ, the supplementary
damping control signal superimposed on the AC and DC voltage control functions,

respectively, to be upssm and upssf , that is
m ¼ m0 þ Kac ðsÞðVsref Vs Þ þ upssm

ð7:139Þ
/ ¼ /0 þ Kdc ðsÞðVdcref Vdc Þ þ upssf
Because the ESS supplementary damping controller associated with upssm and
upssf is implemented through regulating, respectively, the exchange of reactive and
active power between the ESS and the rest of the power system, it is named as ESS
reactive power PSS (power system stabilizer) and active power PSS
correspondingly.
Similarly, denote the transfer functions of the voltage controllers of the VSC
adopting the PAM algorithm generally to be K1 ðsÞ and K2 ðsÞ (see Fig. 7.8 and
Eq. (7.104) where the controllers adopt the I and PI control law, respectively), the
added damping control signal to be upsspam , that is,

1 1
/ ¼ /0 þ K1 ðsÞ K2 ðsÞðVsref Vs þ upsspam Þ þ Vsref Vdc ð7:140Þ
k k
The damping controller associated with upsspam is called the PAM PSS. By using
Eq. (7.122), linearization of Eqs. (7.138)–(7.140) can be obtained as
Kac ðsÞða1f DVdc þ a3f D/ þ a4f DdAB þ a5f DVA þ a6f DVB Þ
Dm ¼
1 þ a2f Kac ðsÞ
1 ð7:141Þ
þ Dupssm
1 þ a2f Kac ðsÞ
D/ ¼ Kdc ðsÞDVdc þ Dupssf
K1 ðsÞK2 ðsÞða4f DdAB þ a5f DVA þ a6f DVB Þ

D/ ¼
k þ a3f K1 ðsÞK2 ðsÞ
ð7:142Þ
K1 ðsÞK2 ðsÞ K1 ðsÞK2 ðsÞa1f þ kK1 ðsÞ
þ Dupsspam DVdc
k þ a3f K1 ðsÞK2 ðsÞ k þ a3f K1 ðsÞK2 ðsÞ
From Eqs. (7.138), (7.141), and (7.142), localized small-signal model of the ESS
is shown in Figs. 7.14 and 7.15, respectively, for the VSC adopting the PWM and
PAM, where DPA ðuESS Þ ¼ hdc DVdc þ hm Dm þ hf D/; uESS ¼ ½ DVdc Dm D/ T .
7.2.3.2 Robustness of the ESS-Based Stabilizer to the Variations

of Line-Loading Conditions
In Fig. 7.13, when the operating point moves along the linearized PA dAB curve,
sign of Dd_ AB and DP_ A is always same and Dd_ AB ¼ 0 when DP_ A ¼ 0. Hence, it can
Δu pss − m 1 Δm
+ hm
1 − a 2f K ac (s)
K ac (s)
1 − a 2f K ac (s)
ΔδAB
a 4f
ΔVA ΔPA (u ESS )
a 5f + a1f +
ΔIdc2
ΔVB
a 6f
1
a 3f
Cdc
Δu pss −f 1
+ Δφ
−k dc + h dc
s
ΔVdc
K dc (s)
hf
Fig. 7.14 Localized small-signal model of the ESS with the VSC adopting the PWM algorithm
ΔδAB
a 4f
ΔVA ΔPA (u ESS )
a 5f +
hf +
ΔVB ΔIdc2
a 6f
K1 (s)K 2 (s) 1
k + a 3f K1 (s)K 2 (s) Cdc
Δu pss − pam -
K1 (s)K 2 (s) + −k dc
1
Δφ
+ h dc
k + a 3f K1 (s)K 2 (s) - s
ΔVdc
a1f K1 (s)K 2 (s) + kK1 (s)

k + a 3f K1 (s)K 2 (s)
Fig. 7.15 Localized small-signal model of the ESS with the VSC adopting the PAM algorithm
be easily concluded that in Eq. (7.126), if the decomposition of DPA ðDupss Þ in

DdAB Dd_ AB coordinate is
DPA ðDupss Þ ¼ CPSS DdAB þ DPSS Dd_ AB ð7:143Þ
the stabilizer associated with Dupss can suppress the line power oscillation if and
only if DPSS [ 0. The bigger DPSS [ 0 is, the more damping is provided by the
stabilizer.
Figures 7.16 and 7.17 show the block diagram of the direct damping contri-
bution from the ESS reactive power PSS and active power PSS to the small-signal
oscillation of line power, DPA ðDupssm Þ and DPA ðDupssf Þ, respectively, when the
Δu pss − m 1 Δm ΔPA (u pss − m )

hm
1 − a 2f K ac (s)
Fig. 7.16 Forward path of ESS reactive power PSS
Δm
hm
K ac (s)
1 − a 2f K ac (s)
ΔPA (u pss −f )
+ a1f +
a 3f
Δu pss −f 1
+ −k dc h dc
Δφ s
ΔVdc
K dc (s)
hf
Fig. 7.17 Forward path of ESS active power PSS
PWM algorithm is used. In Fig. 7.16, if Dm is decomposed in DdAB Dd_ AB

coordinate to be
Dm ¼ C0pssm DdAB þ D0pssm Dd_ AB ð7:144Þ
from Fig. 7.15, Eqs. (7.101), (7.126), (7.143), and (7.144), it can have (/0 ¼ 0)
VA
DPSS ¼ hm D0pssm ¼ akVdc0 sin u0 D0pssm ð7:145Þ
xR
At a higher line-loading condition, u0 and hence DPSS is bigger. This means the
ESS reactive power PSS is more effective to suppress the line power oscillation at a
higher line-loading condition.
In the case of the ESS active power PSS when the PWM control is used as
shown in Fig. 7.17, if the decomposition of Dm, DVdc , and D/ in DdAB Dd_ AB
coordinate, respectively, is
Dm ¼ Cpssfm DdAB þ Dpssfm Dd_ AB

DVdc ¼ Cpssfdc DdAB þ Dpssfdc Dd_ AB ð7:146Þ
D/ ¼ Cpssff DdAB þ Dpssff Dd_ AB
from the above decomposition, Fig. 7.17, Eqs. (7.101), (7.126), (7.143) and
(7.144), then it can have (/0 ¼ 0)
DPSS ¼ hm Dpssfm þ hdc Dpssfm þ hf Dpssff

VA
¼ akðVdc0 sin u0 Dpssfm þ m0 sin u0 Dpssfdc þ m0 Vdc0 cos u0 Dpssff Þ
xR
ð7:147Þ
When the line-loading condition increases, sin u0 increases but cos u0 decreases
in Eq. (7.147). Hence, DPSS does not change as much as it does in the case of the
ESS reactive power PSS. This means that the robustness of the ESS active power
PSS to the variations of line-loading conditions is expected to be better than that of
the ESS reactive power PSS. The same conclusion can be drawn from Fig. 7.18 for
the ESS PAM PSS when the PAM control is used.
The same conclusion has been established in Sect. 4.1.2.2 in the case of a
single-machine infinite-bus power system. Function of the ESS reactive power
stabilizer involves the regulation of magnitude of the terminal voltage of the VSC.
Hence, in essence it works via controlling the exchange of reactive power between
the ESS and the rest of the power system. Both the ESS active power stabilizer and
the PAM stabilizer function through the regulation of phase of the terminal voltage
of the ESS, that is, the exchange of the active power between the ESS and the rest
of the power system. Therefore, the ESS active power stabilizer and PAM stabilizer
suppress the line power oscillation directly by absorbing active power from or
injecting it into the power system when they “see” the excess or lack of active
power delivered along the transmission line. Hence, their capability to damp the
power oscillation is less affected by the change of system loading conditions.
The ESS reactive power stabilizer affects indirectly the variation of active power
delivered along the transmission line through the regulation of injection or
absorbing of reactive power from the ESS. At a lower loading condition, regulation
of reactive power exchange is related less closely with the variation of line active
power. Hence, the stabilizer is less effective in damping the power oscillation.
ΔPA (u pss − pam )

hf +
Δu pss − pam ΔVdc

K1 (s)K 2 (s) + Δφ 1
−k dc h dc
k + a 3f K1 (s)K 2 (s) - s
a1f K1 (s)K 2 (s) + kK1 (s)

k + a 3f K1 (s)K 2 (s)
Fig. 7.18 Forward path of ESS PAM PSS

7.3 An Example of a Multi-machine Power System … 367
7.3 An Example of Multi-machine Power System

with a Grid-Connected FC Power Plan
Fuel cell (FC) power generation is one of the most promising technologies of clean
energy for the future. Among several types of the FC, the high-temperature solid
oxide fuel cell (SOFC) is considered to be the most attractive FC technology for
grid-connected power generation. It is foreseen that, in future power systems,
gird-connected high-capacity SOFC power plants could play certain important role
as one of the main clean power generation sources.
This section presents an example of a multi-machine power system with a SOFC
power plant which is integrated into the power system via a shunt VSC. The
objective is to demonstrate how the linearized mathematical model of the power
system with the shunt VSC-based SOFC power plant can be established for power
system small-signal stability analysis and control. It shows how the localized
small-signal model of the SOFC power plant can be developed and a stabilizer
attached to the VSC of the SOFC power plant can be effectively designed by using
the localized phase compensation method introduced in Sect. 7.2.

System with a Grid-Connected FC Power Plant
7.3.1.1 Dynamic Model of the SOFC Power Plant
Power generation by a SOFC is the conversion of chemical energy in hydrogen

(H2) and oxygen (O2) into electric energy through the following basic reaction at
two nodes of the SOFC:
Anode: H2 þ O¼ ! H2 O þ 2e
O2
Cathode: þ 2e ! O
2
DC current Ifc is generated from the releasing of electrons (2e−) at the anode due
to the above chemical reaction. A dynamic model of the SOFC power plant suitable
for use in power system studies is shown in Fig. 7.19. In Fig. 7.19, electrical
dynamic describes the chemical reaction to restore the charge that has been quickly
drained by the load. This is modelled by a first-order transfer function. To ensure
the FC operation within the safe operating area, current control reference, Ifc–ref, is
limited by the following boundaries:
Umax Umin N0
Ifcrefmax ¼ q ; Ifcrefmin ¼ q ; Kr ¼
2Kr h2in 2Kr h2in 4F
electrical Ifc-ref-max DC/DC converter DC/AC converter

dynamic dc m φ
Idc2 Idc1
Pfc-ref Pfc-ref Ic
I fc-ref = I fc
Vfc AC
DC/DC Vdc Idc Cdc
1 grid
I fc = I fc-ref
1+Te s Vc
fuel reaction
I fc-ref-min q h2-in 1 1
p h2 = (q h2-in -2K r I fc )
K h2 1+Th2 s
q h2-in 1 1
p o2 = (q o2-in -K r I fc )
2K r 1 K o2 1+To2 s
q h2-in = Ifc 1 1 1
U opt 1+Tf s p h2o = 2K r I fc
rho K h2o 1+Th2o s
fuel q o2-in
Ifc p h2 p O2 p h2O
processor
Vfc RT p h 2 × p02 0.5
Vfc = N 0 [E 0 + ln( )] - rI fc
2F p h 2o
Nernst equation
Fig. 7.19 Dynamic model of a grid-connected SOFC power plant
where Umax and Umin are the maximum and minimum fuel unitization, respectively,
N0 the number of cells in series in the FC stack, qh2–in the hydrogen input flow rate,
and F the Faraday constant.
Fuel processor depicts the dynamics of fuel supply which is a process to reform
the natural gas supply to the hydrogen-rich fuel. This can be simply represented by
a first-order lag model of time constant Tf, where Uopt is the optimal fuel utilization.
The chemical process of fuel reaction inside the FC stack is described by three
first-order transfer functions for hydrogen, oxygen, and water, respectively, where
rho is the ratio of hydrogen to oxygen, Kh2, Ko2, and Kh2o the valve molar constant,
Th2, Ko2, and Th2o the time constant and ph2, po2, and ph2o the partial pressure for
hydrogen, oxygen, and water flow, respectively. The internal EMF generated by the
SOFC stack is calculated by the well-known Nernst equation, where E0 is the ideal
standard potential, R the universal gas constant, T the absolute temperature, and r
the ohmic loss.
Grid connection of the SOFC power plant can be implemented by a two-stage
configuration of power electronics interface as shown in Fig. 7.19. In order to avoid
the underused or overused fuel, the DC/DC converter can be controlled to ensure a
constant current output from the fuel cell stack. That is,
dc ¼ dc0 þ Tdc ðsÞðIfcref Ifc Þ ð7:148Þ
where dc is the duty cycle and Tdc ðsÞ the transfer function of the FC current
controller.
A multi-machine power system
x As Node s x sB
VA IAs Vs IsB VB
Node A Ic xs Node B
Vc
SOFC
power plant
Fig. 7.20 A multi-machine power system with a SOFC power plant connected at node s
Figure 7.20 shows the SOFC power plant connected to a multi-machine power
system at node s via a VSC. Voltage at the AC terminal of the DC/AC converter of
the SOFC power plant is (see Eq. (7.96)) given as
c ¼ mkVdc \c ¼ mkVdc \u þ /
V ð7:149Þ
If the DC/AC converter is controlled by the PWM control, m and / are the
modulation ratio and phase of the PWM, respectively. If the control algorithm of
the DC/AC converter is the PAM, then m = 1.
From Eq. (7.108), it can have the following:
Idc1 ¼ isd mk cos / þ isq mk sin / ð7:150Þ
where subscripts d and q denote the d and q components of the corresponding

variable, respectively, in the d–q coordinate of the SOFC power plant as shown in
Fig. 7.20. Because
Vfc
Pfc ¼ Ifc Vfc ¼ Idc2 Vdc ; Vdc ¼ ; ð7:151Þ
1 dc
it can have Idc2 ¼ ð1 dc ÞIfc such that the dynamics of the DC/AC converter can be
expressed as
1
V_ dc ¼ ðIdc1 þ Idc2 Þ
Cdc
ð7:152Þ
1
¼ ½isd mk cos / isq mk sin / þ ð1 dc ÞIfc
Cdc
From the point of view of power system operation, node s can be treated as a PV
bus. Active power output from the SOFC power plant is kept to be pfc-ref because of
the current controller implemented in the DC/DC converter. Voltage control
scheme applied by the DC/AC converter will be very useful as far as the power
system operation control is concerned. This can be achieved by either the PWM or
the PAM algorithm as shown in Fig. 7.22 (see Figs. 7.7 and 7.8), where Kac ðsÞ and
Kdc ðsÞ are the two transfer functions of the PWM- and PAM-controlled DC/AC
converter, and upss is a supplementary damping control signal to improve power
system stability.
7.3.1.2 Linearized Model of the SOFC Power Plant
Similar to Eq. (7.106), from Fig. 7.20, it can have the following:
mkVdc sin / Vs mkVdc cos /

isd ¼ ; isq ¼ ð7:153Þ
xs xs
s ¼ vsx þ jvsy . Hence,

In the common x–y coordinate of the power system, V
linearization of Vs can be obtained as
DVs ¼ vTs0 DVsxy ð7:154Þ

h i
v Dvsx
where vTs0 ¼ vVsx0 Vsy0 ; DVsxy ¼ : From Eqs. (7.153) and (7.154), it can
s0 s0 Dvsy
have the following:
Disd ¼ b1d DVdc þ b2d Dm þ b3d D/

ð7:155Þ
Disq ¼ b1q DVdc þ b2q Dm þ b3q D/ þ bT4q DVsxy
If the linearization of Eq. (7.148) is Ddc ¼ Tdc ðsÞDIfc and the state-space
realization of transfer function Tdc ðsÞ is ðAd ; Bd ; Cd ; Dd Þ, then it can have the
following:
X_ d ¼ Ad Xd þ Bd DIfc
ð7:156Þ
Ddc ¼ Cd Xd Dd DIfc
If the state-space realization of transfer function Kac ðsÞ and Kdc ðsÞ is
ðAAC ; BAC ; CAC ; DAC Þ and ðADC ; BDC ; CDC ; DDC Þ, respectively, then from
Fig. 7.22 it can have that for the PWM control
X_ AC ¼ AAC XAC þ BAC DVsxy

ð7:157Þ
Dm ¼ CAC XAC þ DAC DVsxy þ Dupss

ð7:158Þ
D/ ¼ CDC XDC þ DDC DVdc þ Dupss
and for the PAM control
X_ AC ¼ AAC XAC þ BAC DVsxy þ bpampss Dupss

D/ ¼ CDC XDC þ DDC CAC XAC þ DDC DAC DVsxy DDC DVdc þ dpampss Dupss
ð7:159Þ
By using Eqs. (7.155)–(7.159), linearization of Eq. (7.152) can be obtained as
DV_ dc ¼ a1 Vdc þ a2 DIfc þ aT3 Xd þ aT4 XAC

ð7:160Þ
þ aT5 XDC þ aT6 DVsxy þ a7 Dupss
where Dupss can be Dumpss , Du/pss or Dupampss with different coefficient a7.
Linearization of dynamic model of the SOFC power plant of Fig. 7.19 can be
obtained as
Pfcref0 1
DIfcref ¼ DVfc ; DIfc ¼ DIfcref
V2fc0 1 þ Te s
2Kr 1 1
Dqh2in ¼ DIfc ; Dqo2in ¼ Dq
Uopt 1 þ Tf s rho h2in
1 1
Dph2 ¼ ðDqh2in 2Kr DIfc Þ
Kh2 1 þ Th2 s ð7:161Þ
1 1
Dpo2 ¼ ðDqo2in Kr DIfc Þ
Ko2 1 þ To2 s
1 1
Dph2o ¼ 2Kr DIfc
Kh2o 1 þ Th2o s
DVfc ¼ c1 Dph2 þ c2 Dp02 þ c3 Dph2o þ c4 DIfc
Rearranging Eq. (7.161) can give the following:
DI_fc ¼ d1 Dph2 þ d2 Dp02 þ d3 Dph2o þ d4 DIfc

Dq_ h2in ¼ d5 Dqh2in þ d6 DIfc
Dp_ h2 ¼ d7 Dqh2in þ d8 Dph2 þ d9 DIfc ð7:162Þ
Dp_ o2 ¼ d10 Dqh2in þ d11 Dpo2 þ d12 DIfc
Dph2o ¼ d13 Dph2o þ d14 DIfc
From Eqs. (7.156)–(7.160) and (7.162), the state-space equation of the SOFC
power plant can be obtained as
X_ SOFC ¼ ASOFC XSOFC þ BSOFC DVsxy þ BPSS Dupss ð7:163Þ

Fig. 7.21 Phasor diagram in y

the d–q coordinate of the q
SOFC power plant Vc
Vs d
φ
ϕ
x
where
T
XSOFC ¼ DVdc DIfc Dqh2in Dph2 Dpo2 Dph2o XTd XTAC XTDC
v
Since u ¼ ac tan1 vsysx , it can have the following:
Du ¼ aT0 DVsxy ð7:164Þ

h v i
vsx0
where aT0 ¼ v2 þ v2
sy0
sx0 sy0 v2sx0 þ v2sy0 . From Fig. 7.21, it can have the following:

isx cos u sin u isd
¼ ð7:165Þ
isy sin u cos u isq
Linearization of Eq. (7.165) is given as

Disx cos u0 sin u0 Disd sin u0 cos u0 isd0
¼ þ Du
Disy sin u0 cos u0 Disq cos u0 sin u0 isq0
ð7:166Þ
DIsxy ¼ b1 DVdc þ b2 Dm þ b3 D/ þ bT4 DVsxy ð7:167Þ

Disx
where DIsxy ¼ . By using Eqs. (7.157)–(7.159) and (7.163), Eq. (7.167) can
Disy
be expressed as
DIsxy ¼ CSOFC XSOFC þ DSOFC DVsxy þ DPSS Dupss ð7:168Þ
7.3.1.3 Linearized Model of Multi-machine Power System

with SOFC Power Plant
Similar to the derivation in Sect. 6.1, it is assumed that there are N generators and
M nodes in the multi-machine power system with the gird-connected SOFC power
plant. In the common x–y coordinate of the power system, node voltage and current
are denoted as vxj þ jvyj and ixj þ jiyj , j ¼ 1; 2; . . .; M, respectively. Without the loss
of generality, it can be assumed that node 1 to N is generator node and node s in
Fig. 7.20 is node (N + 1). Defining (see Eq. (5.51))
DVxy ¼ ½ Dvgx1 Dvgy1 Dvgx2 Dvgy2 DvgxN DvgyN T

DIxy ¼ ½ Dix1 Diy1 Dix2 Diy2 DixN DiyN T
Linearized model of generators is (see Eq. (6.33)) given as
X_ gp ¼ Agp Xgp þ Bgp DVxy

ð7:169Þ
DIxy ¼ Cgp Xgp þ Dgp DVxy
where Xgp is the state variable vector of generator dynamics, including state
variables of the AVRs and the PSSs. Denote the voltage at other nodes by the
following vector:
T
DVoxy ¼ DVðN þ 2Þx DVðN þ 2Þy DVMx DVMx
The network equations can be established as

2 3 2 32 3
DIxy Ygg Ygs Ygo DVxy
4 DIsxy 5 ¼ 4 Ysg Yss Yso 54 DVsxy 5 ð7:170Þ
0 Yog Yos Yoo DVoxy
Substituting Eqs. (7.163), (7.168), and (7.169) into Eq. (7.170), it can have the
following:
2 3 2 32 3
Cgp Xgp Ygg Dgp Ygs Ygo DVxy
4 CSOFC XSOFC þ DPSS Dupss 5 ¼ 4 Ysg Yss DSOFC Yso 54 DVsxy 5
0 Yog Yos Y2o DVoxy
ð7:171Þ
Equation (7.171) can give the following:

DVxy
DVoxy ¼ Y1
oo ½ Yog Yos
DVsxy
ð7:172Þ
DVxy Cgp 0 Xgp 0
¼ Y1 þ Dupss
DVsxy gsxy
0 CSOFC XSOFC DPSS

Ygg Dgp Ygs Ygo 1
where Ygsxy ¼ Yoo ½ Yog Yos
Ysg Yss DSOFC Yso
Arranging the first equation in Eqs. (7.169) and (7.163) together, it can have the
following:
" #
X_ gp Agp 0 Xgp
¼
X_ SOFC 0 ASOFC XSOFC
ð7:173Þ
Bgp 0 DVxy 0
þ þ Dupss
0 BSOFC DVsxy BPSS
Substituting Eq. (7.172) into (7.173), system state equation can be obtained as
_ ¼ AX þ BDupss
X ð7:174Þ
where

Xgp
X¼
XSOFC

Agp 0 Bgp 0 1 Cgp 0
A¼ þ Ygsxy ;
0 ASOFC 0 BSOFC 0 CSOFC

0 Bgp 0 0
B¼ þ Y1
BPSS 0 BSOFC gsxy DPSS
System small-signal stability can be improved by the supplementary stabilizing

control superimposed on the normal control functions of the DC/AC converter (see
Fig. 7.22). If the feedback signal of the stabilizer is the deviation of active power
delivered along the transmission line from node A to s in Fig. 7.20, DPA , then it can
have the following:
ac voltage controller dc voltage controller

Vs u pss Vdc u pss
− − φ
+ m' m
+ φ'
K ac (s) + K dc (s) +
Vs-ref Vdc-ref
PWM
Vs Vdc
− − φ
+ K ac (s) + K dc (s)
Vs-ref
+
u pss PAM
Fig. 7.22 Control of DC/AC converter either by the PWM or by the PAM algorithm

VA Vs
PA ¼ Re VA ¼ Re½ðgAs þ jbAs ÞðVA VA Vs VA Þ
ZAs

¼ gAs v2Ax þ v2Ay þ bAs ðvAx vsy vAy vsx Þ
From the above equation, it can have the following:

DvAx
DPA ¼ ½ 2gAs vAx0 þ bAs vsy0 2gAs vAy0 bAs vsx0
DvAy

Dvsx
þ ½ bAs vAx0 bAs vAy0
Dvsy
¼ ps DVsxy þ p0s DVoxy ð7:175Þ

1
DPA ¼ ½ 0 ps p0s Y1
oo ½ Yog Yos Ygsxy

Cgp 0 Xgp 0
þ Dupss
0 CSOFC XSOFC DPSS
¼ CX þ DDupss ð7:176Þ
Equation (7.176) is the output equation of state-space representation of the power

system. By denoting y ¼ DPA , the linearized model of the multi-machine power
system with the gird-connected SOFC power plant is obtained as (Eqs. (7.174) and
(7.176))
X_ ¼ AX þ BDupss
ð7:177Þ
y ¼ CX þ DDupss
This is the linearized mathematical model of the whole power system. It can be
used for system small-signal stability analysis and design of the stabilizer attached
to the VSC of the SOFC power plant.
7.3.2 Design of a Stabilizer Attached to the VSC

of FC Power Plant by Localized Phase
Compensation Method
7.3.2.1 Localized Design of the Stabilizer
Based on the linearized model of Eq. (7.177), power system small-signal stability
can be examined by modal analysis, i.e. computation of system oscillation modes
and associated eigenvectors. For the design of the supplementary stabilizer, various
methods proposed for the design of the PSS or FACTS stabilizers introduced in
Chap. 5 and 6 can be used. However, even the stabilizer is to suppress a locally
observed oscillation, such as the power oscillation along the transmission line from
node A to s in Fig. 7.20, and the mathematical model of the whole power system of
Eq. (7.177) needs to be established. This increases significantly not only the
complicity of the design, but also the possibility of error in obtaining system model
of Eq. (7.177), as it is difficult to gain and validate the accurate system parameters
and operational data when the scale of the power system is large. In Sect. 7.2,
a much simpler method to design the stabilizer by using a localized linearized
model of the VSC-based unit is introduced. The stabilizer is designed by a localized
phase compensation method to suppress the power oscillation along the transmis-
sion line where the VSC-based unit is installed. In this section, a stabilizer attached
to the VSC of the grid-connected SOFC power plant is used as an example to
demonstrate the application of the localized phase compensation method as follows.
By substituting Eq. (7.121) into (7.163), the following localized state-space
equation of the SOFC power plant dynamics can be obtained as
sX_ SOFC ¼ ALSOFC XSOFC þ BESOFC DE þ BPSS Dupss ð7:178Þ
By using Eqs. (7.157)–(7.159), (7.164), and (7.121), Eq. (7.101) can be written
as
DPA ¼ CLSOFC XSOFC þ CESOFC DE þ dSOFC Dupss ð7:179Þ
DPA ¼ DPA ðDEÞ þ DPA ðDupss Þ ð7:180Þ
where
DPA ðDEÞ ¼ ½CLSOFC ðsI ALSOFC Þ1 BESOFC þ CESOFC DE

DPA ðDupss Þ ¼ ½CLSOFC ðsI ALSOFC Þ1 BPSS þ dSOFC Dupss
DPA ðDupss Þ is the direct contribution from the stabilizer to the variation (oscil-
lation) of line active power, DPA . If DPA is used as the feedback signal of the
stabilizer, then the transfer function of stabilizer is constructed by a lead-lag
network

Dupss ¼ Kpss DPA ð7:181Þ
Denote
FSOFC ðsÞ ¼ CLSOFC ðsI ALSOFC Þ1 BPSS þ dSOFC ð7:182Þ
According to Sect. 7.2.2, parameters of the stabilizer can be set at the angular
oscillation frequency, xs , to satisfy the following equation:

FSOFC ðsÞKpss ¼ Dpss s ð7:183Þ
with s ¼ jxs .
The stabilizer will supply positive damping to suppress the line power oscillation
effectively.
7.3.2.2 An Example
Figure 7.23 shows the configuration of a four-machine two-area power system

integrated with a SOFC power plant. This is the modified example power system
used in Sect. 6.3. The system has a poorly damped inter-area oscillation mode.
Parameters of the SOFC power plant are as follows:
T ¼ 1273; F ¼ 96487; R ¼ 8:314; E0 ¼ 1:18; N0 ¼ 384; Kr ¼ 0:966e6;

Umax ¼ 0:9; Umin ¼ 0:8; Uopt ¼ 0:85; Kh2 ¼ 8:43e4; Kh2o ¼ 2:81e4;
Ko2 ¼ 2:52e3; Th2 ¼ 26:1; Th2o ¼ 78:3; To2 ¼ 2:91;
r ¼ 0:126; Tf ¼ 5; Te ¼ 0:08; rho ¼ 1:145; Cdc ¼ 1:0
XT2
G2 L5
PA
X26 6 X56 5
G1 1 2 4 3 XT4
XT1 X12 X45 X34 G3
VA Vs VB
L2 XT3
Is Xs
VC
G4
SOFC power plant
Fig. 7.23 A power system integrated with a PV generation plant

To enhance the damping of the inter-area oscillation mode, a stabilizer can be

added onto the DC/AC VSC of the SOFC power plant as discussed in the above
section. For the purpose of demonstration, the following three cases are studied:
1. The PWM is adopted and the stabilizer is superimposed on the AC voltage
controller (see Fig. 7.22). This is the PWM reactive power PSS.
2. The stabilizer is added on the DC voltage controller (see Fig. 7.22) of the PWM.
This is the PWM active power PSS.
3. The PAM algorithm is used (see Fig. 7.22). The added stabilizer is named the
PAM stabilizer.
Three types of stabilizers are designed at the line-loading condition
PA0 = 500 MW by using the method introduced in Sect. 7.2.2 based on the localized
model of the SOFC power plant of Eqs. (7.178) and (7.179). To confirm the
effectiveness of the stabilizers designed, results of modal analysis obtained by using
the complete mathematical model of the power system of Eq. (7.177) are given in
Table 7.1 and simulation results using the nonlinear mathematical model of the
example power system are shown in Fig. 7.24. In the nonlinear simulation, the
example power system is subject to a disturbance of 1 % increase of mechanical
power input to G4 for 100 ms at 0.2 s of simulation. From Table 7.1 and Fig. 7.24,
it can be seen that the stabilizers are designed successfully by using the localized
phase compensation method to effectively suppress the inter-area power oscillation.
Three types of stabilizers are tested at a lower line-loading condition of the
example power system with PA0 = 300 MW. Results of modal analysis are given in
Table 7.1. Figure 7.25 presents the results of nonlinear simulation (subject to the
same disturbance as that in Fig. 7.24). From Table 7.1 and Fig. 7.25, it can be seen
that at the lower line-loading condition, PWM reactive power stabilizer designed at
the higher line-loading condition with PA0 = 500 MW cannot suppress the inter-area
oscillation effectively. The robustness of the PWM active power and PAM stabi-
lizer is better as they can still function to damp the inter-area power oscillation
when the line-loading conditions of the power system change.
Similar to the case discussed in Sect. 7.2.3 about the ESS-based stabilizer, the
PWM active power and PAM stabilizer function through regulating the phase of
terminal voltage of the SOFC power plant, that is, the exchange of active power
between the SOFC power plant and the rest of the power system. Therefore, they
suppress the inter-area power oscillation directly by absorbing active power from or
injecting it into the power system when they “see” the excess or lack of active
Table 7.1 Inter-area oscillation mode

PA0 = 500 MW PA0 = 300 MW
Without any stabilizer −0.103 ± j3.471 −0.883 ± j3.602
With PWM reactive power stabilizer −0.355 ± j3.169 −0.221 ± j3.331
With PWM active power stabilizer −0.405 ± j3.365 −0.320 ± j3.454
With PAM stabilizer −0.529 ± j3.363 −0.398 ± j3.419
PA(100MW)
Without any stabilizer
With PWM active power stabilizer
With PWM reactive power stabilizer
With PAM stabilizer
Time (second)
Fig. 7.24 Results of nonlinear simulation when the system is subject to a small disturbance
(PA0 = 500 MW)
PA(100MW)
Without any stabilizer

With PWM active power stabilizer
With PWM reactive power stabilizer
With PAM stabilizer
Time (second)
Fig. 7.25 Results of nonlinear simulation when the system is subject to a small disturbance
(PA0 = 300 MW)
power delivered along the transmission line. Hence, their capability to damp the
power oscillation is less affected by the change of system loading conditions.
Function of the PWM reactive power stabilizer involves the regulation of magni-
tude of terminal voltage of the SOFC power plant. Hence, in essence it works via
controlling the exchange of reactive power between the SOFC power plant and the
rest of the power system. The stabilizer affects indirectly the variation of active
power delivered along the transmission line through the regulation of injection into
or absorbing of reactive power from the SOFC power plant. At a lower line-loading
condition, regulation of reactive power exchange is related less closely with the
variation of line active power. Hence, the stabilizer is less effective.
7.4 Damping of Multi-mode Oscillations by Multiple

Stabilizers Attached to a Single UPFC
To suppress power system multi-mode oscillations, the commonly used schemes

are to deploy multiple decentralized PSS or/and FACTS stabilizers. Usually, a
stabilizer is designed to improve the damping of one selected oscillation mode and
hence damping of multi-mode oscillations is achieved by multiple stabilizers
jointly. Under these schemes, parameters of multiple stabilizers are set either
individually or simultaneously. In engineering practice nowadays, strategy of
individual setting is often used because the setting can be easily tuned in field. In
many cases, it is effective and acceptable for the multiple stabilizers to provide
reasonable damping to multi-mode oscillations. Extensive research in the last three
decades has demonstrated that the coordinated design to simultaneously set mul-
tiple stabilizers can provide better results to effectively suppress power system
multi-mode oscillations as introduced in Sects. 5.2.2 and 6.2.4.
However, from the point of view of engineering practice, at the stage of field
tuning after parameters of multiple stabilizers are set in coordination, it could be a
difficult and challenging task to finally implement the schemes of simultaneous
setting of multiple stabilizers. This is because often parameters of stabilizers set in
coordination need to be further tuned carefully in the field to provide required
control performance. However, those stabilizers are often at different geographical
locations in a power system. Cross-location and simultaneous field tuning of
parameters of multiple stabilizers is practically very difficult. In this case, if
parameters of stabilizers are tuned in field one by one in a sequence, the overall
strategy of deploying multiple stabilizers will be the combination of simultaneous
setting and individual field tuning. It is unlikely that the optimality of stabilizer’s
design by setting their parameters simultaneously can be achieved eventually by
implementing individual parameters tuning in the field.
Hence, this section demonstrates a scheme of attaching multiple stabilizers to a
single UPFC to damp power system multi-mode oscillations. The method takes the
full advantage of the UPFC, which can perform multiple control functions of active,
7.4 Damping of Multi-mode Oscillations by Multiple Stabilizers … 381
reactive power, or/and voltage regulation. Multiple stabilizers are superimposed on

different control functions of a single multifunctional UPFC. Each stabilizer is set to
specially meet the need of damping improvement of one selected oscillation mode
and multiple stabilizers are designed in coordination with damp multi-mode
oscillations. This arrangement of multiple stabilizers is as same as that of con-
ventional schemes of coordinated design of multiple stabilizers introduced in
Sects. 5.2.2 and 6.2.4. Hence, the optimal design to suppress multi-mode oscillation
can be achieved. Moreover, being attached to one UPFC, multiple stabilizers are at
a same geographical location in the power system. Therefore, problem of
cross-location field tuning of stabilizer’s parameters is effectively avoided.
7.4.1 Coordinated Design of Multiple Stabilizers

by Artificial Fish Swarm Algorithm (AFSA) [8]
Methods of optimal coordinated design of multiple stabilizers can be classified

under two categories: mathematical optimization and non-mathematical optimiza-
tion. In Sect. 5.2.2, a representative mathematical method based on the nonlinear
programming is introduced. One of the most well-known methods of
non-mathematical optimization is the genetic algorithm (GA) which has been tested
for the coordinated design of multiple stabilizers in power systems. In this section, a
different method of non-mathematical optimization—artificial fish swarm algorithm
(AFSA)—is presented for the design of multiple stabilizers attached to a single
UPFC to suppress multi-mode power system oscillation as follows.
7.4.1.1 Objective Function
The coordinated design of multiple stabilizers is based on the following objective

function
J ¼ fðxÞ ¼ minf1i ; i ¼ 1; 2; . . .; mg ð7:184Þ
where 1i is the damping of the ith oscillation mode and m the number of oscillation
modes to be damped by m stabilizers. Hence, the objective function J is the
damping of the oscillation mode of the power system that is of the least damping.
The coordinated design is to search for the maximum of J = f(x).
Transfer function of the jth stabilizer superimposed on a selected primary control
function of the UPFC is given as
sTwj 1 þ sT2j 1 þ sT4j

Kj
1 þ sTwj 1 þ sT1j 1 þ sT3j
Parameters of the stabilizer, T2j, T4j, and TWj are predetermined, and T1j, T3j, and
gain Kj are to be set by using the AFSA. In the parameter space of J, parameters of
the jth stabilizer involved in the coordinated design are subject to the following
constrains:
Kjmin Kj Kjmax ;
T1jmin T1j T1jmax ; T3jmin T3j T3jmax
The AFSA is an algorithm based on the population evolution similar to the GA.
The basic argument of the AFSA is that in a water area, the place where there are
most fish is where there is most food. The optimal search imitates the behaviours of
fish to seek food. In the n-dimensional parameter space of the objective function
J ¼ fðxÞ defined by Eq. (7.184), the algorithm starts with constructing k artificial
fish and the position of the ith fish is denoted by vector xi ¼ ½ xi1 xi2 xin .
The amount of food at present position is denoted by Yi ¼ fðxi Þ. Distance between
artificial fish is dij ¼ xi xj and the visual distance of artificial fish is denoted by
Visual. Crowding index is CI and maximal moving step is Step. The AFSA is
implemented by three main behaviours of artificial fish as introduced in the fol-
lowing section.
7.4.1.2 The Artificial Fish Swarm Algorithm
Preying behaviour
It is assumed that the present position of artificial fish is xi ðtÞ and the amount of
food at the position is Yi ðtÞ. Within the visual distance of artificial fish, i.e.
dij \Visual, a new position xi ðt þ 1Þ is randomly chosen and Yi ðt þ 1Þ is calculated.
If Yi ðt þ 1Þ [ Yi ðtÞ, then the artificial fish moves one step forwards. Otherwise, a
new position is randomly chosen.
Swarming behaviour
This is to count the number of fish within the visual distance, nf and to calculate the
central position xc of this group of fish. The amount of food at the central position is
Yc ðtÞ. If Yc ðtÞ=nf [ CIYi ðtÞ, this means that there is more food at the central
position and the position is not very crowded. Hence, the artificial fish moves one
step towards the central position. Otherwise, it performs preying.
Following behaviour
If within the visual distance of an artificial fish, there is most amount of food
Ymax ðtÞ at position xmax . Using xmax as the centre, the number of fish nmax within its
visual distance (dij \Visual) is counted. If Ymax ðtÞ=nmax [ CIYi ðtÞ, then there is
more food at xmax and it is not very crowded there. The artificial fish moves one
step towards xmax . Otherwise, it performs preying.
In the AFSA, preying behaviour guarantees the convergence of the algorithm.
The introduction of crowding index CI is to prevent that a local optima is found as
AFSA Initialization
(number of artificial fish k and
their initial position)
Calculation of amount of food
Swarming movement of Following movement of

artificial fish artificial fish
NO
More food found?
YES
Preying movement of
artificial fish
Objective
NO function satisfied?
(amount of food found
satisfied?)
YES
STOP
Fig. 7.26 Flow chart of the AFSA
the final solution. The swarming behaviours are to ensure the stability of the algo-
rithm and global optimization. Following behaviours are to accelerate the conver-
gence of the algorithm towards the global solution. A flow chart of implementing the
AFSA is shown in Fig. 7.26. The AFSA essentially is a random global optimization
algorithm. It is of many attractive features of the genetic algorithm, but it does not
need to perform crossover and mutation. Hence, it is relatively easier and simpler.
7.4.2 Examples
7.4.2.1 An Example of Damping Two-Mode Oscillations by Stabilizers

Added on a UPFC
The 4-machine 2-area power system of Fig. 7.24 is used as an example, where the
SOFC power plant is replaced by a UPFC and G1 and G2 are not installed with the
PSS. The UPFC employs four control functions as defined by Eq. (7.71). In the
system, there are three oscillation modes. One local mode related to G3 and G4 is
well damped. An inter-area mode (G1 and G2 against G3 and 4) and a local mode
(G1 against G2) are poorly damped. Two stabilizers are superimposed on two
normal control functions of the UPFC to suppress the poorly damped local-mode
and inter-area mode oscillation. To select the control functions to be attached with a
stabilizer, the controllability index is calculated and presented in Table 7.2. From
Table 7.2, it can be seen that in order to suppress two-mode oscillations, the first
stabilizer should be added to the function of the AC voltage regulation to damp the
inter-area oscillation mode and the second stabilizer to DC voltage regulation to
damp the local oscillation mode.
Two stabilizers superimposed on the functions of the AC and DC voltage reg-
ulation of the UPFC are designed in coordination based on the linearized model of
the power system presented in Sect. 7.2. Feedback signal of the first stabilizer is the
deviation of active power delivered from node 7–8 and that of the second stabilizer
is the difference between the deviation of active power from node 5–7 and that from
node 2–7. Results of coordinated design of two stabilizers by using the AFSA are
given in Table 7.3. From the third, fourth, and fifth rows of Table 7.4, it can be seen
that two stabilizers can enhance the damping of two oscillation modes effectively.
Table 7.2 Results of controllability computation of the UPFC normal control functions
Control function to attach stabilizer Inter-area mode Local mode
Active power regulation 3.77 0.10
Reactive power regulation 5.85 0.11
AC voltage regulation 49.29 1.41
DC voltage regulation 138.26 7.43
Table 7.3 Results of coordinated design (T2 = T4 = 0.05, Tw = 5)

Parameters The first stabilizer The second stabilizer
T1 0.01 0.37
T3 0.12 0.07
K 0.25 28.36
Table 7.4 Oscillation modes in the example power system with the UPFC
Inter-area mode Local mode
Without any stabilizer −0.12062 ± j3.4785 −0.20745 ± j5.8104
With the first stabilizer −0.70652 ± j3.0758 −0.21014 ± j5.8047
With the second stabilizer −0.05775 ± j3.3794 −1.1759 ± j5.7957
With two stabilizers −0.6812 ± j3.0858 −1.2403 ± j5.6547
Fig. 7.27 Simulation results (a)

when the first stabilizer is 0.46
delta1-3 Without stabilizer
installed. a Inter-area mode 0.44 With the first stabilizer
oscillation and b local-mode
oscillation 0.42
0.4
0.38
0.36
0.34
0.32
0.3
0.28
0 2 4 6 8 10
t/s
(b)
0.155
delta1-2
Without stabilizer
0.15 With the first stabilizer
0.145
0.14
0.135
0.13
0.125
0.12
0.115
0 2 4 6 8 10
t/s
Figures 7.27 and 7.28 present the results of nonlinear simulation when the
example power system is subject to a three-phase to-earth fault at node 8 for
100 ms. Figure 7.27 is that when only the first stabilizer is installed. Because the
first stabilizer is set to damp inter-area mode, in Fig. 7.27 the local-mode oscillation
is not suppressed effectively. From Fig. 7.28, it can be seen that with two stabilizers
being superimposed on the AC and DC voltage regulator of the UPFC, respectively,
two-mode oscillations of the example power system are damped successfully.
Hence, this example demonstrates that it is feasible to suppress multi-mode oscil-
lations by arranging two stabilizers on a single UPFC.
Fig. 7.28 Simulation results 0.46 (a)

when two stabilizers are delta1-3 Without stabilizer
installed. a Inter-area mode 0.44 With two stabilizers
oscillation and b local-mode 0.42
oscillation
0.4
0.38
0.36
0.34
0.32
0.3
0.28
0 2 4 6 8 10
t/s
0.155
(b)
delta1-2
Without stabilizer
0.15
With two stabilizers
0.145
0.14
0.135
0.13
0.125
0.12
0.115
0 2 4 6 8 10
t/s
7.4.2.2 An Example of Damping Three-Mode Oscillations

by Stabilizers Added on a Multi-terminal UPFC
In order to show that the example presented in the above section is not the unique
case, the configuration of the example power system of Fig. 7.24 is modified to
form a 6-machine power system as shown in Fig. 7.29. In this example power
system, there are three weakly damped oscillation modes as shown in the second
row of Table 7.5. They are: (1) the first inter-area oscillation mode (first mode) of
group of machines 1 and 2 against that of machines 3, 4, 5, and 6; (2) the second
inter-area oscillation mode (second mode) of group of machines 3 and 4 against that
of machines 5 and 6; (3) a local oscillation mode (local mode) of machine 1 against
machine 2. Two other local oscillation modes are very well damped and hence are
not listed.
In the example power system, a multi-terminal UPFC (MUPFC) is installed at
node 7. The MUPFC is of one shunt and two series VSCs and its mathematical
model is given as:
G6
13
416MW L 14
1 5 14 15 16 12 G5
G1 160km
25km
7 10 km 25km
MUPFC
160 km 9 10 11 3 G3
10km 25km
405MW
2 L7 L9
G2 4
G4
Fig. 7.29 A 6-machine power system installed with a MUPFC
Table 7.5 Oscillation modes in the example power system of Fig. 7.28
First mode Second mode Local mode
Without any stabilizer −0.115 ± j4.296 −0.113 ± j3.338 −0.230 ± j5.845
With the first stabilizer −0.677 ± j3.987 −0.059 ± j3.395 −0.234 ± j5.825
With the first and second −0.827 ± j3.806 −0.557 ± j3.588 −0.234 ± j5.828
stabilizers
With three stabilizers −0.635 ± j3.792 −0.610 ± j3.663 −0.933 ± j5.623

dvdc 3mE iEd 3mB a iBd a
¼ ½ cos dE sin dE þ ½ cos dB a sin dB a
dt 4Cdc iEq 4Cdc iBq a

3mB b iBd b
þ ½ cos dB b sin dB b
4Cdc iBq b
2 3
cos dB
mB a a vdc
vBtd a 0 xB a iBd a 6 7
¼ þ6 2
4 mB a sin dB
7;
5
vBtq a xB a 0 iBq a a vdc
2
2 3
m B b cos dB b vdc
vBtd b 0 xB b iBd b 6 7
¼ þ46 2 7;
vBtq b xB b 0 iBq b mB b sin dB v
b dc
5
2
2 3
mE cos dE vdc
vEtd 0 xE iEd 6 7
¼ þ6 2
4 mE sin dE vdc 5
7
vEtq xE 0 iEq
2
where mE ; mB a ; mB b and dE ; dB a ; dB b are the modulation ratios and phase

angles of shunt and series VSCs, respectively. The MUPFC is to perform the
primary control functions to regulate the voltage at node 7 and power flow along
line 7–9 and line 7–14 with the following six PI controllers.
DC voltage regulator: dE ¼ dE0 þ ðKdcp þ Ksdci Þðvdcref vdc Þ
AC voltage regulator: mE ¼ mE0 þ ðKvp þ Ksvi ÞðVref V7 Þ
Active and reactive power regulators for line 7–9:

KBapi
dB a¼ dB a0 þ KBapp þ ðP79ref P79 Þ
s

KBaqi
mB a ¼ m B a0 þ K Baqp þ ðQ79ref Q79 Þ
s
Active and reactive power regulators for line 7–14:

KBbpi
dB b ¼ dB b0 þ K Bbpp þ ðP714ref P714 Þ
s

KBbqi
mB b ¼ mB b0 þ KBbqp þ ðQ714ref Q714 Þ
s
In order to improve the damping of those three oscillation modes, three stabi-
lizers are to be added on three of six MUPFC normal control functions. A linearized
model of the system can be established by using the similar procedure presented in
Sect. 7.2. Based on the linearized model, controllability index to superimpose
damping function on the normal control functions of the MUPFC is calculated and
presented in Table 7.6. Maximal controllability predicts the best damping effect
(bold figures in Table 7.6). Hence, the AC voltage control function is selected to
add the first stabilizer and the reactive power control function for line 7–9 is to add
the second stabilizer to damp the first and second inter-area oscillation mode,
respectively. DC voltage control function is chosen to add the third stabilizer to
suppress the local oscillation mode. Locally available signals of active power are
used as the feedback signals and those three stabilizers are set in coordination using
the AFSA. Results of the coordinated design are given in Table 7.7.
Table 7.6 Controllability index of the MUPFC normal control functions

Control functions 1st mode 2nd mode Local mode
AC voltage 62.06 6.93 9.93
Reactive power of line 7–9 9.86 19.52 0.86
Reactive power of line 7–14 9.04 17.85 0.91
DC voltage 72.59 5.47 16.07
Active power of line 7–9 0.89 0.98 0.18
Active power of line 7–14 0.88 0.99 0.18
Table 7.7 Results of coordinated design (T2 = T4 = 0.05, Tw = 5)

Parameters The first stabilizer The second stabilizer The third stabilizer
T1 0.05 0.5 0.3
T3 0.3 0.1 0.05
K 0.6 2.0 20.0
Success of coordinated design is confirmed by the results of computation of

oscillation modes presented in Table 7.5. From the third, fourth, and fifth rows of
Table 7.5, it can be seen that those three stabilizers are set to improve the damping
of three oscillations of the example power system separately and the multi-mode
oscillations are damped effectively by their joint actions.

when only the first stabilizer
is installed to damp the first
0.6 δ13 Without stabilizer
0.5
inter-area oscillation mode.
a The first mode inter-area 0.4 With the 1st stabilizers
oscillation, b the second mode 0.3
inter-area oscillation, and 0.2
c local-mode oscillation
0.1
0
Time (second)
-0.1
0 5 10 15
(b) Without stabilizer

-0.01 δ 35
-0.02 With the 1st stabilizers
-0.03
-0.04
-0.05
-0.06
-0.07
-0.08 Time (second)
-0.09
0 5 10 15
(c)
0.22
0.2 With the 1st stabilizers
0.18
0.16
0.14
0.12
0.1
0.08
0.06 Time (second)
0.04
0 5 10 15
Figures 7.30, 7.31, and 7.32 give the results of nonlinear simulation of the
example power system when it is subject to a three-phase to-earth fault at node 7 for
100 ms. They confirm the results of eigensolution in Table 7.5 that each stabilizer is
designed to damp the targeted oscillation mode and with three stabilizers
three-mode oscillations are suppressed successfully.

when only the first and second 0.6 δ13
Without stabilizer
stabilizers are installed to
0.5 With the 1st and 2nd stabilizers
damp the first and second
inter-area oscillation modes. 0.4
a The first mode inter-area 0.3
oscillation, b the second mode 0.2
inter-area oscillation, and
0.1
c local-mode oscillation
0
Time (second)
-0.1
0 5 10 15

-0.01
δ 35
-0.02 With the 1st and 2nd stabilizers
-0.03
-0.04
-0.05
-0.06
-0.07
-0.08 Time (second)
-0.09
0 5 10 15
(c)
0.22
0.2 With the 1st and 2nd stabilizers
0.18
0.16
0.14
0.12
0.1
0.08
0.06 Time (second)
0.04
0 5 10 15
References 391
Fig. 7.32 Simulation when (a)

all three stabilizers are Without stabilizer
0.6 δ13
installed. a The first mode
0.5
inter-area oscillation, b the With all three stabilizers installed
second mode inter-area 0.4
oscillation, and c local-mode 0.3
oscillation 0.2
0.1
0
Time (second)
-0.1
0 5 10 15

-0.01 δ35
-0.02 With all three stabilizers installed
-0.03
-0.04
-0.05
-0.06
-0.07
-0.08
Time (second)
-0.09
0 5 10 15
(c)
0.24 Without stabilizer
0.22
δ12
0.2
0.18
With all three stabilizers installed
0.16
0.14
0.12
0.1
0.08
0.06 Time (second)
0.04
0 5 10 15
References
1. CIGRE TF 30-01-08 Report (1999) Modelling of power electronics equipment (FACTS) in load
flow and stability programs
2. Wang HF (1999) Phillips-Heffron model of power systems installed with STATCOM and
applications. IEE Proc Part C 146(5):521–527
3. Du W (2009) Power system small signal oscillation stability as affected by static synchronous
compensator (STATCOM) and energy storage system (ESS). PhD thesis, University of Bath,
UK
4. Wang HF (1999) Applications of modelling UPFC into multi-machine power systems. IEE Proc
Part C 3
5. Wang HF (2000) A unified model for the analysis of FACTS devices in damping power system
oscillations part III: unified power flow controller. IEEE Trans Power Del 3
6. Du W, Wang HF, Cao J, Xiao LY (2012) Application of the phase compensation method for the
design a DC/AC converter based stabilizer to damp power system inter-area oscillation. IEEE
Trans Power Syst 27(3):1302–1310
7. Du W, Wang HF, Cao J, Xiao LY (2012) Robustness of an energy storage system based
stabilizer to suppress inter-area oscillation in a multi-machine power system. IET Proc Gener
Transm Distrib 6(4):339–351
8. Du W, Wu X, Wang HF, Dunn R (2010) Feasibility study to damp power system multi-mode
oscillations by using a single FACTS device. Int J Electr Power Energy Syst 32(6):645–655
Index
A I
AFSA, 381 Initial compensation, 88
AVR, 18 Installing location, 97, 246
Inter-area line power oscillation, 349
C
Closed-loop system, 32 L
Coordinated design, 204 Length of transmission line, 96
Correlation ratio, 199 Linearized model, 17
Load condition, 92
D Localized phase compensation, 349
Damping control, 1, 81 Localized small-signal model, 350
Damping torque analysis (DTA), 3, 42, 81,
121, 266 M
Decentralized control, 209 Modal analysis (MA), 3, 199
Mode shape, 201
E Modulation signal, 155
Eigensolution, 4, 199 Multi-machine power system, 183, 235, 329
Eigenvalue, 4, 32
Electric torque, 18 N
Electromechanical oscillation, 38, 90 Non-negatively interactive design, 285
Electromechanical oscillation mode, 2, 189
ESS, 329 O
Excitation system, 18 Optimum searching, 225
Extended Heffron–Phillips model, 81, 121
P
F PAM control, 352
FACTS stabilizers, 8 Participation factor, 200, 272
Feedback signal, 245 Phase compensation, 47, 139
Flexible AC transmission systems (FACTS), 5 Power system oscillations, 1
Forward path, 90 Power system stabilizer (PSS), 2, 17, 183, 245
PWM, 121
G PWM control, 332
General linearized model, 242, 336
R
H Residue index, 276
Heffron–Phillips model, 26, 189, 235, 329 Robustness, 161

DOI 10.1007/978-1-4899-7696-3
394 Index
S Transfer function, 19
Sensitivity index, 272
Single-machine infinite-bus power system, 17, U
81, 121 UPFC, 143, 339
Small-signal angular stability, 1
SOFC, 367 V
SSSC, 133 Voltage source converter (VSC), 6
SVC, 81, 235 VSC-based active power stabilizer, 130
VSC-based reactive power stabilizer, 130
T VSC-based stabilizer, 121, 329
TCPS, 97, 238
TCSC, 98, 238

Analysis and Damping Control of Power System Low-Frequency Oscillations

Diunggah oleh

Informasi Dokumen

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Analysis and Damping Control of Power System Low-Frequency Oscillations

Diunggah oleh

Hak Cipta:

Format Tersedia

Power Electronics and Power Systems

Analysis and Damping

ISSN 2196-3185 ISSN 2196-3193 (electronic)

Library of Congress Control Number: 2016931424

© Springer Science+Business Media New York 2016

Printed on acid-free paper

This Springer imprint is published by SpringerNature

Power system electromechanical low-frequency oscillations threaten the stable

Beijing Haifeng Wang

2.4.4 Equivalence Between the Damping Torque

4.3 Damping Function of a Uniﬁed Power Flow Controller (UPFC)

6.2 Analysis and Damping Control of Thyristor-Based FACTS

7.4 Damping of Multi-mode Oscillations by Multiple

1.1 Power System Low-frequency Oscillations

Power system low-frequency oscillations are the oscillations of active power

© Springer Science+Business Media New York 2016 1

1.2 Linearized Methods for the Analysis and Damping

Study of power system oscillations is in order to investigate the phenomena and to

1.3 FACTS and Grid-Connected ESS

Flexible AC transmission systems (FACTS) are the technologies emerged around

Fig. 1.1 Illustration on the V1 V2

1. Thyristor-controlled static VAR compensator (SVC), which controls the mag-

Three-phase diagram of GTO-based Single-line diagram of GTO-based

Fig. 1.2 An SVS realized by a VSC circuit

Two stage circuit

1.4 Controllers to Damp Power System Oscillations

To increase the damping of power system oscillations, installation of a supple-

Through the effort of and experience accumulated by power engineers and

Occurrence of power system oscillations is essentially due to the response of

1.5 Design of Damping Controllers to Suppress Power

The prime speciﬁcation of designing a damping controller (a PSS or a FACTS

In order to solve the problem of “eigenvalue drifting” caused by interactions

1.6 Organization of the Book

Table 1.1 Chapter arrangement of the book

2.1 Linearized Model of a Single-Machine Inﬁnite-Bus

2.1.1 General Linearized Mathematical Model

2.1.1.1 Full Mathematical Model of a Synchronous Generator

Fundamental equations describing the dynamics of a synchronous generator are

where wd ; wq ; wf ; wD ; wQ , id ; iq ; if ; iD ; iQ , vtd ; vtq ; vf , and ra ; rf ; rD ; rQ are the mag-

© Springer Science+Business Media New York 2016 17

of synchronous generator. Since there is no static coupling between any d-axis

where xd ; xq ; xf ; xD ; xQ are the self-reactance of winding d, q, f, D, and Q,

2.1.1.2 Excitation System and the Automatic Voltage Regulator

Voltage control in a power system is closely related to the regulation of reactive

Fig. 2.1 Arrangement of an if

system and is implemented by solving some optimization problems. A synchronous

Fig. 2.2 Arrangement of a DC excitation system

alternator stationary diode synchronous generator

Fig. 2.3 Arrangement of a stationary rectiﬁer excitation system

rotational armature and

Fig. 2.4 Arrangement of a rotating rectiﬁer excitation system

Fig. 2.5 Arrangement of a potential-source controlled-rectiﬁer excitation system

From Fig. 2.1, it can have

vf ¼ vf0 þ TEðsÞðVtref Vt þ upss Þ ð2:4Þ

2.1.1.3 A Single-Machine Inﬁnite-Bus Power System

Fig. 2.6 A single-machine inﬁnite-bus power system

Fig. 2.7 d–q coordinate of q

vtd þ jvtq ¼ jxt ðid þ jiq Þ þ vd þ jvq ð2:8Þ

where vd ¼ Vb sin d; vq ¼ Vb cos d and the terminal voltage of generator is as

Pt ¼ vdt id þ vqt iq ¼ vd id þ vq iq ð2:11Þ

2.1.1.4 Linearized Model of Single-Machine Inﬁnite-Bus Power

Linearization of Eqs. (2.1) and (2.2) is as follows:

Dw_ d ¼ x0 ðDvtd þ ra Did þ x0 Dwq þ wq0 DxÞ

Equation (2.48) is the following polynomial equation if Ao is an M M matrix

ð1ÞM kM þ aM1 kM1 þ þ a1 k þ a0 ¼ 0 ð2:49Þ

zi ðtÞ ¼ w1i x1 ðtÞ þ w2i x2 ðtÞ þ þ wMi xM ðtÞ ð2:66Þ